As_II an_AT1 example_NN1 ,_, a_AT1 fictitious_JJ doctoral_JJ student_NN1 wanted_VVD to_TO derive_VVI definitions_NN2 for_IF varying_JJ "_" levels_NN2 of_IO saliency_NN1 "_" of_IO parental_JJ behavior_NN1 intended_VVD to_TO direct_VVI the_AT attention_NN1 of_IO their_APPGE children_NN2 with_IW autism_NN1 ._. 
The_AT parental_JJ behaviors_NN2 (_( i.e._REX ,_, cues_NN2 )_) used_VVN to_TO direct_VVI the_AT children_NN2 's_GE attention_NN1 were_VBDR gestures_NN2 ,_, words_NN2 ,_, actions_NN2 ,_, and_CC noises_NN2 of_IO toys_NN2 created_VVN when_CS the_AT parent_NN1 activated_VVD the_AT toy_NN1 ._. 
The_AT doctoral_JJ student_NN1 hypothesized_VVD that_CST more_DAR salient_JJ cues_NN2 would_VM be_VBI more_RGR successful_JJ in_II directing_VVG children_NN2 's_GE attention_NN1 than_CSN would_VM less_RGR salient_JJ cues_NN2 ._. 
She_PPHS1 could_VM not_XX find_VVI an_AT1 operational_JJ or_CC sufficiently_RR elaborated_VVN conceptual_JJ definition_NN1 of_IO "_" salience_NN1 "_" in_II the_AT extant_JJ literature_NN1 to_TO guide_VVI reliable_JJ judgments_NN2 regarding_II parental_JJ attentional_JJ cues_NN2 in_II a_AT1 free-play_JJ session_NNT1 with_IW the_AT parents_NN2 '_GE children_NN2 ._. 
However_RR ,_, she_PPHS1 was_VBDZ confident_JJ that_CST "_" she_PPHS1 knew_VVD a_AT1 salient_JJ cue_NN1 when_CS she_PPHS1 saw_VVD one_MC1 "_" and_CC that_DD1 other_JJ people_NN who_PNQS were_VBDR familiar_JJ with_IW young_JJ children_NN2 with_IW autism_NN1 would_VM be_VBI able_JK to_TO identify_VVI salient_JJ cues_NN2 ,_, too_RR ._. 
She_PPHS1 wanted_VVD a_AT1 content-valid_JJ definition_NN1 of_IO saliency_NN1 so_CS she_PPHS1 thought_VVD it_PPH1 was_VBDZ important_JJ to_TO rely_VVI on_II more_DAR than_CSN her_APPGE own_DA judgment_NN1 of_IO what_DDQ saliency_NN1 meant_VVD ._. 
She_PPHS1 needed_VVD a_AT1 panel_NN1 of_IO "_" experts_NN2 ._. "_" 
The_AT first_MD step_NN1 was_VBDZ to_TO identify_VVI the_AT "_" experts_NN2 ._. "_" 
The_AT definition_NN1 of_IO expert_NN1 varies_VVZ depending_II21 on_II22 the_AT content_JJ area_NN1 (_( Hoffman_NP1 et_RA21 al._RA22 ,_, 1995_MC )_) ._. 
However_RR ,_, it_PPH1 is_VBZ probably_RR useful_JJ to_TO select_VVI people_NN with_IW both_DB2 explicit_JJ (_( those_DD2 who_PNQS teach_VV0 others_NN2 to_TO do_VDI the_AT skill_NN1 )_) and_CC implicit_JJ (_( those_DD2 who_PNQS practice_NN1 the_AT art_NN1 or_CC skill_NN1 being_VBG studied_VVN )_) knowledge_NN1 of_IO the_AT context-dependent_JJ behavior_NN1 or_CC generalized_JJ characteristic_NN1 of_IO interest_NN1 ._. 
The_AT doctoral_JJ student_NN1 selected_JJ faculty_NN1 members_NN2 responsible_JJ for_IF teaching_VVG college_NN1 students_NN2 who_PNQS were_VBDR training_VVG to_TO be_VBI early_JJ childhood_NN1 educators_NN2 of_IO young_JJ children_NN2 with_IW autism_NN1 ._. 
