Social_JJ networking_NN1 sites_NN2 and_CC our_APPGE lives_NN2 Facebook_NN1 users_NN2 are_VBR more_RGR trusting_JJ ,_, have_VH0 more_RGR close_JJ friends_NN2 ,_, are_VBR more_RGR politically_RR engaged_VVN ,_, and_CC get_VV0 more_DAR support_NN1 from_II their_APPGE friends_NN2 The_AT highest_JJT proportion_NN1 of_IO Facebook_NN1 friends_NN2 is_VBZ high_JJ school_NN1 classmates_NN2 ,_, and_CC Facebook_NN1 helps_NN2 revive_VVI "_" dormant_JJ "_" ties_NN2 with_IW lost_JJ connections_NN2 Washington_NP1 (_( June_NPM1 16_MC ,_, 2011_MC )_) --_JJ New_JJ national_JJ survey_NN1 findings_NN2 show_VV0 that_DD1 use_NN1 of_IO social_JJ networking_NN1 sites_NN2 is_VBZ growing_JJ and_CC that_CST those_DD2 who_PNQS use_VV0 these_DD2 sites_NN2 ,_, especially_RR Facebook_NP1 users_NN2 ,_, have_VH0 higher_JJR measures_NN2 of_IO social_JJ well-being_NN1 ._. 
In_II a_AT1 national_JJ phone_NN1 survey_NN1 of_IO 2,255_MC American_JJ adults_NN2 last_MD fall_NN1 ,_, the_AT Pew_NN1 Research_NN1 Center_NN1 's_GE Internet_NN1 &;_NULL American_JJ Life_NN1 Project_NN1 found_VVD that_CST :_: Facebook_NP1 users_NN2 are_VBR more_RGR trusting_JJ than_CSN others_NN2 ._. 
Controlling_VVG for_IF other_JJ factors_NN2 ,_, the_AT research_NN1 found_VVD that_CST a_AT1 Facebook_NN1 user_NN1 who_PNQS uses_VVZ the_AT site_NN1 multiple_JJ times_NNT2 per_II day_NNT1 is_VBZ 43%_NNU more_DAR likely_JJ than_CSN other_JJ internet_NN1 users_NN2 and_CC more_DAR than_CSN three_MC times_NNT2 as_RG likely_JJ as_CSA non-internet_JJ users_NN2 to_TO feel_VVI that_CST most_DAT people_NN can_VM be_VBI trusted_VVN ._. 
Facebook_NN1 users_NN2 have_VH0 more_RGR close_JJ relationships_NN2 ._. 
Controlling_VVG for_IF other_JJ factors_NN2 ,_, the_AT research_NN1 found_VVD that_CST someone_PN1 who_PNQS uses_VVZ Facebook_NN1 several_DA2 times_NNT2 per_II day_NNT1 averages_NN2 9%_NNU more_DAR close_JJ ,_, core_NN1 ties_VVZ in_II their_APPGE overall_JJ social_JJ network_NN1 compared_VVN with_IW other_JJ internet_NN1 users_NN2 ._. 
Facebook_NN1 users_NN2 are_VBR much_RR more_RGR politically_RR engaged_VVN ._. 
The_AT survey_NN1 was_VBDZ conducted_VVN over_II the_AT November_NPM1 2010_MC election_NN1 season_NNT1 ._. 
Compared_VVN with_IW other_JJ internet_NN1 users_NN2 ,_, and_CC users_NN2 of_IO other_JJ social_JJ networking_NN1 platforms_NN2 ,_, a_AT1 Facebook_NN1 user_NN1 who_PNQS uses_VVZ the_AT site_NN1 multiple_JJ times_NNT2 per_II day_NNT1 was_VBDZ an_AT1 additional_JJ two_MC and_CC half_DB times_NNT2 more_RRR likely_JJ to_TO attend_VVI a_AT1 political_JJ rally_NN1 or_CC meeting_NN1 ,_, 57%_NNU more_DAR likely_JJ to_TO persuade_VVI someone_PN1 on_II their_APPGE vote_NN1 ,_, and_CC 43%_NNU more_DAR likely_JJ to_TO have_VHI said_VVN they_PPHS2 would_VM vote_VVI ._. 
