Modeling Face-to-Face and Internet Based Social Activity Participation Decisions

MISSING IMAGE

Material Information

Title:
Modeling Face-to-Face and Internet Based Social Activity Participation Decisions
Physical Description:
1 online resource (99 p.)
Language:
english
Creator:
Dhakar, Nagendra S.
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Master's ( M.S.)
Degree Grantor:
University of Florida
Degree Disciplines:
Civil Engineering, Civil and Coastal Engineering
Committee Chair:
Srinivasan, Sivaramakrishnan
Committee Members:
Yin, Yafeng
Sampson, William M.

Subjects

Subjects / Keywords:
discrete, ghk, internet, multivariate, probit, social, transportation
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre:
Civil Engineering thesis, M.S.
Electronic Thesis or Dissertation
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )

Notes

Abstract:
The proliferation of internet has altered how people carry out their activities and how they interact with one another. Consequently, travel behavior of people is getting affected. As a result, travel forecasts based on the travel behavior may be inaccurate and misleading. Moreover, transportation plans and policies, either on state or national level, are based on these travel forecasts. Thus, transportation facility investments may be poorly directed too. Therefore, the study of the impacts of the internet on the activity-travel behavior of people continues to be a fertile and an important area of research. In response, this study investigates the relationship between internet based and face-to-face social activity participation. An extensive literature review on impact of internet use is presented. Further, activity participation decisions for internet based social activity and face-to-face social activity are modeled using discrete choice models. The 2000 Bay Area Travel Survey (BATS) data is used in the analysis. Tri-variate probit models are estimated using simulated maximum likelihood framework coupled with smooth recursive simulator, known as GHK. Additionally, effect of various socio-economic characteristics and data time scale factors on social activity participation is investigated. Two definitions of face-to-face social activities are constructed and results are compared. Correlations reveal supplementary effects among forms of social activity. Also, effects of longer duration are found stronger.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (M.S.)--University of Florida, 2009.
Local:
Adviser: Srinivasan, Sivaramakrishnan.
Statement of Responsibility:
by Nagendra Singh Dhakar.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2009
System ID:
UFE0024626:00001


This item has the following downloads:


Full Text

PAGE 1

1

PAGE 2

2

PAGE 3

3

PAGE 4

Ithankmyadvisor,Dr.SivaramakrishnanSrinivasan,AssistantProfessor,DepartmentofCivilandCoastalEngineering,UniversityofFlorida,forhiscontinuoussupportandencouragementthroughoutthedurationofmythesis.Thelearningprocessunderhimhasbeeninvaluable,andIreallyenjoyedeverybrainstormingsessionduringthecourseofthestudy.Iwouldalsoliketothankmycommitteemembers,Dr.YafengYin,AssistantProfessor,andMr.WilliamM.Sampson,Associate-In-Engineeringfortheguidanceandfeedbackonthestudy.Iexpressmydeepsenseofgratitudetomyfamilymembersfortheirperennialmoralsupportandencouragement.IalsorecordmyspecialthankstoallmyfriendsformakingmystayatUFaverymemorableone. 4

PAGE 5

page ACKNOWLEDGMENTS ................................. 4 LISTOFTABLES ..................................... 7 LISTOFFIGURES .................................... 8 ABSTRACT ........................................ 9 CHAPTER 1Introduction ...................................... 10 Background ...................................... 10 InternetUseandSocial-ActivityParticipation ................... 11 ResearchObjective .................................. 13 OrganizationofThesis ................................ 13 2LITERATUREREVIEW .............................. 15 DenitionsandConceptualRelationships ...................... 15 InternetforSocializingandTravelImplications .................. 17 SubstitutionofFFSocialActivities ...................... 18 SupplementEectonPhysicalSocialActivities ............... 19 EmpiricalFindings .................................. 21 GapsinLiterature .................................. 36 SummaryandContributionsoftheStudy ..................... 37 3DATA ......................................... 41 DataCleaning ..................................... 41 DeningInternetBasedandFace-to-FaceSocialActivityParticipation ..... 42 DescriptiveAnalysisofFFandIBSocial-ActivityParticipation ......... 45 OverallShareofActivityParticipation .................... 45 ActivityParticipationWithinaDay ..................... 45 ActivityParticipationAcrossDays ...................... 46 4METHODOLOGY .................................. 57 WhyMultivariateProbit? .............................. 57 MultivariateProbitModel .............................. 58 GHKSimulator ................................. 60 PreviousWorkonMVP ............................... 62 5ANALYSISRESULTS ................................ 64 CorrelatedandUncorrelatedModels ........................ 65 5

PAGE 6

............................ 66 HouseholdSocioEconomicCharacteristics .................. 66 IndividualSocioEconomicCharacteristics .................. 68 ActivityParticipation ............................. 71 DayoftheWeek ................................ 72 MonthoftheYear ............................... 72 Land-useCharacteristics ............................ 73 Weather ..................................... 73 Correlations ................................... 74 SingleDayDataEstimation ............................. 74 Summary ....................................... 75 6SUMMARYANDCONCLUSIONS ......................... 90 SummaryofEmpiricalResults ............................ 91 Socio-economicVariablesandSocialActivityParticipation ......... 91 CorrelationAmongSocialActivities ...................... 93 RecommendationsforFutureWork ......................... 93 REFERENCES ....................................... 95 BIOGRAPHICALSKETCH ................................ 99 6

PAGE 7

Table page 2-1EmpiricalFindings .................................. 39 3-1SampleCharacteristicsofthenalestimationdata(48hour) ........... 48 3-2SampleCharacteristicsofthenalestimationdata(24hour) ........... 50 3-3Shareofsocialactivitiesrecordedbasedontheobjective ............. 52 3-4Primaryactivityissocialthenshareofsecondaryactivities ............ 52 3-5Secondaryactivityissocialthenshareofprimaryactivities ............ 53 3-6ShareofFFandIBactivities ............................ 53 3-7ShareofFFandIBactivities ............................ 54 3-8Socialactivities-48hour .............................. 54 3-9Comparisonofsocialactivitiesover2day(Row%) ................ 55 3-10Comparisonofsocialactivitiesover2day(Column%) .............. 56 5-1Empiricalresultsof48houranalysis:Denition1 ................. 78 5-2Empiricalresultsof48houranalysis:Denition2 ................. 81 5-3Eectofmaleandpresenceofchildinthehouseholdonsocialactivities ..... 84 5-4Correlationsamongsocialactivities(48hour) ................... 84 5-5Empiricalresultsof24houranalysis:Denition1 ................. 85 5-6Empiricalresultsof24houranalysis:Denition2 ................. 87 5-7Correlationsamongsocialactivities(24hour) ................... 89 7

PAGE 8

Figure page 2-1SynchronousPresence(SP)(Source:AdoptedfromYuandShaw(2008)). .... 38 2-2SynchronousTele-presence(ST)andAsynchronousTele-presence(AT) ..... 38 5-1IBSocialActivityParticipationwithAge ..................... 76 5-2OHSocialActivityParticipationwithAge ..................... 76 5-3IHSocialActivityParticipationwithAge ..................... 77 8

PAGE 9

Theproliferationofinternethasalteredhowpeoplecarryouttheiractivitiesandhowtheyinteractwithoneanother.Consequently,travelbehaviorofpeopleisgettingaected.Asaresult,travelforecastsbasedonthetravelbehaviormaybeinaccurateandmisleading.Moreover,transportationplansandpolicies,eitheronstateornationallevel,arebasedonthesetravelforecasts.Thus,transportationfacilityinvestmentsmaybepoorlydirectedtoo.Therefore,thestudyoftheimpactsoftheinternetontheactivity-travelbehaviorofpeoplecontinuestobeafertileandanimportantareaofresearch.Inresponse,thisstudyinvestigatestherelationshipbetweeninternetbasedandface-to-facesocialactivityparticipation.Anextensiveliteraturereviewonimpactofinternetuseispresented.Further,activityparticipationdecisionsforinternetbasedsocialactivityandface-to-facesocialactivityaremodeledusingdiscretechoicemodels.The2000BayAreaTravelSurvey(BATS)dataisusedintheanalysis.Tri-variateprobitmodelsareestimatedusingsimulatedmaximumlikelihoodframeworkcoupledwithsmoothrecursivesimulator,knownasGHK.Additionally,eectofvarioussocio-economiccharacteristicsanddatatimescalefactorsonsocialactivityparticipationisinvestigated.Twodenitionsofface-to-facesocialactivitiesareconstructedandresultsarecompared.Correlationsrevealsupplementaryeectsamongformsofsocialactivity.Also,eectsoflongerdurationarefoundstronger. 9

PAGE 10

ThisproliferationoftheInternethasalteredhowpeoplecarryouttheiractivitiesandhowtheyinteractwithoneanother,causingchangesinthespatial(space)andtemporal(time)distributionofpotentialhumanactivities( YuandShaw 2008 ).AccordingtothePewInternet&AmericanLifeProjectsurveyof2008,intheirfreetime,USadultsinvolveinfollowingcommoninternetactivities:Useasearchenginetondinformation(89%),Checktheweather(80%),Getnews(73%),Visitalocal,stateorfederalgovernmentwebsite(66%),Lookonlinefornewsorinformationaboutpoliticsortheupcomingcampaigns(55%),Watchavideoonavideo-sharingsitelikeYouTubeorGoogleVideo(52%),Lookonlineforinfoaboutajob(47%),Sendinstantmessages(40%),Readsomeoneelse'sonlinejournalorblog(33%),UseanonlinesocialnetworkingsitelikeMySpace,FacebookorLinkedIn.com(29%),Makeadonationtoacharityonline(20%),Downloadapodcastsoyoucanlistentoitorviewitlater(19%),Downloadorsharelesusingpeer-to-peerle-sharingnetworks,suchasBitTorrentorLimeWire(15%),andCreateorworkonyourownonlinejournalorblog(12%). 10

PAGE 11

Mokhtarian 1990 2009 )Specically,itcanserveasasubstitutefortravel(forexample,aworktripbeingreplacedbytelecommuting),generateadditionaltravel(byprovidinginformationaboutpreviouslyunknownactivities),andmodifyexistingtravelpatternsincomplexways(forexample,consolidationofseveralshoppingtripsintooneafterperformingon-lineproduct-comparisons).GiventhegrowingadoptionoftheInternettechnologyandthedesireforcongestionalleviation,thestudyoftheimpactsoftheinternetontheactivity-travelbehaviorofpeoplecontinuestobeafertileandimportantareaofresearch.InternetUseandSocial-ActivityParticipation SrinivasanandBhat 2006 )ontime-useandactivity-travelmodeling,activitiesundertakenbyindividualsareclassiedintothreemajorcategories:subsistence(ormandatory),maintenance,anddiscretionary(orleisure).Mandatoryactivitiesincludethoseundertakenforbiologicalsustenance(suchaseatingandsleeping)andeconomicsustenance(work).Maintenanceactivitiesincludehouseholdandpersonalchores,shopping,personalservices,medicalappointments,andpick-upanddrop-oactivities-theseareactivitiesthatarerequiredforthegeneralpersonalandhouseholdupkeep.Unlikethetwoclassesofactivitiesdescribed,theDiscretionaryactivitiesareconsiderablylessregulatedbyexternalfactors.Someexamplesofdiscretionaryactivitiesincluderecreationalpursuits,socialactivities,relaxing,volunteerandcivicwork. Theinternetcanbeexpectedtohavesignicantimpactsonallthesethreetypesofactivities.Whilepaststudieshasexaminedtheimpactsonmandatory(i.e.,telecommuting-seeforexample, Mokhtarian 1998 )andmaintenance(i.e.,e-shopping-seeforexample, Bhatetal. 2003 )literature(especiallywithinthetransportationengineeringeld)ontheimpactsofInternetonDiscretionaryorLeisureactivities,isminimal( Mokhtarianetal. 11

PAGE 12

).Discretionaryactivitiesmaybeundertakenbyindividualseitherindependently(suchasreadingabook,pursuinghobbies)orjointly.Socialactivities,thefocusofthisresearch,areaspeciccategoryofdiscretionaryactivitiesinwhichsocial(non-work-related)communicationsamongtwoormoreindividualsisoneofthekeyreasonsforundertakingtheactivity.Thus,attendingaparty,familyconversations,talkingwithfriendsoverameal,sociale-mailsanduseofchatprogramsallconstitutesocialactivities.Afurtherdiscussionondeningsocialactivitiesispresentedlaterinthisthesis. Theinternetcanbeexpectedtoinuencesociallife-stylesofpeopleandontherelatedleisureordiscretionaryactivity-travelpatternsinmanyways.E-mailandchatprogramsallowforpeopleto"socialize"withoutalltheinvolvedpersonsbeingphysicallypresentatthesamelocation(thisisreferredtoasInternet-based(IB)socializinginthisthesis).Inthiscontext,itmightbepossibleforpersonstosubstitutesocialtripswithequivalentvirtualoron-linemeetings.Alternately,theconnectivityandinformationprovidedbytheinternetcouldenableindividualstobetterplanandco-ordinatemeetingsandtherebypromoteadditionaltripsforface-to-face(FF)socialactivities.Similarly,thetravel-timesavingsfromparticipationinsomeon-lineactivitiesmaybeinvestedinotherface-to-facesocialepisodesrequiringtravel.Finally,itisalsopossiblethatthevirtualsocializationissoughtwithfriendsandfamilywholivefartherawayand,therefore,doesnotsignicantlyimpacttheday-to-daysocialactivitiesandtravelwhichisundertakenwithlocalfriends/family.Giventheabovediscussion,andtheoverallgrowthininternetusage,theneedtounderstandtherelationshipsbetweenvirtualsocialactivityparticipationandsocialtravelassumesimportancefortransportationplanners. Theextenttowhichindividualssubstitutee-mailsandon-linechattingfordirectsocialcontactsmayalsoinuencethesocial/emotionalhealthofthepopulation.Historically,thedevelopmentsintelecommunicationtechnologies(suchasthetelegraph,thetelephone,andthetelevision)havealwaysraisedconcernsofimpactingthecommunityinnegativewaybyhurtingfamilyrelationships,isolatingpeople,andweakening 12