These_DD2 faculty_NN1 members_NN2 had_VHD also_RR spent_VVN at_RR21 least_RR22 100_MC hr_NNU interacting_VVG with_IW children_NN2 with_IW autism_NN1 ._. 
Four_MC faculty_NN1 members_NN2 were_VBDR selected_VVN to_TO allow_VVI evidence_NN1 of_IO convergence_NN1 without_IW relying_VVG on_II a_AT1 sole_JJ expert_NN1 ._. 
Although_CS there_EX are_VBR many_DA2 types_NN2 of_IO materials_NN2 one_PN1 can_VM select_VVI to_TO elicit_VVI information_NN1 ,_, one_MC1 of_IO the_AT most_RGT efficient_JJ ways_NN2 to_TO elicit_VVI information_NN1 is_VBZ to_TO identify_VVI "_" test_NN1 "_" or_CC "_" tough_JJ "_" cases_NN2 (_( Hoffman_NP1 et_RA21 al._RA22 ,_, 1995_MC )_) ._. 
However_RR ,_, such_DA materials_NN2 tend_VV0 to_TO elicit_VVI an_AT1 uncomfortable_JJ feeling_NN1 of_IO being_VBG evaluated_VVN (_( Hoffman_NP1 et_RA21 al._RA22 ,_, 1995_MC )_) ._. 
In_II contrast_NN1 ,_, "_" familiar_JJ "_" materials_NN2 tend_VV0 to_TO elicit_VVI rapid_JJ expert_NN1 judgments_NN2 and_CC do_VD0 not_XX elicit_VVI anxiety_NN1 from_II experts_NN2 (_( Hoffman_NP1 et_RA21 al._RA22 ,_, 1995_MC )_) ._. 
Therefore_RR ,_, it_PPH1 has_VHZ been_VBN suggested_VVN that_CST a_AT1 combination_NN1 of_IO familiar_JJ and_CC test_VV0 materials_NN2 may_VM be_VBI useful_JJ in_II eliciting_VVG cooperation_NN1 and_CC information_NN1 (_( Hoffman_NP1 et_RA21 al._RA22 ,_, 1995_MC )_) ._. 
In_II our_APPGE example_NN1 ,_, the_AT doctoral_JJ student_NN1 edited_VVD twelve_MC 3_MC -_- 10_MC s_ZZ1 video_NN1 clips_NN2 of_IO parents_NN2 directing_VVG their_APPGE children_NN2 's_GE attention_NN1 ._. 
The_AT children_NN2 's_GE responses_NN2 to_II the_AT parental_JJ behavior_NN1 were_VBDR carefully_RR removed_VVN ._. 
The_AT doctoral_JJ student_NN1 located_VVN and_CC edited_VVN two_MC clear_JJ (_( i.e._REX ,_, familiar_JJ )_) examples_NN2 ,_, each_DD1 of_IO "_" high_JJ ,_, "_" "_" medium_NN1 ,_, "_" and_CC "_" low_JJ "_" saliency_NN1 categories_NN2 ,_, based_VVN on_II her_APPGE own_DA intuitive_JJ definitions_NN2 ._. 
The_AT doctoral_JJ student_NN1 selected_VVN three_MC more_DAR "_" test_NN1 "_" examples_NN2 that_CST she_PPHS1 judged_VVD to_TO be_VBI "_" somewhere_RL between_II high_JJ and_CC medium_JJ saliency_NN1 ._. "_" 
Finally_RR ,_, three_MC more_DAR test_NN1 examples_NN2 were_VBDR selected_VVN that_CST the_AT doctoral_JJ student_NN1 considered_VVN "_" somewhere_RL between_II medium_NN1 and_CC low_JJ saliency_NN1 ._. "_" 
Asking_VVG experts_NN2 to_TO sort_VVI materials_NN2 into_II categories_NN2 is_VBZ one_MC1 informative_JJ method_NN1 for_IF eliciting_VVG information_NN1 (_( Hoffman_NP1 et_RA21 al._RA22 ,_, 1995_MC )_) ._. 