Facebook_NN1 users_NN2 get_VV0 more_RGR social_JJ support_NN1 ._. 
The_AT survey_NN1 explored_VVD how_RGQ much_DA1 total_JJ social_JJ support_NN1 ,_, emotional_JJ support_NN1 ,_, companionship_NN1 ,_, and_CC instrumental_JJ aid_NN1 (_( such_II21 as_II22 having_VHG someone_PN1 help_VV0 you_PPY when_CS you_PPY are_VBR sick_JJ in_II bed_NN1 )_) adults_NN2 receive_VV0 ._. 
Controlling_VVG for_IF other_JJ factors_NN2 ,_, a_AT1 Facebook_NN1 user_NN1 who_PNQS uses_VVZ the_AT site_NN1 multiple_JJ times_NNT2 per_II day_NNT1 receives_VVZ more_RGR emotional_JJ support_NN1 and_CC companionship_NN1 ._. 
For_IF Facebook_NN1 users_NN2 ,_, the_AT additional_JJ boost_NN1 is_VBZ equivalent_JJ to_II about_RG half_DB the_AT total_JJ support_NN1 that_CST the_AT average_JJ American_NN1 receives_VVZ as_II a_AT1 result_NN1 of_IO being_VBG married_JJ or_CC cohabitating_VVG with_IW a_AT1 partner_NN1 ._. 
Facebook_NN1 helps_VVZ users_NN2 retain_VVI high_JJ school_NN1 ties_NN2 and_CC it_PPH1 revives_VVZ dormant_JJ relationships_NN2 ._. 
In_II our_APPGE sample_NN1 ,_, the_AT average_JJ Facebook_NN1 user_NN1 has_VHZ 229_MC Facebook_NN1 friends_NN2 ._. 
They_PPHS2 reported_VVD that_CST their_APPGE friends_NN2 list_NN1 contains_VVZ :_: 22%_NNU people_NN from_II high_JJ school_NN1 12%_NNU extended_JJ family_NN1 10%_NNU coworkers_NN2 9%_NNU college_NN1 friends_NN2 8%_NNU immediate_JJ family_NN1 7%_NNU people_NN from_II voluntary_JJ groups_NN2 2%_NNU neighbors_NN2 Over_RG 31%_NNU of_IO Facebook_NN1 friends_NN2 can_VM not_XX be_VBI classified_VVN into_II these_DD2 categories_NN2 ._. 
However_RR ,_, only_RR 3%_NNU of_IO Facebook_NN1 friends_NN2 are_VBR people_NN users_NN2 have_VH0 never_RR met_VVN in_II person_NN1 ,_, and_CC only_RR 7%_NNU are_VBR people_NN who_PNQS have_VH0 met_VVN only_RR one_MC1 time_NNT1 ._. 
The_AT remainder_NN1 is_VBZ friends-of-friends_NN2 and_CC social_JJ ties_NN2 that_CST are_VBR not_XX currently_RR active_JJ relationships_NN2 ,_, but_CCB "_" dormant_JJ "_" ties_NN2 that_CST were_VBDR meaningful_JJ once_RR and_CC have_VH0 been_VBN at_RR21 least_RR22 somewhat_RR maintained_VVN through_II use_NN1 of_IO Facebook_NN1 ._. 
