HUMAN RESOURCES AND SYSTEM DESIGN
Concepts
The human contribution
The human role as a system controller
The business of designing machines, processes and systems can be pursued more or less independently of the properties of people.
Nevertheless people are always involved, the designer himself is a human being and his product will shape the behaviour of many workers and other users.
More fundamentally, the design activity will be meaningless unless it is directed towards serving some human need.
In spite of all this, the design process itself is often thought about and executed without any formal considerations about people.
Inevitably the engineer, architect or other designer devotes most of his attention and expertise to devising mechanisms, buildings and so on which support some human activity more effectively than those currently available.
The new machine or system must not be very different from the old one for a variety of reasons.
The old one did its job, not perfectly but well enough to justify its existence.
The new one is usually designed on the basis of copying the old one but removing as many as possible of the faults.
There are other reasons such as commonality of components and, of course, shortage of imagination which lead to most design and development being a progressive iterative process.
This happens to suit the human operators because most of their skills will transfer along the line of development of the machines and systems.
From the engineering point of view the hardware technology is central and the operators tag along supporting the activity of machines which are basically doing the work.
In nineteenth century transport for example, the coach and horses was not a serious competitor for the steam-engined train, nor was the man with a spade as productive as the operator with a steam shovel.
This was the earliest technology in which power was derived from sources other than animal muscles and the output was so obviously superior that men did not mind putting up with the inconvenience of machines which were uncomfortable or awkward to use.
They even took pride in developing new skills which enabled them to use difficult machines which inexperienced people could not use.
This attitude was reinforced by the second wave of technology which provided instrumentation, ways of sensing and recording data which gave more accurate and reliable data than that which can be detected directly by the human senses.
At this stage the typical machine operator manipulated machine controls on the basis of data presented on instruments.
The machine controls caused power to be applied and consequences to occur which resulted in changed readings on the instruments.
The man was in the control loop (Hick and Bates, 1950).
More recently, in the third wave of technology the man has been removed from the responsibility for continuous control (this is best done by computers working to a moderately flexible range of programs) but he remains as the monitor of the system performance and a selector of the appropriate program determined by the changing short-term objectives.
He may keep the responsibility for setting-up and shutting-down the system or this also may be partially delegated to computers.
This is the ' Supervisory Control ' role (Sheridan and Johannsen, 1976).
Information flow in these three phases of development of system control is shown in fig. 1.1
Supervisory control is by no means universal, in fact it remains restricted to high technology systems such as aircraft, computer controlled machine tools, chemical plants and power stations.
(Edwards and Lees, 1974).
There are still many systems where the man is in the control loop, for example in vehicle driving and in most manufacturing production processes.
There are many tasks where the operator is assisted only by hand-tools and simple powered machines, for example in craft-work and surgery.
There are also plenty of tasks, although these are now perhaps more common in leisure than in paid work, where the man supplies the muscle power, for example the manual labourer and the active sportsman.
In general, the working man may function at any level from the senior partner in high technology operations to the provider of muscle power in physically demanding jobs.
In all cases the ergonomic requirement is that the task as designed should make use of his abilities and be adaptive to his limitations.
The human contribution in high technology systems
All working systems, including those incorporating advanced technology of processes and process control, depend on skilled operators.
There are social, economic and technical reasons for this.
Socially the public will not easily accept automatic systems.
When seeking reassurance that a system is safe they study the people who are responsible for it.
They have no means of assessing the reliability of automatic control systems but they can make some assessment of a skilled individual and without knowing very much about his working procedures they will trust his actions if he seems calm, competent and mature.
The most direct example is the nervous airline passenger who is comforted by the belief that the middle-aged pilot he saw sitting on the flight deck is flying the aeroplane.
Less directly but equally importantly, people living in the vicinity of a large plant would not take kindly to the idea that it was not under the control of human beings who are on the spot and are assumed to know exactly what is happening.
p 3:DIAGRAM/FIGURE
Economically it remains true that it is inordinately expensive to construct a system which must continue to be reliable without direct human intervention, for example under the sea or in space.
Equally it is expensive to ensure that equipment will continue to function under rough handling conditions, for example the necessary ' ruggedising ' of military equipment.
A system can be designed much more economically if it can be assumed that skilled personnel are available to control and take care of it.
Technically the state of the art seems to be that systems are most efficient, reliable and safe when control is largely automatic but the human operator remains in a monitoring and supervisory role.
That is, continuous dynamic dealing with minor perturbations is automatic and so also is the application of basic rules about safety; for example, the system or some part of it might be programmed to go through a step-change function such as a shut-down if certain parameters exceed prescribed limits.
The role of the human operator in these circumstances is largely specified by operating instructions which are mandatory.
Given this and this  he must do that.
Characteristically his intervention in this way is a response to some other human requirement; for example, a maintenance man may wish to have a particular sub-system shut down for his attention or a user of the system product may wish to receive a different input to meet his purposes.
Many of these interactions could be made automatic were it not for the basic need for a human presence for the social/ economic reasons mentioned above and because the human operator has to act as the ultimate back-stop when things go badly wrong.
Things go badly wrong when there has been some extreme untoward event such as a fire, an earthquake or a bomb, alternatively an unanticipated combination of faults may occur within the plant itself.
In either case the human intervention must be a very high level one based on complex diagnosis and innovative design-type thinking about how best to cope so as to avoid a catastrophe.
A catastrophe can happen when the system energy is no longer channelled as intended by the designers or because there is a release of toxic substances or both.
The demands on the operator stem from two kinds of task which are not compatible because they require very different levels of skill.
On the one hand he must cope with routines where the demand is for precise obedience to established instructions and on the other hand he might suddenly be faced with a need to respond in a creative manner totally outside any instructions.
Superficially the solution would seem to be to have two kinds of operators, those who carry out the routines and others who are on call should unforeseen emergencies arise.
There are two snags about this approach.
Firstly there may not be time or it may not be physically possible to suddenly introduce a high level operator, for example we can not put a different pilot on the flight deck when an aircraft gets into trouble.
Secondly, the high level operator will not maintain his skill and familiarity with the system unless he operates regularly within it.
His ability to intervene effectively must be a function of his' hands on ' experience supported of course by his conceptual knowledge of how the system functions.
This last proposition is based on experience of systems where the intervention requires complex manipulative activity, for example a surgeon dealing with an emergency during an operation or an aircraft pilot taking over manually.
It is not so obviously true where the intervention takes the form essentially of a decision to initiate a single direct action such as closing a particular valve or starting up a stand-by pump.
In these circumstances it is not clear whether or not the effectiveness of the intervention is quite so dependent on operating experience.
Taking for example the nuclear power plant control room, the question is whether the desk operators should be expected to cope with all emergencies which appear within the total information presentation or whether, for complex and dangerous situations, a more senior person such as the shift-charge engineer should be called upon to make the decisions.
This policy issue must be cleared before the basic personnel decisions about selection and training of staff at the various levels can be made and before the information presentations can be designed.
The human contribution in low technology systems
The extreme case of man as a source of muscle power is dealt with in the previous book in this series  The Body at Work.
Such working situations are now relatively rare in developed countries and the man is more usually employed in tasks where there is a considerable control element requiring extensive information processing.
Such tasks range from the use of simple hand tools to tracking using complex powered machinery.
Human beings are needed for those tasks for a variety of reasons from their highly dexterous manipulative potential through to their ability to accept informal instructions.
These reasons are given in detail in the discussion of man-machine function allocation (p. 35).
The demands on the human operator are difficult to quantify or even to describe because the process is essentially an interactive one.
The skilled operator is working in harmony with his tools and machines.
The concept of ' demand ' is often not appropriate because it implies confrontation, the process is more aptly considered as one of cooperation and persuasion.
The human operator is pursuing an objective, usually to make something or to go somewhere, and he achieves this with the assistance of the hardware at his disposal.
This hardware, if it is well-designed, aids him in his progress toward his objective.
There may be demands in the sense that there are always obstacles to be overcome but these stem from the nature and the variability of the situation he is in; that is, the materials he has to deal with or the environment he has to move through.
The skilled operator will aim for efficient performance.
Efficiency incorporates not only quantity and quality of achievement but also preservation of his own safety and health and that of others who might be involved as working partners, passengers or the general public in the neighbourhood.
Of course, if his tasks are badly designed they may well make unnecessary demands.
These begin with bodily aspects of the operating posture required and the forces which must be exerted to operate controls.
for a work-space which is used full-time very slight departures from the optimum may lead to problems in the long term, for example in strained ligaments, tendons and muscles.
Posture is constrained not only by the physical dimensions of the work-space but also by the need to be in a position to see certain events and to feel others through control operations or machine movement.
Other demands will arise if the information presented is inadequate in content, e.g. poor lighting may make it difficult to detect some relevant cues, or in structure, e.g. because of poor coding and presentation of information on dials, charts or screens.
Correspondingly there may be output demands created by control operations which result in things happening too quickly or unexpectedly.
The level of demand is to do not only with the tasks as they are done but also with the duration for which they must continue to be done.
This applies particularly to routines and repetitive work where the main operator limitation is not capacity or skill but stamina.
In principle, repetitive work is best left to automatic machines but the flexibility of human performance is often needed because slight changes are required either to modify the product or to cope with different materials.
Batch production work is often of this kind and in general it is by no means as unchanging or boring as it might appear to the casual visitor to the factory.
In all the clothing trades for example there are continuous changes in sizes, materials, colours, styles and so on which provide the appropriate variety for the customers and incidentally for the producer.
Work design must take account of the needs of the consumer as well as those of the worker.
Ergonomics and other aspects of technology
The technical specialist who represents the people point of view implicit in the approach just described is the ergonomist.
In the previous book in this series the relationship of ergonomics to other aspects or kinds of technology was discussed in detail.
The unique feature of ergonomics is its emphasis on the characteristics of human operators and their relevance to the design of work.
The role of the ergonomist is essentially an advisory one.
In the case of high technology design he must provide a service for the engineers and scientists who carry the technical design responsibility.
This service may simply be as a source of information or the ergonomist can have more power in that the design decisions involving people have to be approved by him.
This is usually restricted to obvious cases of man-machine interaction such as the design of displays, controls and work-spaces.
In the case of low technology design again he has an advisory role but with a general emphasis on awareness of the kinds and ranges of people who are being catered for.
for systems which are already operating the ergonomist is likely to cooperate with occupational health and safety specialists and to play a part in the consideration of accidents and occupational diseases which might arise from unnecessarily stressful work situations (p. 294).
It will be appreciated that for most kinds of work and most working organisations it is not feasible to employ specialist ergonomists.
In such cases those who have responsibilities for work design need to have some awareness of the principles of ergonomics.
The philosophy of work
Ergonomics is sometimes confused by questions such as, who is the ergonomist servicing  the employer or the worker? or, to whom is he ultimately responsible  the state, the employer, the producer or the consumer? fortunately it is possible to dispose of these issues without taking up any particular political position.
Ergonomics seems to flourish equally well in capitalist and socialist/communist systems, ergonomics activity can be sponsored with equal validity by employers organisations and by workers organisations.
This is because, as already mentioned, ergonomics is about efficient use of people.
There can be no good reason to object to this when it is recognised that efficiency incorporates personal factors such as safety, health and quality of working life as well as system factors such as productivity and quality of work.
There is rarely any conflict in that good ergonomics is in line with the objectives of the worker, the employer and the consumer or customer.
There are sometimes larger issues of whether what the system does is acceptable or not, for example in weapon systems, and the ergonomist as a citizen will have his own opinion on these matters but the ergonomist as an ergonomist is confined to questions of efficiency.
There is an underlying necessary ethic which is best summarised as the social contract.
An individual is a member of a community from which he obtains considerable benefits, in return he develops special skills which he applies for the benefit of the community.
The application of special skills in this way is called work.
The ergonomist has to believe that work is a good thing and that to conduct it efficiently is always better than to conduct it inefficiently.
The design of systems.
The system concept
A system is a set of interacting parts which has meaning as a whole because it is possible to define a purpose  a reason why the parts are seen to be related to the whole and a justification for the concept of the whole (Singleton, 1974).
Thus, systems theory is essentially teleological  explanations are in terms of consequences rather than causes.
The parts may themselves be complex and may be conceived as having their own subsidiary purposes in confluence with the system purpose, these are sub-systems.
The system in turn is part of a larger complex  the parent system.
Thus, systems are hierarchical and at each level the unit is considered as a functional rather than a physical entity.
The systems approach is often a convenient way of looking at human behaviour either internally  this is man as a set of sub-systems or in terms of the man interacting with mechanisms  man-machine systems, or man interacting with organisations  socio-technical systems.
The enormous flexibility of the systems approach has its penalties in potential confusion due to the highly variable relationship between functional and physical entities.
Sometimes this relationship is fairly close; for example, in the nervous system, the endocrine system, the digestive system and so on, and sometimes it remains almost entirely unknown, for example, the perceptual system and the decision-making system.
These last two are obviously each related to the nervous system and to the brain but not necessarily to a particular part of the brain and taken together they overlap in such obscure ways that they are best regarded as sub-systems in different domains, one can talk in terms of one or the other but not both simultaneously.
Systems theory is a useful way of identifying complex entities, particularly those which have a functional unity and of talking about relationships but it remains at some level of abstraction from reality.
for example, to solve any design problem there has at some stage to be a switch from systems thinking to thinking in terms of physical entities because these are the things that can be created, precisely located and manipulated in the real world.
Nevertheless, there are enormous dividends in being able to discuss functions independently of physical mechanisms and to explain relationships without being restricted to established and readily observable physical connections or separations.
For example, thinking of a power station as a way of translating energy from fuel into electric power enables the designer to consider various options.
Should the fuel be of a fossil type such as coal or oil or should it be the radioactive elements in a nuclear reaction?
When the heat has appeared how best should it be taken away  by pressurised water, by steam, by liquid sodium or by carbon dioxide?
This heat has to be turned into mechanical energy, currently always by a steam turbine and the mechanical energy in a rotating shaft is then translated by a generator into electricity.
The losses in this transfer of energy from one form to another can be calculated and the flow through the various sub-systems can be traced.
The initial consideration of the power station as an energy processing system aids design discussions about the relative advantages and limitations of various mechanisms and their performance in practice.
As the system operates it has to be controlled.
The control of the flow of energy is partly through automatic servomechanisms and partly through human operations.
The design problem of which does what is also best thought about initially in system terms.
Distinct from the flow of energy the designer is, at this stage, thinking about the flow of information.
Energy and information are processed by systems and sub-systems.
In these terms, the human operator is a particular sub-system.
Mechanical devices and human operators are not comparable physically beyond the mundane level of size but they are comparable as different sub-systems whose performance can be described and assessed in terms of the common metrics of energy and information.
Hence the importance of systems concepts to ergonomics.
If the human operator is to control functions within other sub-systems he must receive information about the system state through a man-machine interface.
This interface takes its physical form in the control room.
The control room is part of the communications channel between the human operators and the mechanisms and its design, as a man-machine interface, is one of the key ergonomics problems in power stations.
Office organisations also can be considered as systems.
There will be certain physical mechanisms such as telephones, typewriters and computers but the flow of energy is trivial and the interest centres entirely on the flow of information.
Consider, for example, a travel office.
There is a vast amount of information available either in printed form or through computer networks.
The objectives are to help members of the public to find their way through this information store and to transmit their orders for particular journeys or holidays to the providers of those services.
The manager or designer of the office has to consider the various sub-systems he requires; one for dealing with enquiries, one for handling money, one for receiving new information, one for confirming orders and so on.
The interface between the public and the information store is a counter manned by people.
The counter assistants are parts of the interface.
Information is exchanged verbally and through written documentation.
In this kind of office the variety of requests for service is probably such that it would be necessary to provide a set of categories of the main kinds of activities or functions.
It would then be possible to devise the appropriate system for presenting the required information.
It is a very different situation from a power station control room but nevertheless the same principles of starting by the analysis of the required information apply.
The power station, the travel office and all other man-made systems have in common the employment of people so that sub-systems are required which are associated with personnel functions.
It is necessary to attract and select staff, to train them, to provide them with various services and with potentiality for advancement as their skills increase with experience.
The design of tasks is the common ground between personnel activities such as training and engineering activities such as workplace design.
Consideration of people in system terms provides the possibility of generalisations which apply across great varieties of physical and intellectual activities.
The resulting knowledge is the content of ergonomics.
Fig. 1.2 shows the outline of the systems design process in a way which emphasises the parallel roles of the engineer and the ergonomist concerned respectively with the design of the hardware and the personnel sub-systems and their common design problems of allocation of function and interface design.
The design of high technology systems
The insertion of an adequate Human factors approach into a comprehensive design process is not easy, partly for the reason mentioned already that it has not habitually been regarded as necessary and partly because it cuts across all other decision-making.
The Human Factors specialist can rightly be accused of wanting to have a finger in every pie because, as he sees it, behavioural considerations do affect every design decision.
If ergonomics was confined to, say, the determination of the physical dimensions of the work-space then it would be more readily accepted by traditional designers, but the broad claim that ergonomics has a contribution to make to every design and operational aspect which involves the behaviour of people is a proposition which it is much more difficult for an engineer or other designer to digest (fig. 1.3).
The questions begin with the formulation of the design objectives.
What are the human requirements which the new product or process is intended to meet and how far are they not met by those already available?
This would normally be regarded as a matter for economists and market-research specialists but behind the formidable array of evidence which they will generate are more fundamental questions of consumer preference, customer requirements and so on which are matters of human needs and prejudices.
Competent market research should cover these aspects but the ergonomist can remind the policy-makers that such evidence is needed and that it should be acquired by studies of people which are properly conducted.
If the evidence is favourable then the next step will be to set up a design team containing members covering the relevant range of expertise.
How much design effort is needed in the different areas will not be clear at this stage but unless the system is very simple there are bound to be problems of overlap and of how far to arrange sub-teams by discipline, by function or by problem (Singleton, 1987).
The usual compromise is that there should be problem-orientated sub-teams but with additional support available from discipline-orientated specialists.
If these latter are part of the design team then already we have a matrix organisation but they may well be provided on a service basis from some other part of the parent organisation.
Difficulties will emerge later unless considerable thought is given at this stage to the appropriate range of skills required and the consequent problems of interdisciplinary communication.
The art of good sub-system design is to identify problem areas where work can proceed relatively independently of other areas so that the need for intercommunication is minimised.
There will always be the need for some intercommunication and the appropriate interfaces should be provided.
Nevertheless, if committees, working parties and task forces begin to proliferate within the design team, this is a sure indication that the basic allocation into sub-systems was not optimal.
Communication should be mainly through the central command and control centre which may be no larger than the project leader and his small personal staff as may be a group of heads of sub-systems or both.
These issues are discussed from another viewpoint in Chapter 3 (p. 140).
The analysis of required skills ought in principle to have an ergonomics input but in practice it is usually done on the basis of experience of similar system designs in the past.
There can be a useful discussion between the project leader and an ergonomist where the former is invited to clarify how he sees these intercommunication problems being dealt with including such criteria as what is reported to him, what to the meeting of sub-heads and what occurs directly between sub-teams.
Organisational rules of this kind can be very helpful in steering between the excessive information flow, e.g. memoranda copied to many people, and excessive unilateralism, e.g. refusing further discussion after the issue of formal documents.
Having allocated the human and other resources needed for the overall design task it is then appropriate to develop a more detailed model of the system functions in terms of materials, energy and information flow.
Most designers have an intuitive but unformalised picture in these terms but it is of great utility to produce these pictures in flow diagram form so that any differences of view are revealed and can be reconciled by discussions during the process of agreeing the diagrams.
When there are a number of individuals or groups from different disciplines and pursuing different objectives within some overall specification it is obviously crucial that all their activities should be in line with the common goal.
To this end, it is necessary that there should be intercommunication not only about particular design decisions but also about underlying philosophies so that the ultimate system is conceptually as well as operationally coherent.
Sometimes this is achieved by a dominant project leader who imposes his own philosophy and then makes many of the decisions leaving only detailing for team members.
In recent times it is more usual to have some element of discussion, partly because of the increasingly democratic style of organisations and partly because the variety of relevant technology requires a correspondingly wide range of expertise.
In any case there is a need for widespread intercommunication which may be mainly informal but requires some formality to ensure that clashes of attitude do not delay or distort decisions and to make sure also that finalised decisions are unambiguously recorded.
If this intercommunication is not properly arranged there will be the familiar symptoms of frustrating reiteration, decisions being misunderstood or being constantly revised, political manoeuvring and even concealment of progress within one or more of the teams.
As the overall design progresses the information storage system becomes increasingly important.
If insufficient attention has been devoted to the design of this and also the related access procedures the symptoms are a plethora of documentation which no individual entirely understands, the temptation to do it again rather than waste time finding out how something has already been done and too much reliance on the human memory.
The behavioural symptoms of inadequate organisation of design teams are summarised in Table 1.1.
As the system is made more definitive by the selection of particular physical mechanisms there should be parallel activity in the consideration of how these things are to be used and maintained by the human operator.
Note that the philosophy here is that all the hardware is ultimately an extension of and a support for human functions.
This is the concept of all systems as human beings with various kinds of engineering aids (Singleton, 1967a).
functions are delegated from the human operator rather than allocated to him.
This extreme ergonomics stance serves as an antidote to the equally extreme engineering approach in which systems are regarded basically as hardware which for reasons of economic or technological limitations have to depend occasionally on some human performance.
Even if there is no dependence on human performance in the on-line operational mode it is bound to exist in the maintenance mode.
In either case, there will eventually be a need for the consideration of the kinds of human operators required and how they are to be selected and trained.
