The Niakhar Social Networks and Health Project

(1)

HAL Id: hal-02289832

https://hal-amu.archives-ouvertes.fr/hal-02289832

Submitted on 17 Sep 2019

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of

sci-entific research documents, whether they are

pub-lished or not. The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est

destinée au dépôt et à la diffusion de documents

scientifiques de niveau recherche, publiés ou non,

émanant des établissements d’enseignement et de

recherche français ou étrangers, des laboratoires

publics ou privés.

Distributed under a Creative Commons Attribution| 4.0 International License

The Niakhar Social Networks and Health Project

Valérie Delaunay, Laetitia Douillot, Steven Rytina, Yacine Boujija, Simona

Bignami, Sadio Gning, Cheikh Sokhna, Loubna Belaid, Babak Fotouhi,

Abdourahmane Senghor, et al.

To cite this version:

Valérie Delaunay, Laetitia Douillot, Steven Rytina, Yacine Boujija, Simona Bignami, et al..

The Niakhar Social Networks and Health Project.

MethodsX, Elsevier, 2019, 6, pp.1360-1369.

(2)

Method

Article

The

Niakhar

Social

Networks

and

Health

Project

Valerie

Delaunay

a

,

Laetitia

Douillot

b

,

Steven

Rytina

c

,

Yacine

Boujija

b

_,

_Simona

_Bignami

b

_,

_Sadio

_Ba

_Gning

d

_,

Cheikh

Sokhna

e

,

Loubna

Belaid

c

,

Babak

Fotouhi

f

,

Abdourahmane

Senghor

g

,

John

Sandberg

h,

*

a

LPED,IRD,Aix-MarseilleUniversité,France

b

UniversitédeMontréal,Canada

c_McGill_University,_Canada

d_Université_Gaston_Berger,_Saint-Louis,_Senegal e

VITROME,IRD,Aix-MarseilleUniversité,AP-HM,IHU-MéditerranéeInfection,Marseille,France

f

HarvardUniversity,UnitedStates

g

Prose,Senegal

h

TheGeorgeWashingtonUniversity,UnitedStates

*Correspondingauthor.

E-mailaddresses:valerie.delaunay@ird.fr(V.Delaunay),laetitia.douillot@ird.fr(L.Douillot),yacine.boujija@umontreal.ca

(Y.Boujija),simona.bignami@umontreal.ca(S.Bignami),sadio-ba.gning@ugb.edu.sn(S.BaGning),cheikh.sokhna@ird.fr

(C.Sokhna),Loubna.belaid@mail.mcgill.ca(L.Belaid),babak_fotouhi@fas.harvard.edu(B.Fotouhi),asenghor@sicap.sn

(A.Senghor),jsandber@gwu.edu(J.Sandberg).

https://doi.org/10.1016/j.mex.2019.05.037

ContentslistsavailableatScienceDirect

MethodsX

(3)

ABSTRACT

ThispaperpresentsdetailsofthedesignandimplementationoftheNiakharSocialNetworksandHealthProject (NSNHP),alarge,mixed-methodsprojectfunded bytheU.S.NationalInstituteofGeneralMedicalSciences (NIGMS).Byredressingfundamentalproblemsinconventionalsurveynetworkdatacollectionmethods,the projectisaimedatimprovinginferencesconcerningtheassociationbetweensocialnetworkstructuresand processesandhealthbehaviorsandoutcomes.Fieldedincollaborationwithanongoingdemographicandhealth surveillancesysteminruralSenegal,theNSNHPincludesqualitativedataconcerningthedimensionsofsocial associationandhealthideologiesandbehaviorsinthestudyzone,twopanelsofanewsocialnetworksurvey,and severalsupplementaryandafﬁliateddatasets.

Longitudinalsocialnetworksurveylinkedtopre-existingsurveillancedata

Addressesfundamentalmethodologicalconstraintsinprevioussocialnetworkdata

Enablessocialnetworkanalysesofhealthbeliefs,behaviors,andoutcomes

Methodname:Socialnetworksurvey

Keywords:Socialnetworks,Demography,Publichealth,Surveymethodology

Articlehistory:Received28March2019;Accepted30May2019;Availableonline5June2019

SpeciﬁcationsTable

SubjectArea: SocialSciences

Morespeciﬁcsubjectarea: Populationstudies,socialnetworks Methodname: Socialnetworksurvey