"_" There_EX has_VHZ been_VBN a_AT1 great_JJ deal_NN1 of_IO speculation_NN1 about_II the_AT impact_NN1 of_IO social_JJ networking_NN1 site_NN1 use_NN1 on_II people_NN 's_GE social_JJ lives_NN2 ,_, and_CC much_DA1 of_IO it_PPH1 has_VHZ centered_VVN on_II the_AT possibility_NN1 that_CST these_DD2 sites_NN2 are_VBR hurting_VVG users_NN2 '_GE relationships_NN2 and_CC pushing_VVG them_PPHO2 away_II21 from_II22 participating_VVG in_II the_AT world_NN1 ,_, "_" noted_VVD Prof._NNB Keith_NP1 Hampton_NP1 ,_, the_AT lead_NN1 author_NN1 of_IO the_AT new_JJ Pew_NN1 Internet_NN1 report_NN1 ._. 
"_" We_PPIS2 've_VH0 found_VVN the_AT exact_JJ opposite_JJ --_NN1 that_CST people_NN who_PNQS use_VV0 sites_NN2 like_II Facebook_NP1 actually_RR have_VH0 more_RGR close_JJ relationships_NN2 and_CC are_VBR more_RGR likely_JJ to_TO be_VBI involved_JJ in_II civic_JJ and_CC political_JJ activities_NN2 ._. "_" 
This_DD1 survey_NN1 also_RR showed_VVD that_CST more_DAR people_NN are_VBR using_VVG social_JJ networking_NN1 sites_NN2 --_NN1 the_AT figure_NN1 is_VBZ now_RT 47%_NNU of_IO the_AT entire_JJ adult_NN1 population_NN1 ,_, compared_VVN with_IW 26%_NNU that_DD1 was_VBDZ measured_VVN in_II our_APPGE similar_JJ 2008_MC survey_NN1 ._. 
Among_II other_JJ things_NN2 ,_, this_DD1 means_VVZ the_AT average_JJ age_NN1 of_IO adult_JJ social_JJ networking_NN1 site_NN1 users_NN2 has_VHZ shifted_VVN from_II 33_MC in_II 2008_MC to_II 38_MC in_II 2010_MC ._. 
Over_RG half_DB of_IO all_DB adult_JJ social_JJ networking_NN1 site_NN1 users_NN2 are_VBR now_RT over_II the_AT age_NN1 of_IO 35_MC ._. 
In_II Pew_NN1 Internet_NN1 's_VBZ first-ever_RR reading_VVG on_II specific_JJ Facebook_NN1 activities_NN2 ,_, the_AT survey_NN1 found_VVD that_CST on_II an_AT1 average_JJ day_NNT1 :_: 15%_NNU of_IO Facebook_NN1 users_NN2 update_VV0 their_APPGE own_DA status._NNU 22%_NNU comment_NN1 on_II another_DD1 's_VBZ post_NN1 or_CC status._NNU 20%_NNU comment_NN1 on_II another_DD1 user_NN1 's_GE photos._NNU 26%_NNU "_" Like_II "_" another_DD1 user_NN1 's_GE content._NNU 10%_NNU send_VV0 another_DD1 user_NN1 a_AT1 private_JJ message_NN1 "_" Facebook_NN1 has_VHZ become_VVN the_AT dominant_JJ social_JJ networking_NN1 platform_NN1 in_II31 terms_II32 of_II33 both_DB2 number_NN1 of_IO users_NN2 and_CC frequency_NN1 of_IO use_NN1 ,_, and_CC it_PPH1 is_VBZ striking_VVG to_TO note_VVI that_CST the_AT makeup_NN1 of_IO the_AT population_NN1 is_VBZ changing_VVG ,_, "_" noted_VVD Lauren_NP1 Sessions_NNT2 Goulet_NP1 ,_, co-author_NN1 of_IO the_AT report_NN1 ._. 
We_PPIS2 also_RR found_VVD interesting_JJ variation_NN1 in_II the_AT characteristics_NN2 of_IO users_NN2 across_II different_JJ social_JJ networking_NN1 sites_NN2 ._. 
People_NN pick_VV0 the_AT platforms_NN2 which_DDQ best_RRT meet_VV0 their_APPGE social_JJ and_CC professional_JJ needs_NN2 ._. "_" 
For_REX21 instance_REX22 ,_, the_AT report_NN1 found_VVD :_: Nearly_RR twice_RR as_RG many_DA2 men_NN2 (_( 63%_NNU )_) as_CSA women_NN2 (_( 37%_NNU )_) use_VV0 LinkedIn_NN1 ._. 