In_II our_APPGE example_NN1 ,_, the_AT doctoral_JJ student_NN1 independently_RR asked_VVD each_DD1 expert_NN1 to_TO sort_VVI the_AT video_NN1 clips_VVZ into_II one_MC1 of_IO three_MC categories_NN2 that_CST varied_VVD by_II saliency_NN1 level_NN1 ._. 
No_AT time_NNT1 limit_NN1 was_VBDZ given_VVN and_CC experts_NN2 could_VM change_VVI their_APPGE minds_NN2 ._. 
This_DD1 continued_VVD until_CS each_DD1 expert_NN1 said_VVD that_CST he_PPHS1 or_CC she_PPHS1 was_VBDZ finished_VVN with_IW the_AT sorting_NN1 ._. 
The_AT relative_JJ value_NN1 of_IO individual_NN1 versus_II group_NN1 interviews_NN2 is_VBZ unknown_JJ ._. 
However_RR ,_, empirical_JJ studies_NN2 indicate_VV0 that_CST when_CS experts_NN2 meet_VV0 as_II a_AT1 group_NN1 ,_, they_PPHS2 tend_VV0 to_TO rapidly_RR find_VVI and_CC argue_VVI over_RP the_AT small_JJ number_NN1 of_IO points_NN2 on_II which_DDQ they_PPHS2 disagree_VV0 (_( Hoffman_NP1 et_RA21 al._RA22 ,_, 1995_MC )_) ._. 
It_PPH1 is_VBZ consensus_NN1 that_CST we_PPIS2 seek_VV0 ._. 
Therefore_RR ,_, we_PPIS2 prefer_VV0 that_CST experts_NN2 be_VBI interviewed_VVN independently_RR ._. 
The_AT literature_NN1 on_II eliciting_VVG expert_NN1 knowledge_NN1 indicates_VVZ that_CST structured_JJ interviews_NN2 are_VBR more_RGR efficient_JJ than_CSN unstructured_JJ ones_NN2 (_( Hoffman_NP1 et_RA21 al._RA22 ,_, 1995_MC )_) ._. 
In_II our_APPGE example_NN1 ,_, the_AT doctoral_JJ student_NN1 examined_VVD the_AT sorted_JJ materials_NN2 (_( a_ZZ1 )_) to_TO identify_VVI how_RRQ the_AT test_NN1 cases_NN2 were_VBDR sorted_VVN and_CC (_( b_ZZ1 )_) to_TO identify_VVI any_DD familiar_JJ cases_NN2 that_CST were_VBDR classified_VVN in_II a_AT1 category_NN1 that_CST was_VBDZ unexpected_JJ ._. 
Generic_JJ probes_NN2 were_VBDR asked_VVN about_II these_DD2 items_NN2 (_( Hoffman_NP1 et_RA21 al._RA22 ,_, 1995_MC )_) ._. 
The_AT questions_NN2 and_CC answers_NN2 were_VBDR audio_JJ recorded_VVN for_IF later_JJR analysis_NN1 ._. 
For_REX21 example_REX22 ,_, assume_VV0 clip_NN1 2_MC is_VBZ a_AT1 test_NN1 case_NN1 for_IF classification_NN1 to_II either_RR high_JJ or_CC medium_NN1 and_CC clip_VV0 4_MC is_VBZ a_AT1 familiar_JJ case_NN1 for_IF high_JJ saliency_NN1 ._. 
The_AT doctoral_JJ student_NN1 might_VM ask_VVI "_" Why_RRQ did_VDD you_PPY classify_VVI clip_NN1 number_NN1 2_MC as_CSA highly_RR salient_JJ ?_? "_" to_TO elicit_VVI rules_NN2 that_CST might_VM define_VVI high_JJ saliency_NN1 ._. 