Traditionally these personnel functions were regarded as quite separate from design functions and indeed they were often left until the hardware design and commissioning was virtually complete.
The essence of the systems approach is that personnel decisions should proceed in parallel with engineering decisions, partly to economise on development time and partly to provide scope for trade-offs.
One factor which has given renewed emphasis to this need for cooperation is the increased use of simulation (p. 128).
Simulators were originally considered to be training devices but their use is now extending into other fields such as personnel assessment and as aids to the design of the information interfaces.
It used to be considered that simulator specifications could only be proposed when the on-line system design had been completed but there are considerable advantages in developing the simulator before the on-line system so that it becomes a dynamic mock-up of the real situation.
Detailed evaluations can be carried out on this mock-up before the real system is constructed.
All systems are constructed from bought-in components which may themselves be complex enough to warrant a human factors evaluation.
for example, a motor/pump system may well have its own controls, there will be standard maintenance procedures and there will be an operation/maintenance manual.
All of these can be assessed from the point of view of clarity of presentation and ease of use.
The system being constructed will require its own operating instructions, maintenance procedures and perhaps safety regulations.
The design and preparation of this documentation has a Human factors component.
It is a matter of storage, access to and presentation of information.
Even at the design stage it is useful also to recognise that all this material will require procedures for updating in the light of experience
Finally it must be accepted that technological and operational expertise is never complete.
The operators of plant are bound to discover by experience aspects of the design which could be improved.
There should be a mechanism for feeding this information back to the designers so that the succeeding system designs will avoid these problems.
In particular, human performance within systems is impossible to predict with any precision and details of this performance as it occurs in practice or on simulators is the only basis for prediction on matters such as safety and system reliability either for the current system or for future systems.
Potential human factors inputs to the large scale design process are summarised in fig. 1.4.
In all behavioural matters and indeed in many technical matters also it is important to avoid the impression that when operations in practice contain undesirable features not predicted by designers then the designers have made mistakes.
This may be true in a formal sense but allocation of blame is not usually a productive exercise particularly from a learning view-point.
It is valuable to inculcate the attitude that everyone learns, that is, design and operating staff all improve on the basis of experience with feed-back.
The exposure and remedy of facets of systems performance which are not entirely as predicted is in everyone's interest.
One example of this is matters to do with safety: it is always useful to collect ' critical incident ' data as well as accident data as one part of the feed-back about system performances.
The setting up of routines for this communication is another aspect of ergonomics.
In summary, although the basic contribution of the ergonomist is to the design of information interfaces these interfaces extend far beyond the man-machine interface found in the control room.
They include the documentation of the design itself, communication within the design team, between designers and suppliers and between designers and operators (fig. 1.5).
The operations staff include those who set it up, those who control it during operations, those who maintain it, and eventually those who have to dismantle it.
To give one example of the size of the documentation problem it is not unusual for a large process plant to have two or three hundred volumes of instructions.
These include the operating instructions, the maintenance instructions, the suppliers' instruction manuals, the safety regulations and instructions for particular emergencies.
pp 17C18:DIAGRAM/FIGURE
The design of smaller systems
The design of a machine or device which is assigned as an aid for one person will not usually encounter the problems associated with elaborate design teams just described, but nevertheless some human factors sensitivity will help to provide a better product.
This can be important not only from the point of view of efficiency in design and use but also in improving the sales appeal.
It is increasingly common for advertisements to emphasise features such as' man-machine harmony ' or ' user friendliness'.
The implication is that ergonomics has been properly applied.
There is a need for formal ergonomics in any design where the designer himself is not an experienced user.
Some hand tools are beautifully designed to match the user, e.g. the two-handed scythe.
In these cases it is clear that the design developed by constant trial in practice.
There are others, however, where even after centuries of use this matching is not obviously satisfactory, for example violin playing would seem to be posturally a disaster, but there may be some subtle aids to very precise control of the sound generated in having the source of sound very close to the ear and perhaps some vibration transmission by the direct contact of this instrument and the jawbone.
Ergonomics is concerned with the transmission of information.
for modest design enterprises information exchange is needed between the designer and the potential user, between the designer and the producer and between the designer/producer and the customer/actual user (Fig. 1.6).
The designer needs to know about the tasks which his new device is intended to aid and also about the kind of person who will use it.
Clearly, designing for fit young men in the armed forces has different constraints from designing for middle-aged female workers in manufacturing.
Some of the required information may be susceptible to precise description in numerical terms such as body size and strength, but most of it will be of the ' awareness' kind when the designer recognises broadly that particular kinds of people have particular advantages and limitations.
Similarly the range of tasks for which the device is required may not be known in the kind of detail which is available from comprehensive task descriptions but the designer will consider the extremes of what it is likely to be used for and the environment in which it will be used, for example designing a machine-tool for use in a factory has different requirements from designing a powered garden tool where the user could be wearing heavy gloves, will not be wearing safety-boots and will not receive any formal training.
Communication between the designer and the producer used to be done almost entirely by traditional engineering drawings but there is now a tendency to use sketches, schematic drawings, photographs and models including computer-based models, for example complicated shapes in three dimensions are difficult to represent as rectangular projections and are not easily interpreted.
Communication with the customer/user takes the form of operating instructions which are contained in an instruction book.
The design of this book, booklet or even leaflet is a much neglected art and is frequently very badly executed  hence the common attitude reflected in the aphorism ' when all else fails try reading the instructions'.
The most common failings are to rely too much on words rather than extensively labelled diagrams, to use technical terms and concepts which the user may not understand, to focus on what it is at the expense of what it does and to fail to separate the needs of the operator from the details which are needed to set up the device, to locate faults, to provide remedies and to provide routine maintenance.
An extended booklet should have a very good index or other access system to the information.
The designer is not the person to evaluate the clarity and usability of instructions, a sample of actual users will provide a much better assessment.
further details are given on p. 76.
The design of macro-systems
The larger systems which make up communities and nations are currently in a state of flux because of the impact of technology.
The provision of such a variety of telecommunication and travel facilities makes it impossible for any community however large or small to develop in isolation.
The consequence is an increasingly standardised way of life or at least aspirations towards a particular way of life across the world.
The differences which remain are enforced mainly by differences in average income and in style of government.
Technology continues to change jobs with consequent changes in ways of life and it also changes life-expectation which has profound effects on all communities.
However, just as hardware systems are dominated by engineering thinking, macro-systems are dominated by economic thinking.
There is excessive reliance on the adaptability of people who are required to adjust to whatever is dictated by economic and technological policies.
Politicians in all countries would accept the proposition that the system should be designed to serve the people but whatever the political philosophy behind the system, the people always seem to be more its servants than its masters.
Economic issues are real enough and it is often pointed out that politics is about priorities.
The priorities are assessed on the basis of evidence and unfortunately evidence about people is not so readily quantified as evidence about financial and physical resources.
The result is that systems at every level from the state to the small company are evaluated and policies are developed which are not sufficiently orientated to the needs and aspirations of the people within the system.
This is a very large issue mostly beyond the scope of this book except to note that ergonomics is an exemplar of a style of thinking which is based on characteristics of people.
Ergonomics can be defined as systems design with the attributes of people as the frame of reference.
Although the human operator is for many purposes within ergonomics appropriately considered as a mere information processing device, any design issue must be considered in the context that these human operators are individuals and citizens within communities.
What they believe, what they want, what they like, what they dislike and whether they work efficiently or not are determined by the societal context as well as by the physical environment.
It is therefore appropriate to bear in mind the position and trends in human resources in the community of which the workers are members.
For such macro-systems, the principle that the designer produces what he thinks the users need is being replaced by the principle that the users themselves decide what they want and the professionals assist them in designing as specified by the users.
This approach generates another duty for designers and ergonomists, to present to the users an array of possible options which will aid their discussions and choices.
Knowledge
Demographic data
Currently there are about 4.5 billion people in the world, 1 billion in the developed world and 3.5 billion in the developing world (Table 1.2).
This puts a correct perspective on the contents of this book which are mainly concerned with technologically-based societies although there is a section in Chapter 3 on ergonomics in developing countries.
Most of the evidence and experience reported here has been derived from the 16% of the world population living in Europe and North America.
The two worlds differ markedly in the distribution of populations, developed countries have about 65% of the population in the economically active age range from 15C64 years, about 25% under 15 and the remaining 10% in retirement.
This last group is still increasing quite rapidly.
The retired group is much smaller in developing countries where about 55% are in the economically active age range and more than 40% are under 15.
Life expectation at birth is about 45 years in developing countries and more than 70 years in developed countries.
The longer life expectation of females is more marked in the developed countries (Table 1.3).
The age distribution of the working population for two countries, one developed and one developing, which happen to have the same sized labour force is shown in fig. 1.7.
West Germany has a greater proportion of people economically active but they start work later and retire earlier than do workers in Pakistan.
There are about 4 million research and development scientists and engineers in the world, 3.5 million in developed countries and 0.5 million in developing countries, almost all this latter 0.5 million are in Asia, the numbers in Latin America and Africa are very small (UNESCO statistical year book).
The differences in various amenities are shown in Table 1.4.
The energy consumption is about 100 units (kg of coal equivalent per person) in a developing country and about 10000 units in Europe and the U.S.A. This is affected by the fact that developing countries mostly have tropical or sub-tropical climates while developed countries are in the temperate zone, but there is a similar difference of about 100: 1 in income per head between the rich countries and the poor countries (The Economist  the world in figures).
Within any country the Human Resource situation can be approached by this basic division into the ' economically active ' and others who, for one reason or another, are supported by the workers.
Those supported include young children, students, retired persons, seriously disabled persons, people living on independent means and, somewhat unfairly, women doing domestic work.
The numbers and proportions of those economically active in a sample of developed countries are shown in Table 1.5.
It will be seen that in each country this economically active group is about half the total population.
for males the proportion is usually about 55% and for females about 45%.
A very large economy such as the U.S.A. has more than 100 million workers, the large European countries have about 25 million workers.
Japan is about half-way between with about 60 million workers.
The smaller European countries have less than 5 million workers.
Workers in this sense are a country's Human Resource which is deployed for the benefit of the community generally.
The primary need of any person or population is water and food and in primitive societies the supply of these commodities will absorb the effort of almost all the workforce.
As food supplies improve it is possible to move some of the Human Resource into construction and manufacturing.
In the U.K. the concentration on manufacturing in the nineteenth century was balanced by the importing of food.
France, by contrast, had a smaller manufacturing base but remained self-sufficient in food so that the dramatic fall of proportion of workers in primary industries did not occur until after World War II (fig. 1.8).
In the U.S.A. even at the time of President Lincoln about 1860 there was serious debate as to whether the economy should remain agricultural or whether there should be a positive attempt to develop manufacturing.
In India the total population is currently about 665 million of which 244 million (37%) are economically active.
153 million are engaged in agriculture and fishing, that is 63% of the working population (I.L.O.
data).
In the U.K. less than 3% of the working population are now employed in agriculture, in most advanced countries this figure is now well below 10%, as illustrated in fig. 1.9.
This also shows that the predominant way of earning a living in a developed country now is in the service industries.
fig. 1.
10 shows the current distribution of U.K. Human Resources.
The former basic industries  mining, agriculture, construction and manufacturing  now employ about a third of the total workforce.
In manufacturing between 1980 and 1985 output measured per person-hour or per person employed increased by about 25% but the total output remained roughly constant, that is the manufacturing labour force is still falling rapidly (Economic Trends 1985).
Some consequences of these trends are discussed in more detail in Singleton and Crawley (1980).
p 24:DIAGRAM/FIGURE
p 26:DIAGRAM/FIGURE
From the end of World War II to the early 1970s there was virtually full employment in developed countries, but in the last ten years the position has changed considerably as shown in fig. 1.11.
Apart from Ireland it would seem that the smaller European countries have contained the situation better than have the larger ones, but this is an artefact since other small countries such as Belgium, Netherlands and Denmark follow the same trend as the larger countries.
It should be noted that the trend of an increase in unemployment from about 5% to about 10% over ten years is sufficiently general to indicate that it is not a function of the style of government.
These data for countries do not reveal considerable underlying differences within countries.
In England for example, unemployment in the North is twice that in the South-East, and in most European countries there are lesser employment opportunities in rural areas as opposed to urban areas resulting in depopulation of the countryside and an increase in the size of cities.
This trend is even more marked in developing countries.
Human resource utilisation
This brief excursion into the complex field of labour statistics is intended to illustrate the following points which are relevant to the theory and practice of ergonomics.
1.
Since the beginnings of technology, now about 200 years ago in the case of the U.K., the labour market has been in a continuous state of flux.
2.
The overall shift has not been so much from agriculture to manufacturing but rather a steady rise in the service industries, and a fall in primary industries.
3.
The manufacturing labour force in developed countries is now in accelerating decline as productivity and the investment per worker increases, this is resulting in a further increase in service industries and an increase in unemployment.
4.
There are considerable differences between the problems of so-called developed and developing countries.
This dichotomy is useful but is an oversimplification; degree of development as measured for example by income per head or technology utilisation is best described by a continuous scale.
A particular country is at one point on that scale at one time by one measure.
5.
Accelerating technological development has pushed the government of every country off-balance in the past decade.
This is illustrated in developed countries by the increase in unemployment which is fundamentally a failure to properly utilise the total Human Resource.
In developing countries there are many contrasts in that high technology (airlines, nuclear power, information systems; etc.) exists in small pockets alongside the still unresolved traditional problems of inadequate nutrition and health.
6.
Ergonomics also is pushed off-balance because the variety of human work increases in parallel with technology.
The traditional problems of heat stress and physical work-load remain, particularly in the developing countries, while new problems appear as more complex systems are designed requiring new operator roles and more serious consequences of inadequate performance.
7.
Nevertheless technology now provides the possibility of allocating high energy and repetitive work to machines so that the typical man at work is in a supervising rather than an operating mode.
8.
When machines provide the energy and set the pace the objective of the operator (and consequently of ergonomics) shifts from rates and amounts to quality, reliability and safety.
9.
The general concept of work is also shifting from man-machine interaction to man-man interaction with machines as aids.
10.
Many of the policy issues are mistakenly addressed in that there is too much emphasis on finding uses for what technology can provide rather than driving technology by the needs of people.
The behavioural specialists are responsible for providing data and attitudes for the policy-makers which should redress the balance.
11.
The economic concept of work is too narrow for ergonomics.
There are many ergonomics problems in the non-economically active half of the population.
If ergonomics is regarded as the science of work then work must be defined as purposeful activity rather than merely as activity for economic gain.
12.
This brief statistical survey reveals the scale and diversity of problems associated with the mind at work.
This is daunting but it is also challenging and stimulating.
Principles, methods and procedures
Education, work and leisure
Virtually all human societies have developed the principle that the pattern of personal activity should change with age.
In the early years there is extensive leisure devoted to play in which the basic manipulative and social skills are developed.
There is also more formalised learning through an educational system which concentrates mainly on information processing: reading, writing and arithmetic and more elaborate symbolic and organisational skills.
When the individual has matured he usually accepts responsibilities for child-rearing and for work.
In his declining years the proportion of leisure increases again, sometimes suddenly when he ' retires' but preferably gradually as his energies diminish.
Thus we can normally expect a mature mind in a person at work and although developmental processes of growth and senescence are of some interest, studies of the person working usually assume that we are dealing with a mind which is equipped with the basic skills derived from play and education but one which has not been subject to any serious diminution of capacity.
This is not to suggest that the age variable can be ignored.
It is accepted that ageing is a continuous process from birth to death and there are considerable differences between the average worker in the 25C45 age group and the apparent equivalent in the 46C65 age group.
The age distribution of a particular workforce can materially affect the problems of work design and training.
Correspondingly it must be recognised that an individual has entered work with a particular educational standard.
During his working years he has considerable leisure which is bound to have some interactions with work.
for approximate computational purposes it is normally assumed in the western world that a full working year is no more than 200 man-days.
The remaining 165 days disappear in week-ends, holidays and absence for training, sickness and so on.
There is also a widespread convention that the working day should be 7C8 hours, that is about a third of the total day.
Thus the full-time worker is in fact working for less than 200% of the total time.
The distinction between work and leisure and between work and education is an arbitrary one from the point of view of stress on the person.
for example, many city office workers regard the main stress of the day as the difficulties in getting from home to work and back again with work as a relatively relaxed period between.
Even more strikingly, some working mothers regard a period at work, particularly if it is part-time, as a welcome respite from the great variety of stresses arising in looking after a home and children.
Some of the most exacting university degrees, such as those in the biological and physical sciences demand a level of work load from the conscientious student which is greater than that expected in most occupations.
One would expect self-selected leisure activities to be different from and complementary to those at work.
for example, the manual labourer is unlikely to take up an energetic hobby such as cycling and he is likely to spend his holidays relaxing at the seaside rather than walking in the mountains.
The young office worker in a safe sedentary job may well take up rock-climbing as a dangerous, physically exacting antidote for the dull occupational part of his life.
Although it would be convenient scientifically and sometimes personally to regard work as a separable part of a human life this separation is never entirely valid and can be misleading.
The mind at work is part of the person at work and that same person has many other activities in the past and the future and in parallel which have their interactions with the work.
In system jargon, the working world is one sub-system within a complete set of sub-systems which make up the total environment or life space of the individual.
To study one sub-system in detail without regard to interactions with the others is usually inadequate, but on the other hand, to try to examine simultaneously the whole complex of sub-systems is to invite disorientation.
The remedy is a compromise in which activity in one sub-system such as that which incorporates work is studied but in the context of sensitivity to external influences.
Systems theory can be applied in many ways to behaviour at work.
The two main theories emphasise respectively interactions with other people; socio-technical systems and interactions with machines; man-machine systems.
Socio-technical systems
In the period of technological innovation following the second world war a group of human relations specialists, working from the Tavistock Institute in London, noted that the social behaviour of workers and their organisational structure in groups and teams was influenced by the technical systems associated with that work.
The first studies were in coal mines where there were different degrees of mechanisation associated with the same type of working group and it was also possible to distinguish different social organisations working in similar technical environments (Trist and Murray, 1948).
Some groups were organised so that each worker carried out an independent part-task with little reference to the total task, whereas in other cases each worker had a commitment to the total task and carried out various sub-tasks in this context.
This latter, labelled a ' composite ' group as distinct from a conventional group, had greater productive achievement and lower absenteeism.
More generally, the effectiveness of the production system depends on the way the social system is adjusted to the technical requirements (Emery and Trist, 1960).
Technological decisions necessarily dominate the design of the system in the sense of how the collective task of transforming inputs into outputs is achieved.
Nevertheless, the social system which integrates the activities of the workers and provides the flexibility for relating separate tasks to the needs of the production system as a whole must be considered.
Hence the concept of the socio-technical system where technical and social requirements are harmonised into one effective organisation.
The organisation, like a living organism, maintains a dynamic equilibrium with the environment.
Thus it is an open system rather than a closed one and its behaviour can not be understood by simply looking at the relationship between inputs and outputs.
The adaptability of the enterprise may lie in providing appropriate outputs to match the changes of the market from relatively stable physical inputs or may lie in providing a fixed output with changing sources of inputs.
In either case there is no fixed relationship between physical inputs and outputs, the system responds to the environment.
It follows that one of the main tasks of management is to control the interaction with the environment.
The implication also is that, if management has got the technology right within the system design, it should then attend to the environment and should allow, within the internal activities, scope for self-organising into autonomous groups.
If, on the contrary, there is a rigid internal separation between the roles of setting standards and meeting standards then conflict has been built into the system (Klein, 1979).
This approach can legitimately be regarded as the parent of organisation theory (p. 143), and of the ' quality of working life ' (QWOL), sometimes also called the ' job design ' movement (Davis and Cherns 1975, Davis and Taylor 1972).
It provides considerable insight into the eventual difficulties of the early twentieth century mass production methods as exemplified by the moving belt type assembly line.
On the other hand it does not provide remedies for production problems which can be applied as standard recipes or procedures.
Again, because of the elusive if all-pervading nature of the theme, it is difficult to formalise into legislation on humanising work as several European countries have discovered by experience.
It has emerged in many countries that, as the standard of living rises, it is more and more difficult to run an autocratic production system using the indigenous labour force.
The first management remedy is to import a more docile labour force from another country or from a less developed part of the same country.
This is only a temporary solution and there have been many attempts to organise more humane working systems.
The danger here is that, although things may be pleasant for the workers, the incentive to work very hard is not so direct and it may not be possible to maintain a competitive position in comparison with the same industry in a developing country.
Hence a further development of research on work motivation (p. 144).
Taking a more global view, one opinion is that the manufacturing industries should indeed move to developing countries whilst the advanced industries orientate their labour forces toward information and other service industries.
There is a trend in this direction (Singleton and Crawley, 1980), but, of course, it is not the complete answer.
There are other complications.
Robots, for example, are best suited for the highly repetitive manufacturing tasks which are the ones least suited to human efforts so that robotisation can cure some of the ills of mass production.
Looking at the environment, as market demands become more sophisticated so there is a need for operator flexibility in production system outputs, which has its counterpart in greater variety of work within the production system (Lupton and Tanner, 1980).
Socio-technical system theory has been highly fruitful in generating other academic lines of development: organisation theory, quality of working life concepts and work motivation, in stimulating many innovations in production organisation and in provoking governments to consider legislation in matters such as humanisation of work (Germany) and co-determination at work (Sweden).
Paradoxically it has not developed a formalised set of principles or immutable knowledge which could, for example, be structured into an academic text-book about socio-technical systems.
The problem seems to be that it is fundamentally about attitudes to work and such attitudes are subject to an inextricable and non-monotonic range of influences not only from work itself, with all the pressures of technological change, but also from prevailing economic, social and cultural changes in the community generally.