The_AT average_JJ adult_NN1 MySpace_NP1 user_NN1 is_VBZ younger_JJR (_( 32_MC )_) ,_, and_CC the_AT average_JJ adult_NN1 LinkedIn_NN1 user_NN1 older_JJR (_( 40_MC )_) ,_, than_CSN the_AT average_JJ Facebook_NN1 user_NN1 (_( 38_MC )_) ,_, Twitter_VV0 user_NN1 (_( 33_MC )_) ,_, and_CC users_NN2 of_IO other_JJ social_JJ networking_NN1 sites_NN2 (_( 35_MC )_) ._. 
MySpace_VV0 and_CC Twitter_VV0 users_NN2 are_VBR the_AT most_RGT racially_RR diverse_JJ mainstream_JJ social_JJ network_NN1 platforms_NN2 ._. 
MySpace_VV0 users_NN2 tend_VV0 to_TO have_VHI fewer_DAR years_NNT2 of_IO formal_JJ education_NN1 than_CSN users_NN2 of_IO other_JJ social_JJ network_NN1 services_NN2 ,_, whereas_CS most_DAT LinkedIn_NN1 users_NN2 have_VH0 at_RR21 least_RR22 one_MC1 university_NN1 degree_NN1 ._. 
There_EX were_VBDR several_DA2 other_JJ surprises_NN2 in_II the_AT survey_NN1 the_AT authors_NN2 found_VVD notable_JJ :_: Social_JJ networking_NN1 sites_NN2 are_VBR increasingly_RR used_VVN to_TO keep_VVI up_RP with_IW close_JJ social_JJ ties_NN2 ._. 
Looking_VVG at_II those_DD2 people_NN that_CST social_JJ networking_NN1 site_NN1 users_NN2 report_VV0 as_CSA their_APPGE core_NN1 discussion_NN1 confidants_NN2 ,_, 40%_NNU of_IO users_NN2 have_VH0 friended_VVN all_DB of_IO their_APPGE closest_JJT confidants_NN2 ._. 
This_DD1 is_VBZ a_AT1 substantial_JJ increase_NN1 from_II the_AT 29%_NNU of_IO users_NN2 who_PNQS reported_VVD in_II our_APPGE 2008_MC survey_NN1 that_CST they_PPHS2 had_VHD friended_VVN all_DB of_IO their_APPGE core_NN1 confidants_NN2 ._. 
MySpace_VV0 users_NN2 are_VBR more_RGR likely_JJ to_TO be_VBI open_JJ to_II opposing_JJ points_NN2 of_IO view_NN1 ._. 
We_PPIS2 measured_VVD "_" perspective_NN1 taking_VVG ,_, "_" or_CC the_AT ability_NN1 of_IO people_NN to_TO consider_VVI multiple_JJ points_NN2 of_IO view_NN1 ._. 
There_EX is_VBZ no_AT evidence_NN1 that_CST social_JJ networking_NN1 site_NN1 users_NN2 ,_, including_II those_DD2 who_PNQS use_VV0 Facebook_NP1 ,_, are_VBR any_DD more_DAR likely_JJ than_CSN others_NN2 to_II cocoon_NN1 themselves_PPX2 in_II social_JJ networks_NN2 of_IO like-minded_JJ and_CC similar_JJ people_NN ,_, as_CSA some_DD have_VH0 feared_VVN ._. 
Moreover_RR ,_, regression_NN1 analysis_NN1 found_VVD that_CST those_DD2 who_PNQS use_VV0 MySpace_NP1 have_VH0 significantly_RR higher_JJR levels_NN2 of_IO perspective_NN1 taking_VVG ._. 