It_PPH1 should_VM be_VBI noted_VVN that_CST both_DB2 the_AT clip_NN1 and_CC the_AT category_NN1 were_VBDR indicated_VVN in_II the_AT interviewer_NN1 's_GE question_NN1 ._. 
This_DD1 aids_VVZ later_JJR analysis_NN1 ._. 
Further_RRR imagine_VV0 that_CST clip_VV0 4_MC was_VBDZ unexpectantly_RR classified_VVN as_CSA medium_JJ saliency_NN1 ._. 
The_AT doctoral_JJ student_NN1 might_VM say_VVI "_" I_PPIS1 noticed_VVD you_PPY classified_JJ clip_NN1 2_MC as_CSA medium_JJ and_CC clip_VV0 4_MC as_CSA high_JJ "_" to_TO record_VVI on_II the_AT tape_NN1 how_RRQ the_AT clips_NN2 were_VBDR classified_VVN ._. 
Then_RT the_AT doctoral_JJ student_NN1 might_VM ask_VVI the_AT expert_NN1 ,_, "_" What_DDQ is_VBZ different_JJ about_II clip_NN1 2_MC (_( i.e._REX ,_, the_AT 'familiar'_JJ clip_NN1 that_CST was_VBDZ expected_VVN to_TO be_VBI classified_VVN as_CSA high_JJ but_CCB was_VBDZ classified_VVN as_CSA medium_JJ )_) than_CSN clip_NN1 4_MC (_( i.e._REX ,_, the_AT test_NN1 clip_VV0 that_DD1 was_VBDZ categorized_VVN as_CSA high_JJ )_) ?_? "_" 
The_AT latter_DA is_VBZ meant_VVN to_TO elicit_VVI particularly_RR useful_JJ rules_NN2 that_CST might_VM reveal_VVI "_" conditional_JJ rules_NN2 "_" (_( i.e._REX ,_, those_DD2 that_CST are_VBR used_VVN under_II some_DD conditions_NN2 but_CCB not_XX others_NN2 )_) ._. 
If_CS such_DA conditional_JJ rules_NN2 are_VBR uncovered_VVN ,_, then_RT the_AT doctoral_JJ student_NN1 might_VM test_VVI her_APPGE understanding_NN1 of_IO the_AT condition_NN1 of_IO the_AT rule_NN1 by_II asking_VVG "_" What_DDQ if_CS (_( the_AT condition_NN1 )_) were_VBDR not_XX present_JJ ,_, would_VM (_( the_AT rule_NN1 )_) apply_VV0 ?_? "_" 
This_DD1 type_NN1 of_IO structured_JJ interviewing_NN1 would_VM be_VBI continued_VVN until_CS the_AT interviewer_NN1 felt_VVD she_PPHS1 understood_VVD the_AT experts_NN2 '_GE rationale_NN1 for_IF their_APPGE sorting_NN1 ._. 
After_II interviewing_VVG all_DB experts_NN2 and_CC transcribing_VVG (_( or_CC marking_VVG wave_NN1 files_NN2 of_IO statement_NN1 via_II a_AT1 computer_NN1 software_NN1 such_II21 as_II22 NVivo_NP1 ;_; Bazeley_NP1 &;_NULL Richards_NP1 ,_, 2000_MC )_) ,_, commonalities_NN2 among_II the_AT experts_NN2 '_GE responses_NN2 to_II questions_NN2 are_VBR analyzed_VVN ._. 
This_DD1 is_VBZ sometimes_RT called_VVN "_" theme_NN1 analysis_NN1 "_" or_CC "_" category_NN1 analysis_NN1 "_" (_( Bazeley_NP1 ,_, 2007_MC )_) ._. 
In_II our_APPGE example_NN1 ,_, the_AT doctoral_JJ student_NN1 identified_VVN (_( a_ZZ1 )_) the_AT number_NN1 of_IO sensory_JJ modalities_NN2 and_CC (_( b_ZZ1 )_) the_AT number_NN1 of_IO behaviors_NN2 as_CSA two_MC themes_NN2 that_CST at_RR21 least_RR22 three_MC of_IO the_AT four_MC experts_NN2 used_VMK to_TO justify_VVI her_APPGE choices_NN2 ._. 
