social capital and education outcomes in urban and rural peru · social capital and education...

W O R K I N G P A P E R N O . 2 8

Social Capital and Education Outcomes in Urban and Rural PeruSantiago CuetoGabriela GuerreroJuan LeonMary De SilvaSharon HuttlyMary E PennyClaudio F LanataEliana Villar

Although enrolment in primary schools in Peru is very high, more than half of primary school children are one or more grades below the norm for their age. Furthermore, evaluations have shown that, when tested, Peruvian school children score well below the norms expected for their age. Their scores are also below the average levels of countries with similar socio-economic circumstances. The role of social capital in explaining these findings has not been studied, although research in the USA has suggested positive associations between social capital and educational achievement. Social organisations that focus on childcare are one example of strong community networking resources in Peru. The Young Lives study offers an opportunity to investigate whether social capital is associated with educational progress and achievement.

The results of the study confirmed poor educational outcomes for many Peruvian school children. High proportions were unable to master simple tasks and were in a lower school grade than they should have been for their age group. There is a clear negative association between educational achievement and poverty. Overall, rural students are poorer and are thus more prone to low achievement (lower results in tests) and falling behind their expected grade. However, there seem to be no significant differences between boys and girls in these outcomes.

The results do not support the hypothesis of a positive association between the social capital available to the family and the educational outcomes of their children, except for the association between cognitive social capital (at the community level) and children being in the correct grade for their age. This means that communities which experience more quality in their social relationships (e.g. trust) are more likely to have children who are in the correct school grade for their age. This result applies to how children are progressing at school (whether they are in the correct grade for age) but does not extend to children’s achievement (results in tests).

Perhaps the main interventions to improve the quality of the Peruvian education system are not to be found in the quality and quantity of social relationships within communities, but in improving educational inputs and processes within schools and breaking the strong association found between socio-economic status and educational performance. On the other hand, it might be that the types of instruments (objective questionnaires) and analysis (quantitative) used in this study do not offer the best way to capture the importance of social capital. If this is the case, then qualitative designs and analyses should be used to explore these issues.

Published by

Young Lives Save the Children UK 1 St John's Lane London EC1M 4AR

Tel: 44 (0) 20 7012 6796 Fax: 44 (0) 20 7012 6963 Web: www.younglives.org.uk

ISBN 1-904427-29-4 First Published: 2005

All rights reserved. This publication is copyright, but may be reproduced by any method without fee or prior permission for teaching purposes, though not for resale, providing the usual acknowledgement of source is recognised in terms of citation. For copying in other circumstances, prior written permission must be obtained from the publisher and a fee may be payable.

Designed and typeset by Copyprint UK Limited

Young Lives is an international longitudinal study of childhood poverty, taking place in Ethiopia, India, Peru and Vietnam, and funded by DFID. The project aims to improve our understanding of the causes and consequences of childhood poverty in the developing world by following the lives of a group of 8,000 children and their families over a 15-year period. Through the involvement of academic, government and NGO partners in the aforementioned countries, South Africa and the UK, the Young Lives project will highlight ways in which policy can be improved to more effectively tackle child poverty.

The Young Lives Partners Centre for Economic and Social Studies (CESS), India

Department of Economics, University of Addis Ababa, Ethiopia

Ethiopian Development Research Institute, Addis Ababa, Ethiopia

General Statistical Office, Government of Vietnam

Grupo de Análisis para el Desarrollo (GRADE), Peru

Institute of Development Studies, University of Sussex, UK

Instituto de Investigación Nutricional (IIN), Peru

London School of Hygiene and Tropical Medicine, UK

London South Bank University, UK

Medical Research Council of South Africa

RAU University, Johannesburg, South Africa

Research and Training Centre for Community Development, Vietnam

Save the Children UK

Statistical Services Centre, University of Reading, UK

Social capital and Education outcomES in urban and rural pEru

Social capital and Education outcomes in urban and rural peruSantiago cuetoGabriela GuerreroJuan leonmary de SilvaSharon Huttlymary E pennyclaudio F lanataEliana Villar

WorKinG papEr no.28


ii

ii

prefaceThispaperisoneofaseriesofworkingpaperspublishedbytheYoungLivesproject,aninnovativelongitudinalstudyofchildhoodpovertyinEthiopia,India(AndhraPradeshState),PeruandVietnam.Between2002and2015,some2,000childrenineachcountryarebeingtrackedandsurveyedat3–4yearintervalsfromwhentheyare1until14yearsofage.Also,1,000olderchildrenineachcountryarebeingfollowedfromwhentheyareaged8years.

YoungLivesisajointresearchandpolicyinitiativeco-ordinatedbyanacademicconsortium(composedoftheUniversityofReading,theLondonSchoolofHygieneandTropicalMedicine,LondonSouthBankUniversityandtheSouthAfricanMedicalResearchCouncil)andSavetheChildrenUK,incorporatingbothinterdisciplinaryandNorth-Southcollaboration.

YoungLivesseeksto:

producelong-termdataonchildrenandpovertyinthefourresearchcountries

drawonthisdatatodevelopanuancedandcomparativeunderstandingofchildhoodpovertydynamicstoinformnationalpolicyagendas

traceassociationsbetweenkeymacropolicytrendsandchildoutcomesandusethesefindingsasabasistoadvocateforpolicychoicesatmacroandmesolevelsthatfacilitatethereductionofchildhoodpoverty

activelyengagewithongoingworkonpovertyalleviationandreduction,involvingstakeholderswhomayuseorbeimpactedbytheresearchthroughouttheresearchdesign,datacollectionandanalyses,anddisseminationstages

fosterpublicconcernabout,andencouragepoliticalmotivationtoacton,childhoodpovertyissuesthroughitsadvocacyandmediaworkatbothnationalandinternationallevels.

Initsfirstphase,YoungLiveshasinvestigatedthreekeystorylines–theeffectsonchildwell-beingofi)accesstoanduseofservices,ii)socialcapital,andiii)householdlivelihoods.Thisworkingpaperisoneofaserieswhichconsideranaspectofeachofthesestorylinesineachcountry.Asaworkingpaper,itrepresentsworkinprogressandtheauthorswelcomecommentsfromreaderstocontributetofurtherdevelopmentoftheseideas.

TheprojectreceivedfinancialsupportfromtheUKDepartmentforInternationalDevelopmentandthisisgratefullyacknowledged.

Forfurtherinformationandtodownloadallourpublications,visitwww.younglives.org.uk.

•

•

•

•

•


iiiiii

acknowledgementsTheauthorsaregratefulforthecommentsprovidedbyTrudyHarphamonapreviousdraftofthispaper.Thisprojectwouldnothavebeenpossiblewithouttheextremeeffortandprofessionalismofthefieldteamsandlocalhelpers,underthesupervisionofSofíaMadrid.TheauthorswishalsotothankHéctorVerásteguiforhisdatabasemanagementassistance.

the authorsSantiagoCuetoisSeniorResearcherandExecutiveDirectoratGRADE,Lima.GabrielaGuerreroandJuanLeonareAssistantResearchersatGRADE.MaryDeSilvaandSharonHuttlyarefromtheUniversityofLondonSchoolofHygieneandTropicalMedicine.MaryEPennyisDirectorof,andClaudioFLanataisPrincipalInvestigatorat,theInstitutodeInvestigaciónNutricional,Lima,andElianaVillaristheNationalCo-ordinatorforYoungLivesPeru,SavetheChildrenUK.


iv

abstract AlthoughenrolmentinprimaryschoolsinPeruisveryhigh,morethanhalfofprimaryschoolchildrenareoneormoregradesbelowthenormfortheirage.Furthermore,evaluationshaveshownthat,whentested,Peruvianschoolchildrenscorewellbelowthenormsexpectedfortheirage.Theirscoresarealsobelowtheaveragelevelsofcountrieswithsimilarsocio-economiccircumstances.Theroleofsocialcapital1inexplainingthesefindingshasnotbeenstudied,althoughresearchintheUSAhassuggestedpositiveassociationsbetweensocialcapitalandeducationalachievement.SocialorganisationsthatfocusonchildcareareoneexampleofstrongcommunitynetworkingresourcesinPeru.TheYoungLivesstudyoffersanopportunitytoinvestigatewhethersocialcapitalisassociatedwitheducationalprogressandachievement.

TheresultsofthestudyconfirmedpooreducationaloutcomesformanyPeruvianschoolchildren.Highproportionswereunabletomastersimpletasksandwereinalowerschoolgradethantheyshouldhavebeenfortheiragegroup.Thereisaclearnegativeassociationbetweeneducationalachievementandpoverty.Overall,ruralstudentsarepoorerandarethusmorepronetolowachievement(lowerresultsintests)andfallingbehindtheirexpectedgrade.However,thereseemtobenosignificantdifferencesbetweenboysandgirlsintheseoutcomes.Theresultsdonotsupportthehypothesisofapositiveassociationbetweenthesocialcapitalavailabletothefamilyandtheeducationaloutcomesoftheirchildren,exceptfortheassociationbetweencognitivesocialcapital(atthecommunitylevel)andchildrenbeinginthecorrectgradefortheirage.Thismeansthatcommunitieswhichexperiencemorequalityintheirsocialrelationships(egtrust)aremorelikelytohavechildrenwhoareinthecorrectschoolgradefortheirage.Thisresultappliestohowchildrenareprogressingatschool(whethertheyareinthecorrectgradefortheirage)butdoesnotextendtochildren’sachievement (resultsintests).