"_" Social_JJ networking_NN1 sites_NN2 have_VH0 become_VVN increasingly_RR important_JJ to_II people_NN as_CSA they_PPHS2 find_VV0 ways_NN2 to_TO integrate_VVI check-ins_NNU2 and_CC updates_VVZ into_II the_AT rhythms_NN2 of_IO their_APPGE lives_NN2 ,_, "_" noted_VVD Lee_NP1 Rainie_NP1 ,_, a_AT1 co-author_NN1 of_IO the_AT report_NN1 ._. 
"_" People_NN use_VV0 them_PPHO2 now_RT to_TO stay_VVI in_II31 touch_II32 with_II33 their_APPGE best_JJT friends_NN2 and_CC distant_JJ acquaintances_NN2 alike_RR ._. 
But_CCB the_AT story_NN1 has_VHZ n't_XX ended_VVN ._. 
It_PPH1 's_VBZ clear_JJ that_CST the_AT world_NN1 of_IO networked_JJ individuals_NN2 will_VM continue_VVI to_TO change_VVI as_II the_AT platforms_NN2 and_CC populations_NN2 of_IO users_NN2 continue_VV0 to_TO evolve_VVI ._. "_" 
In_II this_DD1 survey_NN1 ,_, 2,255_MC American_JJ adults_NN2 were_VBDR surveyed_VVN between_II October_NPM1 20-November_NPM1 28_MC ,_, 2010_MC ,_, including_II 1,787_MC internet_NN1 users_NN2 ._. 
There_EX were_VBDR 975_MC users_NN2 of_IO social_JJ networking_NN1 site_NN1 such_II21 as_II22 Facebook_NP1 ,_, MySpace_NP1 ,_, LinkedIn_NP1 ,_, and_CC Twitter_NN1 ._. 
The_AT margin_NN1 of_IO error_NN1 on_II the_AT entire_JJ survey_NN1 is_VBZ plus_II or_CC minus_II 3_MC percentage_NN1 points_NN2 ,_, on_II the_AT internet_NN1 users_NN2 is_VBZ plus_II or_CC minus_II 3_MC percentage_NN1 points_NN2 ,_, and_CC for_IF the_AT social_JJ networking_NN1 site_NN1 users_NN2 is_VBZ plus_II or_CC minus_II 4_MC percentage_NN1 points_NN2 ._. 
Social_JJ networking_NN1 sites_NN2 (_( SNS_NN2 )_) provide_VV0 people_NN with_IW the_AT opportunity_NN1 to_II friend_NN1 members_NN2 of_IO their_APPGE overall_JJ network_NN1 of_IO family_NN1 members_NN2 ,_, coworkers_NN2 ,_, and_CC other_JJ acquaintances_NN2 ._. 
Much_RR has_VHZ been_VBN made_VVN of_IO the_AT use_NN1 of_IO the_AT word_NN1 "_" friend_NN1 "_" in_II this_DD1 context_NN1 ._. 
Those_DD2 who_PNQS are_VBR listed_VVN as_CSA friends_NN2 on_II SNS_NN2 may_VM indeed_RR be_VBI friends_NN2 in_II the_AT traditional_JJ sense_NN1 ,_, but_CCB they_PPHS2 can_VM also_RR be_VBI old_JJ acquaintances_NN2 (_( e.g._REX ,_, from_II high_JJ school_NN1 )_) or_CC very_RG casual_JJ connections_NN2 between_II people_NN who_PNQS have_VH0 never_RR have_VH0 met_VVN in_II person_NN1 ._. 
Some_DD worry_VV0 that_CST as_II a_AT1 result_NN1 of_IO using_VVG these_DD2 services_NN2 ,_, people_NN may_VM become_VVI more_RGR isolated_JJ and_CC substitute_VVI less_RGR meaningful_JJ relations_NN2 for_IF real_JJ social_JJ support_NN1 ._. 
Others_NN2 believe_VV0 this_DD1 might_VM enrich_VVI and_CC expand_VVI relationships_NN2 ._. 