PAGE 13

BarghandMcKenna 2004 ).SuchconcernsareparticularlystronginthecontextoftheInternetasitcombinespracticallyallofthebreakthroughfeaturesofprevioustechnologies(suchasthetelegraph,thetelephone,andthetelevision)inasinglecommunicationmedium. TheabovediscussionunderscorestheimportanceofunderstandingtheimpactoftheInternetonthesocial-activityparticipationbehaviorofpeople.Fromatransportationperspective,substitutionoftravelforsocializingbyinternet-basedepisodesmightbebenecialintermsofcongestionalleviation,reductioningasolineconsumption,etc.However,fromasociologicalstandpoint,suchtrendsmaynotbedesirabletotheextentthatitweakenshumansocialtiesandaectstheemotionalhealthofthepeople.ResearchObjective Specically,studyinvestigatesrelationshipbetweenIBsocialactivityandFFsocialactivity.Further,eectofsocialactivitydenitiononsocialactivityparticipationisexplored.Also,impactsofvarioussocio-economiccharacteristicsandtimescalefactoronsocialactivityparticipationareexamined.OrganizationofThesis 13

PAGE 14

14

PAGE 15

Thischapterdevelopsconceptualunderstandingoftheactivityparticipationindierentspacesandpresentsanoverviewofthepastresearchthatexaminedtherelationshipbetweeninternet-basedactivityandface-to-faceactivity.Firstsectionprovidesdenitionsoftwospaces-physicalspaceandvirtualspace-andrelationshipbetweenactivitiesinthosetwospaces.Subsequentsectiondiscussesinternetuseforsocialactivitiesandrelatedtravelimpacts.Previousresearchontherelatedtopicissummarizedinthenextsection.Lastsectionprovidesthechaptersummaryandhighlightscontributionsofthestudy.DenitionsandConceptualRelationships Literatureprovidesvariouswaysofdeningsocialactivity. Mokhtarianetal. ( 2006 )discussereasonsofthisuncertaintyinsocialactivitymeasurement.She,however,statesthatmanyactivitieshavecharacteristicsofmorethanonecategoryoutofthreeconventionaltypes;henceitgetsdiculttodenethoseactivitiestobeofparticularkind.Specically,activitiesarebrokenintosmalleractivities,combinedwithfragmentsofotheractivities,andspreadacrossalargenumberoflocations.Furthermore,fragmentsofmultipleactivitiesofdierentkindsmayoverlapbecauseoftheabilityofdoingmultipleactivitiessimultaneously,alsocalledasmultitasking.Therefore,activitiesareinterleavedinoneotheranddiculttoidentifyasofonekind. Socialactivitiescanbeconductedineitherthe"physicalspace"orinthe"virtualspace".Thephysicalspacecanbedenedasthematerialworld.Thevirtualspace,also 15

PAGE 16

Giventhatsocialactivitiesinvolvetheinteractionoftwoormorepersons,therearesignicantdierencesintermsofhowtheseinteractionsmighthappeninphysicalandvirtualspaces.Socialinteractionsinthephysicalspacearegovernedby"couplingconstraints"( Hagerstrand 1970 ).Inotherwords,coincidenceinspaceandtimeisrequiredforsocialinteractioninphysicalspace.Therefore,individualsmayhavetotravelinphysicalspacetoreachacommondestinationatthesametimetoengageinface-to-facesocialinteractions.Astheindividuals'abilitytotraversethephysicalspaceisdictatedbythetransportationsystem,physicalsocialactivityparticipationisstronglyinuencedbythetransportationsystemcharacteristics.Thisideawiththespace-timeprismdiagramisillustratedinFigure 2-1 .Slopeoftheprismistheinverseofspeed,therefore,breadthoftheprismrepresentscongestiononthenetwork.Specically,prismwithhighslope,widerprism,willhavelessspeed,suggestingthatnetworkiscongested.Similarly,thinnerprismindicatesthatnetworkisnotcongested.Reiteratingthattoperformsocialactivitycommunicationbetweentwoormorepeopleisrequired,thusspacetimeprismsfortwopersonsareshowninthegure.Theoverlapoftwopersonsactivitiesinspaceandtimerepresentsopportunityforsocialactivityparticipation.However,socialactivityparticipationisfurtherconditionalonthetypeoflocation.Forexample,iftwopersonsmeetatthefreewaythenitisdiculttomakeastopandcommunicate,thuschancesofperformingsocialactivityareless. Socialactivitiesinthevirtualspacerequiretele-presenceincontrasttothephysicalpresencerequirementofsocialactivitiesinthephysicalspace.Tele-presenceisacommunicationmodethatisdierentfromtheconventionalphysicalpresence,which 16

PAGE 17

JanelleandHodge 2000 ).Withtele-presence,peoplecansenseandinuencetheenvironmenttoaspatialextentfarbeyondtheirphysicalproximity.Thus,socialinteractioninvirtualspacerelaxesthecouplingconstraintimposedbyphysicalinteractions{coincidenceinneitherspacenortimeisrequired.Virtualspacecarriestheowofinformation.Ittakesnegligibletimeforinformationtobetransmittedinthevirtualspaceallowingpeopletoexchangeinformationevenwhentheyoccupydierentphysicallocationsandtherebyenhancingconnectionsbetweenpeoplethroughelectroniclinkages,eg.email,chatrooms,forums,discussionboards,onlinegames(multipleplayer)etc. VirtualspaceissupportedbyICTinfrastructureandfacilitieswhicharepresentinthephysicalspace.TherearespecicplacesinphysicalspacewhereanindividualcanaccesstheInternet(virtualspace).Someofthesephysicalaccesspointsincludehome,work,cybercafs,andlibraries.Althoughthereisarapidgrowthintheextentoftheseaccesspointswiththeimprovementinwirelesstechnology,forthepurposesofthisthesis,wewillfocusonhomeastheprimaryaccesspointtotheinternet. Socialinteractioninthevirtualspacewiththespace-timeprismdiagramisillustratedinFigure 2-2 .FigureshowstwopersonsAandBwhocanconnecttothevirtualspaceatanytimethroughtwoaccesspoints.Ifbothpersonsconnecttothevirtualspaceatthesametime,t1,thensocialcommunicationcanbemadeusingonlinechat,multiplayeronlinegamesetc.Thisiscalledsynchronoustele-presence.Nevertheless,iftwopersonsconnecttovirtualspaceatdierenttimes,t1andt2,thenemail,forumsetc.canbeusedforsocialcommunication.Thisiscalledasynchronoustele-presence.InternetforSocializingandTravelImplications 17

PAGE 18

Mokhtarianetal. 2006 ) Kenyonetal. ( 2002 )introducedanalternativetermforconductingactivitiesoninternet,virtualmobility.Theydenevirtualmobilityas:"'Virtualmobility'isashorthandtermfortheprocessofaccessingactivitiesthattraditionallyrequirephysicalmobility,butwhichcannowbeundertakenwithoutrecoursetophysicaltravelbytheindividualundertakingtheactivity.Thus,virtualmobilitycreatesaccessibilityopportunities,bothsubstitutingforphysicalmobilityandenablingaccesswherepreviouslytherewasanaccessibilitydecit." Therefore,baseonthepreviousliteratureimpactofinternetbasedsocializationonface-to-facesocializationcanbesegregatedintotwocategories:substitutionandsupplement.Followingsubsectionsprovidedetaileddiscussionontheseimpacts.SubstitutionofFFSocialActivities Mokhtarianetal. ( 2006 )highlightthat,alternativewaywillbechosenifthenetutilityoftheinternet-basedformoftheactivityexceedsthatoftheotherforms.Locationandtimeindependence,andfragmentabilityarethecharacteristicswhichmaytendtoincreasetheutilityofinternetbasedsocialactivity.Utilityoftheactivitywillincreaseasthetechnologyimproves;hencethepotentialofinternetbasedsocialactivitytocrowdoutotherformsofsocialactivities.Forexample,forsomepeopleinternetbasedcommunicationisaneasierorcosteectivewaytoconnectwithfriends,familyorrelatives,hencesubstitutingtheneedofphysicalpresence. Theinternetalsooersvarietyofopportunitiesfornewactivitieswhichwouldnothaveoccurredotherwise.Forexample,playonlinegame,involveindiscussionswith 18

PAGE 19

Mokhtarianetal. ( 2006 )furtherdescribethenatureoftheseactivitiesingeneraltermsas,"althoughthedisplacementmaybeimmediate,aswhenanindividualdecidesataparticularmomenttospendtimeonanICT-basedactivityratherthansomeotheractivity,itcanalsooccuroverlongerperiodsoftimeandmoresubconsciouslythanconsciously." Thissubstitutionofphysicalactivitiescanhavetwotravelimplications,rstreductionintravel,andsecond,noreductionintravel.Reductionintravelispossibleifinternetbasedsocialactivitydirectlysubstitutesthecounterpartphysicalsocialactivity,henceeliminatingtheneedoftravelandsavingtime. Lyons ( 2002 )emphasizestheimportanceofthissubstitutioneectandmentionsthatreductioninabsoluteleveloftripmakingbythissubstitutioneectistheuppermostdesiredconsequencefromtheperspectiveofthetransportplannerandpolicymaker.SupplementEectonPhysicalSocialActivities 19

PAGE 20

Lyons ( 2002 ),however,wondersaboutthecontributionofvirtualmobilityinsavingthetraveltimeandaskswhethernewlygeneratedtripsareundertakenbycar. Nevertheless,IBsocializationmaynothaveanyimpactatallonphysicalactivityparticipation.Theinternetcanactasanothermediumtointeractwithfamily,friendsorrelatives.Thisimpactdoesnotaecttravel. Variousstudieshaveinvestigatedthefactorsinuencepeople'schoicebetween'physicalspaceactivities'andtheirsubstitutionwith'virtualspaceactivities'andconcludedthattwofactorsplayimportantrolesinthisdecisionmakingprocess-thegeneralizedcostofreachingamenities,andservicesandthequalityorattractivenessofthoseamenitiesandservices.Thetermaccessibility(afunctionofthetwofactors)canthenbeusedtorepresentthenetappealtoanindividual.Generalizedcostrefersto"ameasurecombiningallthemainattributesrelatedtothedisutilityofajourney"( deDiosOrtuzarandWillumsen 2002 ).Itincludestraveltimeandmonetarycostaswellassafety,comfort,convenience,etc. AvirtualtripontheInternetisdrasticallysmallercomparedtothetimespentintravellingorwaitingintracjams( Lyons 2002 ).Themonetarycosttendsnottobeconcernedwithgettingtothe'destination'butwithparticipatingininformationexchangeatthedestinationorwebsiteforexample.(Infactwhenonereferstovisitingawebsitewhatactuallyhappensisthatthepagesfromthatsitearesenttotheindividual-i.e.theactivitycomestotheindividualratherthantheindividualgoingtotheactivity.)ThemonetarycostwhenitexistsfortheindividualcaneitherbeintheformofaxedcostsuchasamonthlysubscriptionforInternetaccessoravariablecostwheremeteredaccesschargestheuserforthetimespentontheInternet.( Kenyonetal. 2002 ) 20

PAGE 21

Lyons 2002 ). Intermsofgeneralizedcostvirtualaccessibilityislikelytowinpreferenceoverphysicalmobility.But,thisraisesaninterestingquestion,oftowhatextenttheattractivenessofactivitiesaccessedvirtuallycomparesfavorablywiththatofequivalentactivitiesaccessedphysicallyandtowhatextentlowergeneralisedcostofvirtualaccesscanbetradedoagainstprobablerelativedecienciesofvirtualactivityattractiveness( Lyons 2002 ). Findingsofthepaststudiesthatinvestigatedimpactofinternetuseonphysicalsocialactivitiesaresummarizedinthenextsection.EmpiricalFindings 2-1 presentssummaryofthestudiesthathaveanalyzedtheimpactofinternetuseonphysicalsocialactivities.Informationaboutdatacollectionyear,surveyarea,samplerestrictions,ifany,andsamplesizeforeachstudyareprovidedinthetable. Krautetal. ( 1998 )wereoneoftherststoexploretherelationbetweeninternetandphysicalsocialactivitiesandindicatedasubstitutioneectofinternetuseonphysicalsocialactivities.TheyanalyzedWeeklyinternetusedataoftheparticipantswhowereintheir1or2yearsofonline.Fourmeasuresofsociallifewereused;familycommunication(timespenteachdayoncommunicationwithfamilymembers),sizeoflocalnetwork(numberoffriendsinPittsburghareaindividualsocializewithatleastonceamonth),sizeofdistantsocialnetwork(numberoffriendsoutsidePittsburghareaindividualsocializewithatleastonceayear),andsocialsupport(howeasyitistogettangiblehelp,advice,emotionalsupport,andcompanionship,andhowmuchtheygetsenseof 21

PAGE 22

Thisstudywasfollowedbynumerousstudieswhofocusedonthisissue,however,onlyfewcouldoersupporttothendingsof Krautetal. ( 1998 ).Thosestudiesinclude, NieandHillygus ( 2002 ), Nieetal. ( 2002 ),and Nieetal. ( 2005 ).AlloftheseusedSIQSS(StanfordInstituteforthequantitativeStudyandsociety)studydata,whichwascollectedforAmericanswhomaintaineddetailedactivity-diary.Dataincludedactivityrecordsofrespondentsforsixrandomlyselectedhoursfrom6timeblocks(onehourfromeachblock)onthepreviousday("Yesterday"). NieandHillygus ( 2002 )segregatedactivitiesoninternetintofollowing:work,education,socialtime,andothers.Theyusedfourmeasuresforsociability:(1)timespentactivelyengagingorparticipatinginanactivitywithfriends,(2)timespentactivelyengagingorparticipatinginanactivitywithfamilyand(3)timespentonsocializingactivities(visiting,parties,etc.).Bivariatecomparisonofinternetusersandnonusersshowedlesstimebyinternetuserswithfriends,withfamilyandonphysicalsocialactivities.Further,multivariateregressionanalysiswasperformedbycontrollingformaritalstatus,gender,age,education,race/ethnicity,numberofchildren,singleparenthoodandlivingalone.Additionally,internetusewasdierentiatedbasedonthelocation(athome,andatwork).Resultsindicatednegativerelationbetweeninternetuseandtimespentwithfriends,familyandonsocialactivities.Homewasemergedasthecriticalplacewheredirecttrade-osbetweenonlineandface-to-facecommunicationwasfacedbytheindividual.Moreover,thisnegativerelationwasfoundstrongonweekenddaysthanweekdays. 22