Man-machine systems
Work is very rarely performed by a man without the aid of some machinery and correspondingly machines rarely function for very long without human intervention.
Man and machine can be regarded as sub-systems each making a distinctive contribution towards the achievement of the purposes of the working system.
The system itself can be described as a set of functions necessary for the overall purpose.
The emphasis on defining the system in terms of its purposes or objectives and analysing it in terms of its functions are the characteristics of the systems approach.
It follows that at some stage in the design process it should be useful to consider how the various functions are most effectively allocated  to man or to machine.
Fig. 1.2 shows the essentials of the system design process but since feed-back paths are omitted this figure does not indicate either the repetition and iteration which goes on in operational design or the different possible priorities and variability in the order of decision-making.
The engineering emphasis will be on the search for mechanisms which can most effectively perform the required functions with the human operator in a support role undertaking functions which are not readily mechanised.
The ergonomics emphasis will be on the human operator as the navigator moving towards the system objectives supported by various mechanisms.
One neutral viewpoint is to consider formally the relative advantages and limitations of men and machines.
These can be listed (Table 1.6) on what is called a ' Fitts List ' after Paul Fitts who originally proposed this approach (Fitts, 1951).
At the time it was hoped that eventually it would be possible to refine the quantitative distinctions between the performance of men and machines so that allocation of function decisions would be unambiguous (Wuffeck and Zeitlin, 1962), but this has not happened.
The reason seems to be that men and machines are fundamentally complementary rather than comparable, this point was originally made by Jordan (1963).
If a required function can be specified precisely then a mechanism can be designed to perform it.
The reasons for incorporating men in systems are to do with properties such as flexibility and adaptability which by their nature are not specifiable in numerical terms.
Incidentally this clarifies why the man-machine system is such a potent combination.
Each provides attributes which are not easily available from the other.
A man without the support of machinery is physically weak and informationally slow.
A machine without human guidance is inflexible and highly limited in recognition and learning facilities.
Nevertheless it is not common practice to consider the man-machine allocation of function as an isolated design decision.
More usually the allocation emerges from a pattern of related decisions including what can be reliably and economically achieved by hardware and software, what the allocated tasks look like as integrated jobs for the human operators, the skills and traditions of the available work-force, the importance of safety and reliability and so on (Table 1.7).
As potential functional weaknesses emerge during the design appraisal further consideration is given to what functions might be allocated to the man or machine to provide greater mutual support.
The result of selective competent design along these lines should be a system with optimal man-machine allocation given the technological state of the art and a comprehensive sensitivity to the feasibility of the human tasks.
A detailed discussion of the historical development of the allocation of function concept is given in Singleton (1974).
This approach contravenes currently fashionable Human Factors methodology that there should be a formal attempt to carry out man-machine allocation.
Operational experience of real systems design reinforces doubts about the allocation of function concept as a basis for a unitary systematic procedure.
The paradigm illustrated in fig. 1.12 has the virtue of simplicity.
Like many block diagrams this conceals the practical problems.
from such a simple diagram it is easy to assume that all the thinking is done in terms of functions.
This can not be sustained because the only way to validate task analysis is to switch from functional concepts to behavioural concepts and the only way to validate hardware concepts is to switch to thinking in terms of specific mechanisms.
Thus, allocation of function requires a continuous juggling of functional, mechanistic and behavioural approaches.
In addition to the separate criteria related to men and mechanisms there are others to do with information acquisition, storage and presentation focussing on the interface design which has its own design criteria.
Note that a real problem usually involves men rather than one man and thus there are extensive man-man communication issues where few of the functions could conceivably be allocated to non-human mechanisms.
In a military command and control situation these will typically be sub-system operators, supervisors and customers using the output (p. 243).
Even when design attention can be focussed on the interaction within a single man-machine sub-system it emerges that the man may have very different roles in setting-up, operation and maintenance of his particular machine and each role may require separate man-machine allocation considerations.
A designer is often restricted by various mandatory requirements based for example on imagined safeguards against accidents.
There are also unspecifiable but none the less potent political restrictions, some from within the organisation, some from outside it.
For example the use of nuclear power in either the military or the civil sense is regarded as qualitatively different from non-nuclear power in its safety, control and management requirements.
Design is never an isolated logical process.
There is a legacy of experience from earlier designs which inevitably guide and inhibit a new design.
The shifts of allocation of function from this point of view are illustrated in Fig. 1.13.
Swain and Wohl (1961) argue that the man-machine concept is oversimplified, it should be man-machine procedures allocation and also that prime equipment should be separated from check-out equipment (Fig. 1.
14)
In equipment involving advanced technology the allocation of the control function is between the on-line operator and the designer.
The latter builds decisions into the hardware and software based on his knowledge of the limitations of the mechanisms.
High technology can also be used to provide the operator with flexibility in the allocation of functions, he can make his own decision about when to control manually and when to leave it to the mechanisms.
This is called supervisory control (Fig. 1.
15)
pp 38C39: FIGURE/DIAGRAM
It emerges from the above discussion that man-machine allocation of function is at best an oversimplification and it may lead to the neglect of some key features of human performance.
Only the human operator can cope with the concept of purposes and goals and more particularly with their modification in the light of changing circumstances.
Knowledge of circumstances is communicated partly by information exchange across man-machine interfaces but also by verbal exchanges between people.
These verbal exchanges can be factual but can also be diffuse in the sharing of attitudes, beliefs, urgencies, doubts and so on.
These features are just as important to efficient performance as are data about plant functions but they can not be incorporated in a man-machine allocation process.
Implicit in the attempt to complete a systematic man-machine function allocation is the principle that some trade-off is possible.
More attention to the hardware and information presentation will yield improved operator support whereas more attention to operator training will minimise the difficulties of interpreting information presentations and thereby making full use of machines (Singleton, 1967b).
These alternative procedures are summarised by the expressions' fitting the man to the job ' or ' fitting the job to the man '.
The systems approach indicates that the two should be balanced on technical, economic and behavioural criteria (Table 1.8.).
To proceed systematically, both require that there should be the greatest possible understanding of what the man is required to do.
The design of the man-machine interface must develop from an understanding of how the man is expected to interact with the machine and correspondingly the design of selection, allocation and training schemes depends on knowing what is required of the operator.
The study of demands on the operator requires task analysis.
In this context the term ' machine ' is used as shorthand for the total hardware and software system or, in other terminology, the plant.
For information systems the allocation of function and interface design issues merge into the procedure for considering the required points of human intervention in the total information network.
From this point of view the allocation of function and interface design are one core design activity based on the man-machine concept.
Task analysis
Terminology
In the context of systems design, task analysis has become the generic term applicable to the field study of human performance.
This is somewhat confusing because the centre of interest may be jobs or skills rather than tasks, and strictly the procedure is often more accurately described as one of synthesis rather than analysis.
For example, a designer may build up a sequence of actions which an operator must follow in order to cause his product to function properly  he has synthesised the task from a knowledge of how the machine or other product was designed to function.
Nevertheless, such activity comes under the heading of task analysis (Fig. 1.16).
Tasks represent the human contribution to system performance but for purposes of personnel selection, training and allocation, it is necessary for tasks to be separated or combined into jobs.
For systems which have evolved on the basis of operational experience rather than those which have been created by a formal design process, job analysis rather than task analysis may be the starting point of human factors studies.
The ergonomist will begin his investigation by the consideration of what the personnel actually do in their jobs which have emerged and been refined by practice and personal preference rather than design.
A task is usually considered to be an integrated sub-activity with a particular purpose as distinct from a job which is a particular kind of work carried out by an individual in a system context.
Jobs are collections of tasks or parts of tasks.
Occupation is the wider term which covers a range of jobs and the career of an individual in a community context.
For example, management is an occupation, a production manager is a job, and work scheduling is a task.
These distinctions are demonstrated in Fig. 1.
17.
Another source of ambiguity is that task analysis is often used as a generic term covering any study and description relating to people at work.
Sometimes this is no more than a description of what has to be done but sometimes it incorporates how the operator does it or should do it.
Strictly, this latter should be called a skill description rather than a task description.
In principle a task description is not necessarily an activity of a person, the task might possibly be completed automatically.
There is no such ambiguity about a skills analysis which is always person-orientated and not just system-orientated.
This confusion of terms obscures even further the difficult technical problems of moving from a task description to a skill description.
The task description should emerge logically from the system design and more particularly from the allocation of functions.
There is no such deductive way of getting at the skill description.
This can only be done by studying operators.
There are two problems in analysing tasks: one is to acquire the evidence on which to base the task description, the other is to record what has been found out in a way such that other people, notably the interface designers and the personnel specialists, can use it effectively to assist in their work.
It aids clarity to reserve the term analysis to mean the acquisition of evidence and description to mean the presentation of the evidence.
The test of a task description is: does it indicate a necessary function within the system performance?
The test of a skill description is: does it indicate how an individual goes about a task, does it allow for individual differences such as those between a highly skilled operator and a mediocre one?
For routine low-level tasks there may be no great distinction between a task description and skill description but for high-level tasks there will be a considerable difference.
Skills analysis is an attempt to get at what is going on in the operator's mind.
For a repetitive task there may be comparatively little going on in the mind which emerges at the level of consciousness.
For example, the assembly of a set of components will be carried out by an experienced operator without thinking about it.
He may be surprised when the analyst reveals the complexities of his hand movements and finger manipulations.
The only way he can recall what he does is by doing it, literally ' going through the motions'.
On the other hand, for a task such as fault diagnosis he will be thinking about it in various structured ways which he can reveal by talking about it and exploring verbally why he looks at particular indications or takes particular actions.
Human tasks and the purposes of task analysis are too varied for there to be one standard procedure for either analysis or description.
Analysing a task where the main requirement is speed of response is different from analysing a task in which problem solving is required.
A task analysis in the context of ' fitting the man to the job ' can be very different from one designed to assist in ' fitting the job to the man '.
As mentioned above the two extremes of human tasks are the procedure and the diagnosis.
Procedural task analysis is usually initiated on the basis of technology and logic by the designer.
A task has been allocated to the human operator and what he must do is a function of the needs of the hardware.
Such tasks may be in operation or maintenance.
In either case the designer writes down the sequence of actions which must be taken coupled with informational indications of what should be happening.
For example, in starting up a machine or system (Table 1.9).
Effectively the operating instructions are the task description.
It is, of course, desirable to check that nothing has been forgotten and to this end it is usual to ask an operator to go through the designed procedure while observing what he does and also observing that the system performs as expected.
This may be done on the machine if it is available, or on a simulator or on some kind of mock-up of what the system will be like.
This is known as the talk through or walk through method of task analysis.
The difference is only in the kind of task and size of hardware.
For example, in the case of starting up a machine at a control panel the operator will be given the instructions and asked to provide a verbal commentary as he actuates the various controls and checks the various indications, this is a ' talk through '.
On the other hand, a maintenance check of a series of related machines may involve moving to and fro between them while activating this and that, this is a ' walk through '.
The operating instructions/task description can be modified if the analysis reveals any problems from the operator's point of view; anything which the designer may have omitted or any difficulties the operator has in moving around, manipulating controls or checking information presentations.
Ultimately, the machine layout or interface design may be modified to make the operating instructions easier to follow.
For complex processes the procedures can include stages of diagnosis (see below) because the operator has to monitor and interpret what is happening in the plant.
In particular he is aware of various conditional constraints which form a multidimensional envelope, and whatever happens he must keep the plant state within this envelope.
Thus he must have extensive knowledge of the parameters and their relationship, e.g. pressure/temperature curves, together with the safety limits of each parameter.
Sometimes the safety limits are fixed, sometimes they are conditional on the state of other variables or plant, sometimes they are merely probabilistic.
This is an extremely difficult task, and the operator requires various aids such as alarm systems and automatic cut-out devices which are also effectively monitoring the parameters approaching limits (see alarm systems p. 225).
Diagnostic task analysis is even more complex and may indeed be virtually impossible to complete.
It depends on the complexity of the task faced by the operator.
The range of diagnostic task difficulties for a control centre is shown in Table 1.10.
The operator is assumed to be reacting to a situation in a control room, containing a great variety of dials, charts, and computer driven displays together with the controls needed to take action in any part of the system.
The simplest thing that can go wrong is the failure of one mechanism such as a pump, valve or thermostat.
This will be indicated in the control room by changes in pressures and temperatures which the skilled operator will trace readily to the particular failure.
Usually there will be some redundancy in the system so that other mechanisms can take over.
If the hiatus is serious he may have to shut down all or part of the system or it may trip automatically and the operator's task is to observe/control the shut-down to a quiescent state.
The worst possible condition that can be envisaged is one in which there is an expert in the plant aiming to do the maximum possible damage and aware of the indications of his actions which are transmitted to the operator  this is not quite science-fiction, it could happen as a form of sabotage,.
A much less demanding condition is one in which the saboteur is attempting to damage a plant about which he has little or no technical knowledge.
However, the difficult condition which requires the most consideration because it is not that unlikely is one in which the operator is faced with a situation where various things have gone wrong naturally or by inadvertent human interference in the plant or instrumentation.
His diagnostic task is somehow to differentiate between the information he is receiving which is false and that which is valid and follow this by the selection of a remedial strategy.
There are further complications in that, because of the plant failures, not all the standard range of remedial actions will be available.
Some automatic reactions may be taking place which might help or hinder, and he must also stay within the mandatory operating instructions.
Unfortunately, because such situations are created by combinations of events the number of possibilities is very large and it would be prohibitively tedious to enumerate all of them.
A complete task description is impossible and the appropriate preparation for the operator is not in direct training but rather in education/understanding of the system and the way it operates.
He must attempt his diagnosis by returning to the relevant scientific and technological principles.
How he does this in general terms is illustrated in Fig. 1.
18.
The task analysis will take the form of a consideration of the most likely or most dangerous failure and combinations of failures and the ways in which these will appear as indications in the control room.
This should occur not only at the design stage but also as experience develops in operating the plant and data on actual failures and failure rates become available.
Such task analysis can only be carried out to some cost-effective limit beyond which the skilled operator must be trusted to get things right by applying his broad expertise.
This is not as much of a gamble as it may appear at first sight, there are general principles of how to cope with the system, some of which can be built into automatic safeguards and some of which can be conveyed to the operator as knowledge and instructions.
In serious cases the aim is to shut down the system safely.
In less serious cases the operator may for a while maintain certain parameters at appropriate levels without diverting his attention to detailed diagnosis, this is called function control.
In dealing with lower technology or less formal system design, there is a range of relatively standard ways of finding out about operator performance by studying it directly.
Task analysis, in common with every other perceptual process, is a matter of setting up mechanisms for categorisation and filtering.
The analyst is faced with an activity which is continuous, homogeneous and full of detail, some of which is significant and relevant and some of which is of no consequence.
For example, the worker may pause at one point and scratch his nose, this may indicate that a critical decision is being made or it may indicate simply that his nose was itching.
The art of task analysis always is to select what does matter, reject what does not matter and separate into categories or stages something which seems at first sight to have no internal boundaries.
As already mentioned what is significant for a task description leading to an interface design may be different from what is significant for personnel allocation and training but the available procedures are the same.
The acquisition of evidence and the presentation of evidence are linked, of course, in that it may be necessary to decide what should be presented before it is acquired although more usually it is a matter of presenting what can be acquired.
The range of human tasks has already been indicated by the description of the extremes of procedural and diagnostic task analysis.
The simplest kind of human task, and the most straightforward to observe and analyse, is that in which the operator goes through a standard sequence of actions.
These are still common in industrial assembly and inspection tasks.
The most complex and most difficult to analyse are tasks requiring original thinking often with many apparently random alternations between deduction and induction, this occurs in fault diagnosis tasks.
There are at least two dimensions of complexity: the kind of task and whether the analysis focuses on material, information or people.
These are illustrated in Table 1.
11.
Each cell in this table represents an enormous variety of human work but this is hardly surprising since the total of eighteen cells covers all human work.
There is a variety of task analysis methods available and the selection of an appropriate set varies with the kind of work (as represented by these cells) and also with the purposes of the analysis.
Table 1.12 gives examples of kinds of task and the probably appropriate method of task analysis for each cell in Table 1.11.
It should be emphasised that this is only a tentative illustration.
The method used for a particular analysis depends on a great variety of factors including the analyst's expertise, the time available, the situations accessible for study, the related information already available, the purpose of the analysis and so on.
Task analysis by observation
Process charts.
Accurate observation of what a person is doing is not easy.
The observer needs some discipline which ensures that he really does focus on the detail and separates the action into a coherent sequence.
The required discipline is provided by the attempt to complete a chart such as a flow process chart using the A.S.M.E. symbols.
These were originally standardised by the American Society of Mechanical Engineers.
The symbols and an illustration of a flow process chart are shown in Fig. 1.19.
It is assumed that whatever an operator is doing can be classified into one of five categories: operation, inspection, transport, delay and storage.
Delay occurs when the operator is doing nothing  usually waiting for a machine to complete a sub-task  it differs from storage where there has been a particular decision to put something aside.
For example, storage for a hand implies some static work  the operator is holding something or exerting a force of some kind although there is no resulting movement.
Process charts may concentrate on material, operator actions or the events in the process.
These are known respectively as material-type, man-type or process-type charts.
More details of these charts and how to complete them are given in ILO (1979).
The same symbols are used for a two-handed process chart in which a detailed description of what each hand is doing separately is provided (Fig. 1.20).
The meaning of the symbols changes somewhat because the separation of events is much more detailed.
The same symbols can also be made to follow the concurrent behaviour of members of a team (Chapanis, 1959).
p 52:DIAGRAM/FIGURE
Note that, apart from the inspection category, the emphasis is entirely on what the operator does.
Little attention is paid to the input side as distinct from the output side, the sensory information and feed-back which the operator is using to control what he is doing is neglected.
This is appropriate enough for highly repetitive routine tasks because, as mentioned earlier, the operator is probably proceeding with little conscious attention and the limitation on performance is in the speed of the action rather than in the informational control.
To broaden the emphasis on a two-handed process chart it is possible to add another column labelled ' attention ' or ' sensory activity ' and to try to describe human activity in these areas at particular points in the sequence.
This goes beyond direct observations and requires some inference or speculation on the part of the observer.
pp 53C54:DIAGRAM/FIGURE
Finer analysis of hand movements is possible using classifications such as Therbligs (Shaw, 1952), or Methods-Time-Measurement symbols (Maynard et al.,
1956).
Analysis at this level of detail is not usually attempted by direct observation but rather by film or video analysis.
Sequence charts.
These are similar to flow process charts except that no standard symbols are used, the chart is simply a list of the actions the operator takes but there may well be additional columns referring to the events which occur and the consequent information which appears.
A sequence chart is more appropriate for a routine interaction with a machine rather than a physical manipulation of objects.
An example is given in Table 1.
1 3.
Sequence charts are particularly useful for the study and description of starting up and shutting down procedures.
They form the basis for procedural training.
Link design charts.
This is a detailed sequence chart orientated more towards operator choice of actions and correspondingly relevant to the study of interface design rather than training (Singleton, 1964).
It is assumed that for every action the operator has a purpose, the choice of the particular action at the particular time or point in the sequence is triggered by some other event, usually the successful completion of an action is indicated by a particular information presentation which also is noted on the chart.
The chart is particularly useful for detecting when information about action completion is not available and when an action is required which is not triggered by information from the situation.
That is, the operator has to depend on his memory (which is unreliable).
An example is given in Table 1.
14.
Flow charts.
These are more graphical presentations as distinct from verbal listings.
They are used to indicate the spatial location and separation of events which are integral parts of the same process.
An illustration of a flow chart is given in Fig. 1.21.
It is possible to obtain a kind of flow chart automatically as it were by using successive photographs of operator positions on the same photographic plate.
More usually, this is done by attaching lights to the wrists or to other parts of the body, if the light is flashing at a known frequency the relative time of particular positions as well as the sequence can be measured.
This is called a chronocyclograph.
Flow charts are a useful record of a process but are not particularly suited as a basis for analysis which might lead to improved job design or training.
On the other hand, they can clearly indicate differences in ways of doing something.
Block diagrams.
These are particularly useful to describe situations where events occur in parallel or in patterns rather than in a single series.
The block diagram is highly versatile as a spatial representation of any process at any level, that is the same process can be represented by one block or by very many blocks which provide more detail.
The art of useful block diagram production is to select the appropriate level of detail.
Most people find such diagrams much easier to understand than paragraphs of explanatory prose but the flexibility itself can lead to the difficulties illustrated in Fig. 1.22.
p 57C58:DIAGRAM/FIGURE
It is important that blocks are at about the same level of complexity within one diagram.
Blocks may represent physical entities, topics, concepts, decisions, actions or processes.
If these are mixed in the same diagram the distinctions between categories should be indicated by using differently shaped blocks.
Many of the figures in this chapter illustrate this convention.
Homogeneity is also desirable in the meaning of lines and arrows.
A line may indicate a causal link but simple cause-effect relationships are destroyed if there are feed-back lines.
Usually the line represents the flow of something but that something might be material, energy, information or time but again these should not be mixed on the same diagram unless differently coded lines are used.
It is usually possible to invert a diagram so that blocks and lines are interchanged.
One of the more subtle failures is to confuse a block diagram with its inverse.
An inverted block diagram can be a path-node diagram.
The advantage of these is that the length of the path can be used to indicate time taken.
This is the basis of ' critical-path analysis' where such a network can be analysed.
The shortest path through a network is the minimum time in which the whole process represented could be completed.
The longest sequence of required activities is the time it will take if no adjustments are made.