Here_RL below_RL are_VBR our_APPGE findings_NN2 on_II all_DB of_IO this_DD1 ._. 
Looking_VVG at_II people_NN 's_GE overall_JJ social_JJ networks_NN2 ,_, not_XX just_RR their_APPGE online_JJ ties_NN2 ,_, the_AT average_JJ American_NN1 has_VHZ 634_MC ties_NN2 in_II their_APPGE overall_JJ network_NN1 ,_, and_CC technology_NN1 users_NN2 have_VH0 bigger_JJR networks_NN2 ._. 
Most_DAT Americans_NN2 overall_JJ networks_NN2 contain_VV0 a_AT1 range_NN1 of_IO social_JJ ties_NN2 that_CST consist_VV0 of_IO friends_NN2 ,_, family_NN1 ,_, coworkers_NN2 ,_, and_CC other_JJ acquaintances_NN2 ._. 
This_DD1 includes_VVZ a_AT1 handful_NN1 of_IO very_RG close_JJ social_JJ ties_NN2 and_CC a_AT1 much_RR large_JJ number_NN1 of_IO weaker_JJR ties_NN2 ._. 
It_PPH1 is_VBZ nearly_RR impossible_JJ for_IF most_DAT people_NN to_TO reliably_RR list_VVI all_DB of_IO the_AT people_NN they_PPHS2 know_VV0 ._. 
This_DD1 makes_VVZ it_PPH1 very_RG difficult_JJ to_TO measure_VVI people_NN 's_GE total_JJ network_NN1 size_NN1 ._. 
However_RR ,_, social_JJ scientists_NN2 have_VH0 developed_VVN methods_NN2 for_IF estimating_VVG the_AT size_NN1 of_IO people_NN 's_GE networks_NN2 ._. 
The_AT approach_NN1 that_CST we_PPIS2 use_VV0 is_VBZ called_VVN the_AT "_" scale-up_NN1 method_NN1 "_" &lsqb;_( 3_MC &rsqb;_) ._. 
This_DD1 approach_NN1 has_VHZ been_VBN embraced_VVN by_II social_JJ network_NN1 analysts_NN2 and_CC its_APPGE history_NN1 and_CC rationale_NN1 are_VBR described_VVN in_II Appendix_NN1 D._NP1 The_AT method_NN1 is_VBZ based_VVN on_II the_AT knowledge_NN1 that_CST the_AT people_NN a_AT1 person_NN1 comes_VVZ to_TO know_VVI in_II a_AT1 lifetime_NNT1 are_VBR made_VVN up_RP of_IO various_JJ subpopulations_NN2 (_( e.g._REX ,_, categories_NN2 of_IO people_NN ,_, such_II21 as_II22 family_NN1 ,_, doctors_NN2 ,_, mailmen_NN2 ,_, people_NN named_VVD "_" Rose_NP1 ,_, "_" etc_RA )_) ._. 
If_CS we_PPIS2 know_VV0 the_AT size_NN1 of_IO a_AT1 subpopulation_NN1 from_II publicly_RR available_JJ statistics_NN ,_, such_II21 as_II22 how_RGQ many_DA2 mailmen_NN2 there_EX are_VBR or_CC how_RGQ many_DA2 people_NN there_EX are_VBR named_VVN "_" Rose_NP1 ,_, "_" and_CC we_PPIS2 know_VV0 how_RGQ many_DA2 people_NN a_AT1 person_NN1 knows_VVZ from_II this_DD1 subpopulation_NN1 ,_, we_PPIS2 can_VM make_VVI an_AT1 accurate_JJ estimate_NN1 of_IO a_AT1 person_NN1 's_GE total_JJ network_NN1 size.2_FO This_DD1 approach_NN1 assumes_VVZ that_CST the_AT composition_NN1 of_IO people_NN 's_GE social_JJ networks_NN2 mirrors_NN2 the_AT presence_NN1 of_IO a_AT1 specific_JJ subpopulation_NN1 in_II society_NN1 (_( e.g._REX ,_, if_CS one_MC1 out_RP of_IO 100_MC people_NN in_II the_AT population_NN1 have_VH0 a_AT1 characteristic_NN1 ,_, 1/100_MF people_NN in_II a_AT1 person_NN1 's_GE network_NN1 should_VM share_VVI this_DD1 same_DA characteristic_NN1 )_) ._. 