PAGE 23

Nieetal. ( 2002 ),selfassessmentquestionsofinternetimpactonrespondent'ssocialliveswereused:howuseofinternetaectedthetimeyouspendwithyourfamily,friends,andattendingevents?(Responseoptions:increased,unchanged,ordecreased).Responsestothesequestionswereanalyzedagainsttheweeklyhoursoninternetanditwasfoundthattimeoninternetisnegativelyassociatedwithsocialcontacts.Inotherwords,lesstimewasspentwithrealhumanbeingsastimeoninternetgrew. Niewasinvolvedinanotherstudytoo, Nieetal. ( 2005 ),wheretheysimplyusedtimespentoninternetactivitiesasmeasureforinternetuse.Theyperformedregressionanalysisofactivetimewithfamiliesandfriendsandconcludedthattimespentoninternetcomesfromtimespentwithfamily,timespentwithfriends. Restofthestudiesdonotsupportthedisplacementhypothesisconcludedbyabovestudiesandindicatedmixedimpacts.Ingeneral,timespentonactivitiesbyinternetusersandnonuserswerecompared.Additiontothis, LeeandZhu ( 2002 ), Mikami ( 2002 ), Choi ( 2008 ),and Smithetal. ( 2008 )usedworldinternetprojectsurveydatatoexploretheimpact. LeeandZhu ( 2002 )studiedtherelationbetweeninternetuseandsociabilitywithtwodatasets,collectedattheendofyear2000.Firstdatasetdrawnfromtwocitiesinmainlandchina(BeijingandGuangzhou)andwascollectedfromface-to-faceinterviewsof2500adults.Thesecond,telephonicinterviewsof1007adults,includesresidentsinHongKong.ThesetwodatasetswerecomparedwithinthemselvesandalsowithUSdatafromtheUCLAstudy.Twosociabilitymeasurementswerecreatedforbothdatasets:(1)theamountoftimespenttalkingwithfamilymembers,and(2)theamountoftimespentonsocializingwithfriendsandrelatives.Interactionwithfamilymemberswasfurtherassessedforminganinteractionscalebycombiningfouritems:no.ofdaysperweekspenttogetherwithfamilymembers(a)dining,(b)watchingTV,(c)playing,or(d)shopping.Sixindependentvariables,age,sex,education,familyincome,marital 23

PAGE 24

Mikami ( 2002 )usedtwo-wavepanelsurveydataofJapancollectedinyears2000and2001.Datafromrstsurveyincluded2555individualsaged12to74.1346individualsoftherstsurveywith1470additionalindividualsaged12to75wereinterviewedinsecondsurveyconductedin2001.However,respondentswithageover18yearswereincludedintheanalysis.Sociabilitywasmeasuredbasedontwoquestions:(1)timespentsocializingwithothermembersofone'shouseholdorfriends,and(2)respondent'sownperceptionoftheinternet'sinuenceontheirsocializingbehaviorwithfamilyandfriends.Additionally,thisdatawascomparedwithUCLAandchina-HongKongdata.CrosssectionalcomparitiveanalysisandMCAmethodanalysisproducedresultssimilarto LeeandZhu ( 2002 ).AdditionalMCAanalysisfurthersuggested"virtualcommunity"eect;higherleveloffriendsocializingforheavierinternetusers. Choi ( 2008 )comparedinternetusersandnonusersbyusingSingaporeInternetProject(SIP)survey2007.Duringsurvey,questionswereaskedtorespondentsabouttheirperceptionofinternetimpactsontheirface-to-faceinteractionswithfamily,friends,orcolleagues.Analysisofresponsesindicatethatinternetusersspendmoretimesocializingface-to-facewithfriendsoutsideofschool/ocehoursandspendlesstimesocializingwithfamily.Choiexplainedthepatternasthepossibleresultofdierentdemographiccharacteristicsofusersandnon-users.Analysisfurthershowedthatabouttwo-thirdoftherespondent'sface-to-faceinteractionswerenotaectedbyinternet,andrestreportedimpactonbothsidesequally-halfsayitincreasedandotherhalffeelitdecreasedtheface-to-faceinteractions. 24

PAGE 25

( 2008 )analyzedanationalprobabilitysamplecollectedthroughtelephonicinterviewinNewZealand.Resultsweredrawnfromthecrosssectionalanalysisbetweendependentvariablesandindependentvariables(gender,age,ethnicity,urban/ruralarea,householdincomeanduserstatus).Duringinterviewtwoquestionswereaskedaboutimpactofinternetonthesociallife:(1)sincebeingconnectedtotheinternetathome,themembersofhouseholdhasspentface-to-facetimetogether(Responseoptions:more,less,orsame).(2)Sincebeingconnectedtotheinternet,howmuchtimewasspentwithfriends(Responseoptions:more,less,orsame).Signicantnumberofrespondentsreporteddecreaseintheirface-to-facetimewithfriendsandfamily.However,asignicantsmallnumberofNewZealandersreportedtoextendtheironlinecontactintotheoineworld. Studiesthatrecordedinternetusebasedonthelastweekusageinclude, Krautetal. ( 1998 ), Kestnbaumetal. ( 2002 ), Pronovost ( 2002 )and Qiuetal. ( 2002 ). Kestnbaumetal. ( 2002 )analyzedthe1998-2001probabilitysurveyof1775respondentsaged18-64.Respondentsusedtimediariestokeeprecordoftheiractivitiesforasingleday.Sociallifewasmeasuredasoneofthefreetimeactivities(othersare:adulteducation,organizationalactivity,recreation,andcommunication).Internetusewasmeasuredasthetimespentoninternetactivities.ControlsfordemographicdierenceswereappliedusingMCA(MultipleClassicationAnalysis)technique.MajorindependentvariableInternetusewasappliedintwoways:singledayandweekly.Forasingleday,timespentoninternetactivitieswasdirectlytakenfromsingledayactivityrecordsreportedbyrespondents.Secondly,forweekitwasmeasuredthroughthedirectquestionaskedtotherespondentsaboutthetimespentontheinternetinaweek.Comparisonsof"yesterday"freetimeactivitiessuggestedslightdecreaseinsocializingandvisitingpeopleoutsidethehome.Onthecontrary,interactionwithinhomewasfoundincreased.Furthercomparisonofinternetusersandnonusersbyusingweeklyinternetusetimesuggestedthatinternetusersspentslightlyhighertimeinbothin-homeandawaysocializing.However,nosignicantdierenceswere 25

PAGE 26

Pronovost ( 2002 )usedsingledayactivityrecordsof10,749personsaged15yearsandoverinCanada.Singledayactivityinformationwascollectedthroughtelephonicinterviewsduring1998-2001andwasthenevenlydistributedoveraweek.Householdswithnotelephoneathomewereexcludedbutdatawasweightedtoaccountforthesehouseholds.However,intheanalysisonlypersonswithaged18-64wereconsidered.Socialactivitiesweremeasuredintermsofsevencategories:Religion,Organizations,socialevents,social/visit,conversation,othersocial,andeatout.Analysiswasperformedfortwotimescales:UsageYesterday(singledayinternetuse),andUsageGenerally(includesthosewhoreportedlong-termusageofinternet).Internetuserswerefoundlessactiveinvisitingand"other"socialactivityevenafterMCAadjustmentsfordemographicvariables(sex,age,children,educationanddayoftheweek).Moreover,Intermsoftotalsocialtime,internetuserswerelesssocialactive.Inusagegenerallyanalysis,resultsshowedlittleevidencefortheinternetdisplacementeectonsociallife.Further,analysisresultsof"withwhom"entriesshowedreductionintimewithfriends. Qiuetal. ( 2002 )comparedtimespentonactivitiesbyinternetusersandnonusersusingthenationalsamplecollectedinyear2001.Thesamplecontainsweeklyrecordofactivitiesbymorethan400workingclassfamilies.Socialactivitiesweremeasuredasthetimespentonactivities,religion,organizations,socialevents,social/visit,conversation/home.Studydidnotndstrongevidencesofconstrainedsociallifeintermsofvisits,meetings,attendingsocialeventsorconversations.However,itwasobservedthatITusersdonotspendlesstimewithfamily,friendsorothers. Alongwithinternetusagelastweek, ZhuandHe ( 2002 )usedinternetadoptionhistoryandmagnitudeofweeklyuseintheiranalysis.Theyanalyzedthesampleof1,007adultsofagebetween18and74inHongKong.Samplewascollectedthroughtelephonicinterview.Amountofonlinetimerepresentedweeklytimespentonfollowingactivities: 26

PAGE 27

Contrasttoabovestudieswhichusedcrosssectionaldata,followinganalyzedpaneldata: LeeandKuo ( 2002 ), Gershuny ( 2002 ), Mikami ( 2002 ), ColeandRobinson ( 2002 ),and Kiesleretal. ( 2002 ). LeeandKuo ( 2002 )usedpaneldataofstudentsfromsecondaryschoolsinSingapore.Firstdata,collectedin1999,included1251students.817studentsofthoseweresurveyedinsubsequentyear2000andwereusedfortheanalysis.Amongthese619studentswereinternetusersinbothyears,43studentsgaveupusinginternet,and155becamenewusers.Weeklytimespentontheinternetwasreportedthroughself-administeredquestionnaires.Further,estimateofweeklytimespentonplayingsportsandexercising,interactingwithfamily,andsocializingwithclosefriendsface-to-faceoutsideofschoolwasprovidedbytherespondentsandwasusedasmeasureforphysicalsocialactivity.Regressionanalysisresultsshowednosignicantchangeinfamilyinteraction,however,increaseinsocializationwithfriendswasobserved. 27

PAGE 28

( 2002 )analyzedtwowavesofanationaltimediaryhouseholdpanelstudy,theHome-on-Line(HoL)study.Intherstwavesurvey,conductedin1999,1109personswithageover10keptweeklyrecordoftheiractivities.Inthesecondwavesurvey,740personsoutofthesamegroupagainprovidedrecordoftheiractivitiesin2000.However,nalestimationdataincludedonlypersonsoverage16.Physicalsocialactivitiesweremeasuredasfollowing:voluntarywork,church,helppeople(notownhome),goingtoconcert,theatres,cinema,sports,clubs,Eatingout,drinking(pubs,restaurants),visiting(ormeeting)friendsorrelatives,beingvisitedbyfriends/relativesinownhome,receivingtelephonecalls,makingtelephonecalls.SimplecomparisonofITusersandnonusersfromtwowavepaneldatadidnotindicatesignicantdierenceinthetimespentonsocialactivities.However,itindicatedthatsocialactivitiesoseteachothertonullifytheeect.Specically,usersweremoreactiveineatingoutanddrinkactivities,butspentlesstimeonvisiting.Further,comparisonofnewusers,whowerenonusersin1999,didnotshowanylossofsociallife.Although,newusersspentmoretimeongoingoutandvisits/beingvisiting.Besides,womenwerefoundtobemoreactivethanmanineatingout. ColeandRobinson ( 2002 )employedMCAtechniquetoanalyzetwodatasetsfromUCLAsurveys.Firstdatawascollectedinyear2000bytelephonicinterviewsof2096individuals.1300individualsoutofthesealongwith700newrespondentswereinterviewednextyear2001.Threesociabilityindicatorswereused:(1)timespentsocializedwithothermembersofhouseholdsorfriends,and(2)thenumberoffriendsandneighborsoneknew,and(3)timespentparticipatinginclubsorvoluntaryorganizations.Nodierencebetweenusersandnonuserswerefoundintermsofsocializing. Kiesleretal. ( 2002 )didnotexactlyusepaneldatabutanalyzedtwosamplestogether.First,collectedin1995to1996,wasthefollow-upstudyoftheoriginalHomeNetstudyandincluded208membersfrom93families.Secondwasfromthesurveyundertakenin1998to1999inPittsburghareaandinvolvedpeoplewhohadpurchasedanewcomputerortelevisionsetinlast6months.Resultsoftherstdataoerednosupport 28

PAGE 29

Although,mostoftheabovestudiesexaminedtheimpactoftheinternetuseasawhole,fewstudieslookedspecicallyattheimpactofinternetuseforcommunicationonphysicalactivities.Thosestudiesinclude, Wellmanetal. ( 2001 ), Cogetetal. ( 2002 ), NeustadtlandRobinson ( 2002 ), Baymetal. ( 2004 ),and Shklovskietal. ( 2004 ). Wellmanetal. ( 2001 )usedNationalGeographicSocietySurvey2000.Estimationsampleconsiststotalof39211NorthAmericanAdults(34,839Americansand4,372Canadians).Onlyinternetuserswereincludedinthesurveyandfollowingactivitiesweredenedasinternetbasedsocialactivities:chat,email,playingmultiusergame,visitingmultiuserdimensions,andotherroleplayingenvironments.Crosssectionalanalysisofemailusewithface-to-facetimewithfriends,andfamilyrevealedthatwithmoreinternetusepeoplemademorefrequentcontactswithfriendsandrelatives.However,thiseectmaybetheresultsofthecommunicationfacilityinternetprovidesinadditiontothetraditionalcommunicationmediums(face-to-face,telephone).Furtherresultsindicatedthatpeoplealsovisitfriendsandfamily,whichconrmsthattraditionalway 29

PAGE 30

Cogetetal. ( 2002 )concludedthatinternetdoesnottaketimeawayfromtheface-to-facecommunication,thus,didnotoersupporttothetimedisplacementhypothesisoftheinternet.Foranalysis,theyusedtherstyearoftheUCLAinternetprojectsampleof2096Americanssurveyedin2000.Datawascollectedbytelephonicinterviewandhasbothinternetusersandnonusersage12andabove.Threemeasuresweretakenforinternetuse:(1)internetuserornonuser,(2)timespentoninternet,and(3)internetexperience.Face-to-facesocializingwasmeasuredthroughtwoitems:(1)numberoffriendsoutsidehouseholdrespondentssaworspoketoeachweek,and(2)timespentwiththem.Tworegressionanalyzeswereperformed.Intherstregressionanalysis,completesamplewasused,andbothinternetusersandnonuserswereusedasthemainindependentvariables.Secondregressionanalysiswasperformedonthesubsampleonlywithinternetusers. NeustadtlandRobinson ( 2002 )refusedtoacceptinternetasbadeectonsociallife,andassociatedinternetwithgreatersociallife.TheyanalyzedtheGeneralSocialSurvey(GSS),apersonalin-homeinterviewsurvey,conductedinyear2000.Dataisnationalprobabilitysampleof2,817adultsofage18andolder.Twomeasuresofinternetusewereused:(1)weeklytimespentontheinternettosendorreceiveelectronicmail.(2)WeeklytimespentonWorldWideWebotherthanelectronicmail.Foursetsoftraditionalsocialquestionswereaskedtoanindividual:frequencyofsocialinteractionswithrelatives,neighbors,friends,andatbars.Annualaverageofthefrequencywascalculatedbyusingtheweightsoffrequencyscaleprovidedtotherespondents.Newchannelquestionswerealsoasked:thetotalnumberoffriendsandrelativescontactedatleastonceayearvia 30