Nameandreferenceoforiginalmethod N/A

Resourceavailability: www.nsnhp.org

Methoddetails

Socialnetworkprocesses,throughthecontentofnetworktiesandtheirstructuralcharacteristics, arekeytounderstandingdiffusionofinnovationandadoptionofnewbehaviorswithregardtohealth. Criticallimitationsinpriordatacollectiondesigns,however,generallyprecludeunbiasedestimatesof sociallearning,inﬂuenceanddiffusionprocessesthroughnetworks,andinferencesconcerningthem. Thesedesignlimitationsincludecensoringofthetypesofties(orrelationships)betweenindividuals elicited,thenumberofnetworkassociates(alters)elicitedwithineachtypeoftie,andrespondent(or ego)reportsofalters’characteristicsandbehaviors.Further,thevastmajorityofsocialnetworkdata that have been collected to date concerning health are egocentric (eliciting network ties from respondentsbutnotbetweenmembersof thefullnetwork),makingitimpossibletoanalyzethe potentiallycriticalimpactofstructuralcharacteristicsofnetworksonhealthbehaviorsandoutcomes [1,2].

TheNiakharSocialNetworksandHealthProject(NSNHP),fundedbytheU.S.NIGMS,wassetupto addresstheseproblemsandprovideauniquesourceofindividual-level,longitudinalsocialnetwork dataonhealthbehaviors.TheNSNHPcollectedtwopanelsofsurveydata(in2014and2016)froma ruralpopulationinEasternSenegalthroughaninnovativeresearchdesignwherebynetworkalters cited by survey respondents are linked to longitudinally-collected data concerning theirhealth behaviorsinalong-standingdemographicsurveillancesystem,theNiakharHealthandDemographic Surveillance System (NHDSS), member of Indepth-Network [3].1 _This _innovation_allows _for _the

1

www.indetph-network.org.

(4)

identiﬁcation of morenetwork members acrossmore typesof ties than generally possiblewith conventionalsurveydesigns.Itfurtherallowslinkageofbothrespondentsandtheirnetworkalters(as wellas theirkin and community members)to high-quality,prospectivelycollected longitudinal health,demographicandeconomicdataavailableintheNHDSS.Together,thesedatamakepossible more_{ﬁne-grained} analyses of social learning and diffusionthan have generallybeen previously possibleinepidemiological,demographic,andpublichealthresearch.

Setting

The NHDSS is locatedinthe Niakharand Diarère districts ofthe departmentofFatick (Sine-Saloum), 135kmeastofDakar,Senegal.Itencompasses8villageswhichhavebeenunderdemographic surveillancesince1963,and22othervillageswhichwerebroughtundersurveillancefortheﬁrsttime in1983(Fig. 1).Thepopulationofthesurveillanceareahasdoubledinthepast30years,coveringmore than44000peopleatthetimeofdatacollectionfortheﬁrstpaneloftheNSNHPin2014.

The economy ischaracterized by rainfedagriculture(with rotationof millet,groundnuts and fallow),andlivestock productionof cattle,sheepand goats[4]. Diversiﬁcationof cropshasbeen ongoingforseveraldecadesduetothegovernment'swithdrawalofsupportfromtheproductionof groundnutsandariseinrainfalllevelsthatfavorsthecultivationofwatermelonandmarketgardening [5]. Circular movements of migration towards urban areas, particularly the capital, Dakar, have intensiﬁedinrecentdecades[6].

(5)

NiakharDemographicandHealthSurveillanceSystem(NDHSS)data Surveillancedata

Asdiscussedabove,thefundamentalinnovationofthisprojectistoﬁeldthesocialnetworksurvey instrument over a population that hasbeen (forthe last 50 years), and continues tobe, under continuous demographic surveillanceby the IRD, throughthe Niakhar Demographic and Health SurveillanceSystem(NDHSS).TheNDHSScollectsdata,longitudinally,onavarietyofdemographic, social,andhealthphenomena.Theseinclude,butarenotlimitedto,pregnancies,maternalhealth visits, antenatal care, weaning, births, stillbirths, miscarriages, traditional and modern medical assistanceatbirth,fever-inducingillness,theuseofhealthfacilities,vaccinationsforchildren,useof medical treatments and prophylaxes such as anti-malarial medications and treated bed-nets, mortality(includingverbalautopsiesforthedeceased),geospatiallocationofresidenceandcomplete migrationhistories.Investigatorsareabletolinkthesedatatorespondents,membersofrespondents’ households and communities, and, using the name identiﬁers collected in the main survey instrument,respondents’networkalters.