An optimum path can be selected and other activities adjusted to phase in with those on this' critical path '.
There are formal techniques of network analysis which occasionally prove useful; for example, in identifying equivalent diagrams or ways of simplifying diagrams (Battersby, 1970).
Algorithm charts.
These are a particular form of block diagram where the emphasis is on binary choices.
Again there can be concentration on events or on operator decisions.
In the latter case these are usually an oversimplification because there is an underlying assumption that decisions are being made in a simple logical sequence.
In practice, the operator jumps forwards and backwards within the pattern indicated on the chart rechecking some decisions and anticipating others.
Nevertheless, by relating required information to decision points any lack of information can be revealed.
It invariably turns out that an operator makes a series of decisions partly on the basis of what is currently displayed (on-line information) and partly on the basis of other material obtained from his knowledge of the system performance, his briefing and his operating instructions (off-line information).
Adequate on-line and off-line information is essential for apposite decision-making.
Fig. 1.23 shows such a chart.
p 60:DIAGRAM/FIGURE
F.A.S. T. diagram.
Function Analysis System Technique was developed by Bytheway (1977) within value engineering.
It imposes useful discipline on the production of a functional diagram in two ways.
Within each block the function should be described by a verb and a noun.
The functions are arranged in a sequence such that an arrow to the right on any line indicates HOW? and an arrow to the left indicates WHY?
If these criteria can not be met then either the functions are in the wrong order or some are missing.
The diagram should be produced by writing each function on a card, the cards are then arranged to meet the criteria.
Unfortunately the resulting diagram is not so easy to interpret as are some of the other forms of chart.
In particular there is no obvious starting or ending point and the lines do not indicate a sequence.
Fig. 1.24 shows an example.
General functional diagrams.
As a Task Description, a block diagram is intended to illustrate the dynamic behaviour of the system with emphasis on human interactions within the system.
For some purposes it is appropriate to use blocks and lines to indicate the flow of energy or material and to indicate all the ways in which the human operator can affect this flow.
This is not a performance diagram but a potential performance diagram (Fig. 1.25).
For other purposes it is appropriate to concentrate on a specific sequential task and illustrate the order of events involving human actions (e.g. Figs. 1.
19 and 1.20).
If a particular dimension (usually left to right or up to down) indicates a time scale as well as sequence this is called a time-line diagram.
The blocks may even be omitted to further emphasise the relative timing of events.
(Fig. 1.26).
Alternatively if the process is highly complex the emphasis may be on the pattern of events as indicated by a network of blocks with a specialist symbol indicating the many points within the system for which information for the operator might appear and another symbol for points where he can interfere in the system dynamics (Fig. 1.27).
Task analysis by discussion
Observation of operator performance within many high technology systems reveals nothing more than a person sitting at a desk scanning various kinds of displays at intervals and just occasionally picking up a telephone, making a note in a log-book or manipulating a control.
In such situations, the task analysis requires extensive discussion rather than observation.
Such discussions will centre on the operators themselves but may also involve others such as the management, the training officers, the safety specialists and the system designers.
There will often be complications in that there can be two or more operators working simultaneously with overlapping tasks.
The supervisor may also monitor the same information or may have his own separate displays or both.
In such a team operation there will normally be a clear standard mandatory difference of responsibility but in practice what they actually do can overlap and overlap differently for different teams.
Inevitably, discussion with different individuals about a particular task will yield very different versions of what is being done or should be done.
If there is a written version of what ought to be being done this is a useful basic document.
However the supervisor will point out that things have changed since that was produced and he will have his own view of the essentials of the task.
The operators themselves will again have a different version.
The differences will be mainly in emphasis on what is important central activity and what is peripheral.
Even between operators carrying out what is ostensibly the same task there will be discrepancies, particularly if the discussion extends from what is required to the way these requirements are met.
Similarly the specialists, the training officers, task designers and safety officers will have their own version structured in terms of the particular difficulties they have had in their specific contributions to the task performance.
Finally, task descriptions differ with the personal biases of analysts and the reasons why they are conducting the analysis.
However, in spite of all these sources of ambiguity the comprehensive task description is the essential starting point for most human factors work.
An approach structured by a particular procedure is of considerable assistance in reducing ambiguity and ensuring that the result is comprehensive.
The approach might consist of any one or any combination of the following techniques.
Responsibility charts.
A useful starting point for the study of any system is a responsibility chart.
This can indicate not only the hierarchy but also the different levels of modelling the system.
Normally, the operators closest to the plant will have the most detailed information and this gets condensed or reduced as one moves up the hierarchy.
For example in a power station the desk operator will monitor the detail of the plant performance, the supervisor will have more summarised information and greater contact with the national grid, while the station manager's on-line information about the plant is presented on one or at the most two dials in his office which indicate station output.
A responsibility chart can indicate not only decisions and duties but also the information needed to discharge these duties (Table 1.
1 5).
Although it is usual to model responsibilities in terms of a hierarchical tree there are bound to be overlaps between what appear to be different levels of the hierarchy.
These shared or duplicated responsibilities are unavoidable and often in fact are desirable because the process is one of mutual monitoring which usually improves reliability.
Occasionally overlap may have the opposite effect due to the ' falling between two stools' phenomenon, two operators at the same level or at different levels may each assume that the other has taken a particular action which they both know is required.
The remedy is adequate information feed-back and continuous checking but the need for this may not be obvious from the chart because the areas of overlap are difficult to illustrate.
It makes for much neater charting to assume that there are no overlaps but this defeats one of the main purposes of the chart.
Hence the most comprehensive and useful charts often have many irregular comments, asterisks, footnotes, directing arrows and so on, which at first sight detract from the clarity of the presentation.
Hierarchical Analysis.
Tasks, in common with responsibilities, are susceptible to subdivision at different levels.
A useful way of describing a complex task is to separate it into sub-tasks and further separate these sub-tasks into even smaller units.
This provides a very clear picture of the total activity although the order of doing things may not be obvious.
Such an hierarchical table (Fig. 1.28) can be supplemented by lists or sequences of operations which define a plan or a subroutine within a total task.
Correspondingly, a tree or network of operations can define a pattern of decision-making.
This technique, originally developed by Annett et al (1971), has been used extensively as a basis for training and retraining requirements (Duncan, 1975, Shepherd and Duncan, 1980), and also for the study of job satisfaction (Crawley and Spurgeon, 1979).
For a particular system there may be many different but equally valid ways of separating tasks into sub-tasks depending on the relative emphasis on training, interface design and other system design aspects.
Sometimes a ' bottom-up ' approach may be better than a ' top-down ' one.
Protocol analysis.
An operator carrying out a task can be asked to talk about what he is doing as he does it.
Clearly this is useful for tasks which involve extensive scanning of information sources and deliberations about what action to take.
The visible action may apparently be trivial  adjusting one or two of very many available controls  or may even be non-existent.
This sort of task is very common in process control situations where the operator often makes a preliminary three-way decision corresponding to the worldwide system of red, amber and green traffic lights.
That is, he does nothing, he does nothing at the moment but prepares for action, or he takes action.
Because so much of the activity is in searching for information and processing it leading to a decision, the observed output is obviously a poor indicator of total activity and his verbal commentary is a much better guide to his task performance.
Such a commentary can be recorded or notes can be taken directly by the analyst.
The snag about recording is that an enormous amount of material is produced which is very tedious to analyse.
The presentation of the result may use any of the charting methods described in this chapter, or may take the form of a descriptive narrative.
M.C.I. Job-process analysis.
Man-Computer Interaction has further complicated the relationship between human operators and other parts of a system but the computer can help to analyse these interactions.
Tainsh (1985) has developed a technique in which a computer-based simulation of a generic information handling task is used to assist in analysing the behaviour of an operator or a team of operators.
There are two V.D.U.s, one displaying alphanumeric data, the other graphical data.
A group of operators (note, these are people with extensive experience of the same or similar tasks, not the unskilled performer normally used in human performance experiments) run through a task using this equipment.
Their behaviour and achievements are observed and discussed amongst themselves and with the analyst (Fig. 1.29).
p 67C68:DIAGRAM/FIGURE
Conclusion
Task analysis has not and can not be reduced to a set of standardised procedures which the aspiring analyst could acquire in a formal training course.
The techniques just described are no more than guidelines in what remains the creative art of studying and describing what an experienced worker is doing or will be required to do.
The danger in being too rigid is that justice will not be done to the flexibility and complexity of human performance.
The danger of being too flexible is that the ergonomists will have an inadequate basis from which to design interfaces, training courses and other operator aids.
If a comprehensive task description can be completed then it is legitimate to enquire why the task can not be allocated to a technical device rather than a human being.
Yet without a reasonably comprehensive task description the total system performance may be unreliable or unpredictable.
There is a necessary compromise which can only be arrived at in the context of a particular system.
As mentioned earlier there are always the twin problems of how to get the information and how to present it.
It is sometimes easier to proceed on the basis of: ' If I were doing the job this is how I would do it ', but this can be delusory even for analysts who have past experience as operators.
For analysts who are designers creating a new task there may be no alternative, but any task description arrived at by this method should be checked against the behaviour of real operators as they gain experience.
A particular difficulty about task synthesis is that there is no easy way of confirming completeness.
A task description which is part of a design process should be fully  documented in terms of the origins of information, the assumptions made and the methods used.
The standard method of producing task descriptions is to write all the separate sub-tasks at various levels on cards and then assemble the cards in a network.
There is a temptation to make these final diagrams very large, but often it is more useful to separate into small units with cross-referencing.
The optimum size seems to be A3, this is easily copied and circulated but is large enough to contain considerable detail.
The procedure can be computerised, there are programmes available for the easy production, manipulation and storage of block diagrams with useful cross-referencing facilities.
There is an inherent difficulty in that often the main dividend from a task analysis is from the act of obtaining it.
Those not involved find it difficult to gain very much from a presentation of a task description.
Like reference books, they have to be used rather than read.
Task analysis can be very expensive in skilled manpower, but in looking at cost/value it is important to include in value the increase in mutual understanding which occurs when a multi-disciplinary team conducts the analysis.
If such a team is used the required man-hours can be optimised by using the Delphi method rather than a long series of group meetings.
Political, economic and social issues which arise in conducting Task Analyses are discussed in the context of Job Analysis (p. 148).
Application
Man-power planning
The ' Human Resource ' concept of man-power as the central profit-generating asset of any organisation has made some progress over the past twenty years.
There was particular interest around 1970 when companies were prosperous, stable and innovative (IPM, 1972, Patten, 1971).
The objective is to revise and almost reverse the traditional accountancy concept of man-power as a resource consuming factor requiring regular payments, overheads in the form of heated, lighted, equipped workspaces and so on.
From the narrow accountancy viewpoint, people are a cost and it is desirable to keep this cost as low as possible.
In these terms it is very difficult to justify, for example, sending a member of staff on a training course.
The training requires expenditure and so also does the replacement for the person away.
Where is the return?
The return is actually in the improved human resource but this is not readily measurable in terms which accountants use.
There have been two attempts to reduce human resources to this form of measurement (Giles and Robinson, 1972).
One is to acquire a human asset multiplier for each level of staff.
The multiplier for manual staff might be unity but it increases with staff level and becomes three or more for management.
The other is to add together all the personnel costs for each kind of worker from the original job advertisement through to the retirement or redundancy payments.
The first does attempt to obtain a value, the second remains essentially a cost and not a value.
It would be unduly optimistic to assume that management skills have risen to a level such that costs undertaken can be assumed to be equivalent to value.
A less direct measure which is applicable only to the most senior management is to observe the fall or rise of the share price when a particular executive leaves or joins a company.
This can be very high, for a large company a change in market value of millions of pounds is not unusual.
The most serious consequence of the accountancy view of man-power is that high level policy decision-making does not give this factor sufficient weight.
For example, the main board of a company will have complete details of cost and values of physical resources such as buildings and equipment but relatively scanty data about the man-power which is their most important asset.
Indeed it will be described almost entirely as a liability because the costs will be known.
Similarly, broad factors such as the state of morale will again only be described by negative features such as absenteeism, stoppages, strikes, low quality output and so on.
It is obvious that company output and projects must reflect the man-power and its performance level, but, if this is acknowledged at all, it is only in brief verbal terms within, for example, the chairman's annual report.
The obsession, of current management, with measurement and the reduction of every variable to numbers for ease of computer manipulation has worsened the situation.
Unfortunately there is no easy remedy.
Human resources are too complex a variable to be easily susceptible to measurement beyond the level of counting heads and trades.
Key variables such as the quality of staff at any level and the current morale are only expressible in general verbal descriptive terms.
It is generally accepted (p. 145) that ' hygiene ' type variables such as pay and working conditions can, if they are inappropriately specified, depress morale and motivation but they will not in themselves result in very high levels of these parameters.
Morale and motivation are influenced more by conditions of work in the sense of work variety opportunities and attitudes of management.
These factors are worthy of emphasis because in practice they are still dealt with very badly in many organisations and yet it takes little trouble or expertise to make an enormous difference.
for example, very few senior managements take the trouble to keep workers fully informed about policies and likely changes.
The result is excessive reliance on rumour and continuous speculation which is not only time-consuming but is also highly deleterious in morale terms.
The two human factors which can be unambiguously measured are age and sex.
The age distribution of a work force is always an interesting and important source of information.
For most purposes a rectangular distribution is the most desirable because it provides the appropriate mix of youthful enthusiasm and ageing experience.
It also makes for regular changes as staff retire and are replaced.
Many distributions show a peak in a particular age group mainly associated with taking on a large number of young people when the organisation was started or reorganised, this is not desirable because it leads to excessive competition for promotion at particular stages.
These comments are obviously most relevant for work forces where there is little tendency to move out, this is characteristic of a surprisingly large number of organisations in European countries and in Japan but much less so in the U.S.A.
Stable work forces have their advantages in terms of familiarity with the enterprise and its peculiar needs but they do result in problems if the market for the products is erratic or otherwise unpredictable.
This is particularly so in high technology e.g. design and construction of ships, aircraft, power stations and chemical plant, where the product unit is large and expensive and the variety of required expertise is extensive.
Some peaks of demand can be dealt with by using consultants, but they tend to be expensive and do not always fit well with ill-defined but important factors such as the favoured style of the organisation.
In well ordered systems it may be possible to calculate the length of in-house experience required to achieve the required level of expertise in particular topics and compare this with the lead time for particular product requirements.
Given time available to train new staff it is not necessary to keep more than a nucleus in that particular expertise.
On the other hand, if there is usually insufficient notice of need it may be necessary to keep staff who have periods with little to do between projects.
It is also desirable to have the possibility of shifting staff between design and operational duties.
This not only provides flexibility in the use of man-power, it also improves communication.
These real difficulties of man-power utilisation are such as to force many organisations to become larger and larger if they are to remain economic.
Size provides a cushion to the vicissitudes of markets but it creates other problems.
For example, difficulties of communications and in particular, how to inculcate the feeling within each individual that he and his performance are important to the organisation.
This is considered in more detail in Chapter 3.
Human resources should be fundamental to subjects such as ergonomics which are essentially people-centred but the concept remains procedurally undeveloped at any level from a continent through countries, regions and companies down to particular working systems.
This is one reason why policy makers never seem to take account fully of the people affected by their decisions.
The design of procedures
The increased importance of standardised informational support for both machines and operators has led to the concept of procedures as a separate design issue within a system (Fig. 1.30).
Much of this information can be incorporated in computer programs but there is also a need for extensive printed paper for the use of the operator.
This takes three main forms: checklists, routines and knowledge-texts.
The terminology in this field is not standardised.
Software might be restricted, as in this book, to computer programs but it is sometimes used as the generic term to include all sources of information.
Procedures is sometimes used for what are here called routines and sometimes to indicate all the operator's information support except software.
The situation is further confused when some but not all the operator information support is itself computer-based.
Sometimes the term firm-ware is used.
However whatever the terminology used there is a requirement for design effort.
The quality of procedures would be much improved if designers followed some simple guide-lines (Table 1.16).
Many documents are written on the implicit assumption that the reader is much like the writer in terms of expertise and style of thinking.
Similarly it is assumed that the conditions under which the reading will take place are the same as those of writing which are usually a clean well-lit office.
They may have to be read in the open air, in rain and darkness or within a plant under oily, cramped, poorly lit conditions.
The user probably has limited expertise and he may be working under various stress conditions.
The material on which the information is printed, the typeface used, the sequence, the structure and the format generally should reflect the conditions of use, (Easterby and Zwaga, 1984).
All these things can be achieved if it is acknowledged that all procedures should a validated by users under the normal conditions of use.
Checklists
Checklists are used in the pre-start mode, for routine maintenance and for fault-finding.
They normally take the form of a series of questions or statements which the operator uses to guide and structure his inspection of the hardware and other system components.
The sequence is dictated by the optimal order in which actions should be taken, for example in starting up an aircraft.
There are many situations when the order is of no great consequence, for example a safety audit of a new machine, and in this latter case some logical structure is needed.
As mentioned above, the logic of the design may not be the optimal basis if the way the operator thinks and acts is different.
Baker (1984) suggests that four kinds of shortcoming emerge when checklists are validated in the real situation.
The operator may deviate from the listed procedure because it may require excessive moving about or because he is interrupted by the requirements of other tasks.
Systematic use of a checklist often reveals faults in work design such as inadequate access, visibility and labelling.
The checklist itself may be inconsistent, incomplete, imprecise or ambiguous.
Finally the checklist may have reprographic limitations such as inadequate vertical or horizontal structure, poor typefaces and insufficient linking of requirements, descriptions, remarks and tick-off points.
The International Ergonomics Association developed a general ergonomics checklist which is reproduced in Edholm (1967).
There is a deceptive simplicity about such lists which requires extensive effort to ensure a comprehensive but coherent content.
The context is, of course, the context of ergonomics as conceived by the designers of the list.
Routines
A routine is the most common form of instructional material.
The reader is required to follow a series of steps which might, for example, introduce a change in the performance of a complex system.
Normally routines are single chains of instructions although it is possible to have some conditional branching.
Each step should be simple and complete in itself  preferably with some indication that successful completion has been achieved, for example the appearance or the change of colour of a signal light, or the noise of a motor starting or a change of pressure which indicates that there has been a change in the flow of some material.
As mentioned earlier a routine can often be regarded as a written task description.
By definition and design intention routines are inflexible.
The user is intended to behave exactly according to their instruction.
No great skill or intellectual effort is required, which is one way of saying that the human operator is being under-used, but the purpose of a routine is usually to avoid common human error such as omitting a step or reversing steps.
In general the value of formal routines is precisely that the operator, left to himself, will not normally function in such a systematic sequential manner.
Knowledge texts
Knowledge is contained in documents which may vary from single sheet presentations to large manuals.
As instructional material it is indirect in that the reader is not necessarily told what to do, as in a routine.
Rather he is informed about the situation and the purpose of informing him is to provide a knowledge base from which he can work out his own actions.
Nevertheless, the relevant action remains important and the knowledge provided should be carefully and stringently selected.
It is too easy to follow the academic tradition and provide knowledge for its own sake.
The ' need to know ' criterion is crucial because unless it is applied, the operator can be overburdened with unnecessary material.
Correspondingly the material should be structured in terms of how it is most likely to be used.
In this context the operator requires a guide for action and not a scientific text-book.
It is invariably better to provide material diagrammatically, systematically or at least in well-structured short paragraphs rather than in flowing prose.
Consider for example how readily information about the relative position of streets is available from a map of a town centre rather than a verbal description of all the streets.
Such a map will enable the reader to work out for himself how to get from one location to another within the town.
Note the enormous flexibility compared with a routine which would tell him only how to get from one particular point to another.
A second important difference is that if he misses his place in a routine he will be totally lost, whereas if he happens to go in the wrong direction while using a map he is still able to make whatever correction is necessary.
However the map does require more intellectual effort than does the routine not only in making decisions about how to proceed but also in filtering out the required information from the massive irrelevance (for a particular task) which is equally available.
Thus, one would expect the use of the maps to be slower than the use of routines with the human performance being more versatile but less predictable.
With two-dimensional representations such as maps it remains important to avoid too much homogeneity and provide substructure in the form of main roads and key buildings or other features such as rivers.
See, for example, Fig. 5.3
Manuals which may contain maps, schematic diagrams and other materials warrant separate consideration.
The design of manuals
A manual is usually in the form of a book or booklet which is sent with the hardware from the maker to the customer.
It is part of the communication package from the designer to the user.
Thus it is a particular form of job aid which provides the operator with information on a system and its performance.
The standard of manual design has traditionally been very low for several reasons.
For this purpose, the designer is too familiar with the product and he rarely takes the trouble to find out what the user really needs to know.
Often the manual has to be provided hurriedly at the end of a project when the budget is running out, and it is regarded as a tiresome chore still required after the essential design work has been completed.
The problem is particularly acute for computers.
Manuals for cars, machine tools and so on are poor but at least it is possible to observe the mechanisms and their functions.
Most computer operators do not know or wish to know about the mechanisms, and in any case these are not accessible to vision and manipulation.
The computer operator is entirely dependent on the manual, at least in the early stages of gaining familiarity with the system and its functions.
He is usually very badly served although the principles of manual design are well known.
For most systems the manual is best divided into parts as indicated in Table 1.17.
The first part should be a general introduction which orientates the readers towards the system, its purpose, its functional structure and mode of operation and its performance limits.
This should be followed by a separate section on how to set the system up in the first place and how to check that it is working properly.
If it is fairly small it may be necessary to describe how to unpack it, fix it into position, connect it to power supplies and start it up.
If it is large, the emphasis will be on the start up procedure.
The section describing the operation of the system should cover not only how the system operates but also how the user can operate it.