By_II "_" behaviors_NN2 ,_, "_" we_PPIS2 mean_VV0 actions_NN2 such_II21 as_II22 "_" gesture_NN1 to_II ,_, "_" "_" moves_VVZ on_RP ,_, "_" "_" talks_NN2 about_II ,_, "_" and_CC "_" operates_VVZ "_" referent_JJ objects_NN2 ._. 
To_TO "_" test_VVI "_" the_AT accuracy_NN1 with_IW which_DDQ the_AT abstracted_JJ themes_NN2 classified_VVN the_AT 12_MC video_NN1 clips_NN2 ,_, clips_NN2 were_VBDR assigned_VVN 3_MC through_II 1_MC1 to_TO correspond_VVI with_IW high_JJ through_II low_JJ classification_NN1 ,_, respectively_RR ._. 
Average_JJ numerical_JJ scores_NN2 were_VBDR derived_VVN across_II experts_NN2 to_TO estimate_VVI "_" expert_NN1 "_" classification_NN1 of_IO each_DD1 of_IO the_AT 12_MC video_NN1 clips_NN2 ._. 
Using_VVG the_AT two_MC themes_NN2 of_IO number_NN1 of_IO actions_NN2 and_CC number_NN1 of_IO sensory_JJ modalities_NN2 ,_, and_CC the_AT average_JJ expert_NN1 sorting_NN1 of_IO the_AT 12_MC video_NN1 clips_NN2 ,_, the_AT doctoral_JJ student_NN1 derived_VVD the_AT following_JJ operational_JJ definitions_NN2 for_IF the_AT three_MC levels_NN2 of_IO saliency_NN1 ._. 
These_DD2 definitions_NN2 classified_VVN the_AT 12_MC video_NN1 clips_VVZ in_II31 accordance_II32 with_II33 the_AT average_JJ expert_NN1 sorting_VVG ._. 
High_JJ saliency_NN1 was_VBDZ defined_VVN as_CSA using_VVG at_RR21 least_RR22 three_MC behaviors_NN2 to_TO draw_VVI the_AT child_NN1 's_GE attention_NN1 to_II an_AT1 object_NN1 that_CST appealed_VVD to_II at_RR21 least_RR22 two_MC sensory_JJ modalities_NN2 ._. 
Medium_JJ saliency_NN1 was_VBDZ defined_VVN as_CSA using_VVG two_MC behaviors_NN2 to_TO draw_VVI the_AT child_NN1 's_GE attention_NN1 to_II an_AT1 object_NN1 that_CST appealed_VVD to_II two_MC sensory_JJ modalities_NN2 ._. 
Low_JJ saliency_NN1 was_VBDZ defined_VVN as_CSA one_MC1 behavior_NN1 to_TO draw_VVI the_AT child_NN1 's_GE attention_NN1 to_II an_AT1 object_NN1 that_CST appealed_VVD to_II at_RR21 most_RR22 two_MC sensory_JJ modalities_NN2 ._. 
This_DD1 complex_JJ definition_NN1 classified_VVN all_DB 12_MC clips_VVZ reliably_RR by_II 2_MC observers_NN2 ._. 
Ultimately_RR ,_, the_AT test_NN1 of_IO such_DA a_AT1 process_NN1 is_VBZ whether_CSW a_AT1 study_NN1 using_VVG a_AT1 systematic_JJ observational_JJ measurement_NN1 system_NN1 relying_VVG on_II its_APPGE derived_JJ definitions_NN2 results_NN2 in_II confirming_VVG hypotheses_NN2 regarding_II saliency_NN1 and_CC the_AT child_NN1 's_GE correct_JJ responses_NN2 to_II parents_NN2 '_GE attentional_JJ directives_NN2 ._. 