Thehouseholdsurveyofhouseholdgoodsandequipment

Investigatorsarealsocapableoflinkingrespondents,theirnetworkalters,andallotherindividuals inthesurveillancezonetoextensiveinformationconcerninghouseholdwealth,materialpossessions, facilities(power,buildingmaterials,waterandsanitationamongothers),andagriculturaland non-agriculturalproductionthroughahouseholdcensusconductedinthezonebytheIRDatthesametime astheﬁrstpanelin2014.

NSNHPinstrumentsanddata

Qualitativedataonsociabilityinthestudyzone

In2007,semi-structuredin-depthinterviewswereconductedwithastratiﬁedrandomsampleof 24adultsintheNHDSSstudyzone.Theinterviewguidewasstructuredusing30commonlyused questionsdesignedtoelicitnamesofindividualswithwhomtherespondenthadacertaintypeoftie, orrelationship(knowsas‘namegenerators’inthesocialnetworkliterature)aspromptsfordiscussion. Theaimofthisinvestigationwastoassessthedimensionalityofsociabilityinthezoneandtoidentify majortypesofandmodalitiesofsocialtiesbetweenindividuals.Availableontheprojectwebsite, thesetranscribedandtranslatedinterviewswereusedtoidentifyadiscretesetofnamegenerators hypothesizedtocapturethemostcomprehensivepersonalnetworkspossibleinthispopulation. Pilotnetworksurvey

In2008,apilotsurveywasﬁeldedoverastratiﬁedrandomsampleof141adultsfromthestudy zone. This pilot contained 15 name generator questions derived from the qualitative study of sociabilityinthezone,across4theoreticallykeydomainsofinteraction.Theseincludedaffectiveties, thoseprimarilyassociatedwithexchangeor support,those whosesaliency wasprimarilydueto frequency ofor time spentin interaction, and role-relational ties, structural orinstitutional ties potentiallyentailingobligation,andoneresidualnamegenerator[11–15].Alterelicitationforeach namegeneratorwasfreechoice,withrespondentsallowed(andprompted)tonameasmanyaltersas theywishedineach,allowingformultiplexity,orthenamingofuniquealtersacrossmultiplename generators.

For each alter elicited in the name generators, the survey instrument gathered additional information.Theseincludedadiscretesetof‘nameidentiﬁers’,orquestions,derivedfromqualitative research concerning how residents in this population uniquely identify each other, that were hypothesizedcollectivelytoprovidedispositiveidentiﬁcationofaltersintheNDHSSsurveillancedata. Inaddition,extensive‘nameinterpreters’,orquestionsconcerningthecharacteristicsofaltersorthe relationship between ego and alter were also collected. These included questions concerning relationshipwiththealter,kinshipstatus,timespent(anddesiredtobespent)withthem,aswellas

(6)

measuresoftiestrength,includingapsychophysicalmeasureofsubjectivevalueofeachaltertothe respondent.

Qualitativediseasenarratives

In 2012,semi-structuredqualitativeinterviewsconsistingofillness ordiseasenarratives were conductedwithasimplerandomsampleof98individualsabovetheageof16inthestudyzonefrom thesurveillancesystemdatabasewiththeprimaryaimofdevelopsurveymeasuresofideational contextcapableofdiscriminating cognitiveschemasrelatedtohealth.Theinterviewguideasked respondentstodiscussarecentepisodewhentheyorsomeoneclosetotheminthefamilyhadbecome sick, and when time allowed, a second episode was discussed. For each episode, respondents discussedtheinitialrecognitionof illness,symptoms,perceivednosologyand etiology(including broader discussion of these topics), sequential therapeutic steps taken (auto-medical and professional,biomedicalandethnomedical)reasonsfortheparticularsequencingpursued,logistical arrangementsandﬁnancialcosts,andperceivedoutcomesoftherapyastheroleofothers(family, neighbors,socialnetworkmembers,medicalpractitioners)ineachoftheseareas.Interviewswere conductedinthelocallanguage,transcribedandtranslatedintoFrench,thenenteredintoNVIVO qualitativesoftwareandcodedforreferencestodetaileditemswithinthebroadcategoriesofcauseof illness,placeoftreatment,illness/symptoms,helpgiven/receivedtypeoftreatment,judgementsof responsibility,perceptionofgoodhealthandotherindividualsimplicated.