PerhapsthemaininterventionstoimprovethequalityofthePeruvianeducationsystemarenottobefoundinthequalityandquantityofsocialrelationshipswithincommunities,butinimprovingeducationalinputsandprocesseswithinschoolsandbreakingthestrongassociationfoundbetweensocio-economicstatusandeducationalperformance.Ontheotherhand,itmightbethatthetypesofinstruments(objectivequestionnaires)andanalysis(quantitative)usedinthisstudydonotofferthebestwaytocapturetheimportanceofsocialcapital.Ifthisisthecase,thenqualitativedesignsandanalysesshouldbeusedtoexploretheseissues.

1 SeeSection1.1foradiscussionofdefinitions.


�

contents

Preface ii

Acknowledgements iii

Theauthors iii

Abstract iv

�. Introduction 2

1.1 Social capital 2

1.2 Social capital and education 4

1.3 community organisations in peru 6

2. Methods ��

2.1 design 11

2.2 Sample characteristics 11

2.3 Variables and procedures 14

3. Results 15

3.1 Social capital in peru 15

3.2 achievement 18

3.3 multivariate analysis 21

4. Discussion 28

References 32

Appendix:Factoranalysisforderivationofsummarysocialcapitalvariables 35


2

1. introductionEventhoughmostchildrenoldenoughtoattendprimaryeducationinPeruaregoingtoschool,morethanhalfofthemareoneormoregradesbelowthenormfortheirage.Theyareclassifiedas‘overage’,resultingfromenteringschoollate(aftertheageofsix)orrepeatingagrade.Asecond,probablyrelated,problemiswithachievement.Atleastonenationalandtwointernationalevaluationshaveshownthatstudentsscorewellbelowthenormsexpectedfortheirage,orbelowtheaveragesofcountrieswithsimilarsocio-economicstatus(egtheUNESCOevaluation,seeUMCandGRADE,2001;PISA,2003).

Althoughtherehavebeenafewstudiesthathavetriedtoexplainthisphenomenonofoverageschoolchildren(INEI,1995;Cortez,2000;Cueto,2004)andachievement(Benavides,2002),nonehaslookedattheimpactofsocialcapital(atleastnotusingthisspecificnamefortheconstruct).TheYoungLivesprojecthasincludedamongitsinstrumentsasetofquestionsrelatedtosocialcapital.Thestudydescribedinthispaperhasincludedthesesocialcapitalvariablesinthe7.5to8.5-year-oldcohorttotrytoexplaintwodependentvariables:(a)schoolgrade(whetherornotthechildisinthegradecorrespondingtoher/hisage),and(b)achievementinmathematicsandlanguage.Thegeneralhypothesisisthathigherlevelsofsocialcapitalwillbeassociatedwithhigherresultsforbothdependentvariables.

Thepaperisdividedintothefollowingsections:firstthereisadefinitionanddiscussionofthetermsocialcapital,includingareviewofrelevantstudies(thoserelatingsocialcapitaltoeducation).ThisisfollowedbyadescriptionofsomeofthenetworksandcommunityorganisationsthatcouldbelinkedtosocialcapitalinPeru(usingdatatakenfromnationalsurveys).EventhoughatpresentwedonothavevariablesrelatedtothesenetworksandorganisationsintheYoungLivesquestionnaire,thedescriptionofthemwillhelpputtheresultsintocontext.ItisexpectedthatfutureroundsoftheYoungLivesprojectwillincludevariablesrelatedspecificallytosocialcapitalorganisationsinPeru.Section2describesthemethodsusedinthestudyandSection3presentstheresults.Thefinalsectiondiscussestheresultsandprovidesananalysisofsomeimplicationsforfutureresearch.

�.�SocialcapitalAccordingtoPortes,theFrenchsociologistPierreBourdieuisresponsibleforthefirstcontemporarydefinitionofsocialcapital,understandingitas‘theaggregateoftheactualorpotentialresourceswhicharelinkedtopossessionofadurablenetworkofmoreorlessinstitutionalizedrelationshipsofmutualacquaintanceorrecognition’(Bourdieu,1980,citedinPortes,1998,p.3).Bourdieu’streatmentoftheconceptisinstrumental,asPortespointsout.Hefocusedon‘thebenefitsaccruingtoindividualsbyvirtueofparticipationingroupsandonthedeliberateconstructionofsociabilityforthepurposeofcreatingthisresource’(ibid,p.3).

JamesColeman’sworkisalsoamongtheearly,moderndefinitionsofsocialcapital.Hedefinessocialcapitalbyitsfunction:‘Itisnotasingleentitybutavarietyofdifferententities,withtwoelementsincommon:theyallconsistofsomeaspectofsocialstructures,andtheyfacilitatecertainactionsofactors–whetherpersonsorcorporateactors–withinthestructure’(Coleman,1988,p.S98).Colemanidentifiedthreeformsofsocialcapital:(a)obligations,expectationsandtrustworthinessofstructures,


3

(b)informationchannels,and(c)normsandeffectivesanctions.Theproblemwiththisclassificationisthatitincludes,underthetermsocialcapital,boththemechanismsthatgenerateit(reciprocityexpectationsandgroupenforcementofnorms,forinstance)andtheconsequencesofpossessingit(suchasaccesstoinformation).

Despitethisfact,Coleman’sworkisveryimportantandhasthemeritofdescribingsomeofthemechanismsthroughwhichsocialcapitalisgenerated.Hepointedouthowcertaintypesofsocialstructureareespeciallyimportantinfacilitatingcertainformsofsocialcapital.Amongthese,hedistinguishedstructureswith‘closure’.Thisistheexistenceofenoughlinksbetweenacertainnumberofpeopletoguaranteetheobservanceofnormsandtherewardofgoodactions(Coleman,1988).AlsoimportantinprovidingsocialcapitaliswhatColemanreferstoas‘appropriablesocialorganisation’.Thegeneralideaisthatanorganisation‘oncebroughtintoexistenceforonesetofpurposes,canalsoaidothers,thusconstitutingsocialcapitalavailableforuse’(ibid,p.S108). Putnam(1995)hassuggestedthatWesterndemocracies,especiallytheUSA,haveseenadeclineinrecentyearsinthecivilsocietymovement(‘civicengagement’),andthusinsocialcapital.Hebasesthisstatementonthedeclineofactivitiessuchasvoting,participatinginpublicorpoliticaleventsand,ratherfamously,bowling:‘ThemostwhimsicalyetdiscomfitingbitofevidenceofsocialdisengagementincontemporaryAmericathatIhavediscoveredisthis:moreAmericansarebowlingtodaythaneverbefore,butbowlinginorganizedleagueshasplummetedinthelastdecadeorso.WhetherornotbowlingbeatsballotingintheeyesofmostAmericans,bowlingteamsillustrateyetanotherformofsocialcapital’(ibid, p.70).Putnamstatesthatlifeis‘easierinacommunityblessedwithasubstantialstockofsocialcapital’(ibid,p.67).

Inamorerecentdevelopmentoftheconcept,JackandJordan(1999)understandsocialcapitalasthe‘culturalpractices,norms,networks,links,know-howandtraditionthroughwhichpeopleconductinformalinteractionsofallkinds.Forinstance,socialcapitalisthetrustthatenablespeopletomakecontracts,ratherthanthecontractsthemselves;theteamworkthatmakesgroupsfunctioneffectively,ratherthantherolesandstructuresofthegroups’(p.243).

Harpham, GrantandThomas(2002)complementthisdefinitionbyhighlightingtheimportanceofthequalityofsocialrelationsindefiningsocialcapital,understandingitas‘thedegreeofconnectednessandthequalityandquantityofsocialrelationsinagivenpopulation’(p.106).

Ascanbeseen,thereisnosetdefinitionoftheconcept,andauthorsoftendefineitslightlydifferently.Perhapsthecommonfeatureofallthedefinitionsistheintangiblecharacterofsocialcapital.AsPortes(1998,p.7)states:‘whereaseconomiccapitalisinpeople’sbankaccountsandhumancapitalisinsidetheirheads,socialcapitalinheresinthestructureoftheirrelationships.Topossesssocialcapital,apersonmustberelatedtoothers,anditisthoseothers,nothimself,whoaretheactualsourceofhisorheradvantage’.Itisbecauseofthisfactthatsocialcapitalisconsideredapublicgoodthat‘existsonlywhenitisshared’(Narayan,1999,p.6).

Differentmodelsofsocialcapitalhaveappearedinrecentyears.In1998,BainandHicksproposedamodelofsocialcapitalconsideringtwocomponents:structuralandcognitive(Harpham,GrantandThomas,2002).Thestructuralcomponentreferstotheextentandintensityofassociationallinksoractivities,citizenshipandsocialsupport,whilethecognitivecomponentincludestrust,reciprocityand


4

sharing.Structuralsocialcapitalissometimesreferredtoasthequantityofsocialrelationships,whilecognitivesocialcapitalisreferredtoasthequalityofthese(DeSilvaet al,2004a).