Defining_VVG Segmenting_JJ Rules_NN2 Once_CS it_PPH1 is_VBZ clear_JJ what_DDQ the_AT lowest_JJT level_NN1 of_IO distinction_NN1 will_VM be_VBI ,_, the_AT onset_NN1 (_( and_CC if_CS needed_VVN the_AT offset_NN1 )_) of_IO instances_NN2 of_IO categories_NN2 may_VM need_VVI to_TO be_VBI defined_VVN ._. 
If_CS the_AT number_NN1 or_CC duration_NN1 of_IO events_NN2 will_VM be_VBI the_AT metric_JJ of_IO interest_NN1 ,_, then_RT segmenting_VVG rules_NN2 are_VBR necessary_JJ ._. 
Interval_NN1 coding_NN1 or_CC time_NNT1 sampling_NN1 behavior_NN1 sampling_NN1 does_VDZ not_XX require_VVI segmenting_VVG rules_NN2 for_IF reasons_NN2 indicated_VVN in_II the_AT next_MD chapter_NN1 ._. 
Briefly_RR ,_, interval_NN1 coding_NN1 means_VVZ that_CST one_PN1 indicates_VVZ whether_CSW at_RR21 least_RR22 one_MC1 instance_NN1 of_IO the_AT key_JJ behavior_NN1 has_VHZ occurred_VVN within_II a_AT1 fixed_JJ interval_NN1 of_IO time_NNT1 (_( e.g._REX ,_, 10_MC seconds_NNT2 )_) ._. 
One_PN1 can_VM see_VVI the_AT need_NN1 for_IF segmenting_VVG rules_NN2 when_RRQ events_NN2 occur_VV0 close_JJ in_II time_NNT1 ._. 
For_REX21 example_REX22 ,_, assume_VV0 we_PPIS2 are_VBR measuring_VVG the_AT number_NN1 of_IO communication_NN1 acts_NN2 ._. 
The_AT child_NN1 reaches_VVZ for_IF a_AT1 ball_NN1 and_CC looks_VVZ at_II an_AT1 adult_NN1 ._. 
Then_RT one_MC1 second_NNT1 later_RRR ,_, the_AT child_NN1 says_VVZ "_" ah_UH "_" while_CS looking_VVG at_II the_AT ball_NN1 and_CC looks_VVZ at_II the_AT adult_NN1 ._. 
Finally_RR ,_, another_DD1 second_NNT1 later_RRR ,_, the_AT child_NN1 says_VVZ "_" ball_NN1 "_" and_CC looks_VVZ at_II the_AT adult_NN1 ._. 
Segmenting_VVG rules_NN2 are_VBR needed_VVN to_TO determine_VVI whether_CSW this_DD1 cluster_NN1 of_IO child_NN1 behaviors_NN2 is_VBZ 1_MC1 versus_II 3_MC communication_NN1 acts_NN2 ._. 
In-seat_JJ behavior_NN1 is_VBZ an_AT1 example_NN1 of_IO a_AT1 behavior_NN1 for_IF which_DDQ duration_NN1 is_VBZ important_JJ ._. 
In_II31 addition_II32 to_II33 precise_JJ definitions_NN2 of_IO onset_NN1 ,_, definitions_NN2 of_IO offset_NN1 are_VBR necessary_JJ to_TO differentiate_VVI examples_NN2 of_IO in-seat_JJ behavior_NN1 from_II near_JJ nonexamples_NN2 of_IO offset_NN1 ._. 
For_REX21 example_REX22 ,_, is_VBZ the_AT momentary_JJ lifting_NN1 of_IO both_DB2 buttock_VV0 cheeks_NN2 from_II contact_NN1 with_IW the_AT chair_NN1 seat_NN1 sufficient_JJ to_TO end_VVI in-seat_JJ behavior_NN1 or_CC does_VDZ such_DA contact_VVI need_NN1 to_TO cease_VVI for_IF more_DAR than_CSN one_MC1 second_NNT1 ?_? 
Segmenting_VVG rules_NN2 almost_RR always_RR involve_VV0 a_AT1 certain_JJ amount_NN1 of_IO arbitrariness_NN1 ._. 