Mainpanelsurvey

TheNSNHP_fieldeditsbaselinesurvey(orpanel1)in2014,consistingtwodatacollectionswith identicalinstruments.Thefirst,or‘populationsample’wasasimplerandomsampleofresidentadults and adolescentsage 16and above drawnfrom the NDHSS surveillancedata. The second was a completecensusofadultsandadolescentsaged16andaboveinonevillage(the‘sociocentricvillage’). Thisvillagewaspurposivelyselectedfromthe30villagesinthesurveillancezonebecauseofseveral desirablecharacteristics.Itisoneoftheoriginal8villageswithsurveillancedatadatingbackto1963.It isarelativelylargecommunity,withestablishedinstitutions(ahealthpost,schools,commercialarea andtowncenter),butisnotofthesamesizeandgeographicdispersionofthetwomajortownsinthe zone. It is also ethnically homogenous but religiously diverse (with a significant community of CatholicsinadditiontothemajorityMuslimpopulation).Mostimportantly,ithasthreecharacteristics whichconstrainsociabilityandinterpersonalinteractiontootherindividualswithinitsboundariesto agreaterdegreethaninothervillagesinthezone,acriticalcriterionforadequateidentificationof sociocentricnetworkstructure[7].First,residencesinthevillageareconcentratedinasmallsetof neighborhoodsaround thecenter ofthe village, and theseneighborhoods arerelativelyisolated geographicallyfrompopulationcentersoftheneighboringvillageswhichsurroundit.Second,while inmanyvillagesinthesurveillancearealandusedforagriculturecanbelocatedatsomedistancefrom the users’ households, in the sociocentric village agricultural land is locatedin relatively close proximitytohouseholds,withinandontheimmediateoutskirtsoftheresidentialneighborhoods, limitingresidents’exposuretothosefromothervillageswhileinthefields.Finally,unlikeotherlarger villagesinthezone,itislocatedononlyonelocalroad,removedfromthehighwayconnectingthe majorregionalpopulationcenters.Thesecharacteristicsincombinationstrengthentheassumptionof arealistboundaryspecificationwithregardtosocialassociationwithinthevillage.

The NSNHP follow-up survey (or panel2), fielded in 2016,attempted re-interviews with all populationsamplerespondentsfromthefirstpanelthestillresidentinthezone.Thissamplewas refreshedwithasimplerandomsamplefromthesurveillancesystemofresidentsnewlyaged16and over,ornewlyarrivedinthezone,tomaintainpopulationrepresentativeness.Interviewswerealso attemptedwithallfirstpanelrespondentsfromthesociocentricvillagewhowerestillresidentthere, aswellasresidentswhohadmovedintothevillageorhadturned16yearsoldbetweenpanels.

Table1presentsthesummaryinterviewresultsforbothNSNHPlongitudinalpanels.Forpanel1, 1310respondentswereinterviewedinthesociocentricvillage,and882respondentsinthepopulation sample.Thosemarked‘absent’wereeithernotreachableafter3attemptsorwereidentiﬁedbykey

(7)

respondentsashavingmigratedoutsideofthezone,thoughtheywereretainedasresidentsinthe surveillance system. Removing these respondents, as well as those too incapacitated to be interviewed, and sample members who had been removed from the surveillance system after compilationofthesamplelist,weobtainresponseratesof96%forthesociocentricvillageand99%for thepopulationsampleamongdefacto,non-incapacitatedresidentsamplemembers.

Informationconcerningretentionandattritioninpanel2,ofthoseinterviewedinpanel1isalso presentedinTable1.Overall,forthoseinterviewedinpanel1,weobtainretentionratesof73.4%and 69%forthesociocentricvillageandpopulationsamples,respectively.Asintheﬁrstpanel,however, samplemembersineachgroupwereabsentduetomigrationbutmaintainedinthesurveillance systemasresidents,haddiedorbecometooincapacitatedforinterview.Additionally,anumberof samplemembersineachgrouphadbeenremovedfromthesurveillancesystemforoneofthese reasons.Takingtheseissuesintoaccount,weobtainretentionratesacrossthepanelsof94%and88% amongthesampleamongdefacto,non-incapacitatedresidentsamplemembersinthesociocentric villageandpopulationsamples,respectively.

Finally,Table1indicatesthat,inpanel2,251newrespondentswereaddedtothesociocentric sample and 258 tothe population sample. These represent response rates among the de facto populationofthoseeligibleasdeﬁnedaboveof94%and98%,respectively.

Surveyquestionsinthemainpanelincluded16namegeneratorsandnameidentifiersasdescribed above concerning the pilot data collection with minor modifications,including two new name generatorsaimedatelicitingaltersfromwhomrespondentsreceivedhelpwhensick,andtowhothey providedsuchhelp.Informationwasalsocollectedforeachalterelicitedconcerningrelationshiptype and duration, kinship affiliation, frequency of interaction, relative socioeconomic position,

Table1

Interviewresultspanels1and2,NSNHP.