Narayan(1999)describesanimportantadditionalclassification:thedifferencebetween‘bonding’and‘bridging’socialcapital.Bondingsocialcapitalreferstothesocialcohesionwithinthegroupstructure,whilebridgingcapitalisdescribedasthetypeofsocialcapitalthatconnectsdifferentcommunitiesorgroups.AccordingtoHarpham,GrantandThomas(2002)thebondingandbridgingconstructpartiallyoverlapswiththehorizontal/verticalconstructofsocialcapital,whichunderstandssocialcapitalaseitherhorizontallybased,becauseitisbasedontherelationshipsbetweensimilarindividualsorgroupsinthesamesocialcontext(egbetweencommunities),orverticallybased,meaningitisbasedontherelationshipsbetweendifferentlevelsofsociety(egcommunity,localgovernment).

AsHarpham,GrantandThomas(ibid)pointout,thebonding/bridgingdistinctionisanimportantonesinceithighlightstheroleofgovernmentandthestatewithinsocialcapital–andthereforetheimportanceofpoliticalcontext–anddemonstratestheneedtobalancebothcomponents.Withoutlinksbetweencommunitiesandlocalgovernmentorgroupswithresources(verticalsocialcapital),socialnetworks,normsandtrustmaynotactuallybeabletoimproveanyaspectofacommunity’swell-being.Likewise,withouthorizontallinksbetweengroupsorcommunities,importantinformationsources,supportchannelsorotheradvantagesofsolidaritywillbelost.

Finally,itisimportanttodifferentiateindividualandecologicalsocialcapital.Individualsocialcapitalreferstothesocialcapitalavailabletooneindividual,whileecologicalsocialcapitalreferstotherelationshipsatthecommunitylevelwheretheindividualsaregathered(DeSilvaet al,2004b).Ecologicalsocialcapitalisusuallyestimatedbyaggregatingindividualsocialcapitaltothecommunitylevel.InPeru,DeSilvaet al(ibid)identifiedninepapersthatexplicitlyincludedsomeformofindividualsocialcapital,2butnoneanalysedecologicalsocialcapital.Theoretically,therehasbeenarecenttrendinthesocialsciencestoanalysesocialprocessesathierarchicallevels,suchasindividual,community,provinceandcountrylevels.Inthefieldofeducation,severalstudieshavefoundthattheindividual-levelvariablesarecorrelatedwhentheindividualscomefromthesamecommunity(BrykandRaudenbush,1992).Hierarchicalanalysisallowstheimportanceofthevariablesatdifferentlevelstobeestimated,andweplantoadoptsuchaframeworkinthisstudybymeasuringbothindividualandecologicalsocialcapital.

Inthepast15to20years,socialcapitalhasbecomeanimportantconceptininternationalresearch,beinglinkedtochildandyouthwelfare(JackandJordan,1999;Runyanet al,1998;FurstenbergandHughes,1995)andself-reportedhealth(Lindstrom,2004).Butnotallsocialcapitalisgood.Authorsfrequentlycitecriminalnetworks,suchasthemafia,asexamplesofnegativeformsofsocialcapital.Inthenextsectionthereisabriefdescriptionofthestudiesrelatingsocialcapitalandeducationintheinternationalliterature,thoughnonewerefoundforPeru.

�.2SocialcapitalandeducationWeidentifiedsixpapersthathaveexploredtherelationshipbetweensocialcapitalandeducation.Teachman,PaaschandCarver(1996)usedtheNationalEducationLongitudinalSurveyfromtheUSAtoexplorethelinksbetweensocialcapitalanddroppingoutofschool.Theyfoundthathaving

2 Sincesocialcapitalisarelativelynewconstruct,itisnotsurprisingthatthereareonlyafewstudiesaboutit.However,thereisalongtraditionofresearchinPeruandotherLatinAmericancountriesaboutsocialmovementsandnetworks,whichwouldincludewhatisnowcalledsocialcapital.See,forexample,EscobarandAlvarez(1992)forstudiesinLatinAmerica,includingPeru,andBlondet(1986)foracasestudyinLima.


5

adivorcedmother,havingastepfatherorfrequentlychangingschoolincreasedthechancesofachilddroppingoutofschool,whileahigherlevelofchild-parentconnectivitydecreasedtheprobabilityofdroppingout.Thisstudyshowsaninterestingcharacteristiccommoninseveralofthestudiesmentionedbelow:theyrefertosocialcapitalastheirmainconstruct,butonlymeasuresomeaspectsofit.Furthermore,itcouldbearguedthattheyarenotmeasuringsocialcapitalatall,butpotentialdeterminantsofsocialcapital,suchashavingadivorcedmother.

Pong(1997)relatedfamilystructure,schoolcontextandeighth-grademathematicsandreadingachievementinarandomnationalUSsample(usingdatafrom1988).Shefoundanegativeassociationbetweenmathematicsandreadingachievementandhavingastepfatherorasingleparent,andapositiveassociationwiththenumberofacquaintedrelatives.Again,itcouldbearguedthatthevariablesincludedinPong’sanalysisarenotstrictlywhattheliteraturewoulddefineassocialcapitalbutareinsteaddeterminantsofsocialcapital.

Israel,BeaulieuandHartless(2001)usedtheUSNationalEducationLongitudinalSurveytoexploretheinfluenceoffamilyandcommunitysocialcapitaloneducationalachievement.Familysocialcapitalwasestimatedthroughananalysisofthenumberofparentspresentinthehousehold,thenumberofsiblings,andthenumberofsiblingswhohaddroppedoutofsecondaryschool.Theyalsoestimatedfamilyprocessesthroughparentalexpectationsfortheirchild’seducation,discussionofschoolmattersbetweenparentsandchildren(referredtoasnurturingactivities),parentalcheckingofhomework,parentallimitsforwatchingTV,andtheamountoftimethechildspentathomewithnoparentalsupervision(referredtoasmonitoringefforts).Communitysocialcapitalwasestimatedthroughananalysisofthesocio-economiccapacityofthecommunity,measuredthroughsixindicators:communityisolation(location);instabilityofthepopulation(iepercentagewhohadmovedinthepastfiveyears);percentageofthestudentpopulationfromaminorityethniccommunity;voterparticipationrate(tomeasureinequalityanddisaffectioninthecommunity);andextentofsocialintegrationinthecommunity;andthedegreetowhichparentsknowtheparentsoftheirchild’sclosestfriend.Theyusedamultilevelmodeltoanalyseachievementinmathematicsandreadingcomprehension(combiningthemintoasinglescore)anddroppingoutofschool.Theirgeneralconclusionisthatbothfamilyandcommunitysocialcapitalhaveaninfluenceoneducationalachievement,althoughtheimpactoffamilysocialcapitalisgreaterthanthatofcommunitysocialcapital.

Horvat,WeiningerandLareau(2003)analysedtherelationshipsbetweenfamiliesandschoolsindifferentsocio-economicgroupsintheUSA.Thisstudyisdifferentfromthepreviousoneinthattheauthorsusedanethnographicapproach.Theirconclusionspointoutthepositiveimpactofthequantityandqualityoffamilynetworksoneducationalachievement.Parentsfrommiddle-classfamiliesweremorelikelythanworking-classparentstousethesenetworkstotheiradvantagewhenconflictsaroseinschool.Also,theyweremorelikelytousethesenetworksinacollectiveway.Theauthorsfoundnodifferencesrelatedtorace.Goddard(2003)analysedthesocialcapitalinasampleofUSschoolstudents.Socialcapitalwasdefinedattheschoollevelbasedon11itemsadministeredtotheteachers.Theitemswereaimedatmeasuring:relationalnetworksthatconnectparents,communityandstudents;trustingrelationshipsbetweenstudentsandparents;andnormsthatsupportstudentlearning.Itwasfoundinthemultilevelanalysisthatfourth-gradestudentsinschoolswithhigherlevelsofsocialcapitalweremorelikelytopassmathematicsandwritingassessmentsthatweremandatory


�

atthestatelevel,eventhoughthesewereschoolswithahighconcentrationofpoorstudents.Thisanalysisisinterestingbecausetheauthorconceptualisedsocialcapitalattheschoollevel,butalsoincludedindividualvariablesintheanalysis.Theauthorsuggestsfurtherstudiestoanalysewhethersocialcapitalisindependentofsocio-economicstatusandwhetheritcanbedevelopedinschoolsservingpoorpopulations.

Ream(2003)usedamixed-methodsapproach(iequantitative,usinglongitudinaldata,andqualitative,performingsemi-structuredfieldinterviewswithstudents)toanalysetheimpactofthesocialmobilityofMexican-Americansoneducationalachievement.Theauthorusedschool-levelsocialcapitalvariables:academicallyrelevantteacher/studentinteraction(fouritems)andschool-initiatedinteractionwithstudents’parents(threeitems).HefoundthatMexican-Americanstudentstendtomovehousemoreoftenthannon-Latino/whitestudents,andthishasanegativeimpactontwelfth-grademathematicsperformance.Furthermore,theauthorfoundthatwhileMexican-Americanshadaslightadvantageonacademicallyrelevantteacher-studentinteractions,thismeasurehadanegativeimpactonreadingperformance,whilethesamevariablehadapositiveimpactforwhitestudents.HenceReamsuggeststhattheremightbesomeformof‘counterfeitsocialcapital’forMexican-Americans.Theauthorrecognisesthattheremightbeotherexplanationsforthisfinding,suchasdifferencesinperceptionbetweenwhiteandMexican-Americanstudents,orthefactthatthetwogroupsmighthavedifferentteachers.Nevertheless,thestudysuggestsinterestingtheoreticalpossibilitiesforinteractionsbetweencultureandtherelevanceofsocialcapital.