Extensive use of line drawings, schematic diagrams, algorithmic lists is desirable, and where prose is used this also should be structured into short sentences with meaningful separation into paragraphs.
The maintenance section should indicate what, if anything, the operator is expected to do and the symptoms which indicate that more specialist maintenance skills are required.
Tables subdivided into Faults/Reasons/Remedies can be extremely helpful.
Parts lists including description reference numbers and sometimes drawings are necessary if the user is expected to deal with his own replacements.
A standard fault is to assume that the reader is familiar with the jargon which the designer habitually uses.
A glossary of terms and acronyms used can be very helpful not only to the reader but also to the writer who is forced to consider carefully how he is using various technical terms.
Finally, all manuals should end with an index which supplements the table of contents at the beginning.
SKILL AND TRAINING
Concepts.
Development aspects
Most laboratory and field studies of human behaviour involve taking a situational snap-shot at a given time in a given place.
It is easy to overlook the continuously changing nature both of people and of work situations.
When the skilled manager encounters what, on the face of it, is an intolerable set of work practices, attitudes and performance he does not necessarily take drastic action, he identifies the natural processes of change and accelerates them.
In a remarkably short time changes in procedures, products and work-load, promotions, resignations, retirements and so on can transform any situation.
The changes within an individual are even more inexorably continuous.
Again it is easy to develop the illusion that a worker is a constant factor in a changing situation particularly for a person in the middle period of working life.
This was encouraged by the traditional view of a skilled man as one who learned his trade by the age of 21 and thereafter practised it for more than forty years until he retired at 65.
This was never the case and in the current working world it is even less so because of developing technology which has its impact on almost all jobs and also on the social context of work.
For example, because of employment levels there is currently an emphasis on earlier retirement from and later entry into the working world.
These trends are particularly marked in manufacturing industry where the work force is falling and the capital investment per worker is rising rapidly.
For these reasons industry is moving towards a concept of the ideal worker as a physically fit adaptable young person.
The labour force is becoming more like the military with the use of a limited age range and the screening out of anyone with any kind of disability who might not be able to work at a pace and with the flexibility and precision which will maintain the return on the very large capital investment.
Even so, it remains true that the performance of the individual is a function of the way he developed skills in his earlier more formative years and the way in which his capacities and skills are still changing in the work environment.
Human development can be described in terms of three attributes: capacities, skills and aspirations (Singleton, 198 1).
The healthy, well-adapted individual keeps them in balance in a dynamic equilibrium.
His natural endowment consists of a particular range of capacities: physical strength, mobility, vision, hearing, intellect and so forth.
These form the basis for the development of a repertoire of skills dictated partly by capacities and partly by environmental circumstances such as the particular family, community and region he is born into and the available facilities for play, education and training.
The balance and integration of capacities and skills is achieved by the third attribute of aspirations.
These are partly determined by and partly determine the other attributes.
The individual is in difficulties if his aspirations do not match his capacities and skills but most people are sufficiently realistic and have sufficient self-knowledge to keep these attributes in line.
The most flexible attribute is the skills, these can be developed or changed in line with capacities and aspirations.
A well-ordered civilisation demands standard basic skills of a physical kind: mobility and manipulation, and also of a mental kind: reading, writing and arithmetic.
On the basis of these standard skills there is enormous flexibility for the individual to develop his own identity and to express it through his specialised skills.
As the individual gets older, he continues to acquire new skills for two reasons.
Firstly, the passage of time provides more and more experience and this is inevitably reflected in the skill repertoire.
Secondly, the ageing individual suffers a steady diminution of capacities, for example the decrease in adaptability to lower lighting levels and the decrease in physical stamina.
Skills develop and change to compensate partially for the changing capacities.
It follows, of course, that decreasing capacities as such are not direct predictors of decreasing performance.
This is true not only for ageing but also for the onset of disabilities (see Chapter 7).
Human skill
Skills are the residues of learning.
To describe behaviour as skilled is to say no more than that it has been influenced by training and experience.
There is often an implication that skilled means highly skilled although this connotation confuses the concept of existence and level.
Unskilled is also used to imply a low level of skill rather than its absence.
Level of skill is usually assessed in terms of extent of formal training although this is only one of the relevant factors: others include length of experience and degree of integration.
Foundation skills are acquired in the first five years of life.
These include not only perceptual-motor skills but also the basic language and social skills.
Thereafter the educational system supplements the interaction with family and friends in providing facilities and exemplars for further symbolic and social skill development.
More details of all these skills are contained in Table 2.1.
The specific skills which a person acquires whilst earning a living are combinations and developments of the standard skills of the normal healthy adult.
This is not to denigrate the uniqueness and quality of working skills.
They are of the underrated because the universal characteristic of high skill is that, to an observer, the performance does not appear to be very difficult.
This is because the skilled individual is normally working well within his capacities, he does not have to search desperately for information because he knows exactly what to look for, his movements are not hurried because they are integrated into smooth chains and he does not react suddenly because he anticipates what is going to happen.
In general, he works on the principle of minimal effort to achieve the required objectives in terms of both speed and quality.
The importance of skill theory is that it relates human performance to systems concepts and to individual differences.
The performer is pursuing an objective and the great variety of his activity has meaning in relation to his objective.
Skill is a functional concept aimed at the understanding of how it is done without recourse to the detail of underlying physiological mechanisms.
Individual differences are obviously a consequence of the particular past experience of each person who has thereby acquired a unique set of skills based on a natural endowment which was also unique.
Skill theory also emphasises the closely integrated nature of behaviour in real situations, it is not readily susceptible to analysis into discrete components.
The internal representation of a skill is called a schema.
A schema is an internal model which can be used to relate external events to a current purpose (Head 1920).
It is modified continuously as external data is received and transformed into information.
Schemata may be person centred or situation centred.
For a more extensive discussion of schemata, how they differ and how they function, see Singleton (1981).
Acquisition of skills
Skills are acquired by a process of trial and error but the trials are not entirely random as is assumed in classical trial and error learning (Thorndike, 191 1).
However, the more primitive the skill the more there must be an element of chance.
For example, one of the skills developed in the first few years of life involves the integration of data arriving through the different sensory channels.
The fact that a visually observed impact between two bodies in the physical world is normally associated with a particular sound has to be learned.
Hence the importance of play in hitting, say, tins with spoons  the visual, auditory and tactile impressions are being integrated.
Similarly the kinaesthetic sensation of a limb moving has to be associated with the visual sensation obtained by observing the limb  hence the countless hours which babies spend just watching their own limbs move.
This is how manipulative skills begin, but it will be noted that even these are not purely ' motor ', they are mainly ' perceptual '.
Greater precision is achieved by more definitive selection and guidance.
Unskilled activity is to do everything rather than to do nothing.
The concept of motor skill is associated, with the development of longer and more complex patterns of activity which can be reeled off following a single trigger and which are monitored on the basis that only feedback which indicates errors needs attention.
Industrial training for manual operations is designed to encourage this kind of development (Seymour, 1966).
Perceptual skills by contrast are developments to do with the greater selectivity of information needed to monitor situations and guide actions.
These begin from an appreciation of the complementary nature of data arriving through the different sensory channels and expand by the acquisition of concepts such as the continuity of the physical world, e.g. a person who walks behind a screen has not vanished, it is accepted that he is still there even though temporarily there is no sense data to confirm it, but a hypothesis will be generated which supposes that, if he walked behind a screen at a constant speed, he ought to reappear at a given time at the other side of the screen.
Similarly it is soon accepted that the ' ostrich ' theory of closing one's eyes and assuming that thereby either oneself or things in the environment will actually disappear is not tenable.
In this phase babies and young children enjoy ' peep ' type games.
There is an important stage at which the person-centred world is superseded or rather supplemented by the world of which the person is one part.
Incidentally this is analogous to the distinction between displays which are ' inside looking out ' and those which are ' outside looking in ' (p. 251).
The perceptual system can function either way which is, of course, why each of the two types of display has its utility.
This development is one utilising pictorial skills as distinct from enactive skills.
The latter are based on the body, its extensions and images, the former are based on internal ' pictures' of the world which are one kind of schema or mental model (see next section).
Social skills are in principle more complex than perceptual-motor skills because the latter assume (correctly) a passive world which will react neutrally, according to the laws of physics.
In spite of occasional feelings to the contrary, usually described as good or bad luck, the physical world is neither for nor against the actions of the individual (except in the robotic sense of Le Chatilier's principle which states that whenever a constraint is placed on a system in equilibrium, the equilibrium is altered in such a way as to annul the efforts of the constraint).
In social situations it has to be recognised that there is at least one other sentient being present who has his own objectives which might accord with or oppose those of the first person.
Thus, social skills involve assessing the skills of the other person, a process known as mutual construing.
The evidence about the constructs of another person is indirect and may have been deliberately obscured or falsified.
Nevertheless, it is true for all skills that a person developing skills is indulging in a process of constructing models of what is happening externally.
These models develop continuously while attempting to modify the external situation (physical or social).
What seems to happen is that an individual, given his arrival in a situation, reviews it in the context of his own objectives and decides that if he takes certain actions, the situation will change in the direction of his objectives.
He then takes the action, observes the changes and compares them with his stored version of those expected.
If they correspond, then all is well, his concept of the situation is confirmed and he can progress to the next step in the context of his objectives.
If, however, the observed changes are not what was predicted, then he has made an error but he has also learned because he must now modify his view of the situation accordingly.
For example, I might decide the angle of light on my desk is not as I require it on the basis of assessing the illumination and the lamp which is the source of it.
I hypothesise that if I move and turn the lamp in a particular way I will get the result I want.
So I reach out, move and turn the lamp and check the result.
If the result is as intended there has been no learning but equally none was needed because I understood the situation, I achieved my objective and I can move on to other problems.
In a more complex case I might sit in on a discussion within a committee but begin to consider that it is not moving in a direction that suits me.
I therefore form a hypothesis about an appropriate contribution based on the proposition being considered, the associated facts and logic, the personalities of my fellow committee members and my own standing within the committee.
I then state my proposition and observe its effects.
If I get the result I wanted there is consolidation rather than extension of my committee skills, but if I do not get the result I wanted then I have learned something and I form a new hypothesis about my next interjection strengthened by this new learning.
It will be noted that errors are the basis of further skill acquisition.
Facilitating the acquisition of skill is a matter of providing situations where the skill can be practised with adequate, sometimes enhanced, feedback so that there can be learning from mistakes.
The necessary conditions for learning are feedback and motivation, the result of actions must matter to the individual.
All this is quite generally true, it applies equally to motor skills, perceptual skills and social skills.
Mental models
The principles of perception indicate that the human operator does not react unambiguously to stimuli which arrive via the sensory mechanisms (p. 196).
To assume that he responds directly to data from the real world is to oversimplify the relationship between his inputs and outputs.
The skill model described above implies that, except in highly restricted artificial situations, his information processing must be described in terms of patterns of stimuli related to patterns of responses.
It is also true that responses are a function of enormous differences between and within individuals.
To encompass all these parameters, it is currently fashionable to describe the person as responding in terms of his mental model of the situation rather than to the situation itself.
The idea is that any person in any situation builds an internal model of the situation which is being reported to him, as it were, by sensory data.
He uses each perceptual datum to update his model of the situation.
The model is his picture of what is going on outside him and it is essentially an hypothesis based on the available evidence.
This theory explains nicely why he sometimes does not react in tune with his own interests, that is, he gets it wrong.
In these circumstances he is considered to have developed a model which does not bridge the gap between reality and his aspirations.
These models are most easy to visualise in pictorial terms  the human operator has his internal picture which topographically matches the real world.
Models can, however, be simpler than these and can take the form of rules or they can be more complex in that symbolism can be used.
Thus a model can be entirely abstract using symbols which have no pictorial quality as in the use of language and mathematics.
Even when a picture is used, it is not complete, it takes the form of an icon which contains what the perceiver considers matters at the time.
For example, a person working in a room with or without other people can easily develop the misconception that he has a complete picture in his mind of the physical environment.
In fact the picture is complete only in the sense that it contains the details which are relevant to his activity in that room, it certainly does not contain all the detail which is available to the senses.
This is readily demonstrated by asking him to close his eyes and answer questions such as' is there a picture on the wall on the left? ' or ' what colour are the eyes of the person you are talking to? '
This is the kind of data which is not relevant to his objectives and thus it is not recorded in his model.
On the other hand he will be able to answer questions such as' is the door behind you to the left or to the right? ' because he has used that door and needs to use it again.
He knows where it is even though he can not currently see it.
Thus the visually-based mental picture is invariably an icon in the literal sense of a ' symbolic representation ' rather than a complete reproduction.
At a more abstract level the mathematician or scientist will perform manipulations within a set of symbols which follow certain rules but neither the symbols nor the rules need have direct representation in reality.
Indeed any attempt to relate them prematurely to reality may restrict the development of the model.
However, at some stage he will return to reality and say, for example, ' what this means in practice or in terms of my next experiment... '
The skilled thinker has an extraordinary facility not only to develop varieties of abstract models but also to switch rapidly between them and between them and reality.
The switching takes place up and down a hierarchy of abstraction and generality.
Consider for example the maintenance engineer operating as a diagnostician.
He will look at a piece of equipment to try to detect signs which indicate what is wrong.
Is there a gap or a hole which indicates a leak?
Is there a part which is too hot?
Most of this information comes not so much from what is happening but from what is not happening.
Basically the thing is not working as it should, the outputs are not as they should be so he will probably begin by checking that the inputs are present, that is the electric power or fuel or other materials are available.
He will then envisage the designed chain of events between input and output and look for the hiatus, he may do this by rules, by a pictorial model or by a model involving some symbolism such as expected voltage levels.
He will shift very rapidly between different representations of the equipment; the thing itself, his maintenance instructions, the manual, the drawings of the system, verbal discussion with a colleague, his recollection of what has happened to it in the past and so on.
He may think in terms of bits of equipment, flows of material, flows of energy or flows of information.
He will shift through levels of generality and levels of abstraction until he has formulated an hypothesis which takes the form of ' if I do this  that should happen '.
He will then perform an action and check its results.
There are many interesting facets of this complex diagnostic process and three are worthy of re-emphasis: firstly most of his information comes not from positive signals, stimuli or cues but from their absence, secondly he is relying on a hybrid model  a mixture of rules, mental pictures and symbolism, thirdly it all happens without very much conscious guidance  one mental or physical event leads to another and each event in the chain plays its part before handing over to the next one.
There is some analogy with a control system used in early naval warfare.
There was not one ship permanently in command, the particular ship in most direct contact with the enemy was in command until the leading position was taken over by another ship which then assumed command as well as action.
The working situation is different from the training school and even more from the educational system.
However it does seem that mental models can be developed in the abstract without the need to resort to action, in other words it is possible to learn while sitting passively in a class-room.
Education or abstract learning does seem to take the form mainly of symbolic modelling and the relating of these models to reality awaits later practical experience.
The great asset of symbolic modelling is just this, learning does not have to depend entirely on practice and it is possible for collective knowledge to accumulate.
This knowledge is communicated between people by the use of shared models.
Conversely, abstract communication can not take place without the development of appropriate models.
For example, the technical teacher is trying to convey various models to his pupils, some of the models will be physically based as in describing particular machines or systems and some will be functionally based as when he talks about thermodynamics or communication.
What the teacher is trying to do is to guide each pupil into developing a set of mental models which have been found by experience or experiment to be useful in understanding technical processes.
Having developed these models the pupil is in a position to communicate much more readily with other technical people who can manipulate similar models.
Conversely, this communication between specialists will be incomprehensible to those who do not have the expertise which resides in this particular set of models.
Even between technical peers as in a design team there can be difficulties in communication because the models which individuals are using are not identical.
Debate, demonstration and other communications take place until a common modelling is accepted.
When there is disagreement, there is much to be said for retreating specifically to consider the model rather than the problem.
Correspondingly, when there is a complex situation to discuss there can be considerable dividends in attempting to get an agreed model ' on the table ' before considering specific issues.
This is what briefing sessions are intended to accomplish.
This approach also explains why engineers and ergonomists sometimes have difficulties in reaching agreement.
The former is likely to be using mechanism-based models with people as ancillaries while the ergonomist will be using models of people with mechanisms as ancillaries.
For good cooperation it is essential that the engineer has some grasp of ergonomics and that the ergonomist understands at least in general terms the technical situation in the particular industry.
There can of course still be disagreements even when there is clear mutual understanding because underlying values result in the assessment of very different priorities.
In summary, communication, whether between a person and the physical world or between people, involves each person in developing and manipulating a mental model by interacting with the situation.
Modelling the physical world and modelling other people's models share the same developmental procedures.
Understanding what a person is doing requires an appreciation of the kind and content of the models that that person holds.
Knowledge
Individual differences
Capacities, abilities and aptitudes
It is useful to discriminate between a capacity which is a natural endowment and an ability which is a competence to act.
A capacity relates to an ability through an intervening phase of education, training and experience.
Thus an ability represents what a person can do now, whereas a capacity is essentially a potential.
An aptitude is demonstrated by success in learning which results in the acquisition of an ability.
An aptitude is therefore a capacity plus an inclination or interest.
Ability and skill are often used interchangeably although ability is the wider term.
An ability is made up of a repertoire of skills and is thus a higher level skill.
Ability and capacity are often confused for several reasons.
Human resources might be expressed in terms of either since capacities can be transformed into abilities by training.
Strictly we can only measure abilities but much psychometric testing aims to infer capacities from the measured abilities, e.g. intelligence tests.
All three terms are attributes of individuals intended to describe their distinction from other individuals.
All are used loosely because there are no taxonomies which are either comprehensive or rigorous.
That is, it would be unduly ambitious to claim to be able to describe a complete range of human abilities and the descriptors used for particular abilities are not always separable.
The same is true for capacities and aptitudes.
The best known attempt to provide a comprehensive structure of human abilities is that due to Guilford (1967).
His model contains about 120 different abilities, it is not currently very popular with psychometricians.
The only unambiguous categorisations of people are by physical size, age and sex.
None of these is uniquely related to abilities but nevertheless age and sex are useful parameters.
Age in particular at least correlates with some of the parameters which affect choice of work and performance at work.
Intelligence and personality
There is no agreed definition of intelligence although a century of experience has accumulated in attempts to measure it (Friedman et al.,
1981).
Intelligence defined as that which is measured by intelligence tests is obviously circular and also variable as the fashions in intelligence testing change.
For example, the attempt in the past thirty years to incorporate more divergent factors changes the emphasis towards inductive and creative abilities.
This supplements the consideration of deductive and logical abilities measured by the traditional convergent questions for which there are unique correct answers.
In spite of the absence of a sound theory, the concept is operationally useful.
Intelligence is a descriptor of the quality of mental functioning, the ceiling of intellectual manipulation to which the individual can rise.
High achievers in technically demanding fields are always intelligent although the converse is not true.
That is, there are many intelligent individuals who are not high achievers.
It follows that a particular score from an intelligence appraisal can be a useful cut-off point in that those who do not attain the cut-off should not be expected to cope with the demands of the particular task.
However, a high score does not necessarily predict success in a high level job.
The intelligence test can be a useful source of evidence about individuals who are considered for lower level jobs.
For more senior jobs individuals will have already demonstrated an appropriate level of intelligence by their educational standards and successful work experience.
In line with the principles of defining the relevant abilities in terms of reliable test measures, intelligence can be divided into verbal ability, arithmetical ability, reasoning ability and spatial ability.
For more extensive discussion see Eysenck (1979).
There is a long established distinction in psychology between cognitive and conative aspects of behaviour.
Cognition is to do with the rational domain: logic, reasoning and problem solving uncomplicated by irrational behaviour.
Conation is to do with the emotional domain: feeling, caring and striving.
Intelligence is connected with cognition but personality incorporates conation.
Thus Freudian theory is regarded as a personality theory.
Clearly if we wish to describe the whole person conative aspects must be included.
Unfortunately personality testing is much less well developed than intelligence testing.
The best known personality tests are the Minnesota Multiphase Personality Inventory (M.M.P.I.), Catell's test (16 P.F.), and the Eysenck Personality Inventory (E.P.I.).
For details see Kline (1976).
These are self-report inventories where the testee has the possibility of cheating in that he can respond with an answer which he considers will give him a good score rather than providing a completely truthful one.
Such manipulation of answers is less feasible with projective tests where the testee is asked to respond to a highly ambiguous situation and his response is presumed to reflect his personality as well as the situation.
The presentation might be a picture (thematic apperception test), an ink-blot (Rorschach test) or prose (sentence completion test).
The difficulty about using these tests is how to score the responses with any reasonable hope of reliability let alone validity.
There are other tests which relate indirectly to personality, namely interest tests.
Here the testee is presented with a set of questions or checklists which steer him into trying to think carefully about his own interests.
The best known ones are the Kuder Preference test and the Strong-Campbell interest inventory.
They are particularly useful in occupational guidance (Nelson-Jones 1982).
Age differences
The extremes of the age range, children and old people, are outside the scope of this book.
The working age range is 15 to 65 years.
This is not to suggest that those outside this range never work, but rather that in advanced countries gainful employment is considered to be mainly the responsibility of those in this particular age band of 50 years.
This band is currently contracting even further in European societies.
Because of the rise in unemployment it is now becoming standard that almost all those in the 15C18 age range are in education or training schemes and retirement is increasingly common from 55 years upwards, particularly from manufacturing industry.
Nevertheless, from both physiological and psychological viewpoints, ageing is regarded as a continuous process so that considerable changes can be expected between 1 8 and 55 years.
Using the information processing model of receptor, central and effector processes (Welford, 1958; Birren and Schaie, 1977), changes occur in all these processes.