Sociocentricvillagesample Populationsample Panel2 Panel2 Panel

1

Completepanel 1

Supplement Total Panel 1 Completepanel 1 Supplement Total Complete 1310 966 258 1224 882 606 251 857 Incomplete partial 14 9 3 12 48 12 6 18 absent 193 184 114 298 290 93 120 213 incapacitated 32 7 8 15 0 6 3 9 dead – 11 3 14 – 1 6 7 notstarted 0 23 6 29 12 62 6 68 refusal 48 29 16 45 6 12 6 18 removedfrom surveillance 26 81 – 81 – 90 – 90 Total 1623 1310 408 1718 1238 882 398 1280 Source:compiledbyauthors.

Table2

Mean(std.dev)numbersofalterselicited,uniquealters,altersidentiﬁableintheNDHSSandproportionofaltersidentiﬁableby sampleandpanelwave,NSNHP2014–16.

Named Alters Identiﬁable Proportionidentiﬁable Populationsample Panel1 45.61(15.83) 26.48(9.1) 21.94(8.37) 0.828(0.138) Panel2 47.02(18.88) 28.19(11.42) 23.58(10.02) 0.838(0.140) Sociocentricvillage Panel1 40.11(14.55) 23.58(8.19) 19.51(7.58) 0.821(0.132) Panel2 43.55(17.32) 25.09(10.4) 20.57(9.07) 0.822(0.142) Source:compiledbyauthor.

(8)

psychometrictiestrengthandindicatorsoftypeofhealthaidreceivedorgivenfrom/toaltersinthose speciﬁcnamegenerators.

Table 2 presents summary statistics for number of alters elicited from respondents for the populationsampleandsociocentricvillagebysurveypanel.Thefirstcolumn‘named’,referstothe totalnumberofnameselicitedregardlessofrepetition(ormultiplexity)acrossnamegenerators.The column‘alters’referstouniquealterscitedeliminatingmultiplexity,and thatfor ‘identifiable’to uniquealterswhohadlivedorcurrentlyliveinthesurveillancezoneandthereforarepotentially identi_fiable in the NDHSS data. The _final column presents estimates of the respondent-level proportions of alters potentially identifiable. As seen here, relatively small, but non-negligible proportionsofuniquealterselicitedarefromareasoutsidethesurveillancezone,andthereforehave noNDHSSdataassociatedwiththem.Onaverage,about3–4outof20–24uniquelycitedaltersarenot identifiableintheNDHSS.

In addition to the network instruments, the ﬁrst panel survey included a large respondent questionnaire including of 20 questions concerning health ideation and behavior derived from analysisofthequalitativediseasenarrativesaimedatmeasuringcognitiveandbehavioralschemasof health and illness in this population. The respondent questionnaire also included batteries of questionsconcerninghealthanddemographicphenomenaofwidepublichealthinterestincluding familyplanningandreproductivehealth,thestatusofwomenandperceptionsoftheacceptabilityof intimatepartnerviolence(IPV)aswellasmentalhealth.

Socialnetworksinboththesociocentricandpopulationsamplesarefoundtoradiateoutwardwith a decaying density function from the household, to the residential compound, residential neighborhood, villageof residence and otherareas in Senegal and beyond [8]. For both panels, approximately90%ofalluniquealterselicitedwereresidentsofthesamevillageastherespondent. Thenumbersofrespondents_’alterswhowereadministeredthemainpanelinstrumentsarenegligible inthe(randomlyselected)populationsample.However,since96%ofallrespondentsovertheageof 16wereinterviewedinthesociocentricvillage,thevastmajorityofalterscitedwerealsointerviewed withthe main survey instruments. Thisallows the full complement of characteristics from the respondentquestionnaire tobeemployedas characteristicsof respondents’ within-villagesocial networks.

Supplementarydata

In2015,betweenthetwopanelsofthemainsurvey,asmallerre-interviewsurveywitharandom sampleof300_ﬁrstpanelrespondentswasconductedusingthemainsurveyinstrument,augmented with questions to test a variety of issues concerning social network survey methodology and reliability.Theseincluded questionsconcerningrespondents knowledgeof alters’ characteristics, includinghealthideation,fertilityhistory,andhouseholdcharacteristicstoassessthevalidityofalter reportsasgenerallycollectedinsocialnetworksurveys.