Inconclusion,itcouldbesaidthatmoststudiesthathaveanalysedtherelationshipbetweensocialcapital(atthefamily,communityorschoollevel)andeducationalachievementhavefoundapositiveassociation.However,noneofthestudiesusesmorewidelyacceptedmeasuresofsocialcapitalsuchascivicparticipation,trustandreciprocity,whichareusedintheYoungLivesstudy.Instead,measuresrelatingtofamilystructureandthequalityofparent-childandteacher-childinteractionarecommonlyused–reflectingtheworkofColeman(1988)–whichmanyresearcherswouldarguearenotsocialcapitalmeasuresatall.

ThestudiesreviewedherehaveallbeenperformedintheUSA.Itisnotknown,therefore,whetherthesameconclusionwouldbetrueindevelopingcountriessuchasPeru.Furthermore,noexperimentaldesignswereusedinthesestudies(suchdesignswouldmakeforastrongercause-effectargument).Someofthestudiesalsosuggestthattheeffectofsocialcapitalmightbedifferentaccordingtotheenvironmentinwhichthestudentsliveandstudy(seeforexampleBlackwellandMcLaughlin,1999,forurban/ruraldifferences).Beforeturningtomethodsandresults,thefollowingsectionpresentssomebackgroundoncommunityorganisationsinPeru,whichwillprovideageneralframeworkforunderstandingthesocialcapitaldatapresentedlater.

�.3CommunityorganisationsinPeruInrecentdecades,severalcommunityorganisationshaveappearedandhavesolidifiedtheirworkinPeru,mostlyinimpoverishedcommunitiesandoftenrunbylocalmothers.Yetthereisverylittleresearchontheimpactthesemayhaveoutsidetheirspecificrangeofobjectives(whichareusuallyrelatedtolocalfeedingorgovernanceprogrammes).Thissectionprovidesabriefdescriptionofthesituationoftheseorganisationsaroundthecountry.


�

Atthebeginningofthe1980s,duetotheseriouseconomicandsocialcrisis,diversesocialgrass-rootsorganisationsbegantoemerge.Theseorganisationswereformedbyindividualsorgroupswiththegeneralgoaloffightingpoverty,buteachhadspecificaims.Mostoftheseorganisationsarecomposedofandrunbylocalwomen,withthepurposeofmeetingtheirfamilies’feedingneeds.Therearethreetypesoforganisationsrelatedtofeeding–communalkitchens,glass-of-milkcommitteesandmothers’clubs–andtheyemergedinbothruralandurbanareasofPeru.Inaddition,neighbourhoodorganisationsappeared,aimedatimprovinglocallivingconditions,suchaselectricityandwatersupplies,andschools.Finally,self-defencecommitteeswerecreated,inbothruralandurbanareas,devotedtodefendingthecommunityfromthethreatofterrorismthathasaffectedthecountrysince1980.3

Thecommunityorganisationsaimedatfeeding–mothers’clubs,glass-of-milkcommitteesandcommunalkitchens–arethemostnumerous.BlondetandMontero(1995)definedsuchcommunityorganisationsasorganisationsthathaveincommonthecollectiveactionofthedailypurchase,preparationanddistributionoffood,withthepurposeofreducingthecostoffamilyfeeding.Theauthorsalsopointoutthattheseorganisationsreducethetimewomeninvestindomesticactivitiesandconstituteasourceofsocialisation,trainingand,eventually,incomegeneration.

Communalkitchensemergedinthe1980sasaninitiativeofgroupsofmothers(onaverage20to30womenpergroup)whoorganisedthemselvesjointlytopreparethemealsfortheircommunity.Themealswerethensoldatlowprices,makingthemavailabletoeverybody.Lowpriceswerepossiblebecauseofthedonationofsomeofthesupplies,thewholesalepurchaseoftherest,andtheunpaidworkofthemothers(CuevaandMillán,2000).Communalkitchensemergedasawomen’sinitiativetodealwiththesevereeconomiccrisissufferedbylow-incomefamilies.Inaddition,thistypeoforganisationallowedthememberstosavetimeoncookingwhichtheycouldtheninvestinotheractivities(studying,working,etc).

Glass-of-milkcommitteesarealsoanimportantsourceofsocialcapitalamongcommunities.TheGlass-of-MilkProgrammeisawelfareprogrammewiththeobjectiveofdeliveringmilkandoatstochildren,seniorcitizens,andpregnantandnursingmothers.Thecommittees,encouragedbythelocalcouncilsandcomposedofmothersfromthecommunity,areinchargeofthedailypreparationanddeliveryofthemilktothebeneficiaries.Itisimportanttopointoutthatthecommitteesreceivepowderedmilkandhavetoprepareitbeforedelivery,aconditionthatguaranteesthecontinuityofthecollectivework.

Mothers’clubsarethethirdtypeofwomen’s organisationdevotedtofeeding.Theyaredefinedasgroupsofmotherswhoorganisethemselvestochannelthebenefitsfromcharitiesandfromwelfareactivitiesorganisedbyreligiousinstitutionsandwealthypersons,andbypoliticalpartiesinterestedinexpandingtheirgrass-rootsorganisations.

AnotherimportantsocialorganisationworthcommentistheWawa Wasidaycareprogramme.Itbeganasalocalmothers’initiative,andbecauseofitssuccessthePeruvianGovernmentdecidedtoturnitintoasocialprogramme.Forasmallfee,workingparentsleavetheirunder-threesinadaycarecentre,wherethereisa‘mother-in-charge’whoistrainedinhealthcare,earlychildhoodstimulationand

3 Inthe1980sandthefirsthalfofthe1990sterrorismwasaseriousprobleminPeru.Therewereatleasttwomajorarmedgroups: Shining Path (Sendero Luminoso) and MRTA (Movimiento Revolucionario Tupac Amaru).The number of deathsaftertwodecadesofterrorismwasrecentlyestimatedtobe69,000(CVR,2003).Nowadays,terroristactivityhasdeclinedconsiderably.


8

basicnutrition.MealsintheWawa Wasiareorganisedthroughcommunalkitchensorglass-of-milkcommittees,whichtakestheburdenofcookingawayfromthemaincaregivers.

Thereareotherimportantsocialorganisationsatthecommunitylevel,forinstancepeasantcommunities,especiallyinruralareassuchasCuscoandPuno,inthePeruvianhighlands.Peasantcommunitiesare‘territorialunits’ofclansorgroupsrelatedbykinshiporhistoricalneighbourhood,whoseekthewell-beingofalltheirmembers.Thecommunityistheownerofthelandonwhichitislocatedandmustberegisteredinordertohavelegalrecognition.Peasantcommunitiesareimportantbecausetheyenablethesocial,culturalandeconomicdevelopmentofalargesectorofthepopulation,usuallymarginalisedbystatepolicies.Forinstance,insomeruralareascommunitieshavebuiltmoreschoolsandruralroadswiththeirownresources,thanthestate.

Thetwoothersocialorganisationsdevotedtoimprovingcommunitywell-beingareneighbourhoodorganisationsandself-defencecommittees.Theformerisdevotedtoimprovingthequalityoftheservicesofferedinthecommunity,suchaselectricityandwater,andalsothequalityoftheschoolsandotherimportantinstitutionswithinthecommunity.Self-defencecommitteesemergedwithincommunitieswiththepurposeofprotectingtheirlandandfightingagainstviolence,especiallyterrorism.Thesesocialorganisationscontributetothedevelopmentoftheircommunities,andprovidesecurityfortheirmembers.

Table1showsthepercentageofhouseholdsthathaveatleastonememberwhobelongsto,participatesin,orisregisteredinagroup,organisation,associationand/orsocialprogramme.Twenty-threepercentofhouseholdshaveamemberwhoparticipatesintheGlass-of-MilkProgramme.Itisimportanttonotethatparticipationinthisprogrammeismorefrequentinruralareas(34.8percent)thaninurbanareas(17.4percent),andthatmostofthehouseholdswithmemberswhoparticipateare‘extremelypoor’or‘non-extremelypoor’.4

Table 1: Percentage of households that have at least one member who belongs to, participates in, or is registered in a group, organisation, association and/or social programme

Group, organisation, association and/or social programme

Rural Urban National

EP NEP NP Total EP NEP NP Total

Sports club and associations 5.4 6.3 8.6 6.6 5.3 3.8 7.9 6.6 6.6

Cultural associations (dance, music etc) 0.6 0.6 0.4 0.6 0.6 0.7 2.2 1.7 1.3

Neighbourhood associations 3.9 4.0 4.3 4.0 2.4 3.8 3.5 3.5 3.7

Peasant groups (peasant self-defence organisations) 12.2 10.3 7.0 10.1 1.1 0.6 0.5 0.6 3.9

Irrigation associations 5.7 8.3 13.6 8.9 0.8 1.8 1.0 1.2 3.9

Professional associations 0.1 0.1 0.7 0.3 0.1 0.2 4.4 2.9 2.0

Labour unions 0.3 1.1 2.8 1.3 1.0 2.1 4.5 1.2 2.8

Mothers' clubs 7.9 7.2 4.7 6.7 4.8 3.2 1.1 1.9 3.6

Parents association 4.8 3.3 2.1 3.6 1.9 1.1 1.2 1.2 2.0

Glass of milk 44.2 35.9 21.0 34.8 35.3 28.9 10.6 17.4 23.5

4 The poverty variable was constructed according to households’ per capita expenditure. Extremely poor households arethosethatcannotaffordabasicfoodbundle(orrequirements),andnon-extremelypoorarethosethatcannotaffordabasicconsumptionbundle.