The receptors diminish in efficiency: there are consistent changes in visual acuity and changes in sensitivity to light, the threshold of hearing increases particularly for higher frequencies.
The effector processes become slower and less powerful but these effects are not so marked or so consistent.
The main change in central processes seems to be decreased short-term memory which results in performance which is less rapid and more deliberate; measured intelligence falls slightly but not very much within the working age range.
This may seem to be a gloomy picture but it must be noted that, in relation to work, age is not a large aspect of individual differences compared with natural endowment, and that increases in ability can more than compensate for small decreases in capacity.
Wellord (1979) suggests that work on the shop-floor is likely to peak in the thirties  those in their twenties lack experience and social skills, those in their forties are beginning to notice the effect of sensory changes and are less tolerant of fast or very heavy physical work.
On the other hand, from the forties onwards there is likely to be greater stability in the sense that individuals are more settled in their jobs and ambitions and are supported by less rapidly changing domestic and leisure situations.
The effectiveness of training can be maintained if the training scheme incorporates features such as smaller learning steps and ' discovery learning ' where the trainee proceeds at his own pace using his own strategies (Belbin 1965).
For more senior positions there are more subtle attributes such as self-knowledge and the ability to see ' the big picture ' which increase with age, these are the reasons why most institutions and societies rely on older leaders (Singleton, 1983b).
Sex differences
There is surprisingly little research on the differences in work performance which are traceable to sex differences, and the situation is complicated in many countries by the inclusion of sex as one of the variables where discrimination is legally inadmissible.
Incidentally, in this book ' man ', ' he ' and ' him ' are used as generic terms covering both sexes.
There can be no argument that, on average, women are physically smaller than men, and there are other features such as the greater proportion of fatty tissue which give men the advantage in relation to tolerance of heat stress and physical effort involving fast or heavy work or awkward posture.
On the other hand in Russia, for example, where so many males were lost in two world wars women do do much heavier work than is traditional in Western countries.
Similarly there seem to be many cultures in tropical countries where the women get on with the necessary work while the men sit around discussing matters.
There appear to be no important differences in performance on intelligence tests in the working age range and as the influence of physical effort in work becomes less important so it is more feasible that jobs can be performed equally well by either sex.
The overall situation was reviewed by Hoiberg (1982).
Equal opportunity legislation exists in most advanced countries but this is not yet reflected in equal pay rates.
The average wage of women workers is two-thirds that of men.
Part-time work seems to be a prerogative of women presumably as a consequence of their central role in the home.
Educational opportunity varies, depending mainly on family attitudes.
In some cultures sons are given priority, others give daughters priority, following the McIver principle ' when you educate a man you educate an individual: when you educate a woman you educate a whole family '.
Ethics, cultural, climatic and economic differences
This is a sensitive topic on which to conduct research or establish principles with the consequence that, in spite of the extensive effort (Triandis et al.,
1980) there is not a great deal of knowledge established beyond argument.
Climate and economics need to be included because these factors may well determine features which at first sight might be attributed to ethnic or cultural factors.
Clearly the work that people do is determined by the level of technology which is related to overall wealth.
Ways of working and locations of work are also affected by climate.
For example, in a tropical country shoes may still be made in the open air by a craftsman who uses his feet as well as his hands to hold and manipulate materials.
In a temperate country shoes are made in a factory by workers who are process controllers rather than craftsmen, and they touch the materials only when loading or unloading machines.
For these and many other reasons the practice of ergonomics in developing countries is different from that in advanced countries (p. 155).
The acquisition of skills is governed by the current needs of particular societies and by the adaptation of individuals within the societies in which they happen to find themselves.
Research on the influence of cultural factors on intelligence testing and test results have been conducted on a wide scale and the results are endlessly discussed because of the inextricable mixture of technical and value judgments.
As Crombach puts it in the foreword to Irvine and Berry (1981) ' the progress is as much in the psychology of the investigators as in the investigations being reported '.
As in other fields of behavioural study there are three phases.
In the first phase techniques and theories are developed in the West.
In the second phase Western practitioners try out their ideas in other cultures and many new insights emerge by comparisons and contrasts, in particular it becomes clear that some of the principles thought to be valid for ' man ' turn out to be valid only for man in Western cultures.
In the third phase scientists from other cultures acquire these concepts, develop them further in their indigenous environment and provide penetrating comments on Western ideas and their application from an external viewpoint (Crombach and Drenth, 1972).
The dominant influence of culture on the direction of human development implies that descriptions of human abilities can not be context free.
Specific cultures are themselves dynamic, in particular they change under the influence of technology.
Ergonomics can not be simply the adaptation of technology to suit invariant properties of ' man ', people change with technology.
Learning
Learning has been a major topic within academic psychology for the past century or so.
The nineteenth century work of Pavlov on the conditioned reflex generated vain hopes that psychology, and in particular learning, might be studied by a rigorous methodology closely allied to physiology.
It was assumed that all learning could be reduced to a series of conditioned reflexes.
There was extensive research in this behaviourist tradition (Hilgard and Marquis, 1940) mostly on the laboratory rat.
So called verbal learning in man was studied by a similar process of scoring repetitive performance.
Reliance on such data produced the systematic behaviour theories of Thorndike, Hull and Skinner.
There was a parallel although less popular theoretical development based on Gestalt psychology which resulted in the field theories of Lewin, Wheeler and Tolman (Hilgard, 1948).
There were many attempts to devise taxonomies of learning which used particular theories as descriptive of different kinds of learning (Melton, 1964).
At this stage there seemed to be little connection between the vast experience of learning in practice within educational and training systems and learning as conceived by psychologists on the basis of laboratory experiments.
The psychologists attempted to polarise various issues such as massed versus spaced learning, whole versus part learning and transfer of training which had some face validity in appearing to identify general principles, but in practice it always seemed that the generalities vanished into the enormous variety of specific issues.
The answer always differed with the particular situation.
Gagne (1970) attempted to produce a theory which was germane to educational problems by switching the emphasis from categories to conditions of learning.
The deliberate attempt to link learning to the design of instruction provided the necessary discipline to treat learning as a practical as well as a theoretical issue.
The eight types of learning which Gagne distinguishes are shown in Table 2.2.
Gagne also separated four phases of a learning sequence; apprehension, acquisition, storage and retrieval.
The apprehension phase includes attending and perceiving, that is accepting new information.
Acquisition is not easy to separate from either apprehension on one side or storage on the other, but it implies that learning has taken place in that something new can now be done if appropriate.
The distinction from apprehension is necessary to incorporate the phenomenon that it is possible to understand a situation in one sense without knowing what to do about it.
Correspondingly acquisition is different from storage in that the latter implies a formal recording possibly in short-term or possibly in long-term memory.
Finally retrieval is a separate process because of another familiar phenomenon that it is possible to know something, that is it must be in storage, but not actually succeed in retrieving it at a particular time.
These phases and the difficulties of separating them reflect the fact that mental processes are not subject to clearly defined distinctions and boundaries.
In common with all educational psychologists Gagne is also restricted by reliance on evidence which comes mostly from laboratory experiments, usually on lower animals, with all the limitations and artefacts which this implies.
More subtly he is also restricted by attempting to set the knowledge in another artificial situation, the educational process of sitting in classrooms learning by rote leavened by some understanding and being examined and tested in standardised situations.
Thomas and Harri-Augstein (1985) attempt a more autonomous approach to learning in two senses.
That is, learning need not be restricted to the orthodox educational system and also that learning, as a process, is essentially self-centred, and self-organised.
This approach follows the tradition of Rogers (1951) who developed client-centred therapy and Kelly (1955) who developed personal construct theory.
The emphasis is on the individual who inevitably has a unique view of the world which he continuously modifies in the light of experience.
Each view of the world can be described using personal constructs which can be elicited by the use of a Repertory Grid.
The individual reflects on his own experience and is thus continuously involved in both internal and external conversations.
Learning takes place as a result of these conversations.
Traditional distinctions between teaching, training and therapy disappear, so also does the conceptual distinction between cognition and conation.
One attractive aspect of this theory is that it highlights the way in which formal instruction can inhibit learning by not allowing sufficient scope for self-expression and self-development  alienation following submission to being taught rather than learning how to learn.
The emphasis on self does not limit the application to individual activity, shared meaning can be negotiated and developed within groups, this is group learning.
Correspondingly the teacher is not redundant, a competent teacher is a learning manager and so also is every parent, manager or counsellor, their task is to provide the conditions and facilitate the process of self-development.
If learning is defined as the improvement of specific performance following experience of that performance then organisations as well as individuals can be said to learn.
Paradoxically much of the more abstract work on the mathematical theory of learning curves has its application at least in principle in the requirements of accountants and production engineers.
The process of making anything from a switch to a civil airliner is subject to improvements in speed, quality and cost reduction which follow the characteristic learning curve.
It is obviously important to the managers of such organisations to be able to predict by how much the process will improve after particular lengths of time and numbers of products made.
Conversely, because of the universal very steep learning curves with early experience, any novel enterprise is bound to have extensive teething troubles, which should be allowed for in general even though they are not predictable in particular.
Experiments, trials, prototypes and the running in parallel of old and new systems are all valuable, almost essential procedures  when something radical is being attempted on any scale from a new car suspension to a new educational system.
Technology and society generally is continuously in trouble because this elementary principle is forgotten or ignored by enthusiastic innovators.
Memory and thinking
Thinking is essentially an internal activity although of course it is stimulated and aided by interaction with the external world including other people.
Since it takes place ' inside ' it must be related to memory.
This is one of several complications of the simplistic view that memory is just a store.
Another is that a store is useless without means of access and egress.
In other words there must be processes of recording and retrieval and the underlying mechanisms must be closely integrated with the mechanism of the store.
The store or stores are almost certainly active rather than passive records of events and knowledge so that these also are best considered as processes.
Early studies of memory by psychologists concentrated on apparatus such as the memory drum which was used to measure the ability to associate verbal items.
In the pursuit of generality these items were often nonsense syllables.
This is another example of the way in which the attempt to standardise human performance can negate the purpose of study.
There is a built-in assumption that memory is about association and a removal of any possibility of demonstrating that memory might be dependent on meaning.
Other early studies of memory attempted to measure capacity by data such as the length of a number series which could be accurately reproduced after one presentation.
The answer turns out to be about seven (Woodworth, 1938).
These studies of immediate memory span are of more than historical interest because the fact that there is such a severe limit is important in the design of information presentations of technical data  a standard ergonomics problem.
Other significant phenomena which have emerged from such studies are that simple reversals of the numbers are a common form of error, accurate reproduction is facilitated by deliberate grouping in twos or threes and the ends of a span seem to be less prone to error than the middle.
With the advent of information theory (Attneave, 1959; Edwards, 1964) other interesting issues arose such as whether the performance of the store could better be measured in terms of bits of information or chunks of material (the bits-versus-chunks controversy (Miller, 1956)) and the possibility that memory processes might distinguish between content and order.
The simple memory span measure confounds these variables.
Clearly there is more to memory than the reproduction of numbers or lists and another distinction which has arisen is between episodic and semantic memory.
The former is concerned with specific events and the latter is relatively context free (Baddeley, 1976).
A parallel but not identical distinction is between short-term and long-term memory.
Short-term memory has capacity limits and is readily purged, it probably keeps a record of the current situation, whereas the long-term memory is the repository of principles and strategies.
Both these distinctions imply that registration in storage continues long past the time of reception of information.
Whether or not this implies shifting to different stores is not known.
In functional terms there do seem to be storage facilities relatively near the entry channels through the senses, those for the visual and auditory systems are called respectively the iconic store and the echoic store (Wickens, 1984).
These stores have available a literal record of the stimulus so that the observer can, as it were, internally inspect them but such impressions decay very rapidly, typically in less than ten seconds.
In an elegant series of experiments Conrad (1964) demonstrated that in transposing visually presented data the errors are more likely to be in numbers or letters of similar sound rather than those which are of similar appearance.
This suggests that this short-term store is auditorily rather than visually based even when the task is visual-motor.
The effect diminishes if there is irrelevant articulatory activity at the time, this raises the possibility that there may be pre-output stores as well as post-input stores, almost literally memory in the fingers.
If an experienced operator of a complex console of switches is asked where a particular switch is he will often touch it and check what he has touched, the action precedes the recall and recall precedes recognition.
The distinction between recall and recognition is clearly an important one, in that the human memory is so much superior in the latter function.
We may be totally unable to recall an event or a fact but recognise it and all its context immediately if it is presented.
Again there seems to be a distinction between what and when.
Any busy person will recall that he has a particular engagement but will have only a sparse internal record of when this commitment will arise, hence the importance of a diary.
If asked to recall a date such a person will proceed by inference rather than by a direct introspective search, e.g. ' it can not be next week but it must be within the next month ' or ' this kind of event is always on a Tuesday ' and so on.
This is one illustration of the process which Bartlett (1932) called ' turning round on one's own schemata '  ' I am in this state therefore this must be so ' rather than the direct retrieval from a store.
In Bartlett's view long-term memory involves the reconstruction of events, such remembering is subject to bias and distortion not only from idiosyncrasy but also from the context and even the culture.
It is in this sense that memory is a consequence of the continuously modified record of experience (p. 100) rather than a separate function.
The analogy of the ego or self, the internal little man consulting personal records is attractive but misleading.
One dubious consequence is that the short-term memory, of limited capacity compared with the virtually infinite capacity of long-term memory, is conceived as the internal stage on which exhibits from the store are manipulated.
This is one view of thinking but again it is thinking restricted by the limitations of the laboratory situation where the problem must be easily assimilated and the proposed solution is readily measurable.
Psychological research on thinking has been the province of the field theorists, in particular the gestalt psychologists (Humphrey, 1951).
There was considerable debate within this school about the overlapping concepts of motive, determining tendency and set.
Some such concept is required to explain why different individuals reach different solutions in diagnosing and providing remedies for particular situations.
These ideas have recently been revived by the concept of ' mind-set ' to describe why errors are made in coping with complex system failures.
Craik (1943) considered that thinking paralleled reality in that within the mind there must be a model of reality which can be manipulated independently of the outside world.
The advantages of such a mechanism are clear, it is possible to manipulate the model and thus understand the outside world in the sense of predicting what will happen without the potential costs of attempting to manipulate the real world and allowing things to happen, some of which might be considered to be unfortunate.
Using technology, this process can now be partially externalised in that, for example, a simulation of a situation can be run in parallel with a real situation and can be manipulated independently of what happens in real time.
Bartlett (1958) regarded thinking as a form of skill in that it has the characteristics of organising information.
A situation can be structured and comprehended so as to indicate a direction for proceeding.
In technological situations there is often also a timing characteristic, evidence is developing and at some point in the future, which itself has to be determined, something will have to be done.
One attribute of thinking as opposed to perception is that the evidence is relatively less complete, gaps have to be filled either within or beyond the available data.
This is achieved on the basis of past experience and interests.
Inevitably the product of thinking is a function of the thinker as well as of the situation.
Past experience is brought to bear on a new situation by a categorisation, a process regarded as central to thinking by Bruner and his colleagues (1956).
Classification must be a function not only of the situation but also of the existing internalised classes and their relative accessibility.
Bruner (1957) discusses the classical problem of determining tendency as' perceptual readiness' and notes also the problem that the first classification can mask a more appropriate classification.
This relates to a currently important limitation in fault diagnosis for technological systems where a first hypothesis about the cause of the problem can inhibit flexibility in considering other possible causes.
The above ideas were generated mainly on the basis of experiments and it is not surprising that many of the key characteristics of thinking in real situations were not given sufficient weight.
These include the use of evidence arising from the absence of events rather than their presence, the cost-effectiveness criterion in pursuing evidence and reasoning processes based on extensive experience rather than intellectual analysis.
More recent work on observation of, rather than experiment with thinking (Johnson-Laird and Wason, 1977) goes some way to restoring the balance in appraising all the relevant parameters in a real situation where thinking is taking place.
There is also a dearth of evidence about thinking by groups.
This is surprising in view of the time which people spend in committees, task forces and the like.
There is experimental work on issues such as kinds of group organisation, e.g. autocratic versus democratic versus laissez-faire (Sherif, 1948) and on the group dynamics of people interactions (Lewin, 1935) but little work on the process of problem solving by groups.
The collective mind in action is clearly an important topic but it seems to be a challenge which psychologists have not yet taken up.
It is important in education as well as in gainful employment (Singleton, 1983a).
Language
There has been extensive work on the psychology of language (Campbell and Smith, 1976; Gerver and Sinaiko, 1978), but with a greater emphasis on language as an interpersonal communication vehicle rather than language as a supporter of thinking.
The work of Piaget (1950) and Chomsky (1957) is essentially about thinking and language or psycholinguistics, but these authors did not concern themselves with the level of problem which arises when, for example, a physiologist and a psychologist attempt to exchange mutually supportive ideas.
Herriot (1970, 1974), however, makes some interesting suggestions about the relationship between language, schema, strategies and deep structure.
As he sees it, deep structure is represented in schemata which relate non-linguistic and linguistic thinking.
Language and indeed behaviour generally is necessarily ordered and is represented internally as strategies, schemata are not inherently sequential and thinking can thus be concerned with context before the constraint of achieving is introduced in order to generate an output.
Thinking regarded as internal manipulation of events and ideas does not necessarily depend on language, but if it is not entirely ego orientated, that is not concentrated on the relationship of the person to the scheme of things, then it usually invokes iconic representation.
Designers of computer based support for thinking beyond the level of merely providing additional information will hopefully begin to simultaneously use and develop these ideas.
Effective computer support must be matched to human thinking so that communication is readily established and yet different in providing functions which supplement human thinking (p. 233).
Learning and memory
The learner must have the drive to learn and he must also have knowledge of the results of his activities.
1.
Behaviour at a higher skilled level is characterised by selectivity in choice of inputs and outputs and by increasing influence of the past and the future rather than the immediate situation.
2.
Learning is an activity of an individual, no one else can do it for him.
Educators and trainers can do no more than provide the conditions, the content and the materials.
3.
Social skills and communication generally depend on the detection and definition of mutually agreed internal models of reality.
4.
The human memory, in common with every other store, has to be positively consulted before it will function.
This is not a limitation, it is essential for coherent behaviour, there would be chaos if material emerged without prompting.
The trigger is usually some external stimulus, not necessarily an obvious one.
It might be a particular smell, a particular melody or a brief glimpse of a person resembling one of the actors in an event which is then recalled.
This facility for associations which are neither obvious nor apparently logical is extremely valuable in creative thinking and problem solving.
The converse of positively trying to recall some fact such as the name of a person is subject to curious blockages which are not understood.
5.
Given the alternative of storage internally in the mind or externally in a file (paper-based or computer-based) external records are much superior for literal reproductions at least in western societies (there is anecdotal evidence concerning illiterate traders in other cultures who seem to have remarkable memories for detailed facts and numbers connected with their personal business).
The capacity and flexibility of desk-top computers is now such that the optimal sharing of processing, storage and presentation of material between computer and operator may well be quite different from the traditional combination of human memory and material on paper.
This topic is too recent for there to be many guidelines available but for a particular topic the designer can assume that he need not be inhibited in allowing his imagination to run to many kinds of novel solutions.
6.
There is a considerable art in developing the most effective combination of external and internal stores and reminders which every highly skilled person deliberately cultivates.
For example, there is no need to write in a diary ' check in-tray and consult secretary ', these actions take place as routine following arrival at the office.
On the other hand a diary entry such as' ring X ' is effective and if the entry is some way in the future it may be necessary to amplify with a trigger of the form ' ring X re Y ' and to have easily accessible some factual data such as the X telephone number and some data re Y. In a different context a car driver will develop a strategy to ensure that he does not run out of petrol, he may rely on a light which appears when the tank is nearly empty or he may calculate from his expected travelling that he need not concern himself about the issue until at least the next weekend, or he may programme himself to react to the fact that he is approaching a particular garage.
This self-programming goes on continuously with an implicit acknowledgment of the need for triggers and the relative advantages of various kinds of stores and stored materials.
7.
This self-organised work design has its parallel in designing tasks for others.
In studying an elaborate sequence of events within a task the designer will systematically consider the need for appropriate triggers for the memory at specific points in the task.
In particular he will never rely on the operator spontaneously remembering to do something which has no natural place in the developing task (p. 56).
8.
There are at least three and probably many more levels of storage.
The sensory stores decay very rapidly, items are lost in less than ten seconds.
The short-term store has a limit of about twenty seconds unless there is opportunity for rehearsal, e.g. this is the maximum time available between looking up a telephone number and then dialling it unless the operator repeats it to himself.
Some kinds of long-term memory seem to endure indefinitely even though they have apparently been lost for a long time, e.g. an older person may find it easy to recall childhood events with great clarity even though there has been no use of the material for fifty or sixty years.
9.
There seems to be a memory for elaborate patterns and sequences which takes a long time to develop.
For example a chemist will learn about compounds by laboriously memorising as a student the properties of each one separately but eventually sets of compounds are perceived as a pattern and thereafter all the relevant material is readily available because it fits together neatly.
The corollary for the trainer is to look for patterns which exist in complex material and introduce the trainee to the patterns as well as to the separate facts.
Similarly an information presentation will be more readily comprehended if there is a detectable structure such as the ' christmas tree ' method of working through a checklist.
10.
Errors in recall can be similar for different individuals.
for example the propensity for simple reversals within a number sequence.
It follows that attempting to reduce error rates by one person checking another's performance or automatically comparing the output of the individuals simultaneously performing the same task is not as effective as it would be if errors in the sources were independent (Chapanis et al.,
1949).