PAGE 31

Similartotheauthorsofabovetwostudies, Baymetal. ( 2004 )alsosuggestednothreattoface-to-facecommunicationfrominternetcommunicationandcalledinternetassupplementtoface-to-facecommunication.Analysiswasperformedfortwostudiesofcollegestudents,whosocializeoverinternet,inUS.Study1wasadiarystudyof51studentsattwoMidwesternuniversities.Participantswereaskedtokeepinteractiondiariesandrecordeach"signicantvoluntarysocialinteraction"theyinvolvedinnext3-5daysperiod.Itwaslefttoparticipantstodetermine"signicantsocialinteraction".Theyalsorespondedtothequestionaboutthefrequencyof"signicantvoluntarysocialinteraction"throughdierentmediums:face-to-face,andinternet.Responsesshowedthatface-to-facecommunicationisstillthemostdominantcommunicationmode.Study2,asurvey,involved496studentsfromMidwesternuniversities.Inthisstudy,participantswereaskedtoreportabouttheirmostsignicantsocialinteractions.Medium(face-to-face,telephone,internet)ofinteractionandtypeoftherelationships(acquaintances,friends,familymembers,andpartners)werecontrolled.Therefore,12versionsofthesurveywereformed;eachparticipantwasgivenonlyoneversion.Overall,versionsofinteractionsthrougheachmediumwereequallydistributedamongstudents.Participantsalsoanswered 31

PAGE 32

Shklovskietal. ( 2004 )performedcross-sectionalandlongitudinalanalysisofthepaneldata.Firstsurveywasconductedinyear2000andconsisted2533adultsage18orolder.1501adultsoftherstsurveycompletedfollow-upsurveyinyear2001.Inbothsurveys,respondentswereaskedtheirperceivedavailabilityofsocialsupportbyspecifyingnumberofpeopletheycangotoforthehelp(options:many,justafew,orhardlyany).Theywerefurtheraskedabouttheirsocialactivityparticipation,suchaswhethertheypaidvisittofriendsorfamilymembersyesterday(options:yesorno).Threemeasuresofinternetusagewereadoptedinthestudy:breadth(rangeofactivities),frequency(howmanytimes),andhistory(forhowlongusinginternet).Extroversionvariable,askedonlyinyear2001,wasalsousedintheanalysis.Gender,age,levelofeducation,andracewereusedasindependentvariables.CrosssectionalanalysiswasundertakenusingOrdinaryLeastSquare(OLS)Regressionmodels,wherelongitudinalanalysisusedHierarchicalLinearGrowth(HLG)models.IncontrasttoOLSregressionmodel,HLGmodelassumesthatresponsesarenotindependent.Crosssectionalanalysisresultsshowedthatbreadthandfrequencyofinternetusewerepositivelycorrelatedwiththefrequencyofgoingouttodinnerwithfamilyandfriends.Internetusefrequencywasnotfoundtobeassociatedwiththefrequencyofvisitingwithfamilyandfriends,doingvoluntarywork,attendingreligiousservices,orlikelihoodofvisitingafriendorfamilymember.However,womenweremoreinvolvedinsocialactivitiesthanmenandolderadultswerelessactiveinvisitingthanyoungeradults.Forextroverts,internetusewasnegativelyassociatedwithdiningout.Incontrast,longitudinalanalysisconcludesthatbreadthwasnegativelyassociatedwiththevisitfriendsorrelatives.Additionally,withbreadthoftheinternetuselikelihoodofvisitwaslessinyear2000thaninyear2001.Moreover,extrovertswerefoundtohavemoredeclineinvisitingwithmorebreadthofinternetuse.Authoralsoexaminedwhether 32

PAGE 33

Mostlystudiesareconductedbymeasuringphysicalsocializationintermsoftimespentonactivities.However, Wellmanetal. ( 2001 ), Williams ( 2007 ),and Robinsonetal. ( 2000 )adoptedotherformsofmeasurementforphysicalsocialization. Williams ( 2007 )communicationoverinternetcanhelpincreasingyourrealworldrelations.Heanalyzedanationalrepresentativesampleof884Americanscollectedthroughweb-basedsurvey.Internetusewasmeasuredasweeklytimespentusinginternetoremailotherthanwork.Individualslocalsupportnetworkwasdeterminedbysummingthestandardfeelingthermometersforindividual'seachofsixclosestfriends.Additionally,bridgingandbondingwereusedtomeasuretobothonlineandoinesocialcapital.Resultsfromthestepwiseregressionsupportthehypothesisthatinternetuseisnegativelyassociatedwithoinesocialcapital.Evenaftercontrollingfordemographicvariables,internetusehasnegativeeectonbothoinebridgingandbonding.Extrovertsandstrongfriendshipnetworkpeoplewerebenetedinbothoinesocialcapitals.Negativeassociationbetweeninternetuseandsocialcapitalwasobservedforwomen.Furthermore,theconverseeectwasobservedformen.Additionally,analysisshowedthatbridgingsocialcapitalwasmoreonlinethanoine. Robinsonetal. ( 2000 )usedasampleof3993individuals,whichconsisted2000respondentsofage18orolderfromthePewCentre's1998newtechnologysurveyandanother1993onlineusers.SamplewasweightedtorepresenttheentireUSpopulation.Singleday("yesterday")activityinformationwasrecordedthroughsurveyapproachsimilartothetimediary.Timespentyesterdayoninternetwasalsorecordedinthe 33

PAGE 34

Fromthetransportationpointofview,impactofinternetontripmakingisimportant,whichisnotdiscussedinabovestudies.However, SrinivasanandReddyAthuru ( 2004 ),and TonnandHemrick ( 2004 )associatedtheimpacttointernettoindividual'stripbehavior. SrinivasanandReddyAthuru ( 2004 )usedthe2000BayAreaTravelSurvey(BATS2000)toexaminethetrade-osbetweenphysicalandvirtualactivitieswithinaday.BATSdatacontainstwodayactivityrecords.Physicalactivitieswerefurthersegregatedintotwocategories,discretionary(recreationalandsocialactivities,relaxing,volunteerorcivicwork,andnon-worknon-shopinternetuseactivities)andmaintenance(householdandpersonalchores,shopping,personalservices,appointmentsduetosicknessorothermedicalissues,andpick-upanddrop-oactivities).Additionally,physicalactivitiesaredenedin-homeaswellasout-ofhome.Virtualactivitywereconsideredtobeundertakenathomeonly.Thereforethreedependentvariables,in-homevirtualactivity,in-homephysicalactivity,andout-of-homephysicalactivityweremodeledusingbinarylogitmodelsandestimatedwithmaximumlikelihoodframework.Itwasassumedthatvirtualactivityparticipationdecisionsacrossdierentpurposesweremutuallyexclusive.Followingexogenousvariablesusedinthemodels:ICT,person,worker,householdrelatedvariablesandalsotimeofdayanddayofweekrelatedinformation.ResultsconcludedthatthesubstitutionofphysicalsocialactivitywithvirtualactivitymaybeencouragedbytheavailabilityandtheneedforICTresourcesathome.Additionaltravelisexpectedwithincreasingnumberofvehiclesinthehousehold.Moreover,substitutionoftravelwithvirtualactivitieswasobservedforprofessionalsandexecutives(maintenancerelated),full 34

PAGE 35

TonnandHemrick ( 2004 )performedanalysisonthedatafromwebbasedsurveyinKnoxville,Tennessee,metropolitanregion.Surveywasconductedinyear2002andcompletedby118respondentswhowereusersofemailand/orinternet(itwaslefttotherespondentsdecidewhethertheyareeligiblewiththiscriteriaornot).Questionsaskedinthesurveywerecategorizedinfour:(1)personalemailandinternetuse,(2)personaltripmakingbehavior,(3)eectsofe-mail/internetuseonpersonaltripmakingbehavior(substitutionorcomplementary),and(d)demographic.Responsestothequestion,howoftenemailand/orinternetwasusedintheplaceoftripsduringthe2weekspriortothesurvey,indicatedtripsubstitutionforactivities(triptowork,bookstores,otherstores,andlibraries,triptoclothingstores,musicstores,friend'shouse,governmentocesandspecialinterestsorganizations).However,internetdidnotndtohavemuchimpactongroceryshopping,school,medicalcare,church,entertainment,orvisitingrelatives.Studyalsoinvestigatedifinternetgeneratesnewtripsanditwasfoundthatinternetwasusedtocommunicatewithcoworkers,friends,andfellowmembersofspecialinterestorganizationswhichresultedintrips.Overall,tripreductioneectwasfoundhigherinmagnitudethanthetripgenerationeect.Thus,resultsestablishedemailorinternetasmeansofreducingtracduringrushhoursandoverall.Further,tworegressionanalyzeswereperformedtoinvestigatetheimpactofinternetusefordierentdemographiccharacteristics.Firstregressionmodelusedfrequencyofemailandtheinternetusetosubstitutenon-workandnon-schooltripsasdependentvariable.Ageandnumberofaccesspointwerefoundtobepositivelyassociatedwithtripsubstitution.Moreover,educationwasnegativelyassociatedwithtripsubstitution.Otherindependentvariables,suchasgender,race,internetexperience,householdincome,workingfulltime,workingatall,presenceofyoung 35

PAGE 36

Becauseofthedatalimitationsandmeasurementissueswithsocialactivity,variousassumptionshavebeenadoptedinpreviousstudies.Thoseassumptionscanfurtherberelaxedandtakenupfornewresearch.Followingarefewofthoseassumptions(orgaps)identiedintheliterature: Shklovskietal. 2004 ). 36

PAGE 37

ItisimportanttonotethattheBATSdatausedintheanalysiswascollectedinyear2000,thusdoesnotreectthepresentinternetusage.Sinceyear2000,signicantdevelopmentshavebeenobservedintermsoftechnologyandinternetbasedactivities.Variousnewdimensionshavebeenexploredandcreatedforsocialpurposes,forexample,socialnetworkingwebsites(facebook,orkut,myspaceetc.),blogs,discussionforumsetc.Furthermore,duringthatperioddial-upwasmorecommonlyusedforinternetaccess,whichrequireddisconnectingphoneline.However,presentlynumerousmethodsareavailable,suchas,landlinebroadband,wi-,satellite,and3-Gtechnologycellphones.Therefore,activityparticipationbehaviorobservedinthestudymaybecompletelydierentthanthepresentbehavior. 37

PAGE 38

SynchronousPresence(SP)(Source:AdoptedfromYuandShaw(2008)). SynchronousTele-presence(ST)andAsynchronousTele-presence(AT) 38

PAGE 39

EmpiricalFindings No. CitationDataYearAreaSampleCharacteristics 1 ( Krautetal. 1998 )1997Pennsylvania,US169personsfrom73householdsAge10yearsorabove2 ( NieandHillygus 2002 )NAUS6000persons,Agebetween18-64years3 ( Nieetal. 2002 )1999US4113personsfrom2689households4 ( Nieetal. 2005 )2005US4,384peopleAgebetween18-64years5 ( LeeandZhu 2002 )2000MainlandChina&HongKong1st:2500adults2nd:1007adultsAgebetween18-74years6 ( Mikami 2002 )2000&2001Japan1st:2555persons2nd:2816personsAge18yearsorabove7 ( Choi 2008 )2007Singapore884personsAge13yearsandabove8 ( Smithetal. 2008 )2007NewZealand1430personsAge16yearsorabove9 ( Kestnbaumetal. 2002 )2000-2001US1775persons18-6410 ( Pronovost 2002 )1998-2001Canada10,749persons,Agebetween18-64years11 ( Qiuetal. 2002 )2001USMorethan400familiesAge18yearsorabove12 ( ZhuandHe 2002 )2000China1007adultsAgebetween18-74years13 ( LeeandKuo 2002 )1999&2000Singapore1st:1251secondaryschoolstudents2nd:817students14 ( Gershuny 2002 )1999&2001UK1st:1109persons2nd:740personsAge16yearsorabove15 ( ColeandRobinson 2002 )2000&2001US1st:20962nd:2000Agebetween18-74years16 ( Kiesleretal. 2002 )1995-96&1998-99 Pittsburgh,US1st:208membersof93families2nd:446membersof216households10orabove17 ( Wellmanetal. 2001 )2000US&Canada34839Americans&4372CanadiansAge18yearsorabove 39

PAGE 40

2-1 .Continued No. CitationDataYearAreaSampleCharacteristics 18 ( Cogetetal. 2002 )2000US2096personsAge12yearsorabove19 ( NeustadtlandRobinson 2002 )2000US2817individualsAge18yearsorabove20 ( Baymetal. 2004 )-US1st:51students2nd:496studentscollegestudents21 ( Shklovskietal. 2004 )2000&2001US1st:2533adults2nd:1501adultsAge18yearsorabove22 ( Williams 2007 )-US884persons23 ( Robinsonetal. 2000 )1998US3993individualsAge18yearsorabove24 ( SrinivasanandReddyAthuru 2004 )2000BayArea,US4214respondentsAge14yearsorabove25 ( TonnandHemrick 2004 )2002-03Tennessee,US118people 40

PAGE 41

Theprimarysourceofdatausedinthisresearchisthe2000BayAreaTravelSurvey(BATS).ThissurveywasconductedbyMORPACEInternational,Inc.,(Source-SurveyDocuments)onthebehalfofBayAreaMetropolitanTransportationCommission(MTC).Over34,000individualsfromapproximately14,500householdstookpartinthistime-diarysurveyandprovidedadetailedaccountoftheiractivityandtravelpatternsforatwo-dayperiod.Therestofthischapterprovidesanoverviewofthisdatasetanddescribesthedataassemblyprocedure.Firstsectiondescribestheelementsofthesurveyrelevanttothisresearchandoutlinesthedata-cleaningprocedure.NextsectionaddressestheissueofidentifyingFace-to-Face(FF)andInternetBased(IB)social-activityparticipationsfromthesurveydata.Finalsectionpresentsdescriptiveanalysisoftheface-to-faceandinternetbasedsocialactivityparticipationsofthesurveyrespondents.DataCleaning Agerestriction(18.4%)andnoaccesstointernetathome(25.1%)leadtomostofthedataloss. Descriptivecharacteristicsoftheexplanatoryvariablesfromthetwoestimationdatasets(48hoursand24hour)arepresentedinTable 3-1 andTable 3-2 .Notethatmeanand 41