�

Group, organisation, association and/or social programme

Rural Urban National

EP NEP NP Total EP NEP NP Total

Communal kitchen 5.9 7.0 5.3 6.0 5.8 6.2 2.1 3.5 4.4

Participative roundtable (mesa de concertación) 0.1 0.4 0.4 0.3 0.0 0.0 0.2 0.1 0.2

Local Administrative Health Committee 0.1 0.2 0.2 0.2 0.0 0.0 0.1 0.1 0.1

Temporary (urban) employment A Trabajar Urbano 0.2 0.2 0.1 0.1 1.7 0.8 0.3 0.5 0.4

Temporary (rural) employment A Trabajar Rural 4.4 3.5 1.8 3.4 0.2 0.1 0.1 0.1 1.2

Other 28.0 21.0 14.9 22.0 13.5 7.0 6.6 7.2 12.3

Source: EnaHo, 2002 (national survey of homes, database) – iV trimester note: Ep = extremely poor; nEp = non-extremely poor; np = non-poor

Therearesomeformsofsocialorganisation,suchaspeasantself-defenceorganisationsandirrigationassociations,thatareconsiderablymorecommoninruralareas.However,participationinpeasantself-defencecommitteesisgreateramongtheextremelypoor,whileparticipationinirrigationassociationsisgreateramongthenon-poor(inruralareas).Ingeneral,participationinmostofthesocialprogrammesororganisationslistedinTable1ismorecommoninruralareas.ThismightberelatedtothefactthatthereisalonghistoryofcollectiveworkamongthePeruvianruralpopulation,andpovertyismoreprevalentinruralareas.Participationincommunityorganisationsaimedatfeedingisgreateramongtheextremelypoorandthenon-extremelypoor,especiallyinurbanareas.

Finally,itisimportanttonotethatmosthouseholdsdonothaveamemberwhoparticipatesinprofessionalassociationsorlabourunions,especiallyamongtheextremelypoorandthenon-extremelypoor,bothinruralandurbanareas.Onlythenon-poorfromurbanareashavearelativelyhighlevelofparticipationinthiskindofsocialorganisation,possiblybecausetheyhaveformaljobswhereastheextremelypoorandnon-extremelypoorareinformalworkers.

SofarwehavereviewedthecurrentsituationofsomeofthemostimportantsocialorganisationsinPeruandprovidedabriefdescriptionofthem.Aswehaveseen,communityorganisationsaimedatfeedingareimportantsourcesofnetworking,especiallyinruralareasandamongtheextremelypoor.Sincethesetypesoforganisationscanenablemembersofthecommunitytoimprovetheirqualityoflife,forinstance,byimprovingtheirnutrition,itwouldbeinterestingtofindoutiftheseorganisationsalsohaveanimpactonspecificeducationaloutcomes,suchasachievementandtheproblemofoveragestudents.Noneoftheabove-mentionedorganisations,however,hasanobjectivetoimprovetheeducationaloutcomesofthecommunity’schildrenwhoareenrolledinprimaryorsecondaryeducation.TheYoungLivesquestionnaire,forthefirstround,includedsomequestionsonsocialcapital.Inthecurrentstudywewilltakeintoaccountwhethertheindividuallivesinanurbanoraruralarea,giventhattherearelargesocio-economic,languageandculturaldifferencesbetweenurbanandruralareasinPeru.

HowcouldsocialcapitalaffecteducationaloutcomesforPeruvianstudents?EventhoughthisisthefirststudyweareawareofthatusesthesocialcapitalconstructinPeruinrelationtoeducation,internationalliteraturesuggestssomepossiblemechanisms.Harpham(2002)proposesatheoreticalmodel,wheresocialcapitalinsideandoutsidethefamily(suchastheextentofnetworks,support


�0

receivedfromnetworks,perceivedtrust,reciprocityandsharednorms)wouldhaveanimpactonintermediatevariables.Thesewouldincludetheparents,suchasanincreaseinresourcestoinvestinthechild,parentaldecisionstoinvestinthechild,andthevalueparentsgivetoeducation.Inturn,intermediatevariableswouldhaveanimpactonchildwelfareoutcomes,suchasthosestudiedhere.Thecurrentstudy,however,willnottesttheseprocesses,onlytheassociationsbetweensocialcapitalandeducationaloutcomes.ThereisstrongevidenceinPerutosupporttheideathateducationishighlyvaluedbythepopulation,eveninimpoverishedareas(seeforexampleAnsiónet al,1998).Inthiscontextitisreasonabletoassumethatmoreclose-knitcommunities(iethosewithhigherlevelsofsocialcapital)wouldcaremorefortheeducationofalllocalchildrenthancommunitieswheresocialtiesaremoreloose.Again,however,thespecificprocessesbywhichthiswouldoccurwerenotlookedatinthecurrentstudy.


��

2. methods

2.�DesignThedatapresentedherewasgatheredinthefirstroundoftheYoungLivesprojectinPeru,betweenAugustandDecember2002.Thereweretwocohortsintheproject;thecurrentreportreferstothechildrenwho,atthetimeofdatacollection,wereaged7.5–8.5years.Asmentionedbefore,theYoungLivesprojectisplannedasalongitudinalstudythatshouldlastuntil2015;thesecondroundofdatacollectionshouldoccurin2006.Thus,thedatapresentedhereiscross-sectionalandtheanalysisshouldbeinterpretedintermsofassociationsbetweenvariablesandnotcause-effectrelationships.Mostoftheanalysisseparatesoutthedataforurbanandruralchildren,and,withinthesegroups,separatesboysfromgirls.

2.2SamplecharacteristicsThePeruYoungLivesdatasetusedhereconsistsofanationwidesampleof704childrenagedbetween7.5and8.5yearsatenrolment.ChildrenwereenrolledforthefirstphaseoftheprojectbetweenAugustandDecember2002.Thesamplewasdistributedbetween20sentinelsitesinurbanandruraldistricts.All1,818districtsofPeruwererankedaccordingtotheirpovertylevel.Eliminatingthewealthiestfivepercent,asampleofdistrictswastakenacrossthepovertyspectrum.Withineachdistrictsampled,arandomdwellingwasselectedandthenalldwellingsnearbyweresystematicallycheckedtoidentifychildrenoftherelevantage.Inafewcases,aftercoveringthewholecommunityoreventhewholedistrict,therewerefewerchildrenthanrequired,sochildrenlivingincontiguouscommunitiesanddistrictsweresurveyed.AdetaileddescriptionofthedatasetandsamplingprocedurecanbefoundinEscobalet al(2003).Table2presentssomecharacteristicsofthesample.

Table 2: Number and percentage of children in the sample, by school grade and area

Area

TotalRural Urban

First grade 26 (14%) 27 (5%) 53 (8%)

Second grade 119 (66%) 271 (52%) 390 (55%)

Third grade 34 (19%) 215 (41%) 249 (35%)

Fourth grade 2 (1%) 10 (2%) 12 (2%)

Total 181 (100%) 523 (100%) 704 (100%)

Aswouldbeexpectedfromnationalstatistics,itismorelikelythatruralchildrenwillbeinalowergradefortheirage(ieoverage).InPeruitiscompulsoryforsix-year-oldstoattendfirstgrade,buttherearenolegalconsequencesforanybodyifachildfailstodoso.TheschoolyearinPerurunsfrommid-MarchorearlyApriltoearlyormid-December;childrenshouldbeatleastsixyearsofageby30Junetoenrolinfirstgradeinagivenyear.Usingtheaboveinformation,weestimatedthegradeinwhichthechildshouldhavebeenstudying.ThisinformationispresentedinTable3.


�2

Table 3: Students’ status by sex and area (number and percentage)

Students’ status

TotalOverage On-age Below the age for

their grade

Rural Female 23 (28%) 51 (61%) 9 (11%) 83 (100%)

Male 30 (31%) 60 (61%) 8 (8%) 98 (100%)

Total 53 (29%) 111 (61%) 17 (9%) 181 (100%)

Urban Female 35 (15%) 174 (73%) 30 (13%) 239 (100%)

Male 54 (19%) 181 (64%) 49 (17%) 284 (100%)

Total 89 (17%) 355 (68%) 79 (15%) 523 (100%)

All children

Female 58 (18%) 225 (70%) 39 (12%) 322 (100%)

Male 84 (22%) 241 (63%) 57 (15%) 382 (100%)

Total 142 (20%) 466 (66%) 96 (14%) 704 (100%)

Arelativelysmallgroupofchildrenarebelowtheagefortheirgrade.Thisisbecause,insomecases,schoolswillallowyoungerchildrentoenrolinfirstgrade.Table3showsaslightlyhigherpercentageof‘on-age’girlsthanboys.Thisisaninterestingtrendalsofoundinotherstudies.WhileamongtheolderpopulationinPerumostnon-educatedpeoplearewomen,theyoungergenerationsseemtobemovingtowardseducationalequalityforboysandgirls(atleastintermsofenrolmentandschoolgrade).Infurtheranalysis,theon-ageandbelow-agegroupswillbecombinedandcalled‘on-age’.

Table4presentssomeofthegeneralcharacteristicsofthesample(itwastestedwhethertherewerestatisticallysignificantdifferencesbetweenon-ageandoveragechildrenineachgroup).