For_REX21 example_REX22 ,_, we_PPIS2 may_VM decide_VVI to_TO treat_VVI the_AT potentially_RR three_MC clusters_NN2 of_IO behavior_NN1 in_II our_APPGE example_NN1 above_RL as_CSA one_MC1 act_NN1 because_CS the_AT onset_NN1 of_IO the_AT first_MD behavior_NN1 occurred_VVD within_II 3_MC s_ZZ1 of_IO the_AT onset_NN1 of_IO the_AT last_MD behavior_NN1 ._. 
The_AT use_NN1 of_IO a_AT1 temporal_JJ criterion_NN1 ,_, the_AT use_NN1 of_IO "_" onset_NN1 "_" instead_II21 of_II22 "_" offset_NN1 "_" as_II the_AT boundaries_NN2 for_IF the_AT temporal_JJ criterion_NN1 ,_, and_CC the_AT decision_NN1 to_TO call_VVI the_AT reach_NN1 and_CC gaze_NN1 ,_, the_AT vocalization_NN1 and_CC gaze_NN1 ,_, and_CC word_NN1 and_CC gaze_NN1 ,_, all_DB "_" parts_NN2 "_" of_IO the_AT same_DA act_NN1 are_VBR all_DB questionable_JJ ,_, but_CCB defensible_JJ ._. 
Our_APPGE empirical_JJ and_CC theoretical_JJ knowledge_NN1 about_II most_DAT social_JJ phenomenon_NN1 of_IO interest_NN1 is_VBZ almost_RR never_RR sufficiently_RR specific_JJ to_TO guide_VVI this_DD1 level_NN1 of_IO decision_NN1 making_VVG ._. 
But_CCB ,_, these_DD2 decisions_NN2 must_VM be_VBI made_VVN ;_; hence_RR ,_, there_EX is_VBZ some_DD degree_NN1 of_IO arbitrariness_NN1 ._. 
Usually_RR ,_, the_AT best_JJT we_PPIS2 can_VM do_VDI is_VBZ to_TO make_VVI sure_JJ that_CST our_APPGE decisions_NN2 are_VBR defensible_JJ and_CC consistently_RR applied_VVN ._. 
If_CS they_PPHS2 are_VBR carried_VVN out_RP consistently_RR ,_, the_AT potential_JJ lack_NN1 of_IO content_JJ validity_NN1 that_CST may_VM occur_VVI from_II their_APPGE existence_NN1 will_VM be_VBI worth_II the_AT almost_RR certain_JJ gain_NN1 in_II reliability_NN1 due_II21 to_II22 reduced_JJ inter-observer_JJ disagreement_NN1 on_II segmenting_VVG ._. 
Defining_VVG When_RRQ to_II Start_NN1 and_CC Stop_VV0 Coding_NN1 Having_VHG clearly_RR defined_VVN the_AT lowest_JJT level_JJ coding_NN1 distinction_NN1 and_CC with_IW guidance_NN1 about_II how_RRQ to_II segment_NN1 potential_NN1 events_NN2 ,_, the_AT next_MD step_NN1 is_VBZ to_TO decide_VVI when_RRQ observers_NN2 should_VM begin_VVI and_CC end_VVI their_APPGE coding_NN1 of_IO an_AT1 observation_NN1 session_NNT1 ._. 
This_DD1 section_NN1 will_VM include_VVI two_MC types_NN2 of_IO start-stop_JJ signals_NN2 ._. 
First_MD ,_, there_EX are_VBR those_DD2 at_II the_AT beginning_NN1 and_CC end_NN1 of_IO the_AT observation_NN1 session_NNT1 ._. 
Because_II21 of_II22 the_AT inconsistency_NN1 of_IO behavior_NN1 among_II many_DA2 observational_JJ sessions_NNT2 ,_, it_PPH1 is_VBZ useful_JJ to_TO make_VVI explicit_JJ the_AT signal_NN1 for_IF beginning_VVG and_CC ending_VVG coding_NN1 ._. 