At the same time, a roster-based version of the main network instrument was fielded on 500individualswhohadbeencitedasnetworkaltersbyrespondentsinthefirstpaneltoenablean alternativeestimateofnetworkdensityfromthealter’sperspective.Inthisinstrument,respondents werereadalistofnames,randomlyselectedfromthepopulationtobeproportionallyrepresentative oftheresidentialandgeographicdistributionofaltersamongrespondentsinthemainsurvey.Oneof thesenameswassubstitutedwiththenameofthefirstpanelrespondentwhohadcitedtheroster respondentasanalter.Foreachnamereadtotherespondent,theywereaskedifthatpersonwouldbe citedeachofthenamegeneratorsusedinthefirstpanelinstrument,andifso,tocompletethename interpretersfromthatsurveyforthosealters.

LinkingalterscitedtoNDHSSsurveillancerecords

Oneoftheprincipalchallengesofthisresearchhasbeenlinkingnetworkaltersidentiﬁedinthe surveyinstrumenttotheircorrespondingrecords(iftheyexist)intheNDHSSsurveillancesystem.This isa deceptivelysimple problemuntil oneconsiders thepotential sourcesof erroronboth sides, especiallyconcerningnames.Asdiscussedabove,aseriesof‘nameidentiﬁer’questionswereasked

(9)

concerningeachalterelicitedtoaccomplishthis.Theseincludedalters’andalters’parents’names,the namesoftheheadofalters’residentialcompoundandtheircurrentresidentiallocalization(inthe sameresidentialcompound,inadifferentcompoundinthesamevillage,inadifferentvillageinthe studyzone,inthecapitalDakar,somewhereelseinSenegal,international,ordeceased).Inthecaseof localizationsoutsidethesurveillancezone,priorresidentiallocalizationwithinit,ifany,wasalso ascertained.Alsocollectedwere basicinformation onsex,age relativetotherespondent (older, younger,aboutthesameage),matrilinealclan,whetherthealterhad everbeenmarried,and for current migrants, their duration of absence. An algorithm was developed which evaluated the likelihood of identical information in each of these identifiers in surveillance system records conditionalonthejointthedistributionofotheridentifiersintheset.Inpanelone,anidentifier weightingmatrixwasdevelopedfromthepilotdataandthealgorithmwasimplementedaspartofthe CAPIsurveydatacollection,evaluatingthelikelihoodofimprovementincandidatematchingwithany subsequentpieceofinformationaftereveryresponsetoanameidentifier.Thiswasdonewiththegoal of minimizing survey time, stoppingcollection after furthername identifiers wereunnecessary. Matchingresultsusingthismethodwereunsatisfactory,however,asrespondenterrorratesinearly identifiers elicited were not adequately accounted for, and it was discovered that substantial heterogeneityintheidentifierweightmatrixexistedbyresidentiallocation.Toremedythis,trained localresearchassistantsevaluatedasampleof 2000altersincomparison tothefullsurveillance systembyhandtoidentifycertainandhighlikelihoodmatches.Monte-Carlomethodswerethenused todevelop identifier weighting matrixescapable ifidentifyingthehighest percentage of correct matchesateachcategoryofcurrentorpriorresidentiallocalization.Basedonthisexperience,inthe reinterviewandpanel2datacollection,allnameidentifierswerecollectedconcerningeachaltercited, andthematchingalgorithmwasimplementedinanidenticalfashioninpost-processing.

Estimatesoftheproportionofalterscorrectlyidenti_fiedbythealgorithminrandomveri_fication samplesofmatchesstratifiedbycurrent(orprior)residentiallocalizationinthesurveillancezonefor bothpanelsofthesurveyandfortheinterveningreinterviewinstrumentarepresentedinTable3. Overall,thesensitivityofthematchingprocesswaslowerinpanelone(81.8%)thanthereinterviewor secondpanel(91%and86%,respectively)forreasonsoutlinedabove.Inbothmainpanels,correct linkageofaltersdiminshedwithdecreasingresidentialproximity.Theestimatedproportionsofalters correctlymatchedforthesecondpanelwerelowerthanthoseforthereinterviewlargelybecause alters’ parents’ names were not collectedat the same rate as in thereinterview instrument, a deficiencyininterviewertrainingandqualitycontrol.Currentresearchisproceedingwiththeaimof

Table3

ProportionofidentiﬁablealterssuccessfullymatchedtoNDHSSsurveillancerecordsbylocalizationofaltersandpanel,NSNHP 2014–2016(95%CI).