PAGE 42

1. Browsing/Informationseeking 2. Browsing/Shopping 3. Entertainment/games 4. Social-e-mail/chatrooms 5. Banking/nancial 6. Travelbrowsing/reservations 7. Studyingoron-linecoursesforschool 8. CDsandvideos Thoserespondentswhoindicatedusingtheinternetforactivitytype4,"Social-e-mail/chatrooms"aredenedtohaveparticipatedinIBsocialactivitiesduringtheday.Itisimportanttonotethattheactualnumberofepisodesandthedurationofactivityparticipationarenotavailable.Further,theactuallocationwhereinternetwasusedisalsounknown.WeimplicitlyassumethatIBsocialactivitiesareundertakenathome. IncontrasttodeningtheparticipationinIBsocialactivities,deningFFsocial-activityparticipationismoreinvolved.Althoughthesurveyinstrumentexplicitlyincludesa"socialactivity"purpose(activitypurposecode11islabeled"socialactivities"andis 42

PAGE 43

1. Thereisonlyonepurposedenedfortheepisodeandthisisactivitycode11,"socialactivities" 2. Therearemultiplepurposesdenedforthisepisodebuttherstpurposeisactivitycode11,"socialactivities" 3. TherearemultiplepurposesdenedforthisepisodeandtherstpurposeisNOTactivitycode11,"socialactivities"whereasoneoftheotherpurposes(second/third/fourthorfth)is. Intherawdataset,7702activityepisodeswereidentiedforwhichtherstpurposewasindicatedas"socialactivities"(Table 3-3 ).Afurther7342activityepisodeswereidentiedinwhichthesecond,third,fourth,orthefthactivitypurposewasindicatedas"social"(butnottherst).Thisclearlyhighlightsthatonecannotfocusonlyontherstactivitypurposeindicatedindeningsocialactivities.Further,thetableindicatesthat,inthecaseofout-of-home(OH)episodes,socialactivitiesaremostlikelytobeidentiedastherstpurpose.However,inthecaseofin-homeepisodes,socialactivitiesmaybeidentiedastherst,second,third,orthefourthactivitypurpose. Table 3-4 identiesthesecondactivitypurposeforthoseepisodesforwhichtherstpurposewasindicatedas"social".Wendthatinmostofthecases(about81%),therearenootherpurposesdened.Whenasecondpurposeisalsodened,thisisfoundmostlikelytobemeals,recreation,orresting/relaxingwhichseemsreasonableasitiseasytohavesocialconversationswhilepursuingthesetypesofactivities. Table 3-5 examinestherstactivitypurposereportedfortheepisodesforwhichthesecondpurposeisindicatedas"social".Almost70%oftheepisodeshave"meals"or"recreation"astheprimaryactivitypurpose.About11%ofthecases(mostofthembeingin-homeepisodes)alsoreporthouseholdchoresastheprimaryactivitypurpose. 43

PAGE 44

Theaboveempiricalanalysisindicatesthatactivityepisodescanbereportedaseitherbeingforpurely"social"purposesorformultiplepurposes,oneofwhichmightbe"social".Inthelattercase,thesocialactivitymayormay-notbereportedastherstpurpose.Thus,apersonhavingamealwithfriendsmayreporttherstpurposeas"social"andsecondpurposeas"meals"ortherstpurposeas"meals"andsecondpurposeas"social".Onecouldspeculatethattheorderofreportingreectstheperceivedorderofimportance(i.e.,intherstcase,theintentwasreallytochatwithfriendswithhavingamealbeingincidentalwhereasinthesecondcase,theintentwasreallytohaveamealwiththecompanyoffriendsbeingincidental).However,itisimportanttonotethattherespondentsarenotspecicallyaskedtoidentifya"primary"purposeforeachepisode(theyaresimplyaskedtoindicateasmanypurposesasappropriateuptoamaximumofve).Therefore,theorderofchoosingtheactivitytypescouldalsosimplyreecttheorderinwhichtheactivitytypesarepresentedtotherespondent(mealsiscodedas3whereassocialis11)inthesurvey.Further,whenmultiplepurposesarereportedforthesameepisode,therelativeamountsoftimespentineacharenotrecorded(forexample,whenmealsarehadwithfriends,thedurationspentineatingversusconversationsisnotknown). 44

PAGE 45

Denition1:Anactivityepisodeisdenedassocialiftherstactivitypurposeiscodedas11"SocialActivities" Denition2:Anactivityepisodeisdenedassocialifanyoftheveactivitypurposesiscodedas11"SocialActivities" ApersonisthendenedtohaveparticipatedinFFsocialactivitiesduringthedayifthepersonhasatleast1socialactivityepisodeduringtheday. Aseparatedatasetwascreatedtoanalyzeindividual'sactivitypatternover48hourperiod.Inthedataset,socialactivitywasconsideredvalidifindividualundertooktheactivityoneitherday.Sameconceptwasusedtodetermineotheractivities(suchasschool,employment,rainetc.)oftheindividualover48hourduration.Afterward,resultsforthe48hoursperiodwerecomparedwithresultsofthe24hoursperiodanalysis.DescriptiveAnalysisofFFandIBSocial-ActivityParticipation OverallShareofActivityParticipation 3-6 identiestheactivityparticipationshareforeachday.Forboththedenitions,oneitherday,peopleundertooklessin-homesocialactivitiesascomparedtoothersocialactivities(OHandIB).Fordenition1,FFsocialactivitiesoneitherdaydonotdiermuchintermsofactivityshare.Butfordenition2,moreFFsocialactivities-IH(7.8%onday1and8.4%onday2)andOH(10.4%onday1and11.3%onday2)wereperformedonday2.However,IBsocialactivityparticipationismoreorlesssameonbothdays.ActivityParticipationWithinaDay 3-7 .Fordenition1,oneitherday,outof16530individuals,69.4%didnotundertakeanysocialactivity.IBsocialactivitieswereundertakenbymostoftheindividuals(20.4%onday1and20.2%onday2).Overall,27.2%and27.7%individualsperformedonlyonekindoftheFFsocialactivityon 45

PAGE 46

Nevertheless,comparedtodenition1,indenition2fewerfractionsofindividualsundertakenosocialactivity(64.3%onday1and63.7%onday2).ThedierenceisreectedbecauseoftheindividualsmoreinvolvementinFFsocialactivityparticipation-OHsocial(7.0%onday1and7.4%onday2)andIHsocial(5.0%onday1and5.1onday2).ThisresultsinslightlylessindividualswhoparticipatedonlyinIBsocialactivities(18.7%onday1and18.1%onday2).Furthermore,alittleincreaseofactivityparticipationinallthreesocialactivitieswasobserved(0.3%forday1and0.4%forday2). Abovediscussionsuggeststhatlargefractionofthesample,irrespectiveofthedenition,didnotperformanyofthesocialactivitiesoneitherday.Moreover,IBsocialactivitiesalonewereperformedthemost.Frequencyanalysisof48hoursperiod(2day)exhibitssimilaractivitypatterns,Table 3-8 .Asexpected,moreindividualsperformedonekindofsocialactivity(40.7%fordenition1and47.4%fordenition2)andfewerindividualsperformednosocialactivity(59.2%fordenition1and52.6%fordenition2).FFsocialactivityparticipationwasalsoconsiderablyhigherforbothdenitionsascomparedtosingledaysocialactivityparticipation.ActivityParticipationAcrossDays 3-9 andTable 3-10 Fordenition1,withanexceptionofIBsocialactivity,nonesocialactivitycategorydominatedday2,irrespectiveofthesocialactivityundertakenonday1.Further,17.2%and25.3%undertookOHandIHsocialactivitiesonbothdaysrespectively.Moreover,66.1%individualsperformedIBsocialactivitiesonbothdays. Denition2exhibitsthesimilarpatternforIBsocialactivities(62.2%onday2)asdenition1,butFFsocialactivityparticipationisnoticeablydierentfortwodenitions. 46

PAGE 47

47

PAGE 48

SampleCharacteristicsofthenalestimationdata(48hour)

PAGE 49

3-1 .Continued AttributeStatisticAttributeStatistic *Descriptivefordurationiscalculatedoverallindividualsirrespectiveofwhethertheywenttowork/school.Asaresult,averagedurationislessthantheexpected16hours(8hoursaday)duration.**Onlyoneoutofthesevariableswastriedatatime.***TemperatureVariation=(MaxTemp-MinTemp)/MinTemp.

PAGE 50

SampleCharacteristicsofthenalestimationdata(24hour)

PAGE 51

3-2 .Continued AttributeStatisticAttributeStatistic Friday14.3Suburban(AreaDens.=6-30)73.5Weekend12.9Rural(AreaDens<6)5.6MonthoftheYear25.7Accessibility**Spring21.5Population/distance2.96(1.09)Summer41.4Populationdensity/distance0.75(0.60)Fall37.1LN(Popdensity)/distance1.07(0.58)Weather28.5(Popdensityin15miles)/distance0.43(0.61)RainLN(Popdensityin15miles0.61(0.56)No89.4)/distanceYes10.6Amount(inch)0.03(0.15)TemperatureVar.***0.25(0.14) *Descriptivefordurationiscalculatedoverallindividualsirrespectiveofwhethertheywenttowork/school.Asaresult,averagedurationislessthantheexpected16hours(8hoursaday)duration.**Onlyoneoutofthesevariableswastriedatatime.***TemperatureVariation=(MaxTemp-MinTemp)/MinTemp.

PAGE 52

Shareofsocialactivitiesrecordedbasedontheobjective ActivitytyperecordlocationCountTotal(%ofallepisodes)IH(%ofallIHepisodes)OH(%ofallOHepisodes) First77021.827.4870.25 Second31410.719.6221.89 Third20360.522.686.19 Fourth15880.421.961.39 Fifth5770.18.250.29 Primaryactivityissocialthenshareofsecondaryactivities HOUSEHOLDCHORESandPERSONALCARE5.309.492.73 MEALS(athome,take-out,restaurant)20.6615.1524.04 RECREATION/ENTERTAINMENT20.2814.9523.54 SLEEP2.763.842.11 WORKorWORKRELATED,(inoroutofhome)1.311.011.49 SCHOOLorSCHOOLRE-LATED/College/DayCar1.691.821.61 SHOPPING(ATHOME),(orbrowsing)0.080.200.00 SHOPPING(AWAYFROMHOME)0.840.001.36 PERSONALSERVICES/BANK/GOV'T2.461.013.35 RELAXING/RESTING26.3444.8514.99 VOLUNTEER/CIVIC/RELIGIOUSSER-VICESORAC7.301.2111.03 SICKORILL/MEDICALAPPOINTMENT0.380.000.62 NON-WORK(NON-SHOPPING)INTERNETUSE2.234.440.87 PICK-UP/DROPOFFPASSENGER8.372.0212.27

PAGE 53

Secondaryactivityissocialthenshareofprimaryactivities HOUSEHOLDCHORESandPERSONALCARE10.8622.362.57 MEALS(athome,take-out,restaurant)37.3831.6341.51 RECREATION/ENTERTAINMENT32.0929.4334.01 SLEEP0.641.060.33 WORKorWORKRELATED,(inoroutofhome)1.851.292.25 SCHOOLorSCHOOLRE-LATED/College/DayCar4.843.046.13 SHOPPING(ATHOME),(orbrowsing)0.130.300.00 SHOPPING(AWAYFROMHOME)1.880.233.07 PERSONALSERVICES/BANK/GOV'T3.151.674.22 RELAXING/RESTING4.177.981.42 VOLUNTEER/CIVIC/RELIGIOUSSER-VICESORAC1.020.151.64 SICKORILL/MEDICALAPPOINTMENT0.030.000.05 NON-WORK(NON-SHOPPING)INTERNETUSE0.380.680.16 PICK-UP/DROPOFFPASSENGER1.590.152.63 ShareofFFandIBactivities IH(%)OH(%)IB(%) Denition1Denition2Denition1Denition2 Day12.67.87.910.422.9Day22.48.48.411.322.8 53

PAGE 54

ShareofFFandIBactivities Day1Day2Day1Day2 Frequency%Frequency%Frequency%Frequency% None1166669.41146869.41062564.31052263.7OnlyIB336720.4333620.2309618.7299818.1OnlyOHsocial9435.79966.011527.012257.4OnlyIHsocial2851.72531.58215.08425.1IBandOH3151.93312.03652.23902.4IBandIH920.6830.52671.63031.8OHandIHsocial410.2500.31490.91781.1All90.1130.1550.3720.4Total16530100165301001653010016530100 Socialactivities-48hour Denition1Denition2 Frequency%Frequency% None979859.3868752.6OnlyIB379122.9326019.7OnlyOHsocial14528.816349.9OnlyIHsocial3992.410456.3IBandOH7444.57984.8IBandIH1751.14832.9OHandIHsocial1150.73982.4All560.32251.4Total1653010016530100

PAGE 55

Comparisonofsocialactivitiesover2day(Row%) Denition1Denition2 NoneIHSocialOHSocialIBSocialTotalNoneIHSocialOHSocialIBSocialTotal Day1None85.41.66.26.8100.0081.84.17.86.3100.00IHsocial50.917.213.318.6100.0037.435.215.411.9100.00OHsocial53.31.825.319.6100.0044.18.930.916.1100.00IBsocial23.22.08.766.1100.0021.65.410.862.2100.00Total69.42.08.420.2100.0063.76.911.318.1100.00

PAGE 56

Comparisonofsocialactivitiesover2day(Column%) Denition1Denition2 NoneIHSocialOHSocialIBSocialTotalNoneIHSocialOHSocialIBSocialTotal Day2None85.4053.951.423.569.482.638.444.722.264.3IHsocial1.7019.33.62.12.33.933.49.04.36.6OHsocial6.17.123.87.77.97.213.428.59.210.4IBsocial6.819.621.266.720.46.414.717.964.218.7Total100.00100.00100.00100.00100.00100.00100.00100.00100.00100.00

PAGE 57

Discretechoicemodelsareusedtodescribedecisionmakers'choiceamongallalternativesavailable.Inourproblem,individualhastomakethreechoiceseachwithtwoalternatives-internetbased(IB)social,in-home(IH)social,andout-of-home(OH)social.Anynumberofthealternativescanbechosenbytheindividual.Thealternativesarenotmutuallyexclusivefromdecisionmakers'perspective.Further,probitmodelsareusedfortheanalysisasitcanhandletherandomtastevariation,andallowanykindofsubstitution.Inthisreport,probitmodelwiththechoicesetofthreealternativesiscalledastri-variateprobit(TVP).SoftwarepackageLIMDEPisusedtoanalyzeTVPmodels. Remainderofthechapterprovidesanoverviewofmultivariateprobitmodelandderivesspecicationforthestudy.Firstsectionexplainstheneedofmultivariateprobitmodel(MVP).Nextsectionderivesspecicationfortri-variateprobitanddiscussesthechoiceprobabilitysimulator.LastsectionpresentstheearlierworksonMVP.WhyMultivariateProbit? GolobandRegan 2002 ). MVPisnotconstrainedbyIIApropertyandallowsindividualtochoosemultiplechoicessimultaneously,whilemultinomialdiscretechoiceonlyallowasinglechoice.