This_DD1 assumption_NN1 is_VBZ generally_RR true_JJ ,_, but_CCB can_VM be_VBI further_RRR adjusted_VVN to_TO increase_VVI accuracy_NN1 ,_, which_DDQ depends_VVZ on_II four_MC other_JJ factors_NN2 ._. 
The_AT first_MD is_VBZ network_NN1 knowledge_NN1 (_( e.g._REX ,_, you_PPY may_VM know_VVI someone_PN1 ,_, but_CCB not_XX know_VV0 they_PPHS2 are_VBR a_AT1 mailman_NN1 )_) ._. 
The_AT second_NNT1 is_VBZ recall_NN1 accuracy_NN1 (_( e.g_REX ,_, people_NN tend_VV0 to_TO overestimate_VVI the_AT number_NN1 of_IO people_NN they_PPHS2 know_VV0 from_II small_JJ subpopulations_NN2 and_CC underestimate_VV0 from_II larger_JJR ones_NN2 )_) ._. 
The_AT third_MD is_VBZ knowledge_NN1 of_IO a_AT1 large_JJ number_NN1 of_IO subpopulations_NN2 ,_, and_CC the_AT fourth_MD is_VBZ exposure_NN1 or_CC social_JJ mixing_NN1 (_( e.g._REX ,_, older_JJR women_NN2 may_VM have_VHI been_VBN exposed_VVN to_II more_DAR people_NN named_VVD "_" Rose_NP1 ,_, "_" than_CSN ,_, say_VV0 ,_, younger_JJR men_NN2 )_) ._. 
To_TO maximize_VVI the_AT accuracy_NN1 of_IO our_APPGE estimate_NN1 we_PPIS2 did_VDD four_MC things_NN2 :_: 1_MC1 )_) we_PPIS2 asked_VVD about_II subpopulations_NN2 that_CST have_VH0 high_RR recall_VV0 --_JJ people_NN 's_GE first_MD names_NN2 ,_, 2_MC )_) we_PPIS2 chose_VVD names_NN2 that_CST represent_VV0 between_II 0.1%-0.2%_FO of_IO the_AT population_NN1 --_NN1 subpopulation_NN1 sizes_NN2 that_CST has_VHZ been_VBN found_VVN to_TO minimize_VVI recall_NN1 errors_NN2 &lsqb;_( 5_MC &rsqb;_) ,_, 3_MC )_) we_PPIS2 used_VVD a_AT1 relatively_RR large_JJ number_NN1 of_IO subpopulations_NN2 --_NN1 12_MC unique_JJ names_NN2 ,_, 4_MC )_) and_CC we_PPIS2 selected_VVD a_AT1 balance_NN1 of_IO male_NN1 and_CC female_JJ names_NN2 that_CST were_VBDR popular_JJ at_II different_JJ time_NNT1 periods_NN2 --_NN1 they_PPHS2 roughly_RR balance_VV0 each_PPX221 other_PPX222 out_RP in_II31 terms_II32 of_II33 likelihood_NN1 of_IO exposure_NN1 over_II time_NNT1 and_CC minimize_VV0 any_DD bias_NN1 as_II a_AT1 result_NN1 of_IO age_NN1 and_CC gender_NN1 ._. 
In_RR21 addition_RR22 ,_, mobile_JJ phone_NN1 users_NN2 average_VV0 664_MC ties_NN2 ,_, and_CC those_DD2 who_PNQS have_VH0 internet_NN1 access_NN1 through_II a_AT1 mobile_JJ device_NN1 like_II a_AT1 smartphone_NN1 or_CC tablet_NN1 computer_NN1 tend_VV0 to_TO have_VHI about_RG 717_MC ties_NN2 ._. 