Principles, methods and procedures
Interviewing
An interview is a formal conversation in which there is a non-reciprocal relationship between the parties.
It is formal in that there are purposes which define the reasons why the transaction is taking place.
There is conversation which is usually face to face but may sometimes involve telecommunication.
It is non-reciprocal in that the two parties are never equivalent; one is the interviewer who has the initiative, the other is the interviewee.
The interviewing party may be a group facing one interviewee, e.g. a selection board.
The interviewed party may be a group facing one interviewer, e.g. a group of witnesses facing a reporter.
Any transaction involving two groups or two equivalent individuals is regarded as a negotiation rather than an interview.
The interview can be initiated by either side, e.g. by the interviewer when a more senior person calls a subordinate or by the interviewee when a patient calls a doctor.
The information transmission will inevitably be two-way but the main purpose may be transmission from interviewer to interviewee as in an appraisal interview, from interviewee to interviewer as in an opinion survey, or it may change direction during the course of the interview as in a patient-doctor interview.
In this last case the interviewee begins by providing information which facilitates a diagnosis and the interviewer thereafter provides information about possible causes and remedies.
In view of the universality and range of the interview situation it is not surprising that there has been extensive research, but for the same reasons the research is of little operational consequence.
Checks of interview effectiveness by, for example, having the same list of candidates ranked independently by two selection boards invariably indicates rather poor consistency.
On the other hand no manager willingly accepts a new member of staff without the use of a selection interview, no doctor will express a confident diagnosis without an interview with the patient, and no counsellor could function without meeting his clients face to face.
The key to the issue is that interviews except in the simplest case of surveys, to be discussed later, are not about acquiring factual evidence.
This is best obtained by asking interviewees to complete standard forms or to take tests before or between interviews.
Information obtained during an interview is about assessable but unquantifiable attributes such as attitudes, motivation, reliability, integrity, social ease and so on.
In skill terminology the interviewer uses the interview to create a model in his mind which he can use to predict the behaviour of the individual interviewee in a work situation, a stress situation, an illness or whatever an interview is about.
Obviously he can create a better model if he knows what the interviewee looks like, how he dresses, talks and responds non-verbally.
Equally the interviewee forms a model of the interviewer(s) and this can be a considerable aid to his decision-making about whether he can respect, trust, work with or tolerate guidance from the interviewer.
This interaction may or may not achieve a successful conclusion for both sides depending on their level of skill.
Attempting to do research on the interview per se is doomed to failure because the skills of the actors will dominate the success or otherwise of the proceedings.
It should be remembered also that the interview is only one incident within a series of communication functions which aim towards one objective.
For example, a work placement involves trawling for candidates, testing and form-filling and preliminary sifts before the interview followed by induction and training after the interview.
Those responsible are interested in an effective total process not just an effective interview and all the stages in the process interact.
For these reasons it is difficult to generate principles of interviewing, there are always alternative viewpoints and exceptions.
for example:
1.
Always try to put the interviewees at ease.
For senior posts it can be instructive to do the opposite to see how far the candidate can cope with the stress.
2.
Always ask open-ended questions.
If the interviewee is very nervous it can help to ask some questions which can readily be answered with a yes or no, this provides the opportunity to adjust the tension.
3.
Do not make jokes.
Sometimes the interviewee exhibits too much respect creating a communication barrier, a demonstration that the interviewer is only human by a weak joke can remove this barrier.
4.
Always structure the interview to ensure that all essential points are covered.
On the contrary a free-wheeling interview with direction changes as key points emerge can yield unexpected but valuable evidence.
5.
Assess the interviewee under various headings to obtain a comparable profile score.
Part of the skill of the interviewer is in extracting different salient points from individuals and appropriately weighting totally disparate facets.
On the other hand it must be acknowledged that there are many bad interviews and, as always, improvement is possible by design, consideration of procedures and training.
The design of the situation and the process should adhere to standard practices which have been found to be helpful by experience and commonsense.
1.
Obtain all the relevant factual data (biography and testscores) before the final interview.
2.
Provide a waiting room which has accessible toilet facilities including mirrors.
3.
The interviewer should collect the interviewee, not have him sent for from the waiting room.
4.
Provide the interviewee with a written job aid which indicates the name and function of the interviewer(s).
5.
The relative positioning of the two main interactors and the direction of lighting are significant.
The obvious' third-degree ' lighting should be avoided but not being able to see the interview clearly is equally deleterious.
Face to face is the most formal, side by side the least formal, seating at right angles is often a suitable compromise.
6.
Provide the interviewee with opportunities to make points which have not been covered in questions and to ask other questions on matters important to him.
This is not mere courtesy, what the interviewee regards as important can be very revealing.
7.
Provide clear if indirect indications that the interview is ending.
8.
Tell the interviewee what the next stage in the process will be and when.
9.
Escort the interviewee at least to the door, again the behaviour at the parting point can be significant.
In view of the above discussion training of interviewers and interviewees must necessarily be based on principles of training (p. 121) rather than principles of interviewing.
Role playing and case studies are often used and they can be effective in reducing self-unease (self-awareness and self-analysis are necessary attributes of the good interviewer in action) and highlighting idiosyncratic features such as mannerisms and biases.
The most powerful tool of the interview trainer is the closed circuit television system.
So powerful that it needs to be used circumspectly to avoid reducing self-confidence.
Normally the trainer will go through a reproduced interview and discuss the good and bad points which are usually clearly evident.
This provides salutary knowledge of results of the interview itself if not of its degree of success in terms of a real objective.
Randell (1978) suggests that an interviewing at work course can reasonably take two days covering ' unfreezing ', conceptual analysis, technique development and practice with feedback.
Rutter (1979) devised an interview training programme for medical students involving a history taking scheme and two video-taped 15 minute diagnostic sessions one week apart, with discussion feedback based on a nine-point method of scoring information elicited.
These scores were also used to validate the training by comparing experimental and control groups.
Psychological testing
The design and development of tests is a well established formal procedure involving checking for internal consistency and absence of ambiguity, efficiency of administration and marking procedures and the production of norms for definitive populations.
There is a comprehensive list and description of available tests in the ' Mental Measurement Yearbook ' which is updated at intervals of a few years.
The reliability of a test is measured by correlation between the test scores, obtained at two different times, between different forms of the same test or between samples of questions in different parts of the same test.
The interpretation of this measure of consistency is not as straightforward as it seems at first sight.
For example, it has been pointed out that one person can not be asked the same question twice, the second time he is not the same person because he has already been asked the question once.
The validity of a test is an even more complex concept.
It refers in general to the confidence the tester can attach to decisions based on the test scores.
There are at least three kinds of validity: construct validity which refers to the degree to which the test seems to be based on psychological theory, content validity which describes how well the test measures what it is intended to measure and criterion validity which indicates its value as a predictor using as criteria other measures of relevant performance.
The problems of criterion validity nicely illustrate the difficulty of finding any solid foundation on which to construct a psychometric measure.
Suppose that the purpose of a particular test is to predict success in a particular job.
Even with hindsight this is not easy to determine.
The measure of success in a training course depends on the validity of the training course itself, supervisors' opinions based on experience of job performance are subject to distortion by various biases and measured performance such as productivity is subject to many vicissitudes independent of the ability of the producer.
Faced with all these uncertainties the strategies adopted in the U.S.A. and Europe are in general different but usefully complementary.
The American approach is to be as formal and rigorous as possible with insistence on systematic sampling, correlating and so on.
The European approach is less tidy and more humanistic with continuous questioning of philosophies and methodologies.
A good example of the latter is Heim (1970) in which intelligence and personality testing is discussed in the context of the meaning of intelligence, creativity, personality, test taking as a skill and the different but inseparable limitations of speed and accuracy based measures.
For further discussion of these elusive topics see Super and Bohn (1971), Drenth (1978) and Jessup and Jessup (1975).
The aim of psychological measurement is optimal discrimination between individuals, for this purpose the difficulty of the test should match the range of abilities of the testees.
For example an intelligence test may be too easy so that all testees have almost perfect scores, or too difficult in that all testees have very low scores, in both cases discrimination will be poor.
One answer to this problem is tailored testing in which questions are computer-based and the test is adaptive in focussing the range of presented questions at about the level of success/failure which is appropriate for each testee.
In this way discrimination is much improved without excessively lengthening the test time (Killcross and Cassie, 1975).
The value and extent of the use of tests is subject to other changes in the history of the particular community.
Obviously testing can be extensively utilised in war-time when many people are changing jobs both in military and civilian life.
In peace-time the selection ratio, the ratio of applicants to available jobs is a key feature, there tends to be more testing in times of economic depression when jobs are scarce and also in post-war periods when there is extensive experience of using tests.
The use of tests is not, of course, confined to selection procedures.
Indeed there is currently much greater utilisation in guidance and therapy.
Surveys
In common with tests, surveys are required to be as reliable and valid as is feasible in the circumstances.
Surveys are however different from tests in their objective: a test is usually carried out to find out about an individual, a survey is carried out to find out about an issue which is across people.
The issue itself must be specified as precisely as possible usually in terms of both what we want to know and why we want to know.
' Why ' will clarify ' what ' and will also contribute to the practical question of how much to spend on finding out.
For reasons of economy the population of individuals approached is usually a small sample of the total relevant population.
The size of the sample is less important than its structure.
The first rule of sampling is that each member of the relevant population must have some chance of being included.
If the probability of being included is equal for each member of the relevant population then the sample is said to be a random one.
Often the relevant population is stratified to ensure that the sample will be more representative in certain respects.
for example a survey of hospital staff would not usually be carried out by selecting names at random from the total staff list, it might concentrate on doctors, nurses or cleaners only, or might sample medical, paramedical, administrative and support staff in proportions of their total numbers depending on the purpose of the survey.
Sampling might also be carried out progressively involving different stages.
For example a survey concerned with the whole National Health Service staff might begin by separating the hospital and the general practice services, then sample key regions of the country, separate urban, suburban and rural services and then select within organisations as mentioned above.
Sampling might also involve time phasing as in attempting to detect opinion trends leading up to a general election.
It is possible to estimate sample sizes required if the tolerable error is known and vice versa (Yates 1960).
Data may be obtained purely by observation but usually involve asking questions.
The design of questionnaires is itself an art with certain ground rules.
Clearly the questions must be such that the relevant population can be expected to understand them and to make informed responses.
The questions must avoid ambiguities which is not as easy as it might seem, it is desirable to obtain several opinions about the interpretation of each question from the same kind of people as those who will be in the survey.
The sequence of questions may bias the answers e.g. a likely alternation of Yes/No answers is more ' natural ' than a uniform series of Yes or No answers.
The style of the question can also be unintentionally ' loaded ' towards a particular response.
There is a temptation to put in questions which will check against each other but this can irritate the responder who will easily detect redundancy.
There is also the temptation to make the questionnaire too long in the hope of getting just that little bit more information, this can more than defeat its object by reducing the response rate.
A new questionnaire should always be piloted, that is tried out on a small sample, before it is finalised.
The main factor influencing response rate is whether the survey is postal or personal.
The response rate to a verbal questionnaire can easily be 90% or more but a mailed questionnaire can have a response rate as low as 30% even with follow-up reminders.
A low response rate is serious particularly if there is reason to suspect that persons likely to give particular kinds of answers might be less or more likely to respond than others.
Telephone based surveys are currently popular as a reasonable compromise, this increases the speed of data collection rather than the insight which is revealed by personal interviews.
Large scale impersonal surveys can be supplemented by more personally based interviews either with individuals or groups.
The inference from samples may be deductive or inductive.
Deductions can be made by statistical comparisons of samples to determine whether they are from different populations.
Inferences from samples to populations can be made within quantitative confidence limits.
Much of the labour which used to be involved in manipulating and presenting data can now be delegated to specialised computer software but the basic issues of coding and structuring data still depend on the skills of the analyst.
The survey is the most extensively used method of obtaining generalisations about people.
It can vary in precision from quantitative anthropometric surveys to attempts to describe the attributes of products which particular users prefer.
The systematic survey is the antidote to excessive reliance on personal experience which is usually biased by exposure to idiosyncratic samples and by selective perception.
The required systematic process is summarised in Table 2.3.
Task training
The distinction between task training and skill training is one of degree rather than kind.
In task training the objective is to guide the trainee into following rules and essentially to make the performance more automatic.
In skill training the opposite can apply, the trainee is persuaded to be flexible and to orient himself towards the end rather than the means.
The difference between the two kinds of training is not as great as would appear at first sight because it turns out that task training has in common with skill training an emphasis on the perceptual side of human functioning.
Even in what appear to be straightforward motor skills the changes in learning are associated with greater selectivity and more economical use of the evidence needed to guide performance.
These distinctions are summarised in Table 2.4.
Task training is often called operator training or industrial training.
The essential principles are:
1.
The trainee must wish to learn, nothing can be achieved unless the trainee is adequately motivated.
2.
The learning must be steered by adequate knowledge of performance.
The behavioural characteristic which facilitates both of these requirements is knowledge of results.
This is the most potent and consistent performance shaping factor ever uncovered by psychologists.
It simultaneously promotes motivation and performance.
Knowledge of how well one is doing will not provide the basic motivation in that the performer must care about his performance but if this is established, knowledge of performance is a most vigorous factor in developing and maintaining the will to perform.
Carmichael and Dearborn (1950) demonstrated that continuous reading was likely to result in increased failure of concentration within an hour, but if the performance of the reader was checked and confirmed by asking questions at half-hour intervals then performance could continue for six to eight hours without signs of fatigue.
Mackworth (1950), within his classical studies of vigilance, showed that without knowledge of results the proportion of detected signals decreased in succeeding half-hours compared with the first half-hour, but for subjects given knowledge of results there was no deterioration compared with the first half-hour through the full two hours of the study.
Correspondingly the effectiveness of performance will improve with knowledge of performance.
This is as much a logical as a psychological principle.
Clearly if a person has no knowledge of the success of performance the performance can not improve because there can be no guidelines to steer the improvement.
There have been many experiments demonstrating the validity of this principle and exploring detailed variables such as the timing and quality of feedback.
For details see Welford (1968), Bilodeau (1969), Annett (1969).
Miller et al.
(1960) formalise it into what they call the TOTE principle (Test Operate  Test  Exit).
The first test is a review of the situation, operate is the action which follows, the second test is a check as to whether or not the desired result has been achieved and this is followed either by another operation or by an exit if the desired result has been achieved.
The TOTE is regarded as a fundamental unit of behaviour.
Broadly speaking the more immediate the feedback the more effective it is except that there comes a point at which, to use Miller's (1953) distinction, learning feedback becomes action feedback, the latter improves the current response whereas the former improves succeeding responses.
There is a risk that what is intended to be an aid only during training can become an undesirable crutch in that performance comes to depend on it but this is rare and can be avoided by using a suitable time delay, thus ensuring that it is really learning and not action feedback.
The accuracy of the feedback need not be greater than the discrimination ability of the trainee but equally its value will diminish if it is too approximate.
Feedback of the discrepancy between desirable and achieved performance is superior to feedback of absolute performance.
Holding (1965) distinguished between successive sub-classifications of feedback into intrinsic/artificial, concurrent/terminal, immediate/delayed, non-verbal/verbal, separate/accumulated.
The superior value of direct, rapid, unambiguous feedback dominates the choice of length and content of learning periods.
There was an extended controversy between supporters of part and whole learning (Blum and Naylor 1968).
The advantage of whole learning is that the connections between the parts are also learned simultaneously.
However the advantage of part learning is that the feedback can be of much higher quality.
An excellent compromise is the progressive-part method (Seymour, 1966) in which parts are learned separately but combined in groups within a tree-like structure which converges on the total task.
See, for example, Singleton (1959), King (1964).
The identification of the sub-tasks which are appropriate for intensive training with feedback is achieved by task analysis.
Unfortunately there is confusion in the literature and the required procedures may be referred to as either skills analysis or task analysis (Seymour 1968, Singer and Ramsden 1969, Miller 1962).
In the terminology used in this book task analysis is the correct description of this activity.
It will be evident from the dates of most of these references that this kind of training and the supporting research flourished in the 1960s.
The need for it has reduced but has not vanished over the past twenty years.
Technology has shifted the general allocation of function slightly both in industry and in defence.
Many processes have changed so that there is less direct human involvement in the production cycle, e.g. metal working, highly repetitive cycles can be conducted by robots and similar devices, e.g. in car assemblies, and the human operator gets better support even when still involved directly, e.g. the word processor as a substitute for the typewriter.
Nevertheless within most jobs there are at least some tasks which are amenable to this kind of training and the benefits are considerable.
If task training is neglected the larger job may be interfered with, e.g. the manager attempting to use a personal computer with inadequate key-board and data search performance.
Key-board manipulation is a skill relevant to many tasks and is typical of psycho-motor skills in that the perceptual elements of identify and select dominate the motor element of actual key pressing.
As mentioned earlier improvements in speed and accuracy result from changes in the control of performance.
Skill training
The simplest criterion for the distinction between a skill and a task is that a skill description is susceptible to individual differences but a task description is not.
Skill implies idiosyncratic performance and as such can transfer between many tasks whereas a task meets a particular requirement within a system activity.
Thus it follows that skill training is more generalised and generalisable and is better fitted to the overall role of human operators.
Education is the ultimate form of skill training where no attempt is made to specify the particular purposes for which the skills might be utilised.
This limits the potential rigour of design because the anchor of skill training is the specification of the objectives.
Incidentally this accounts for the interminable debate about education, since objectives are not specifiable the content can only emerge as a consensus within the current zeitgeist.
Skill descriptions are generalisable in that human operators have characteristic abilities and limitations and therefore have tendencies to perform in the same way.
Differences in level of skill are partly but not wholly attributable to different locations on a learning curve which has some uniformity for that particular skill.
Skills can not be transferred directly from a trainer to a trainee, the function of the trainer is to provide conditions and guidelines which will facilitate learning.
To do this the trainer must have a coherent model of the form of the skill, this is attained by skills analysis.
Inevitably skills analysis is more general and less definitive than task analysis although many of the same procedures for extracting evidence might be used.
Skills analysis may also involve experiments to study more closely how the required objective is achieved (Drury and Fox, 1975, Singleton, 1978).
It requires a broader approach than task analysis in that it often involves tracing the acquisition of the skill starting from the kind of persons who undertake training (the selection criteria) through the training procedures to the end product (the training success criteria).
Such training often has an emphasis on features such as style and quality rather than mere speed of activity.
This can extend beyond skill training to attitude training (see below).
These differences are summarised in Table 2.5.
There is emphasis on the end purpose or objective with relatively less attention to the means, partly so as to allow for individual differences and partly because the evidence accumulating in particular cases may change the direction of progress.
For this reason skill training is sometimes called strategic training.
Such activities are found in the broad range of tasks concerned with diagnosis.
These can vary from dealing with faults in equipment (Tilley, 1967) to dealing with patients in a surgery or clinic (Rutter, 1979).
In all cases although the reason for the activity is clear cut the means of achieving the desired end vary widely in terms of the path taken and the time required.
The medical interview has much in common with counselling (Joanning et al.,
1979) and occupational guidance (Davis and Shackleton, 1975).
Medical diagnosis involves social skills (Melhuish, 1979) but as discussed earlier the basis and development of these skills is not dissimilar from other cognitive skills and correspondingly the training can follow the same principles (Singleton et al.,
1980; Argyle, 1981).
Attitude training
Attitudes are much more difficult to change than are skills and there can be ambiguity about what constitutes an improvement.
Attitudes involve values which vary between individuals, societies and cultures.
For example ' brain washing ' is a form of attitude training which attempts to change the beliefs and values acquired within a particular culture (Biderman, 1967).
The extreme techniques of isolation, starvation and other forms of stress induction give some indication of the pressures required to positively change attitudes.
Generation of standard useful attitudes has always been one of the main purposes of military training.
The objective is to cultivate the willingness to accept or ignore personal risk (Page, 1987).
In the civilian context the purposes of attitude training are to do with care about quality and safety.
Apprenticeship training is essentially about ensuring standards of workmanship and characteristically it appears at first sight to take an unconscionably long time.
A common fallacy is the attempt to assess apprenticeship schemes in terms of task and skill training.
Safety training is also about attitudes and such schemes rarely have the hoped for success because the difficulty and time required are invariably underestimated.
This is also the reason why road accidents are so difficult to approach by training, safety on the road is much more to do with attitudes than with driving skill (Parry, 1968).
In summary, we have no useful theory about the design of attitude training and success in practice is based on indirect traditional procedures.
Skills analysis
It will be appreciated from the earlier sections of this chapter that skills analysis is not reducible to simple procedures or recipes.
In principle it is impossible because skills are by their nature totally integrated and interactive so that they can not meaningfully be separated into independent parts.
Nevertheless, as in many other situations the analyst, himself a skilled performer, has some success in practice.
A skill description relates to how a task is performed and what is achieved rather than what actions are taken.
The reference is to the end and the means is described in person terms rather than task terms.
This is because the end is more consistent than the means.
Although there may be considerable idiosyncrasy in the actions the situation is not anarchical.
The human operator has characteristic ways of doing things and detection of these is one aspect of the skill of the skills analyst.
There are differences in skills analysis procedure depending on the kind of skill.
For the skills of interacting with the physical world the procedure summarised in Table 2.6 is appropriate.
This involves observation of the skill in practice and during training and discussion with practitioners and trainers.
There are useful ways of focusing such observation by contrasting good/poor performers and experienced/inexperienced performers.
Discussion with practitioners can be focused by asking them to talk while performing and by jointly discussing performances observed on film or tape.
For management skills there is greater emphasis on the objectives and how they are arrived at.