PAGE 58

Where,X=fXnIB;XnOH;XnIHgisavectorofexogenousvariables,and=fIB;OH;IHgisacorrespondingvectorofparameters.Ingeneralterms;Uni=iXni+"ni;i=IB;OH;IH Where,Vni=iXniand"niaretheobservedandrandomcomponentsoftheutilityrespectively.Probitmodelassumesthat"n=f"nIB;"nOH;"nIHgisdistributednormalwithmeanvectorzeroandcovariancematrix(foridenticationpurpose,procedureinLIMDEPnormalizedvariancesto1),=0BBBB@1121321123313211CCCCA O-diagonalelements,ij,intheabovecovariancematrixrepresenttheunobservedcorrelationbetweenthestochasticcomponentoftheithandjthtypesofsocialactivity. Also,thedensityof"nis("n)=1 (2)3 21 2e1 2"0n1"n 58

PAGE 59

(4{7) Now,letyni=1ifalternativeischosen,and0otherwisefori=IB;OH;IH.Ingeneralterms,choiceprobabilityofchoosinganycombinationsofsocialactivities, Where,3istri-variatenormalCDFand3istri-variatestandardnormaldensity.And,qni=(2yni1)=8>><>>:1ifyni=0;+1ifyni=1; (4{9) Integralinequation(4-8)isevaluatedoverallvaluesof"n.Therefore,integralis3-dimensionaloverthethreeerrors.Furthermore,theintegraldoesnothaveaclosedform; 59

PAGE 60

Beforerecentdevelopmentsinthesimulationmethods,numericalmethodswereusedfortheevaluationofmultipleintegrals.Insimulationmethods,choiceprobabilitiesaresimulatedusingMonteCarloMethods.Simulatedmaximumlikelihood(SML)andthemethodofsimulatedmoments(MSM)aretheprimarysimulationapproaches,whichhavebeenappliedtobothmultivariateandmultinomialchoicemodels.However,forMVP,MSMisfoundtobemoreburdensomethanMSL( KeaneandMott 1998 ).Ingeneral,simulationmethodsarecombinedwithchoiceprobabilitysimulatorfortheestimationofMVP. ProbabilitysimulatordevelopedbyGeweke,Hajivassiliou,andKeane(GHK)(describedin Geweke 1989 ; Hajivassiliou 1991 ; Keane 1994 ),alsocalledasSmoothRecursiveSimulator(SRC),isthemosteectiveandwidelyusedmethodtosimulatechoiceprobabilities,originallyappliedtotheMNPmodel( GolobandRegan 2002 ).GHKisstrictlyboundedbyzeroandone,smoothintheparameters,unbiased,andconsistentinthenumberofreplications( ContoyannisandJones 2004 ).Moreover,MonteCarloevidencesubstantiatesthelowvariancesshownbyGHK( ContoyannisandJones 2004 ).Softwarepackageusedfortheanalysis,LIMDEP,usesSMLframeworkcombinedwithGHKsimulator.Therefore,abriefdescriptionoftheGHKsimulatorisprovidedinthenextsection.GHKSimulator 60

PAGE 61

Where,ni=nIB;nOH;nIHisavectorofstandardnormal,orniN(0;1) So,fortheindividualn,setofutilities0BBBB@UnIBUnOHUnIH1CCCCA=0BBBB@VnIBVnOHVnIH1CCCCA+0BBBB@c1100c21c220c31c32c331CCCCA0BBBB@nIBnOHnIH1CCCCA Now,forillustrativepurpose,choiceprobabilityofchoosingallalternatives c11)Prob(nIH>(VnIH+c31nIB+c32nOH) c11&nOH>(VnOH+c21nIB) (4{13) Draw3randomvalueofnIB,rnIBfromatruncatedstandardnormaltruncatedat(VnIB=c11)andcalculatethersttermintheequation(4-13).Similarly,takedrawsrnOHandrnIHandcalculatesecondandthirdterms.Repeatthesestepsforrtimes,whereris

PAGE 62

ThesimulatedprobabilityfromGHKisthenusedintothelog-likelihoodfunctionandstandardmaximizationtechniquesareusedtogetparameterestimates.Log-likelihoodfunctionis,LogL=Log(Pn) (4{15) Theoptimizationproblemtobesolvedhereisnotnecessarilygloballyconvex,soconvergencetotheoptimumisnotguaranteed( deDiosOrtuzarandWillumsen 2002 ).NextsectiondiscussespreviousworksthatusedMVPframeworktomodelmultipleresponsechoiceset.PreviousWorkonMVP ( 1970 )wereoneoftherststousethemultivariateprobitmodel.Theydevelopedanexactmaximumlikelihoodforthebi-variatecase.Sincethen,otherworkshaveappliedMVPinvariouscontexts. ChibandGreenberg ( 1998 )discussedsimulationtechniquesformultivariateprobitmodelsandemployedthesemethodstodatasetswithbi-variateandseven-variatebinaryresponses.Inanotherwork, ContoyannisandJones ( 2004 )modeledsevenbinarydependentvariablesusingMVP.MaximumlikelihoodestimationcoupledwithGHKsimulatorwasusedfortheestimation. SomeresearchersusedsoftwarepackagesfortheestimationofMVP,suchasLIMDEP,STATA,ADF-WLSetc.OtherswrotetheirownprograminGAUSSorMATLABtoestimatemodelsaspertheirrequirements. GolobandRegan ( 2002 )usedADF-WLStoidentifytheadoptionofinformationtechnologyinthetruckingindustry.Sevendierentinformationtechnologies(endogenousvariables)and20exogenousvariablesweremodeledusingstructuredmultivariateprobitmodel. Baltas ( 2004 )calledMVPassimultaneousprobitmodels(SPM)andusedittodescribesimultaneousselectionofmultiplebrands. 62

PAGE 63

Youngetal. ( 2009 )usedSTATAtoestimatetri-variateprobitmodel. Becauseofthelargedataset,numerouschoicesandcomplexityofthemethod,multivariateprobitmodelsarenotoftenusedinthetransportationeld.Onlyfewstudieshavedemonstratedmultivariateprobitapplicationtothetransportationdemandmodeling. ChooandMokhtarian ( 2008 )constructedtri-variateprobittomodelthreetravelrelatedstrategiesforcommutingworkerslivinginthreedistinctSanFransiscoBayareaneighborhoodsandperformedestimationsinLIMDEP.However,intransportationeldMNLismostcommonlyusedtomodelmultiplechoices. Bolduc ( 1999 )estimated9-modetransportationchoicemodelsandusedSMLframeworkviaGHKsimulatorfortheestimation.SimilarMNLframeworkapplication,althoughnotintheeldoftransportation,wasillustratedby Lawrence ( 1997 ).Heappliedtheframeworktoathreechoicedatasetforthe1998democraticSuperTuesdayPrimary. 63

PAGE 64

EstimationsarecarriedoutinLIMDEP.Thetri-variateprobitmodel(TVP)isestimatedusingthemethodofsimulatedmaximumlikelihood(SML)usingasmoothrecursivesimulator,calledasGHK,toevaluatemultivariatenormalprobabilities.Thedetailsofthesimulatorareprovidedin"Chapter4-Methodology".Moreover, CappellariandJenkins ( 2003 )mentionedabouttheasymptoticalconsistencyoftheSMLestimatorasthenumberofdrawstendstoinnity.Parameterestimatesarerobusttodierentinitialseedsaslongasnumberofdrawsisgreaterthanthesquarerootsofthesamplesize. Arangeofexogenousvariablesareemployedtoobtainasubstantiallyrichspecication.Listofexogenousvariablesalongwiththeirdescriptivecharacteristicsareprovidedinchapter3-data(Table 3-1 andTable 3-2 ). Estimationsarecarriedoutbyadopting90%condenceinterval.Therefore,exogenousvariableswitht-statisticslessthan1.60areassumedtohaveinsignicanteectonendogenousvariablesandaredroppedintheprocesstoarriveatthebestspecication.TVPmodelsareestimatedfortwodenitionsofface-to-face(FF)socialactivities-denition1anddenition2.Asmentionedin"Chapter3-Data",denition1denesFFsocialactivityassocialactivityundertakeninrstactivitypurpose.Ontheotherhand,ifsocialactivityisperformedinanyoftheveactivitypurposesitisdenedasFFsocialactivityindenition2.However,internetbased(IB)socialactivityissameforbothdenitions. Estimationsarecarriedoutfortwodatasetsofdierenttimescales:48hourperiodand24hourperiod.Bestspecicationsobtainedfor48hourperiodwerefurtherestimatedfor24hourperioddataset.Estimationresultsfrombothdatasetsarethencomparedtoinvestigatetheeectoftemporalscaleonmagnitudeofresults.However,resultsarealsocomparedacrossdenitionsandacrossdependentandindependentmodels. 64

PAGE 65

Where,LogL0andLogLcorrespondtothelog-likelihoodvaluesoftherestrictedandtheunrestrictedmodelrespectively.Thevalueinequation(5-1)iscomparedagainstthecriticalchi-squarestatisticobtainedusingthedegreesoffreedomequaltothenumberofrestrictionsontheparametervectors.Log-likelihoodvaluesandteststatisticsalongwiththeirdegreesoffreedomarepresentedattheendoftheTable 5-1 ,andTable 5-2 .TheindependentTVPinvolvesthreerestrictionsonthecorrelationmatrix. Inspectionofchi-squarestatisticsrevealsthatlikelihoodfunctionforbothdenitionsimproveremarkablyforthegeneralized,uncorrelatedTVPmodel.Inviewoflargechi-squarestatistics,thenullhypothesiscanberejectedat90%condenceinterval.Inthelightofthepresentempiricalevidences,weadopttheunrestrictedcovariancestructureembodiedintheunrestrictedTVP.ItshouldbenotedhoweverthatthecoecientsoftheindependentTVParealsoconsistentandtheadvantageofthegeneralized,unrestrictedmodeliswithregardtoeciencyandtheinformationobtainedfromestimatingcovariancestructure.Empiricalresultsofunrestricted(correlated)modelarediscussedinthenextsection. 65

PAGE 66

5-1 andTable 5-2 .Followingsubsectionspresentinterpretationsfortheestimatesandarearrangedinfollowingorder;householdsocioeconomiccharacteristics,individualsocioeconomiccharacteristics,activityparticipation,dayoftheweek,monthoftheyear,landusechar.,andweather. Ingeneral,interpretationsarediscussedfordenition1.Sinceinterpretationsforbothdenitionsarefoundsimilar,arenotdiscussedseparately.However,infewcaseswhereeectsaredissimilar,discussionfordenition2isalsoprovided.HouseholdSocioEconomicCharacteristics Asmentionedpreviously,householdsselectedforthesurveyhadatleastonephoneline.Therefore,presenceofmultiplephonelinesonsocialactivitieswasinvestigatedanditwasfoundthatmultiplephonelinesencouragesIBsocialactivities.Inotherwords,onemediumofcommunication(telephone)supplementsanothermediumofcommunication(internet).Furthermore,presenceofbicycleshaspositiveeectonOHsocialactivities.Inotherwords,probabilityofundertakingOHsocialactivitiesishigherwiththepresenceofbicyclesinthehousehold.Thiscouldbeexplainedwiththefactthatbicycleisoneofthetransportmodes,thus,itspresenceencouragesindividualtoplanout-of-hometrips,henceOHsocialactivities.However,ifindividuallivesintherentedhousethenheismorelikelytoperformIBsocialactivitiescomparedtotheindividualwhoownsthehouse. Signicantvariationinthesocialactivitypatternwasobservedfordierenthouseholdstructures.SinglepersonfamilyhouseholdismorelikelytoperformIBandOHsocial 66

PAGE 67

Reiteratingthatforthepurposeofthisanalysisestimationdatawaslteredtocoveronlyadults,whichweredenedasindividualsolderthan15years.Thus,individualsyoungerthanorequalto15yearswerecategorizedaschild.PresenceofthesechildinhouseholdwasthenanalyzedforsocialactivityparticipationandresultsindicatethatithasanegativeeectonIHsocialactivities.However,itdoesnotaectIBandOHsocialactivities.Child'spresenceinthehouseholdimposesresponsibilityonhouseholdmembersasnowtheyhavetotakecareofthechildrelatedworkssuchas,dropof/pickupfromschooletc.Thisextraresponsibilityaectstheirtimespendathome,thus,arelesslikelytoperformIHsocialactivities.However,fordenition2,socialactivityparticipationisnotaectedbythepresenceofchildreninhousehold. 67