Table 4: On-age and overage children’s characteristics by area

Rural(N = 181)

Urban(N = 523)

All children(N = 704)

Overage On-age Overage On-age Overage On-age

Height-for-age (z-score) -2.3** -2.0** -1.7** -1.3** -1.9** -1.4**

Weight-for-age (z-score) -1.4** -1.0** -0.8** -0.4** -1.0** -0.5**

Body mass index 16.2 16.3 16.8 17.1 16.6 16.9

The child attended school in 2001 (%) 88** 98** 86** 99** 87** 99**

Family size 5.5* 4.9* 4.9 4.5 5.1** 4.6**

Student’s age when he/she first went to school 6.2** 6.0** 6.2** 5.7** 6.2** 5.8**

He/she goes to public school (%) 98 99 97** 87** 97** 90**

The child attended preschool (%) 81 88 79** 91** 80** 90**

Mother speaks indigenous language (%) 42* 28* 9** 3** 21** 9**

Mothers with at least secondary education (%) 16 22 44** 69** 33** 58**

* significant at 10%; ** significant at 5%


�3

Theinformationaboveshowsthatmostchildrenoldenoughtoattendprimaryschooldoso.Theproblemisthatabouthalfofthemwillnotfinishsecondaryeducation(Guadalupe,2002).ThisissomethingthatcouldbestudiedinfutureroundsoftheYoungLivesproject.

Table4shows,ingeneralterms,thatchildrenfromruralenvironmentsarepoorer(asshownbytheanthropometricandmothers’educationvariables)thanchildreninurbanenvironments.Withineachgroup,overagechildreningeneralcomefrompoorerenvironmentsthanon-agechildren.On-agechildrenwereyoungerwhentheystartedschool.Overagechildrenaremostlikelytohavemotherswhospeakanindigenouslanguage.On-agechildreninurbanenvironmentsaremorelikelytoattendprivateschoolsthananyothergroup(around87percentofthenationalpopulationattendspublicschool:Guadalupe,2002).

Theanthropometricindicatorsshow,asexpected,thatruralchildrentendtohavepoorernutritionalstatusthanurbanchildren.Inbothruralandurbanareas,overagechildrenshowpoorermeanheight-for-ageandweight-for-agez-scoresthanon-agechildren.However,nodifferencesbetweengroupswerefoundforbodymassindex,anindicatorofacutemalnutrition.

Table5presentsinformationonthecharacteristicsofthechild’shomeandhouseholdassets,producingasimilarpatterntoTable4.Urbanchildrenshowgenerallyhigherindicators,and,withineachgroup,on-agechildrenshowhigherindicators(althoughthepatternisclearerinurbanchildren).

Table 5: Fixed household assets and house infrastructure (percentage)

Rural(N = 181)

Urban(N = 523)



Has a television 40 36 69** 83** 58** 72**

Has an iron 9 16 28** 61** 21** 51**

Has a gas stove 9 15 38** 66** 27** 55**

Has a plough 47 52 17** 8** 29** 18**

Has a machete 94 91 36** 24** 58** 39**

Has an axe 92 85 39** 25** 59** 39**

Wall material: brick/concrete 2 5 28** 53** 18** 42**

Roof material: concrete/cement/tiles/slates 25 20 21** 41** 23** 36**

Floor material: cement/laminated material 4 9 31** 58** 21** 47**

Has piped water into dwelling/yard/plot 64 67 69** 86** 67** 82**

Uses gas/electricity for cooking 0* 6* 31** 54** 20** 43**



�4

2.3VariablesandproceduresThevariableswereobtainedfromthequestionnairesadministeredtothechildren,theirparentsandcommunityleaders.Achievementwasmeasuredthroughthereading,writingandnumeracyitemsdescribedbelow.Socialcapitalvariableswereanalysedaccordingtotheframeworkdevelopedforthisstudy.Thisframeworkusedthedistinctionbetweenstructuralandcognitivesocialcapitaldescribedabove(Harpham,GrantandThomas,2002).Thedistinctionbetweenbondingandbridgingsocialcapital(Narayan,1999)wasnotintroduced,giventhatallquestionswererelatedtointra-communityrelationships(bondingsocialcapital),andnotinter-communityrelationships(bridgingsocialcapital).Thespecificquestions(tobefoundinthe7.5-8.5year-oldquestionnaire,availableontheYoungLiveswebsitehttp://www.younglives.org.uk/)usedtomeasurethestructuralandcognitivesocialcapitalaredescribedbelow.Fieldworkincludedseveralverificationprocedurestoassurethevalidityoftheinformation.

TheinstrumentusedisashortenedversionoftheAdaptedSocialCapitalAssessmentTool(A-SCAT).DeSilvaet al(2004b)conductedastudyonthevalidityofthisinstrumentinthePeruviancontext.Forthis,theyanalysedthedatafromtheYoungLivesproject(specificallycheckingforcriterion-relatedvalidity)andalsoconductedin-depthqualitativeinterviewswithrespondentswhowererepresentativeofthoseenrolledintheYoungLivesproject.Ingeneral,theresultssupportthevalidityofthequestionnaire,althoughproblemswiththeinterpretationofspecificquestionswerefound(seeSection4formoredetails).Thestudyalsoshowed,throughafactoranalysis,thattheindividualitemswerecapturedbysummaryitemsrepresentingstructuralsocialcapital(includingmeasuresforgroupmembership,socialsupportandcitizenship)andcognitivesocialcapital(iemeasuresoftrust).Thesefactorswillbedescribedlaterinthisanalysis.


�5

3. results

3.�SocialcapitalinPeruThesocialcapitalquestionsweredividedintoquestionsoncognitivesocialcapitalandquestionsonstructuralsocialcapital.Thefirstresultspresentbivariateanalysis,toshowafirstlevelofassociations.Animportantissuetoconsiderinallanalysisiswhorespondedtothesocialcapitalquestions.Theproceduresspecifiedthatthemothershouldrespond,butifshewasnotavailablethequestionsweredirectedtoanothermemberofthefamily.Thisdistinctionisimportantsincetheresultsforaparticularfamilyregardingsocialcapitalmightvarydependingontheinformant.Table6givesdetailsofwhotherespondentswere.

Table 6: Person providing information in response to social capital questions

Frequency Percentage

Father 53 8

Mother 602 86

Stepfather 1 0

Stepmother 7 1

Grandfather 6 1

Grandmother 19 3

Other 15 2

Total 703 100

Whilethemajorityofrespondentswerethemothersofthechildren(asexpected),justover14percentoftherespondentswereotherfamilymembers(mostlyfathers).Inthenextsectionswepresentresultsforallrespondentstogether,butinthemultivariateanalysiswepresentresultsbothforsituationswhereonlythemothersareconsideredasrespondents,andwhereweincludedallrespondents.Table7presentstheresultsforcognitivesocialcapitalquestions.

Table 7: Cognitive social capital in Peru by area and child’s status (percentage)

Rural(N = 181)

Urban(N = 523)



Can the majority of people in this community be trusted? 51 47 44** 26** 47** 31**

Do you consider yourself similar to most other households in this community? 29** 63** 56** 69** 46** 67**

Do the majority of people in this community generally get along with each other? 88** 73** 72** 60** 78** 63**

Do you feel as though you are part of this community? 91 90 88* 79* 89** 81**

Would the majority of people in this community try to take advantage of you if they got the chance? 38 42 47 50 44 48



��

Thereisaninterestingtrend:caregiversofon-agechildren,inbothcontexts,considerthemselvesmoresimilartoothermembersofthecommunitythancaregiversofoveragechildrendo.Ontheotherhand,thecaregiversofoveragechildrenaremorelikelytobelievethatthemajorityofpeopleinthecommunitygetalongwitheachother.Table8presentstheresultsforstructuralsocialcapitalquestions.

Table 8: Structural social capital (participation in community-level organisations) in Peru, by area and child’s status (percentage)

Rural(N = 179)

Urban(N = 512)


Respondent belongs to: Overage On-age Overage On-age Overage On-age

Labour union/trade union 2 4 1 2 1 3

Community associationa/co-op 11 5 6** 1** 8** 2**

Women’s group 11 13 16 10 14 11

Political group 4 1 1 2 2 1

Religious group 17 13 9 8 12 9

Credit/funeral group 0 0 0 0 0 0

Clubs/sport association 0 2 2 4 1 3

Health, water, school association or committee 2** 11** 8 8 6 9

Other 11 5 4 4 7 4

** significant at 5% a community group or association

Overall,participationincommunityassociationsisquitelow.Table9presentsthedataforthestructuralsocialcapitalquestionsrelatedtosupport.


��

Table 9: Structural social capital (support), by area and child’s status (percentage)

Rural(N = 179)

Urban(N = 513)



Received support from family and relatives 30 29 52 46 44 42

Received support from household members 42 36 53 47 49 45

Received support from neighbours 19 17 22 18 21 18

Received support from friends who are not neighbours 11 17 20 22 16 21

Received support from community leaders 4 6 8** 2** 6** 3**

Received support from religious leader 15 12 16* 10* 16* 10*

Received support from politicians 0 3 5* 2* 3 2

Received support from government officials/civil servants 2 6 5 2 4 3

Received support from charitable organisation/NGO 2 2 1 2 1 2

Received support from other groups 0 2 1 1 1 2


Therearethreesignificantdifferencesinsupport,allinurbanareas.Overagechildrenaremorelikelythanon-agechildrentoreceivesupportfromcommunitymembers,religiousleadersorpoliticians.Overall,supportseemstocomemostlyfromrelatives(includingmembersofthehousehold),followedbyneighbours,friendsandreligiousgroups.