Proportionmatched Proportionalters Proportionaltersmatched Panel1(n=200/strata)

sameresidentialcompound 0.875(0.831,0.919) 0.387 0.339(0.322,0.356) samevillage 0.793(0.737,0.848) 0.518 0.411(0.382,0.439) othervillage 0.709(0.648,0.769) 0.096 0.068(0.062,0.074)

Total 0.818(0.766,0.869)

Reinterview(n=100/strata)

Total 0.910(0.856,0.964)

Panel2(n=200/strata1,3,500strata2)

Total 0.861(0.823,0.897)

Source:compiledbyauthors.

(10)

increasingtheproportionofalterscorrectlymatchedtothesurveillancedatathroughtheuseofanew algorithmdevelopedtoevaluateconcordanceofaltermatchesacrosssurveyspanels.

Strengthsandlimitations

ThemainstrengthsoftheNSNHPdataarethebroadandcomprehensivenetworkcomponent,and thelinkagebetweenthiscomponentand thesurveillancedataoftheNHDSS.Elicitationofsocial networkstookplaceacrossawide,theoreticallywarranted,andculturallyappropriaterangeoftypes ofinteraction,withoutplacingconstraintsonthenumberofalterselicitedasinthevastmajorityof priornetworksurveydesigns.Detaileddatawasalsocollectedconcerningeachnetworktie,aswellas awiderangeofhealthanddemographicbeliefsandbehaviorsfromrespondents,and,asdiscussed above,inthesociocentricvillagesample,theirnetworkalters.

Thepopulationsamplecanbeusedtoproduceestimatesofassociationsbetweenhealthbeliefsand behaviorsandnetworkcharacteristicsthatarerepresentativeofthebroaderruralpopulationofthe surveillancezoneinamanneridenticaltodatafromconventionalegocentricnetworkdesigns.The sociocentricsample,which identiﬁesnetworktiesbetweenallmembersofthepopulationof the sociocentricvillage,allowsadditionallyfortheestimationofeffectsassociatedwiththestructural propertiesofnetworks.

Fig.2presentsthesociogramfromthefirstpaneloftheNSNHPforthesociocentricvillage.Thisisa graphicdepictionofnetworkconnectionsbetweenrespondents.Thisfiguredepicts8distinctnetwork clusters, or sub-networks identi_fied through the Louvain modularity method [9], roughly correspondingtoneighborhoodswithinthevillage.

Sociocentricdatasuchasthesearecapableofgeneratingcriticalmeasuresnetworkstructureand positionthatcannotbeobtainedwithegocentricdata.Theseinclude(butarenotlimitedto,network density,reciprocityandtransitivity,aswellnetworkcentrality,allofwhichhavebeenhypothesizedto haveimportantimplicationsfortheexplanationofhealthbeliefsandbehaviors[10].Thistypeofdata, concerningbroadpersonalnetworksacrossmultipledomainsofassociationhasrarelycollectedinthe contextofhealthresearch,andwhereithasbeen,ithasbeenlimitedbydesigninthewaysdiscussed above[1],andhasneverbeencollectedbeforeasaprospectivepanel.

Ofequalimportance,theNSNHPnetworkdataislinkedtotheongoing,prospectivedatacollection intheNHDSSforallmembersof thepopulation,includingrespondents,membersoftheirsocial networks, households, kin groups, neighborhoods, and communities as well as extensive supplementarysocioeconomic data collectedon the same population. In combination with the networksurveypanelsandsubstantiveelementsoftheNSNHPsurveys,thismakespossiblesystems analysesofchangeinsocialcontextonhealthandhealthbehaviors,includingtheirdiffusion,not possiblebefore,includingestimationoftheimpactofchangesinthestructureofsocialnetworksand thesubstantivecharacteristicsofalterswithinthemovertime.

(11)

The main weaknesses of the NSNHP data are the limited interval between panels and the integrationwiththeNHDSSdata.TheintervalbetweentheﬁrsttwopanelsoftheNSNHPwasonlytwo years, limiting the ability to assess the association between changes in network structure and composition,andchangeinhealthbeliefsandbehaviors.Fundinghasbeenrequestedtoextendthe NSNHPcohorttotwofurtherpanelstoaddressthisissue.Likeanycomplexlongitudinaldatasource, the NDHSS data themselves can be cumbersome to use, and doing so generally requires close collaborationwithIRDdatamanagersorotherexpertsinitsstructuretoensurequalitymeasurement. Thoughlinkingthisvastdatatothenetworksurveydataisasigni_ﬁcantstrength,italsopresents challengesforinferencesaboutassociationsbetweenindividualandsocialnetworkcharacteristics whicharethefocusoftheproject.Asdiscussedabove,theprocessoflinkingalterstotheirsurveillance recordswasnotperfect.Thiswillhavenoeffectonestimatesrelatedtonetworkcharacteristicsand structure derivedsolelyfromthesurvey,but isa sourceofmeasurementerrorinindicators and structuralcharacteristicsderivedfromlinkagetotheNDHSSdata.