PAGE 68

Variousmeasureswereattemptedtocapturethehouseholdincomeeectonsocialactivityparticipation,suchas,continuousincome,incomeincategories,incomeperpersonetc.However,incomeperpersonisincludedinthenalspecicationasanindicatorforhouseholdincome.EstimationresultsrevealthathouseholdwithhigherincomeperpersonismorelikelytoinvolveinIBsocialactivitythanthehouseholdwithlessincomeperperson.IndividualSocioEconomicCharacteristics Ingeneral,forage,categoriesof10yearsareformedbutyoungerindividualsarefurtherbrokenin5yearsagecategorywithananticipationofobservingmorevariationinsocialactivityparticipation.Overall,ageisfoundtobenegativelyassociatedwiththesocialactivityparticipation.MarginaleectonutilityisplottedagainstagecategoriesforallthreeformsofsocialactivityandpresentedinFigure 5-1 ,Figure 5-2 andFigure 5-3 .Agecategoriesatthesameutilitylevelrepresentsequalpropensityofparticipationinrespectivesocialactivity.Specically,forIBsocialactivityparticipation,rstdropintheutilityisobservedinagecategory25-35years.Afterthat,anotherdropisfoundinagecategory65-75years.LikelinessofOHsocialactivityparticipationdecreasestill 68

PAGE 69

ResultsfurthersuggestthatCaucasians(non-hispanic)aremorelikelytoundertakeIBsocialactivitiesthanotherraces.Moreover,OHsocialactivityparticipationisleastexpectedamongAsian/pacicislandersandIHsocialactivityparticipationisleastexpectedamongHispanics.Denition2resultsshowdissimilarFFsocialactivityparticipationpatterns.Now,OHsocialactivityparticipationismostlikelyinCaucasiansandIHsocialactivityparticipationisleastlikelyinAsian/pacicislanders. Male,comparetofemale,arelessexpectedtogetinvolveinFFsocialactivities.FFsocialactivityparticipationofmaleisfurtherdierentiatedifchildrenarepresentinthehousehold.Combinedeectofthreerelatedvariables,male,presenceofchildren,andmalewithchildreninthehousehold,isexaminedandpresentedinTable 5-3 .Inhouseholdwithchildren,malearelesslikelytoundertakeFFsocialactivities.Thisactivityparticipationismorelikelyformaleifhouseholddoesnothavechildren.Nevertheless,OHsocialactivityparticipationoffemaleissameirrespectiveofthechildpresenceinthehousehold.ProbabilityofundertakingIHsocialactivityislessforfemaleifchildrenarepresentinthehousehold.However,fordenition2,likelinessoffemaleperformingFFsocialactivitydoesnotchangewhetherchildispresentinthehousehold.Additionally,IHsocialactivityparticipationofmaledoesnotchangewiththepresenceofchildinthehousehold. WorkerandstudentvariablesarenotfoundsignicantforIBandOHsocialactivityparticipation.However,fordenition2,individual'sworkerstatusindicatespositiveimpactonOHsocialactivityparticipation,suggestingthatworkerismorelikelytoundertakeOHsocialactivitiesthanthenonworker.Furthermore,employmenteecton 69

PAGE 70

Tondanappropriatemeasuretoreectthebehavioroflicenseddriversonsocialactivitieswasabigchallenge,therefore,numerousmeasureswereconstructedforthispurpose.Asourrstattempt,westartedwithnumberoflicenseddriversinthehousehold.Further,ashouseholdcharacteristicnumberofvehiclesperlicenseddriverinthehouseholdwascalculatedandusedinthespecication.Still,thespecicationcouldnotexposethetrueeectoflicenseddrivers.Thenwesegregatedlicenseddriversbasedonthevehiclesandlicenseddriversinthehousehold.Threehouseholdcategories-licenseddriversbutnovehiclesintheHH,licenseddriversaremorethanthevehiclesintheHH,andlicenseddriversarelessthanthevehiclesintheHH-wereformedandestimated.Buttooursurprise,resultsshowedthatlicenseddriversfromnovehiclehouseholdsaremorelikelytoperformOHsocialactivitiesthanlicenseddriverslivinginhouseholdswithvehicles.CrosstabulationofnovehiclehouseholdswithareatypeindicatedthatmostofsuchhouseholdsarelocatedinCBD(37.2%)andsuburban(45.9%)areas.Moreover,suchhouseholdsaredominatedbysinglepersonfamily(45%).Therefore,specicationsbasedonfamilystructureandareatypesweretriedseparatelybeforearrivingonthenalspecicationwhichclassiesnovehiclehouseholdsbasedontheareatype. Asdiscussed,novehiclehouseholdsarefurthercategorizedforthehouseholdlocation(areatype):cbd,suburban,urban,andrural.Amongsuchhouseholds,itwasfoundthatthelicenseddriverfromthehouseholdlocatedinCBDareaismorelikelytoperformOHsocialactivitiesthanthelicenseddriverfromthehouseholdlocatedinurbanarea.However,licenseddriverinthelattercategoryismorelikelytoperformOHsocialactivitiesthanthelicenseddriverlivinginthehouseholdlocatedinsuburban.Moreover,licenseddriverfromthehouseholdinruralareaislessexpectedtoundertakeOHsocialactivitiesthanthehouseholdinotherareas.Also,nolicensedindividualandlicenseddriverwithnovehiclesinthehouseholdareequallylikelytoundertakeIBsocialactivities. 70

PAGE 71

Nevertheless,licenseddriverwhohastosharevehicleismorelikelytoperformIBsocialactivitythanthedriverwhodoesnothavetosharevehicle.However,licenseddriverinthelattercategoryismorelikelytoperformIBsocialactivitythantheindividualwhodoesnothavedrivinglicense.ActivityParticipation Ingeneral,resultsindicatethatifindividualgotoworkin2days(48hours)periodthenheislesslikelytoperformsocialactivities.ItisfurtherobservedthatindividualgoingtoworkonsingledayislessprobabletoperformIBsocialactivitythantheindividualwhogoestoworkonbothdays,butperformsmoreFFsocialactivitiescomparetothelatterindividual.Thiseect,however,illustratestheuseoftheinternetatwork.Individualwhogoestoworkonbothdays,hastouseinternetatworkifhewantstoperformsocialactivity,however,individualwhogoestoworkonlyononedaycanuseeitherofthetwomediums(IBandFF)ontheremainingday. However,thebehaviorofstudentsinIBsocialactivityparticipationissurprising.Accordingtotheresults,likelinessofperformingIBsocialactivitiesisequalforindividualwhogoestoschoolonbothdaysandwhodoesnotgotoschoolatall.Further,itisobservedthatIBsocialactivityismostlikelytobeperformedbyindividualwhogoestoschoolonlyononeday.AsIBsocialactivitycanalsobeperformedatschool,moreIB 71

PAGE 72

Overall,FFsocialactivitiesaremostlyundertakenondaysinvolvingFridayandIBsocialactivitiesareperformedduringweekdayswithnoFridayinvolved.Duringweekdaysindividualisexpectedtobebusywithhiswork/school,therefore,socialactivitiesaremorelikelytobeperformedovertheinternet.Nevertheless,Fridayisthelastdayoftheworkingperiodthus,individualsliketogooutandsocializewithothers.MonthoftheYear 72

PAGE 73

Parameteronfractionofsinglefamilydwellingunitsisestimatedinsignicantfordenition1.But,fordenition2,itindicatespositiveeectonOHsocialactivityparticipation.Therefore,OHsocialactivitiesaremorelikelytobeundertakeninthezonewithhighernumberofsinglefamilyhouseholds.Further,populationdensityrevealspositiveeectonIBandIHsocialactivityparticipation.Therefore,ifhighernumberofpersonsareavailableinthesurroundingareathanprobabilitiesofundertakingIBandIHsocialactivityarehigher.However,fordenition2,populationdensityhaspositiveeectonallsocialactivityparticipation. AveragehouseholdincomeinthezoneisnegativelyassociatedwithIBsocialactivityparticipationofthehousehold.Inotherwords,lesspropensityofIBsocialactivityparticipationisobservedinhigheraveragehouseholdincomezone.Weather Tempvar=(Maxtemp-Mintemp)/(Mintemp). However,temperaturevariationisnotfoundtohaveanyimpactonthesocialactivityparticipation.Nevertheless,asexpected,rainontheactivitydayrevealsnegativeimpact 73

PAGE 74

5-4 .However,fordenition1,correlationbetweenIHandIBsocialactivityisestimatedinsignicant.Further,correlationsbetweenOHandIBsocialactivityindicatesthatpropensityofperformingOHsocialactivityincreaseswiththehigherparticipationinIBsocialactivityorvice-versa.Therefore,communicationonlinehelpsincreasingyourface-to-faceorrealworldsocialinteractions( Williams 2007 ).Also,face-to-facecommunicationleadstoonlinecommunication.Similarly,correlationsrevealthathigherparticipationinOHsocialactivityincreasesthepropensitytoundertakeIHsocialactivityorvice-versa.Ifindividualismorewillingtocommunicateface-to-faceoutsidehomethenheisalsomorelikelytobeinvolvedinsocialinteractionsathome.Fordenition2,allthreecorrelationsareestimatedsignicant.Therefore,resultsindicatethatsocialactivitiescomplementeachother. Further,EstimatesofthecorrelationssuggeststrongercorrelationbetweenIHandOHsocialactivitythanthecorrelationbetweenOHandIBsocialactivity.Additionally,denition2indicatesweakercorrelationbetweenIHandIBsocialactivitythantheremainingtwocorrelations.SingleDayDataEstimation 5-5 ,andTable 5-6 .Itisobservedthatmanyvariablesinthespecicationareestimatedinsignicantat90%condenceinterval.Specically,forbothdenitions,veoutofthenineteenparametersonIBsocialactivityarenotsignicant,thoseinclude,presenceofmultiplephones,singleparenthousehold,multipleworkers, 74

PAGE 75

Moreover,Inspectionofcorrelationsbetweensocialactivityparticipation,presentedinTable 5-7 ,revealsthatmagnitudeofcorrelationsisweakforsingledayestimationscomparedtothetwodayperiodanalysis.Explicitly,onlyonecorrelationisfoundsignicantforeitherdenitionthatis,betweenIHsocialandOHsocial.CorrelationsbetweenFFsocialandIBsocialactivityarenotsignicant.Inotherwords,formsofFFsocialactivitiesinuenceoneotherandarenotaectedbyothermodeofcommunication(IB).Inthe48hourperiodanalysisat-leasttwocorrelationswereestimatedsignicantforeitherdenition,thus,correlationsarestrongerin48houranalysiscomparedtothe24houranalysis.Summary 75

PAGE 76

IBSocialActivityParticipationwithAge Figure5-2. OHSocialActivityParticipationwithAge 76

PAGE 77

IHSocialActivityParticipationwithAge 77

PAGE 78

Empiricalresultsof48houranalysis:Denition1 AttributeIBSocialActivityOHSocialActivityIHSocialActivity Estimatet-statisticEstimatet-statisticEstimatet-statistic

PAGE 79

5-1 .Continued AttributeIBSocialActivityOHSocialActivityIHSocialActivity Estimatet-statisticEstimatet-statisticEstimatet-statistic Male*child-0.0838-1.99530.19692.5038EmployedLicensedDriver(notalicenseddriverisbase)Driver*NovehinHH*cbd0.64303.3849Driver*NovehinHH*urban0.59153.5973Driver*NovehinHH*suburban0.57962.0136Driver*NoVehinHH*ruralDriver*(vehiclesdriversinHH)0.13122.54170.41636.0606 Numberofcases16530

PAGE 80

5-1 .Continued AttributeIBSocialActivityOHSocialActivityIHSocialActivity Estimatet-statisticEstimatet-statisticEstimatet-statistic LogLikelihoodatconvergence-18985.73LogLikelihoodatconvergenceforconstantsonlymodel-19722.06LogLikelihoodatconvergenceforuncorrelatedmodel-18995.1No.ofrestrictions3Chi-squarestatistics-18.74Criticalchi-squarestatistics7.81

PAGE 81

Empiricalresultsof48houranalysis:Denition2 AttributeIBSocialActivityOHSocialActivityIHSocialActivity Estimatet-statisticEstimatet-statisticEstimatet-statistic

PAGE 82

5-2 .Continued AttributeIBSocialActivityOHSocialActivityIHSocialActivity Estimatet-statisticEstimatet-statisticEstimatet-statistic Male*child-0.1053-2.6574Employed0.17153.9666LicensedDriver(notalicenseddriverisbase)Driver*NovehinHH*cbd0.43862.3386Driver*NovehinHH*urban0.38582.4369Driver*NovehinHH*suburbanDriver*NoVehinHH*ruralDriver*(vehiclesdriversinHH)0.13192.55910.30254.9457 Numberofcases16530

PAGE 83

5-2 .Continued AttributeIBSocialActivityOHSocialActivityIHSocialActivity Estimatet-statisticEstimatet-statisticEstimatet-statistic LogLikelihoodatconvergence-23116.55LogLikelihoodatconvergenceforconstantsonlymodel-24127.76LogLikelihoodatconvergenceforuncorrelatedmodel-23160.4No.ofrestrictions3Chi-squarestatistics-87.7Criticalchi-squarestatistics7.81

PAGE 84

Eectofmaleandpresenceofchildinthehouseholdonsocialactivities Denition1Denition2 Out-of-HomeIn-HomeOut-of-HomeIn-Home ChildNoChildChildNoChildChildNoChildChildNoChild Male-0.1401-0.0563-0.1938-0.1642-0.1691-0.0638-0.1749-0.1749Female00-0.226500000 Table5-4. Correlationsamongsocialactivities(48hour) Denition1Denition2 Estimatet-statisticEstimatet-statistic OHSocialvsIBSocial0.05153.26170.05613.7690IHSocialvsIBSocial-0.0024*-0.10530.05033.0963IHSocialvsOHSocial0.10274.20670.13828.1155 *Notsignicantat90%condencelevel 84

PAGE 85

Empiricalresultsof24houranalysis:Denition1 AttributeIBSocialActivityOHSocialActivityIHSocialActivity Estimatet-statisticEstimatet-statisticEstimatet-statistic

PAGE 86

5-5 .Continued AttributeIBSocialActivityOHSocialActivityIHSocialActivity Estimatet-statisticEstimatet-statisticEstimatet-statistic Male*child-0.0503*-1.02830.27342.8416EmployedLicensedDriver(notalicenseddriverisbase)Driver*NovehinHH*cbd0.58702.9137Driver*NovehinHH*urban0.46132.2892Driver*NovehinHH*suburban0.78172.7459Driver*NoVehinHH*ruralDriver*(vehiclesdriversinHH)0.10581.95930.37014.6585 Numberofcases16530LogLikelihoodatconvergence-15024.29 *Notsignicantat90%condencelevel