Table10presentstheresultsforadifferentsetofquestionsinthestructuralsupportcategory:citizenship.

Table 10: Structural social capital (citizenship), by area and child’s status (percentage)

Rural(N = 181)

Urban(N = 523)



You have joined with other community members to address a problem of common concern

34 34 19 19 25 22

You have talked with a local authority or governmental organisation about problems in this community

30 30 13 14 20 18

Therearenostatisticallysignificantdifferencesbetweenthecitizenshiplevelsofcaregiversofon-ageandoveragechildren.


�8

3.2AchievementAsalreadymentioned,reading,writingandnumeracyassessmentitemswereadministeredtothechildren.Thereadingitemsrequiredthechildrentoreadthreeletters(N,A,P);oneword(‘pan’,whichisSpanishfor‘bread’);andonesentence(‘El pan es rico’,whichisSpanishfor‘thebreadisdelicious’).TheresultsarepresentedinTable11.

Table 11: Reading achievement by child’s status, area and sex Overage (%) On-age (%) Total

Rural

Female

Cannot read or reads only letters 65 28 36

Reads words 0 0 0

Reads sentences 35 72 64

Total 100 100 100

Male


Reads words 0 5 4


Total 100 100 100

Urban

Female


Reads words 0 1 0


Total 100 100 100

Male


Reads words 2 3 2


Total 100 100 100

Thepercentageofstudentswhocouldreadwordsandsentencesismuchlowerforruralstudents.ThisisacommonpatterninallPeruvianevaluations.Theresultsarealsopoorerforoveragechildrenthanforon-agechildren.Resultsforboysandgirlsareverysimilar.


��

Table12presentstheresultsofthewritingassessment.Thisitemrequiredchildrentowritethesentence‘me gustan los perros’ (whichisSpanishfor‘Ilikedogs’).

Table 12: Writing achievement by child’s status, area and sex

Overage (%) On-age (%) Total

Rural

Female

Cannot write or writes with difficulty 81 55 62

Writes without difficulty 19 45 38

Total 100 100 100

Male



Total 100 100 100

Urban

Female



Total 100 100 100

Male



Total 100 100 100

Theaboveshowsbetterresultsforurbanthanforruralchildren,foron-agethanforoveragechildren,andforfemalethanformalechildren.

Table13presentstheresultsforthenumeracyassessmentitem.Thisitemrequiredchildrentosolveabasicmultiplication(2x4).


20

Table 13: Numeracy achievement by child’s status, area and sex

Overage (%) On-age (%) Total

Rural

Female

Cannot solve or incorrect 95 57 67

Correct 5 43 33

Total 100 100 100

Male


Correct 21 53 44

Total 100 100 100

Urban

Female


Correct 17 69 62

Total 100 100 100

Male


Correct 27 77 68

Total 100 100 100

Theresultsshowbetterperformanceforurbanthanforruralchildren,foron-agethanforoveragechildren,andformalethanforfemalechildren.

Table14presentsthecorrelationsamongthedependentvariables.

Table 14: Spearman correlations between dependent variables

Reading Numeracy Writing

Numeracy 0.366

N 599

Writing 0.405 0.356

N 612 655

Overage -0.284 -0.521 -0.254

N 611 653 666

All correlations significant at 1% (two-tailed)


2�

Thehighestcorrelationisbetweennumeracyandoverage,followedbyreadingandwriting(forthestrengthofthecorrelationthesigndoesnotmatter,thesignonlyindicatesdirection).Overall,giventhateachachievementareahadonlyoneitem,andthecorrelationsbetweenthemwerepositiveasexpected,afactoranalysiswasperformedtogenerateasingleachievementscore,composedofthethreeareastested.Moreinformationonthisispresentedbelow.

3.3 MultivariateanalysisTheresultspresentedsofarhavecomefrombivariateanalysisonly.Themultivariateanalysispresentedbelowintroducesseveralpredictors,includingcognitiveandstructuralsocialcapitalvariables,toestablishtheuniqueassociationofeachwiththedependentvariables(achievement,andon-ageversusoverage).Fortheanalysisofsocialcapitalwecombinedvariablesbyusingafactoranalysis(seeAppendixfordetails).Theliteratureonthistopicsuggeststhattheeffectofsocialcapitalmightnotcomefromseveralindependentvariables,butfromacombinedeffect.

Threefactorswereidentified,broadlyconcurringwithwhattheoreticalconsiderationswouldpredict.Factor1groupsthequestionsrelatedtocognitivesocialcapital;factor2referstostructuralsocialcapital,specificallycitizenship;andfactor3alsoreferstostructuralsocialcapital,specificallyparticipation.

Themultivariateanalysisincludedinformationattheindividuallevel(childorchild’sfamily)andatthecommunitylevel.5Themodelsusedtoexploretherelationshipbetweeneducationalsuccessandsocialcapitalvariableswereordinaryleastsquares(forchildachievementresults)andlogisticregression(foron-ageresults).Thisoptionallowsustoobtainrobuststandarderrorsandtoindicatethatobservationsareclusteredintocommunities.Inthiscase,weassumethattheobservationscanbecorrelatedwithincommunitiesandareindependentbetweencommunities.Thenumberofchildrenperlocation(community)wasnotlargeenoughforahierarchicallinearmodel.Itisimportanttonotethenumberofchildrenfromthesamelocation(community6)inthesample.Overall,thesampleincluded75communities.Theanalysispresentedintheprevioussectionincludedallchildren,butthemultivariateanalysisincludesonlychildrenwholivedinacommunitywhereatleasttwochildrenlived,sothattheaverageforthecommunityhadsomesense.Thisprocedureresultedintheexclusionfromtheanalysisof20children,fromasmanycommunities(seeTable15forthedistributionofnumbersofchildrenpercommunity).

Table 15: Number of children per community

Number of children Communities Percentage of communitiesOne 20 27Two 10 13Three 9 12Four 7 9Five 3 4Six or more 26 35Total 75 100

5 The community level refers to the average of responses of children within the same locality. Entered as a cluster intothe regressions, the Stata software allows the cluster option to be introduced into the estimation.This option allows theidentificationofgroups–communities–ofcorrelatedobservationsandgivestheadjustmentforthislevel.

6 Communitiescorrespondtoadministrativeareas.FordetailsseetheYoungLivescommunityquestionnaire,whichcontainsthedefinitionof‘community’(availablefromwww.younglives.org.uk).


22

Thecovariatesinthemodelwereselectedbasedontheliteraturereviewpresentedinthispaper.

Ordinary least squaresChildachievementcouldbeassociatedwithchildcharacteristics(egsex,height-for-age,schoolgrade);familycharacteristics(egparents’education,numberofpersonsperroomathome);socio-economiccharacteristics(egnumberofpossessionsathome,agriculturaltools,housematerial);andsocialcapital(egnumberofgroupsthatthehouseholdparticipatesin).

Yj = Bo + B1Xj + B2Cj + B3Ĉ + Ej Ej~N(0, σ2)

Yj = child achievement

Xj = child and family background

Cj = household social capital

Ĉ = community social capital (aggregated from individual level)

Ej = random component

Logistic regressionSincetheoutcomevariableon-age/overageisbinary,logisticregressionwasusedforthismultivariateanalysis.

ln [p/(1-p)] = Bo + B1 Xj + B2Cj + B3Ĉ

p = probability that the event Y occurs, p(Y=1)

p/(1-p) = odds ratio

ln [p/(1-p)] = log-odds ratio (logit)

Xj = child and family background

Cj = household social capital

Ĉ = community social capital (aggregated from individual level)

Variables included in the analysis

Dependent variables

Achievement:thisvariablewascreatedusingthreescores.Thefirstonereferstothechild’sreadinglevel.Thescoresassignedwere1ifthechildcouldreadsentencesand0iftheycouldnot.Thesecondreferstothechild’swritinglevel.Thescoreswere1ifthechildcouldwritecorrectlyand0ifthechildcouldnotwrite,orwrotethespecifiedtextincorrectly.Thelastonereferstonumeracy.Thescorewas1iftheexercisewascompletedcorrectlyand0ifincorrectorblank.Theachievementvariablewasbuiltthroughafactoranalysisthatcombinedthesescoresandprovidedastandardisedscorethatcombinedachievementinreading,writingandnumeracy.Thesinglefactoraccountedfor58percent

a)


23

ofthetotalvariance.Theweightsofthevariableswere0.45forreading,0.44forwritingand0.43fornumeracy.

On-age:thisvariablewascoded1ifthechildwasthecorrectageoryoungerfortheirgrade,and0iftheywereolder.

Predictor variables

Sex:thiswascoded1formalesand0forfemales.

Height-for-age: thisvariableisthestandardisedz-score.Itisanindicatorofchronicmalnutrition.

Child works: thisvariablewascoded1ifthechildwasinvolvedinanyactivityaimedatobtainingmoneyorothercommoditiesand0ifnot.Thisdatawasgatheredthroughthechildquestionnaire.

Crowding index: thisvariablereferstotheratioofthenumberofpeopleperroomathome.

The mother has at least secondary education: thisdatawasgatheredinthehouseholdsurveyandwascoded1ifthemotherhadatleastsecondaryeducationand0ifshehadalowerlevelofeducation.