One_MC1 ill-advised_JJ way_NN1 to_TO do_VDI this_DD1 is_VBZ to_TO ask_VVI the_AT adult_NN1 administering_VVG the_AT procedure_NN1 to_TO say_VVI when_RRQ to_TO start_VVI and_CC stop_VVI coding_NN1 ._. 
Alternatively_RR ,_, a_AT1 start_NN1 signal_NN1 might_VM be_VBI when_CS the_AT clock_NN1 of_IO the_AT media_NN file_NN1 turns_VVZ from_II 0_MC to_II 1_MC1 s_ZZ1 ._. 
This_DD1 requires_VVZ that_CST the_AT administrator_NN1 or_CC cameraperson_NN1 be_VBI consistent_JJ in_II giving_VVG the_AT verbal_JJ signal_NN1 or_CC in_II beginning_VVG the_AT clock_NN1 relative_II21 to_II22 the_AT onset_NN1 of_IO the_AT observation_NN1 procedure_NN1 ._. 
The_AT problem_NN1 with_IW these_DD2 approaches_NN2 is_VBZ that_CST it_PPH1 shifts_VVZ the_AT responsibility_NN1 for_IF providing_VVG the_AT signal_NN1 consistently_RR to_II the_AT administrator_NN1 or_CC cameraperson_NN1 ._. 
Investigators_NN2 rarely_RR check_VV0 on_II the_AT consistency_NN1 of_IO the_AT timing_NN1 for_IF such_DA signals_NN2 ._. 
Instead_RR ,_, it_PPH1 is_VBZ better_JJR to_TO use_VVI a_AT1 behavior_NN1 that_CST the_AT participant_NN1 does_VDZ or_CC one_PN1 that_CST the_AT examiner_NN1 does_VDZ in_II the_AT course_NN1 of_IO conducting_VVG the_AT procedure_NN1 to_TO mark_VVI the_AT beginning_NN1 and_CC ending_NN1 of_IO coding_NN1 ._. 
For_REX21 example_REX22 ,_, one_PN1 might_VM indicate_VVI that_DD1 coding_NN1 begins_VVZ when_RRQ the_AT adult_NN1 first_MD speaks_VVZ about_II the_AT objects_NN2 in_II the_AT session_NNT1 or_CC first_MD speaks_VVZ to_II the_AT child_NN1 ._. 
Similarly_RR ,_, stop_VV0 signals_NN2 might_VM be_VBI when_CS the_AT examiner_NN1 removes_VVZ all_DB toys_NN2 from_II the_AT table_NN1 ._. 
Whatever_DDQV the_AT signals_NN2 ,_, they_PPHS2 need_VV0 to_TO be_VBI included_VVN in_II the_AT recording_NN1 of_IO every_AT1 session_NNT1 and_CC occur_VV0 at_II times_NNT2 that_CST do_VD0 not_XX exclude_VVI many_DA2 codeable_JJ acts_NN2 ._. 
Another_DD1 use_NN1 of_IO start-stop_JJ signals_NN2 is_VBZ to_TO define_VVI the_AT duration_NN1 of_IO the_AT codeable_JJ sections_NN2 of_IO the_AT observation_NN1 session_NNT1 ._. 
When_CS such_DA start_NN1 and_CC stop_VV0 coding_VVG rules_NN2 are_VBR provided_VVN ,_, they_PPHS2 are_VBR done_VDN so_BCL21 as_BCL22 to_TO potentially_RR reduce_VVI the_AT measurement_NN1 error_NN1 due_II21 to_II22 unexpected_JJ events_NN2 or_CC events_NN2 that_CST are_VBR known_VVN to_TO inhibit_VVI or_CC interfere_VVI with_IW the_AT occurrence_NN1 of_IO key_JJ behaviors_NN2 in_II a_AT1 way_NN1 that_CST does_VDZ not_XX reflect_VVI the_AT phenomenon_NN1 of_IO interest_NN1 ._. 
For_REX21 example_REX22 ,_, the_AT participant_NN1 may_VM be_VBI off-screen_JJ if_CS the_AT session_NNT1 is_VBZ being_VBG recorded_VVN for_IF later_JJR coding_NN1 ._. 