PAGE 87

Empiricalresultsof24houranalysis:Denition2 AttributeIBSocialActivityOHSocialActivityIHSocialActivity Estimatet-statisticEstimatet-statisticEstimatet-statistic

PAGE 88

5-6 .Continued AttributeIBSocialActivityOHSocialActivityIHSocialActivity Estimatet-statisticEstimatet-statisticEstimatet-statistic Male*child-0.0617*-1.3527Employed0.11982.6949LicensedDriver(notalicenseddriverisbase)Driver*NovehinHH*cbd0.3036*1.5435Driver*NovehinHH*urban0.2215*1.1440Driver*NovehinHH*suburbanDriver*NoVehinHH*ruralDriver*(vehiclesdriversinHH)0.10661.97750.23233.4362 Numberofcases16530LogLikelihoodatconvergence-18475.9 *Notsignicantat90%condencelevel

PAGE 89

Correlationsamongsocialactivities(24hour) Denition1Denition2 Estimatet-statisticEstimatet-statistic OHSocialvsIBSocial0.0085*0.41800.0071*0.3779IHSocialvsIBSocial-0.0131*-0.44420.0308*1.5335IHSocialvsOHSocial0.07031.92280.05752.4337 *Notsignicantat90%condencelevel 89

PAGE 90

Widespreaduseoftheinternet,particularlyforsocialactivities,hasestablishedthefearofaectingtraditionalwayofconductingsocialactivitiesincomplexways.Thisisfurtherexpectedtoimpactindividual'stravelbehaviorandhaspolicyimplicationsontransportationplanning.Hence,relationshipbetweeninternetbasedsocialactivityandface-to-facesocialactivityisimportanttoinvestigateandissetasthefocusofthisstudy.However,mixedimpactsarefoundinpreviousstudiesandthoseimpactscanbesegregatedintothreecategories:substitution,supplement,andneutral.Thisstudymodelsactivityparticipationdecisionsofinternetbased(IB)andface-to-face(FF)socialactivityusingdiscretechoicemodels.Tri-variateprobitmodels(TVP)areestimatedusingthe2000BayAreaTravelSurvey(BATS).Therawdataisfurtherstructuredintotwoestimationdatasetsofdierenttimescale,48hourand24hour.Additionally,twodenitionsforFFsocialactivityareconstructed.Foreachdataset,correlatedanduncorrelatedmodelsareestimated.Bestspecicationderivedfrom48houranalysisisfurtherestimatedfor24hourandempiricalresultsarecomparedtoquantifytheeectoftimescaleindataset.TVPmodelsareestimatedineconometricsoftwareLIMDEPusingsimulatedmaximumlikelihood(SML)viasmoothrecursivesimulator,knownasGHK.NotethattheBATSdatausedinthisresearchwascollectedinyear2000,sincethentherehavebeenlotofadvancementsintermsofinternetusageandtechnology,suchas,useofvarioussocialnetworkingwebsites(facebook,myspace,andorkut),mobileinternet,wi-etc.Thesedevelopmentsmayhavecompletelychangedthesocialactivityparticipationbehaviorofthepeople.Therefore,empiricalresultspresentedinthisreportdonotreectthepresentsocialactivityparticipationbehavior. Remainingofthischapterdescribesourunderstandingoftheempiricalresultsandpresentsdirectionsforfuturework.Firstsectionprovidessummaryandbriefdiscussionof 90

PAGE 91

Ingeneral,IBsocialactivityparticipationisaectedthemostbythehouseholdcharacteristics.MoredetailedinspectionrevealedthatIBsocialactivityparticipationismorelikelyinrentedhouseholdsandhouseholdswithmultiplephonelines.Nevertheless,highpropensityofOHsocialactivityparticipationisobservedifbicyclesarepresentinthehousehold.PresenceofchildrenindicatesadverseeectonIHsocialactivityparticipation.Furthermore,IBandOHsocialactivityparticipationisfoundtobemorelikelyinsinglepersonfamilyhouseholds.However,singleparenthouseholdsaremorelikelytoperformIBsocialandOHsocialactivitythancouple,nuclearandotherhouseholds.FurtherexaminationofIBsocialactivitiesindicatedthatparticipationismorelikelybycouplehouseholdscomparedtonuclearandotherfamily.FFsocialactivityparticipation,however,isequallylikelyamongthesethreekindofhouseholds.Presenceofkidsyoungerthan5yearsandmultipleworkers,bothreducetheprobabilityofhousehold'sparticipationinIBsocialactivity.Presenceofmultipleworkersalsoreducestheprobabilityofthehousehold'sinvolvementinIHsocialactivity. 91

PAGE 92

Asanticipated,duringweekdayshighprobabilityofIBsocialactivityparticipationandlowprobabilityofFFsocialactivityparticipationisobserved.Further,ifFridayisinvolvedinanyofthosedays,propensityofFFsocialactivityparticipationimproves.IBsocialactivityismostlikelytobeundertakenduringspring.Moreover,higherprobabilityofIBandIHsocialactivityparticipationisexpectedinhighpopulatedzones.Also,in 92

PAGE 93

Comparisonofdenition1anddenition2empiricalresultsrevealstheimportanceofdeningFFsocialactivities.Fewvariableswhichdonothavesignicantimpactindenition1areestimatedsignicantindenition2orviceversa.CorrelationAmongSocialActivities Firstly,themodelsdevelopedinthisresearchcouldbeenhancedmethodologicallyinseveralofthefollowingways. 93

PAGE 94

Thirdly,currentstudyinvestigatesthetrade-osamongsocialactivitiesoverasingleactivityperiod.Afurtherapplicationofthesamewouldbetoidentifytrade-osofsocialactivitiesoverdaysorweeks. 94

PAGE 95

Ashford,J.R.,andR.R.Sowden.1970.Multi-VariateProbitAnalysis.Biometrics26:535{546. Baltas,G.2004.Amodelformultiplebrandchoice.EuropeanJournalofOperationalResearch154:144{149. Bargh,J.,andK.McKenna.2004.TheInternetandsociallife. Baym,N.,Y.Zhang,andM.Lin.2004.Socialinteractionsacrossmedia:Interpersonalcommunicationontheinternet,telephoneandface-to-face.NewMedia&Society6:299. Bhat,C.,A.Sivakumar,andK.Axhausen.2003.Ananalysisoftheimpactofinformationandcommunicationtechnologiesonnon-maintenanceshoppingactivities.TransportationResearchPartB37:857{881. Bolduc,D.1999.Apracticaltechniquetoestimatemultinomialprobitmodelsintransportation.TransportationResearchPartB33:63{79. Cappellari,L.,andS.Jenkins.2003.Multivariateprobitregressionusingsimulatedmaximumlikelihood.TheStataJournal3:278{294. Chib,S.,andE.Greenberg.1998.Analysisofmultivariateprobitmodels.Biometrika85:347{361. Choi,A.2008.InternetinSingapore:FindingsfromaNationalSurvey.Observatorio(OBS*)2. Choo,S.,andP.Mokhtarian.2008.Howdopeoplerespondtocongestionmitigationpolicies?Amultivariateprobitmodeloftheindividualconsiderationofthreetravel-relatedstrategybundles.Transportation35:145{163. Coget,J.,Y.Yamauchi,andM.Suman.2002.TheInternet,socialnetworksandloneliness.ITandSociety1:180{201. Cole,J.,andJ.Robinson.2002.InternetuseandsociabilityintheUCLAdata:AsimpliedMCAanalysis.IT&Society1:202{218. Contoyannis,P.,andA.Jones.2004.Socio-economicstatus,healthandlifestyle.Journalofhealtheconomics23:965{995. deDiosOrtuzar,J.,andL.Willumsen.2002.Modellingtransport.Wiley. Gershuny,J.2002.SocialLeisureAndHomeIT:APanelTime{DiaryApproach.IT&Society1:54{72. Geweke,J.1989.BayesianInferenceinEconometricModelsUsingMonteCarloIntegration.Econometrica57:1317{39. 95

PAGE 96

Hagerstrand,T.1970.Whataboutpeopleinregionalscience?PapersinRegionalScience24:6{21. Hajivassiliou,V.1991.SimulationEstimationMethodsforLimitedDependentVariablesModels.CowlesFoundationDiscussionPapers:100757:1317{39. Janelle,D.,andD.Hodge.2000.Information,place,andcyberspace:Issuesinaccessibility.SpringerVerlag. Keane,M.1994.AComputationallyPracticalSimulationEstimatorforPanelData.Econometrica62:95{116. Keane,M.,andR.Mott.1998.Astructuralmodelofmultiplewelfareprogramparticipationandlaborsupply.InternationalEconomicReviewpages553{589. Kenyon,S.,G.Lyons,andJ.Raerty.2002.Transportandsocialexclusion:investigatingthepossibilityofpromotinginclusionthroughvirtualmobility.JournalofTransportGeography10:207{219. Kestnbaum,M.,J.Robinson,A.Neustadtl,andA.Alvarez.2002.Informationtechnologyandsocialtimedisplacement.IT&Society1:21{37. Kiesler,S.,R.Kraut,J.Cummings,B.Boneva,V.Helegeson,andA.Crawford.2002.Internetevoluationandsocialimpact.IT&Society1:120{134. Kraut,R.,M.Patterson,V.Lundmark,S.Kiesler,T.Mukopadhyay,W.Scherlis,andL.Green.1998.InternetParadox::ASocialTechnologyThatReducesSocialInvolvementandPsychologicalWell-Being?Americanpsychologist53:1017{1031. Lawrence,E.,1997.SimulatedMaximumLikelihoodviatheGHKSimulator:AnApplicationtothe1988DemocraticSuperTuesdayPrimary.inAnnualMeetingoftheMidwestPoliticalScienceAssociation,Chicago. Lee,B.,andJ.Zhu.2002.InternetuseandsociabilityinmainlandChinaandHongKong.IT&Society1:219{237. Lee,W.,andE.Kuo.2002.Internetanddisplacementeect:Children'smediauseandactivitiesinSingapore.JournalofComputer-MediatedCommunication7. Lyons,G.2002.Internet:investigatingnewtechnology'sevolvingrole,natureandeectsontransport.TransportPolicy9:335{346. Mikami,S.2002.InternetuseandsociabilityinJapan.IT&Society1:242{250. Mokhtarian,P.1990.Atypologyofrelationshipsbetweentelecommunicationsandtransportation.InstituteofTransportationStudies. 96

PAGE 97

Mokhtarian,P.,I.Salomon,andS.Handy.2006.TheimpactsofICTonleisureactivitiesandtravel:Aconceptualexploration.Transportation33:263{289. Mokhtarian,P.L.2009.IfTelecommunicationisSuchaGoodSubstituteforTravel,WhyDoesCongestionContinuetoGetWorse?TransportationLetters1:1{17. Neustadtl,A.,andJ.Robinson.2002.SocialcontactdierencesbetweenInternetusersandnonusersinthegeneralsocialsurvey.IT&Society1:73{102. Nie,N.,andD.Hillygus.2002.TheimpactofInternetuseonsociability:Time-diaryndings.IT&Society1:1{20. Nie,N.,D.Hillygus,andL.Erbring.2002.Internetuse,interpersonalrelations,andsociability.TheInternetineverydaylifepages215{243. Nie,N.,A.Simpser,I.Stepanikova,andL.Zheng.2005.TenyearsafterthebirthoftheInternet,howdoAmericansusetheInternetintheirdailylives.StanfordInstitutefortheQuantitativeStudyofSociety. Pronovost,G.2002.TheInternetandtimedisplacement:ACanadianperspective.IT&Society1:44{53. Qiu,Y.,T.Pudrovska,andS.Bianchi.2002.SocialActivityandInternetUseinDual-EarnerFamilies:AWeeklyTime-DiaryApproach.IT&Society1:38{43. Robinson,J.,M.Kestnbaum,A.Neustadtl,andA.Alvarez.2000.MassmediauseandsociallifeamongInternetusers.SocialScienceComputerReview18:490. Shklovski,I.,R.Kraut,andL.Rainie.2004.TheInternetandsocialparticipation:Contrastingcross-sectionalandlongitudinalanalyses.JournalofComputer-MediatedCommunication10:1{23. Smith,P.,N.Smith,K.Sherman,K.Kripalani,I.Goodwin,C.Crothers,andA.Bell.2008.TheInternet:SocialanddemographicimpactsinAotearoaNewZealand.Observatorio(OBS*)2. Srinivasan,K.,andS.ReddyAthuru.2004.ModelingInteractionBetweenInternetCommunicationandTravelActivities:EvidencefromBayArea,California,TravelSurvey2000.TransportationResearchRecord:JournaloftheTransportationResearchBoard1894:230{240. Srinivasan,S.,andC.Bhat.2006.Amultiplediscrete-continuousmodelforindependent-andjoint-discretionary-activityparticipationdecisions.Transportation33:497{515. 97

PAGE 98

Wellman,B.,A.Haase,J.Witte,andK.Hampton.2001.DoestheInternetincrease,decrease,orsupplementsocialcapital?:socialnetworks,participation,andcommunitycommitment.AmericanBehavioralScientist45:436. Williams,D.2007.TheImpactofTimeOnline:SocialCapitalandCyberbalkanization.CyberPsychology&Behavior10:398{406. Young,G.,E.Valdez,andR.Kohn.2009.Multivariateprobitmodelsforconditionalclaim-types.InsuranceMathematicsandEconomics44:214{228. Yu,H.,andS.Shaw.2008.Exploringpotentialhumanactivitiesinphysicalandvirtualspaces:aspatio-temporalGISapproach.InternationalJournalofGeographicalInformationScience22:409{430. Zhu,J.,andZ.He.2002.Diusion,useandimpactoftheInternetinHongKong:Achainprocessmodel.JournalofComputer-MediatedCommunication7:1{26. 98

PAGE 99

NagendraSinghDhakarwasborninIndia,in1983.Hereceivedhisbachelor'sdegreeincivilengineeringfromtheIndianInstituteofTechnologyBombay,Mumbai,Indiain2005.Mr.Dhakarwasamaster'scandidateandresearchassistantintheTransportationResearchCenter,attheUniversityofFlorida,DepartmentofCivilandCoastalEngineering,andhereceivedhisMasterofSciencedegreeinAugust2009. 99