Objectives in an organisational context are subject to many constraints.
Some such as financial resources are easily specifiable, others such as not contravening the mores of the organisation are more elusive but a key aspect of these skills is sensitivity to limitations and obstacles as well as positive assets (Singleton, 1981).
The variety of management tasks is often so great as to warrant a task analysis with separate skills analyses of particular tasks or groups of tasks.
At this level the practitioner is often the conscious monitor of the practice and facility in self-monitoring is a useful indicator of skill.
Social skills can involve not only self-monitoring but also monitoring of what another person is thinking as well as doing.
For example, analysing a transaction between two skilled negotiators requires the study of the overt interaction and the identification of what each one is thinking about what he himself is doing and what his opponent is probably thinking about in the context of what he is saying (Singleton, 1983a).
There are no standard formats for skill descriptions but usually they can be separated into two parts: the performance and the antecedents to the performance.
That is respectively the key aspects of behaviour which contribute to success and the kind of education, training and experience normally required.
At high levels of idiosyncratic skill (flair see p. 245) analysis is particularly difficult because the practitioner may have little conscious awareness of how or why he achieves unusual success.
In fact it almost seems to be a condition of such performance that the exponent must not proceed logically but must temporarily inhibit his conscious critical faculties and accept control by intuition.
This has been graphically described as' switching on the autopilot ' (Drasdo, 1979).
The commonsense view that to proceed with care is to proceed slowly can also be reversed in high level skills, there are instances where speed seems to be an essential accompaniment of delicate discrimination (Lacy, 1978).
Thus, the observer can not simplify his task by asking the practitioner to go more slowly nor can he rely on the introspection of the skilled performer.
Skill is demonstrated by persistent and efficient pursuit of an objective and the skill can usually be understood although not necessarily written down with any great precision in terms of a goal and the path towards that goal.
The path is generated by navigating through a conceptual map containing landmarks and decision points (key features of the skill) and by the avoidance of obstacles and misleading features which have been sensitively detected.
Applications
Selection and guidance
These processes are variations on the general theme of matching a worker to an appropriate job.
In the case of selection the given point is the job and the population is searched to find the most appropriate worker, in the case of guidance one worker is the starting point and the range of jobs is searched to identify the best one.
The two complementary processes are shown in Fig. 2. 1.
They can usefully be considered together because they are similar and interactive.
Thus, in the selection situation the process involves not only analysing and describing the job but also considering the availability of potentially appropriate workers.
In the guidance situation the process involves analysing and describing the worker and consideration of the availability of suitable jobs.
At the point of a successful match the worker is described in terms of the kinds of job he could do and the job is described in terms of the kind of worker needed to do it, Table 2.7.
For most people and jobs the whole process takes place within a localised community.
For the highest level jobs a whole country or even an international community might be considered, but in European countries most workers attempt to find jobs which do not require either moving house or excessive travelling, and such proximity has corresponding advantages for the potential employers.
Thus, the selection process involves describing the job first by analysis (task and skills analysis) and then the translation of these data into worker relevant terminology.
Consideration of the potential worker availability may lead to a redesign of the job either by the use of a different task combination or by introducing different technology.
It is then necessary to communicate with the market by publicising the available job usually by advertising.
This is itself an important stage in the process because a well prepared advertisement will enable the recipients to engage in self selection which does not, at this stage, involve further selector resources.
To this end it is useful to have a multi-stage process as shown in Fig. 2.2.
Having developed some interest on the basis of a necessarily brief advertisement a potential candidate can apply for further written details which should contain a comprehensive description of the job, the task and skill content, the responsibilities, the salary and the details of the organisation including its history and anticipated future, and an application form.
A name with job title and telephone number should be included so that informal interaction and discussion of details of particular interest to the candidate can take place before the commitment to transmission of the application form.
All this can be done at minimal cost.
If the job description has been properly constructed there should not be an excessive number of applicants so that the next stage of a paper sift is also economical.
There then follow the more expensive stages of final selection which may involve testing and interviewing as described earlier.
The choice of the interviewers and the interview process depends on the particular job.
At one extreme it may be one company representative, at the other extreme there may be a series of interview boards.
The interview board can have covert purposes where the senior person who is the chairman of the board will use the opportunity to orientate the precise expectations from the post to be filled including settling differences of opinion between his subordinates who are also members of the board.
Ideally this should be done before candidates are interviewed but it can happen that differences in preferences for candidates reveal different opinions between board members which the chairman can detect and deal with.
Guidance also involves testing and interviewing but this is invariably a two person situation of a counsellor and client.
The problem is that most clients have only the vaguest, undefined ideas about the most suitable choice and direction of career.
The counsellor can help by identifying the client's job-relevant characteristics and by contributing his knowledge about occupations (Fig. 2.3.).
The former may be supported by testing although test results are essentially evidence to enable the individual to clarify his own self concept.
The latter may start from national classifications of occupations such as the American Dictionary of Occupational Titles (DOT) and the British Classification of Occupations and Dictionary of Occupational Titles (CODOT), but the counsellor's suggestions are likely to converge very rapidly into a more comprehensible repertoire of possibilities using evidence about the particular client (Singleton, 1975).
This process can be supported by a computer with which the client may interact alone or it may be used by the counsellor as a database and prompter.
A guidance episode necessarily takes place at one point in time within the developing skills and experience of the client.
To emphasise the continuous nature of career development and the changing aspirations of people at different ages Super has proposed a division into five stages: Growth (0C14 years); Exploration (15C24 years); Establishment (25C44); Maintenance (45C64 years); and Decline (65 onwards) (Super and Bohn, 1971).
Placement
Placement is a mixture of selection and guidance, a process of mutual matching takes place within a known range of persons and jobs.
The problem arises when a batch of recruits enter military service or when a large company takes on a new set of graduates.
It is one application of psychometrics to man-power planning (see previous chapter).
It becomes of particular significance within the currently fashionable company policy of separating core and peripheral employees.
Given the vagaries of the market and rapid changes in processes and products stimulated by changing technology the strategy has evolved of taking on many employees for a short term only.
These may be production workers but they may also be senior people; managers hired for a fixed term and technologists and others used as consultants.
The placement task is to describe the requirements and the appropriate sources of individuals with the relevant skills (Fig. 2.1).
For the core workers who are permanent staff again there may well be a mixture of people at all levels and here there is more emphasis on career progression and job satisfaction.
Core employees are given written annual assessments which can be discussed with the immediate senior who is probably the writer of the assessment.
For this purpose he is provided with a well-structured form by the personnel department who will also advise on the timely use of in-house or external courses to extend the range of skills of a particular employee.
By contrast the peripheral employee is judged entirely on his past record or that of the consultant company which employs him on a semi-permanent basis.
He is hired to undertake a specific task without reference to longer term features such as job satisfaction or career development.
The peripheral employee will be compensated for his disadvantages by a much greater pay rate per hour and the company gains by reduced overheads and greater work-force flexibility.
The whole strategy works very well for younger workers with well documented training and experience.
This reinforces the view, developed in the previous chapter, that manufacturing industry in particular is following the Armed Forces in relying more and more on fit, skilled, mobile workers.
Psychometrically the problem is to develop appropriate descriptive taxonomies so that a person's actual or potential contribution to an organisation is succinctly expressible.
These remain primitive in that, on the whole, non-technical English is used with all the possibilities of ambiguity and communication at cross purposes which this implies.
The accountants are ahead of the psychologists in developing ideas such as profit centres applicable to small groups or even to individuals.
The increased awareness of the importance of the individual worker to profit and success generally, particularly at higher levels in the organisation, has stimulated the growth of a specialist placement profession colloquially known as the head hunters.
These are companies which specialise in assisting a client in clarifying what kind of person they need and how to locate him within the potentially suitable worker population.
They may also start with an individual and search for a location.
This also is done mainly by interview although some use is made of psychometric tests, profiles and rating scales.
The design of training schemes
Designing and conducting training schemes is expensive and is not undertaken without good reasons.
Among possible reasons are:
1.
An inadequate reservoir of competent workers for particular tasks
2.
A workforce sufficient in numbers or supply but not in skill level as evidenced usually by quality of work and less frequently by failures to meet production targets, danger to people and damage to equipment
3.
Changes in technology which result in changes in required skills
4.
An organisational commitment to continuous improvement of the skills of members
5.
To meet statutory requirements in certain dangerous industries and services
6.
As a service to customers buying equipment
7.
As a service to the community paid for by public funds
The remit of the training scheme and those responsible for it may end at the transition point to operational activity or may continue in those periods of work regarded as a continuation of training and supervised as such.
In either case there needs to be close liaison between those responsible for training and those responsible for operations.
On the other hand training is always vulnerable because the benefits are long-term and it often needs protection from the demands of management and operations where there are other priorities.
The simplest protective barrier is a geographical one, namely to conduct the training away from other centres of activity within the organisation.
Sometimes a separate room is sufficient, in other situations a different part of the country may be chosen to provide adequate isolation.
The essential vulnerability reinforces the requirement that the starting point of design should be a clarification of objectives followed by a consideration of the criteria of assessment that the objectives have been achieved.
None of this need be numerical, sometimes numbers are used to provide a pseudo precision which can be counterproductive because the validity of the numbers can so easily be questioned.
Precise verbal statements and descriptions avoid this pitfall.
The sensitivity of training regimes to external pressures is demonstrated by the erratic interest and activity in this area observable at the national level.
Inevitably there was a burst of activity during and immediately after World War II as shown for example by military research and development (Wolfie, 1951) and by the ' Training within Industry ' (TWI) movement (War Man-power Commission, 1945).
Although research continued the next burst of application in the U.K. took place following the Industrial Training Act of 1964 (Robinson and Barnes, 1968, Barber 1968).
The Act concentrated on training oriented by industry, the other possibilities considered were occupations and geographical regions.
It resulted in the creation of a Central Training Council and a comprehensive range of Industrial Training Boards.
Each Board was financed by a levy on the relevant industry.
Not surprisingly this created a demand for experienced trainer manpower which could not be met, there was some disillusion and retrenchment in the 1970s.
In the 1980s there has been another resurgence of activity following the realisation that the country is simultaneously suffering from a high rate of unemployment and a shortage of workers with specific skills.
Extensive government resources are channelled by the Department of Employment through the Training Commission into the Youth Training Schemes, job training, skill training and so on.
The objectives and criteria of success of any training scheme are best described in system terms.
The next step is to translate these into a person oriented description of what the successful scheme graduate will have acquired.
This is more than skills because the total expertise will include also rules, knowledge and contextual information.
This last will vary from some acquaintance with the parent company to the basic technology behind the tasks.
Together these make up the content or syllabus of the course.
Decisions are then needed on how to train, which will include whether or which parts should be on-line or off-line, whether the operator is required to function in a mainly programmed (i.e. rule-following) mode and how far he may need to function in a conceptual (i.e. actions based on understanding) mode.
Consideration will be given to the use of training aids such as technical notes, schematic diagrams, video tapes, closed circuit television and simulators.
The most important training aids are the instructors, the selection and training of these is a vital secondary scheme supporting the primary scheme.
These many facets of training scheme design are shown in Table 2.8.
Programmed instruction
The origins of this movement are in the teaching machines first developed systematically about 1960.
These consisted of film strips with associated projection devices and control panels which enabled the student to examine a particular screen presentation and respond by moving to another frame.
The succeeding frame might take up another topic or might contain answers and comments on questions posed on the previous one.
There was some scope for interaction in that the series of presentations was determined by the student's choice of responses.
If the student demonstrated by his choices that he did not fully understand a particular point then the programme could send him round an additional explanatory loop.
This was called a branching programme as distinct from a linear programme which contained no loops.
It will be appreciated that this technique of automated teaching seemed to have enormous potential.
Classroom teaching is inherently skilled man-power intensive.
There are obvious benefits in allowing each student to go at his own pace.
It is not surprising that teaching machines had considerable appeal to the military and to large scale industry.
Some years of intensive research and development resulted in a more realistic view (Glaser, 1965; Wallis et al.,
1966, Kay et al.,
1968).
It became clear that the quality of the programme was the key to degree of success.
The machine itself and its style of operation was incidental, in fact it was possible to achieve the same procedure if not quite the same discipline on the student by using programmed books with directional references to other pages at the end of each page.
If a well-structured book is the answer then we have gone full-circle and back to the benefits of a systematic text-book.
In research terms the study of programmed instruction was instrumental in encouraging a more systems view of the teaching/training process with associated trends towards greater consideration of objectives, task analysis, cost/ benefits and so on.
Correspondingly from a practical training viewpoint there developed a more systematic approach to taxonomies of training and of instructional methods together with their inter-relationships.
Tables 2.9 and 2.10 are examples from an overview of military training.
Regarded in systems terms, programmed instruction can not be an isolated continuously developing activity because it depends on the skills of the programme writer and these skills can only be developed on the basis of experience in the classroom.
On the other hand teaching machines are one way of utilising the skilled teacher on a larger scale than his personal activity in the classroom.
There is a different form of automatic teaching device called a pacing machine which can facilitate the development of motor skills.
It consists of a machine which emits bleeps at timed intervals.
The intervals are predetermined by the elements of a motor task or more simply by the target cycle time of a task.
The trainee performs his task and tries to keep in phase with the bleeps which are intended to pace his performance.
The required rate of work can be increased by shortening the time intervals.
The second main phase of programmed instruction became feasible with the development of low cost computing.
Computer aided instruction (C.A.I.) strictly should cover other techniques such as simulation and computer based procedures but it is usually intended to mean a situation where a student sits in front of a screen on which the presented formats relate to his developing cognitive activity as expressed by his manipulations of the associated key-board.
Even with this restricted interface there is enormous flexibility and the training tasks can vary from reading and simple arithmetic to diagnosing faults in the complex plant used in process industries.
The use of computer aided instruction in primary schools has created a new generation of what have been called ' television children '.
They regard the screen/keyboard interface as no more unnatural than a book is to earlier generations.
This has far reaching consequences at the cultural level not only for learning but also for the design of jobs and even for leisure pursuits which increasingly depend on the same kind of interface.
Many of the earlier forms of computer aided instructions used a central computer facility and a set of work-stations where a whole class of trainees could function simultaneously.
The development of micro-computers resulted in the use of more isolated dedicated systems for just one trainee.
In either case it is feasible to collect data about a trainee's performance and modify the presented information accordingly.
It is also possible to collect learning data and feed it to an instructor either on-line or off-line.
The widespread use of computer based learning and testing situations has stimulated the development of ' student modelling '.
All the data about one student is collected into one bank which changes dynamically with his learning achievements (and failures).
In this way a student model containing a profile of attainment together with a history of progress to date can be made available to the trainer and to the trainee.
However, communication links between the trainee and the trainer via a computer are not a substitute for the two-way conversation and observation which normally takes place in a class-room.
Computer programmes are notoriously literal in their requirement for and response to operator responses.
Trainee-computer communication failures for very simple reasons can lead to a complete hiatus unless they are detected and restored by an experienced trainer who can explain to the trainee why the system is failing to respond adequately.
The third phase of programmed instruction called Tutorial Expert Systems (TES) involves artificial intelligence (see p. 234).
To a trainee at the interface this may not seem very different from a teaching machine or computer aided instruction but the logic or architecture of the driving system is different so that even more flexibility is available.
Instead of responding directly by logical rules and paths which have been envisaged and stored by the programmer an expert system training programme contains a knowledge base and an inference engine which can consult the knowledge base and respond more adaptively to the idiosyncratic needs of the trainee.
It can also explain why it is proposing a certain strategy and why the detected strategy of the trainee is appropriate or not, and, if not, why not.
From the user's point of view these systems are usually menu driven, that is the operator can select his path of consultation by choices within ranges of possible choices which are presented to him.
Technically the task of the programme designer is made easier by the use of training skills and declarative programming languages.
A training shell is a generalised tutorial system which can operate with a variety of knowledge bases.
A declarative programming language is one which can accept statements of relationships or rules for relationships contrasting with more traditional languages which encompass only step by step logic.
It will be appreciated that this rapidly developing field of expertise contains extensive new jargon.
At this stage the jargon is an obstacle in that anyone wishing to find out about the field has first to interpret the jargon.
This is doubly difficult in that there is much overlap in terminology and concepts and specialists are not consistent with each other in the use of these new terms.
No doubt the field will eventually clarify and mature and the terminology will be designed with precision so that meaningful concepts, tools and methods are unambiguously clarified and distinguished but this has not yet happened.
From a behavioural standpoint the issues are the same as in any other field where technology is simultaneously providing many new facilities for supporting and measuring human performance.
1.
Technology provides trainer aids not trainer substitutes.
Thus computer assisted instruction (CAI) is more aptly descriptive than computer based training (CBT).
2.
Technology enforces rigorous thinking to the point of revealing embarrassing gaps in learning, teaching and training theory.
On the other hand technology also opens up new possibilities for monitoring and measurement which should support the development of better basic theory.
3.
It is essential to develop the basic taxonomies behind the learning situation.
for example any diagnostic task can be aided by a taxonomy of symptoms and a taxonomy of causes together with connections between them.
4.
The need for interdisciplinary approaches is reinforced but the abiding difficulties of interdisciplinary communications remain at every level from different values and concepts to different terminologies.
Communication between information technologists and behavioural specialists is a continuous struggle which can eventually be rewarding.
5.
The design of the man-machine interface (MMI) or the human-computer interface (HCI) or more simply human interfaces (HI) remains one of the most difficult design problems but a crucial one in that poorly designed interfaces will wreck the performance of any system.
6.
Evaluation is an essential procedure within the implementation of any new system.
for training systems this can be very expensive because of the requirements to monitor the learning process and compare procedures which are anyway never completely comparable.
Apparently extraneous variables such as trainee motivation and trainer expertise continue to dominate other factors such as differences in technology and differences in learning tasks.
7.
Pedagogically the need to be much clearer about what is being taught and why is a mixed blessing.
The formal discipline can sometimes tidy up muddled thinking but at other times it can interfere with the creatively developing interaction between trainer and trainee.
In general, programmed instruction is most useful for tasks which can be comprehensively specified.
This implies tasks based on rules for example arithmetic, or on logic, for example fault finding.
The use of simulators
When serious consequences can follow from human errors, e.g. in flying an aircraft, or when on-line operation is too expensive to be trusted to the hands of trainees, e.g. in process control, there is obviously a case for providing training devices which simulate the performance of on-line systems.
In this way risks and costs are reduced, the performance of trainees is more easily measured and the presented situation is precisely controlled by the trainers.
It is feasible to provide adequate exposure to events which in practice might be rare but very serious.
High fidelity simulation is an interesting challenge to technological design.
Simulators can provide demonstrations which always impress visitors.
With all these advantages it is not surprising that a great deal of effort and resources go into the provision of simulators for high technology industry.
Paradoxically the more advanced the technology the easier it is to provide good simulation.
For example, a process control plant which consists of dials, charts and visual display units responding to activation of switches is much easier to simulate than, say, an opponent in a game of tennis.
The perfect simulation which has been suggested but not yet implemented in process control, power generation and air traffic control would be to have two identical control rooms and two crews coming on duty.
One crew would control a real system and the other would think they were so doing but would in fact be in a control room driven by software.
This seems to be the only way of avoiding the main snag of simulation, namely that the trainee is aware that his performance is not important in that reality out there is not changing as a consequence of his actions.
However, he will always know that his performance on a simulator is being assessed and, except for the business of risks, he may well take more care than he would in a real plant.
In many if not most industries where simulators are used the benefits to training are not confined to dynamic system control and fault diagnosis.
The control function is exercised by a team and the efficient interaction between the team members is also subject to improvement with practice (see p. 95).
Transfer of training from a simulator to a real situation is never complete and does not necessarily increase with degree of fidelity.
Simple simulations can be as effective for training purposes as more elaborate ones and can certainly be more cost effective.
Degree of transfer is not easy to measure and needs to be carefully defined.
For example, in training manually guided missile operators the degree of transfer is appropriately measured by ' first shot ' success whereas in training and retraining flight deck crew the long term contribution to safety is the appropriate measure.
Gagne (1962) provides an excellent review of earlier work in this field.
More recent experience is reviewed in Stammers (1983).
For details of experimental studies using large scale defence simulations see Parsons (1972).
Moraal and Kraiss (1981) contains chapters dealing with simulations of a ship entering harbour, car driving, tank driving, the flight deck and air traffic control.
MSC (1986) reviews the current use of simulators in twenty U.K. industries.
As a research tool the simulator provides a happy medium between the precision of laboratory studies and the realism of field studies.
It is not widely used for research outside the military because of the cost of high fidelity simulator time.
A complex dedicated simulator can cost several million pounds and it needs its own crew of skilled operators.
On the other hand standard micro-computers have considerable potentiality for quite elaborate simulation and effective simulation training on procedures may only require simple static mock-ups.
Simulator development has been driven by technology rather than by the behavioural sciences with the standard result that there is a high reliance on face validity with relatively little resource devoted to systematic evaluation.
Key features such as the timing and extent of simulator usage are based on experience in particular industries rather than on more formal evidence.
The ergonomics approach to their use is shown in Fig. 2.4 but it must be admitted that this is as yet rarely followed in practice.
Simulation should be regarded as one tool within the repertoire of the training specialist.
There are other uses of simulators.
They can be used for assessment independently of training.
Data collected during simulator trials is a useful source of risk and reliability data.
In the nuclear power plant design they are also used as the final stage of pre-construction design evaluation (Williams and Story, 1987).
They can also be used as a resource for checking operating instructions and for conducting detailed task analyses.
For these purposes the simulator must be available before the main control room is constructed.
The full range of potential simulator usage is summarised in Table 2.11.