Once_CS we_PPIS2 control_VV0 for_IF demographic_JJ factors_NN2 ,_, most_DAT types_NN2 of_IO technology_NN1 use_NN1 are_VBR not_XX related_VVN to_II having_VHG either_RR a_AT1 larger_JJR or_CC smaller_JJR number_NN1 of_IO overall_JJ social_JJ ties_NN2 (_( see_VV0 Appendix_NN1 C_ZZ1 ,_, Table_NN1 C1_FO ,_, for_IF the_AT regression_NN1 analysis_NN1 )_) ._. 
For_REX21 example_REX22 ,_, LinkedIn_NP1 and_CC Twitter_VV0 users_NN2 have_VH0 more_RGR overall_RR social_JJ ties_NN2 because_II21 of_II22 the_AT demographics_NN1 of_IO their_APPGE users_NN2 ._. 
When_CS we_PPIS2 control_VV0 for_IF demographic_JJ factors_NN2 ,_, we_PPIS2 find_VV0 no_AT difference_NN1 in_II the_AT size_NN1 of_IO people_NN 's_GE overall_JJ networks_NN2 based_VVN on_II which_DDQ SNS_NN2 they_PPHS2 use_VV0 ._. 
LinkedIn_NN1 users_NN2 tend_VV0 to_TO have_VHI more_DAR friends_NN2 because_CS ,_, unlike_JJ most_RGT social_JJ media_NN ,_, they_PPHS2 are_VBR disproportionately_JJ male_NN1 ,_, and_CC they_PPHS2 also_RR tend_VV0 to_TO have_VHI more_DAR years_NNT2 of_IO formal_JJ education_NN1 ._. 
At_II the_AT same_DA time_NNT1 ,_, while_CS Twitter_NN1 users_NN2 are_VBR more_RGR likely_JJ to_TO be_VBI women_NN2 than_CSN users_NN2 of_IO any_DD other_JJ SNS_NN2 ,_, they_PPHS2 are_VBR also_RR disproportionately_RR more_RGR educated_JJ ._. 
As_II a_AT1 result_NN1 ,_, on_II average_JJ Twitter_NN1 users_NN2 tend_VV0 to_TO have_VHI larger_JJR social_JJ networks_NN2 ._. 
Mobile_JJ phone_NN1 use_NN1 and_CC instant_JJ messaging_NN1 users_NN2 are_VBR associated_VVN with_IW having_VHG a_AT1 larger_JJR overall_JJ network_NN1 ._. 
Unlike_II the_AT use_NN1 of_IO specific_JJ SNS_NN2 platforms_NN2 ,_, the_AT use_NN1 of_IO a_AT1 mobile_JJ phone_NN1 and_CC the_AT use_NN1 of_IO instant_JJ messaging_NN1 services_NN2 (_( IM_NP1 )_) are_VBR associated_VVN with_IW having_VHG more_DAR overall_JJ friends_NN2 ,_, even_CS21 when_CS22 we_PPIS2 controlled_VVD for_IF demographic_JJ factors_NN2 ._. 
Mobile_JJ phone_NN1 users_NN2 have_VH0 social_JJ networks_NN2 that_CST are_VBR on_II average_JJ 15%_NNU larger_JJR (_( an_AT1 additional_JJ 73_MC ties_NN2 )_) than_CSN those_DD2 who_PNQS do_VD0 not_XX use_VVI a_AT1 mobile_JJ phone_NN1 ._. 
Those_DD2 who_PNQS use_VV0 instant_JJ message_NN1 tend_VV0 to_TO have_VHI 17%_NNU more_DAR social_JJ ties_NN2 than_CSN those_DD2 without_IW the_AT internet_NN1 and_CC those_DD2 who_PNQS do_VD0 not_XX use_VVI IM_NP1 (_( an_AT1 additional_JJ 85_MC ties_NN2 )_) ._. 