Dataaccess

Datacollectedfromthequalitativeinterviewsanddeidentifiedispubliclyavailableontheproject website(http://www.nsnhp.org).Allsurveydatafromthepopulationsample,strippedofidentifying information will be made available for restricted use through existing data access protocols maintainedbytheNSNHPandNDHSS(requiringappropriateusejustificationandinstitutionalreview boardapprovalbytheIRDandrequestinginstitution).Datafromthesociocentricsamplearemore sensitive.Thisdatawillalsobemadeavailabletotheresearchcommunity,buta morestringent screeningandsecurityprocesswillbeimplementedfollowingNSNHPandNDHSSprotocolsforaccess andcon_{fidentiality.}Thedatacontainedinthesurveillancesystemandthesurveywhichwillbelinked tothenetworksurveyarehighlysensitiveandarethepropertyoftheNDHSS.Theywillbemade availabletotheresearchcommunityatthediscretionofandfollowingNDHSSconventionalrestricted accessprotocolsaswell.

Acknowledgement

ResearchreportedinthispublicationwassupportedbytheNationalInstituteofGeneralMedical SciencesoftheNationalInstitutesofHealthunderawardnumberR01GM096999.

References

[1]J.M.Perkins,S.V.Subramanian,N.A.Christakis,Socialnetworksandhealth:asystematicreviewofsociocentricnetwork studiesinlow-andmiddle-incomecountries,Soc.Sci.Med.125(2015)60–78SpecialIssue:SocialNetworks,Healthand MentalHealth.

[2]J. Sandberg, Analysedes réseauxsociaux: l’apportdes systèmes desuivi démographique,Niakhar, Mémoires et Perspectives.RecherchespluridisciplinairessurlechangementenAfrique,IRDEditions,Dakar,Senegal,2018. [3]V.Delaunay,L.Douillot,A.Diallo,D.Dione,J.-F.Trape,O.Medianikov,etal.,Proﬁle:theNiakharHealthandDemographic

SurveillanceSystem,Int.J.Epidemiol.42(4)(2013)1002–1011.

[4]A.Lericollais,Paysanssereer:dynamiquesagrairesetmobilitésauSénégal,IRDEditions,1999.

[5]D.Masse,R.Lalou,C.TIne,M.Ba,J.Vayssieres,LestrajectoiresagricolesdanslebassinarachidierauSénégal:élémentsde réﬂexionàpartirdel’observatoiredeNiakhar,Niakhar,MémoiresetPerspectives.Recherchespluridisciplinairessurle changementenAfrique,IRDEditions,Dakar,Senegal,2018,pp.311–339.

[6]R.Lalou,V.Delaunay,Migrationssaisonnièresetchangementclimatiqueenmilieururalsénégalais:formeouéchecde l’adaptation?in:B.Sultan,R.Lalou,M.AmadouSanni,A.Oumarou,M.A.Soumaré(Eds.),Lessociétésruralesfaceaux changementsclimatiquesetenvironnementauxenAfriquedel’Ouest,Synthèses,IRD,Marseille,2015,pp.287–313. [7]E.O.Laumann,P.Marsden,D.Prensky,Theboundaryspeciﬁcationprobleminnetworkanalysis,in:M.Minor(Ed.),Applied

NetworkAnalysis,SagePublications,BeverlyHills,CA,1983.

[8]S.Rytina,J.Sandberg,V.Delaunay,R.Lalou,Animprovedegocentricnetworkmethodology:theNiakharSocialNetworks PilotSurvey,PopulationAssociationofAmericaAnnualMeetings(2008).

[9]V.D.Blondel,J.-L.Guillaume,R.Lambiotte,E.Lefebvre,Fastunfoldingofcommunitiesinlargenetworks,J.Stat.Mech. TheoryExp.2008(10)(2008)P10008.

[10]T.W.Valente,SocialNetworksandHealth:Models,Methods,andApplications,1ed.,OxfordUniversityPress,Oxford,New York,2010.

[11]A.Marin,K.N.Hampton,Simplifyingthepersonalnetworknamegenerator:alternativestotraditionalmultipleandsingle namegenerators,FieldMethods19(2)(2007)163–193.