Mother speaks an indigenous language: thisvariablewascoded1ifthemotherspokeanindigenouslanguageashermothertongue,7and0ifsheonlyspokeSpanish.

Child’s age when she/he first went to school: thisreferstothecaregiver’sanswerwhenaskedtheageatwhichthechildbeganher/hisprimaryeducation.

Type of school:thiswascoded1forpublicschoolsand0forprivateschools.InPeru,moststudieshavefoundhigherachievementforstudentsinprivateschools(egUMCandGRADE,2001).

Both biological parents present at home: thisvariablewascoded1ifthebiologicalparentslivedwiththechildand0ifonlyoneornoparentslivedathome.

Parents help with the child’s homework: thisvariablewascoded1ifatleastoneoftheparentshelpedwiththehomeworkand0ifnot.

The child attended preschool: thiswascoded 1forchildrenwhohadattendedpreschooland0forchildrenwhohadnot.

Socio-economic indices: thesearevariablesconstructedfromseveralindicators,includinghouseholdpossessions,agriculturaltoolsandhousematerials.ThisinformationwasprovidedbytheindividualslistedinTable6.Thevariableswereincludedinafactoranalysiswithvarimaxrotation.Fourfactorsresultedfromthisanalysis,whichaccountedfor62percentofthevariance.Theresultingfactorswere:

7 InPeru,QuechuaandAymaraarethemostwidelyspokenindigenouslanguages,butthereareover40acrossthecountry(Pozzi-Escot,1998).

b)

a)

b)

c)

d)

e)

f)

g)

h)

i)

j)

k)

l)


24

Household assets index:thisfactorgroupedseveralindicatorsrelatedtohouseholdpossessions–radio,bicycle,television,iron,blenderandgasstove.Itaccountedfor20percentofthetotalvariance.

Agricultural index:thisfactorgroupedseveralindicatorsrelatedtothehousehold’sagriculturaltools–plough,wheelbarrow,spade,rake,macheteandaxe.Italsoaccountedfor20percentofthetotalvariance.

House infrastructure index:thisfactorgroupedseveralindicatorsrelatedtohousematerials–brickorconcretewall,cementortiledroof,andcementorlaminatedfloor(weassignedascoreof1forthemoreexpensivematerialand0forthecheaper).Thisfactoraccountedfor12percentofthetotalvariance.

Basic services index:thisfactorgroupedtwovariables–whetherthehousehadelectricity,andwhetherthehousehadpipedwaterinthedwelling.Itaccountedfor10percentofthetotalvariance.

Social capital

Asnotedbefore,thehouseholdsurveyincludedasectionrelatingtostructuralandcognitivesocialcapital.Foreachvariable,aspecifichypothesisofassociationisraised,basedonthegeneralhypothesisthatthehigherthequantityandqualityofsocialcapitalavailabletothefamilyandthecommunity,thehighertheeducationaloutcomes.

Thestructuralsocialcapitalvariablesaddressedwere:

You have joined with other community members to address a problem of common concern: thisvariablewascoded1foryesand0forno.Itishypothesisedtohaveapositiverelationshipwitheducationaloutcomes.

You have talked with a local authority or governmental organisation about problems in this community: thisvariablewascoded1foryesand0forno.Thisvariableishypothesisedtohaveapositiverelationshipwitheducationaloutcomes.

Number of groups the respondent/primary caregiver participated in:theorganisationsincludedwere:labourunion,communityassociation,women’sgroup,politicalgroup,religiousgroup,creditorfuneralgroup,clubsorsportassociation,health/water/schoolassociationorcommittee,andothertypeoforganisation.Eachwasscoredwith1foryesand0forno,thusproducingasumscore.Thisvariableishypothesisedtohaveapositiverelationshipwitheducationaloutcomes.

Number of persons or institutions that give support to the household: thisvariablerepresentsthenumberofpersonsorinstitutions(family,neighbours,friends,communityleaders,religiousleaders,politicians,governmentofficialsandcharitableorganisation)thatgivesometypeofsupporttothehousehold.Thisvariableishypothesisedtohaveapositiveassociationwitheducationaloutcomes.

•

•

•

•

a)

b)

c)

d)


25

Thecognitivesocialcapitalvariableswere:

The majority of people in this community can be trusted:thisvariablewascoded1foryesand0forno.Itishypothesisedtohaveapositiverelationshipwitheducationaloutcomes.

You consider yourself similar to most other households in this community: thisvariablewascoded1foryesand0forno.Itishypothesisedtohaveapositiverelationshipwitheducationaloutcomes.

The majority of people in this community generally get along with each other:thisvariablewascoded1foryesand0forno.Thisvariableishypothesisedtohaveapositiverelationshipwitheducationaloutcomes.

You feel you are a part of this community:thisvariablewascoded1foryesand0forno.Again,therelationshipishypothesisedtobepositive.

The majority of people in this community would try to take advantage of you if they got the chance: thisvariablewascoded1foryesand0forno.Itishypothesisedtohaveanegativerelationshipwitheducationaloutcomes.

BasedonthefactorloadingspresentedinTableA1intheAppendix,threesocialcapitalvariablesweregeneratedforeachindividual:

Factor 1: thisfactorgroupsthevariablesrelatedwithcognitivesocialcapital.

Factor 2:thisfactorgroupsthevariablesrelatedwithstructuralsocialcapital,specificallycitizenship.

Factor 3: thisfactorgroupsthevariablesrelatedwithstructuralsocialcapital,specificallythegroupsthatthecaregiverisamemberofandthenumberofindividualsorinstitutionsthatgiveher/himsupport.

Allarehypothesisedtohaveapositiveassociationwitheducationaloutcomes.

Theregressionspresenttheresultsforallrespondentsandalsoformothers’responsesonly.Thejustificationforthisisthatresultsmightdifferdependingonthesocialcapitalofthefamilymemberresponding(forinstance,therespondentmayhaveprovidedinformationonsocialcapitalavailableforhimorherself,notforthewholefamily).Theanalysisisperformedfirstfortheindividuallevelonly,adjustingthenforthecommunitylevel(Table16).

a)

b)

c)

d)

e)

•

•

•


2�

Tabl

e 16

: Coe

ffic

ient

s in

line

ar r

egre

ssio

ns fo

r so

cial

cap

ital

var

iabl

es, a

t ind

ivid

ual a

nd c

omm

unit

y le

vel,

for

child

ren’

s ac

hiev

emen

ta

Who

le sa

mpl

e (in

divi

dual

leve

l onl

y)W

hole

sam

ple

(add

ing

com

mun

ity

vari

able

s)M

othe

rb(in

divi

dual

leve

l onl

y)M

othe

r(a

ddin

g co

mm

unit

y va

riab

les)

Coe

f.C

IC

oef.

CI

Coe

f.C

IC

oef.

CI

Low

erU

pper

Low

erU

pper

Low

erU

pper

Low

erU

pper

Soci

al ca

pita

l var

iabl

es (

indi

vidu

al le

vel)

Fact

or 1

(cog

nitiv

e)-0

.035

-0.1

140.

044

-0.0

45-0

.127

0.03

7-0

.043

-0.1

250.

039

-0.0

51-0

.136

0.03

3

Fact

or 2

(str

uctu

ral –

citiz

ensh

ip)

-0.0

65*

-0.1

360.

006

-0.0

31-0

.120

0.05

8-0

.053

-0.1

330.

027

-0.0

17-0

.123

0.09

0

Fact

or 3

(str

uctu

ral –

mem

bers

hip)

0.02

4-0

.049

0.09

7-0

.006

-0.0

840.

072

0.04

9-0

.033

0.13

10.

027

-0.0

500.

104

Soci

al ca

pita

l var

iabl

es (c

omm

unit

y lev

el)

Fact

or 1

(cog

nitiv

e)0.

133

-0.1

700.

436

0.13

7-0

.187

0.46

1

Fact

or 2

(str

uctu

ral –

citiz

ensh

ip)

-0.2

02-0

.549

0.14

5-0

.224

-0.6

110.

163

Fact

or 3

(str

uctu

ral –

mem

bers

hip)

0.17

7-0

.051

0.40

40.

134

-0.0

980.

366

Con

stan

t-1

.745

***

-2.9

16-0

.573

-1.7

71**

*-2

.974

-0.5

67-1

.804

***

-3.0

29-0

.579

-1.8

80**

*-3

.207

-0.5

54

Num

ber o

f obs

erva

tions

463

463

420

420

R-s

quar

ed0.

360.

370.

340.

35

Num

ber o

f com

mun

ities

5353

5151

* si

gnifi

cant

at

10%

; **

sign

ifica

nt a

t 5%

; ***

sig

nific

ant

at 1

%

No

tes

a t

he r

egre

ssio

n co

ntro

lled

for

the

follo

win

g va

riab

les

at t

he in

divi

dual

leve

l: se

x, g

rade

, hei

ght-

for-

age,

chi

ld w

orks

, cro

wdi

ng in

dex,

mot

her’s

edu

catio

n, m

othe

r’s la

ngua

ge, c

hild

’s ag

e w

hen

he/s

he fi

rst

wen

t to

sch

ool,

livin

g w

ith b

iolo

gica

l par

ents

, par

ents

hel

p w

ith c

hild

’s ho

mew

ork

at h

ome,

att

ende

d pr

esch

ool,

type

of s

choo

l, ar

ea a

nd s

ocio

-eco

nom

ic s

tatu

s.

b

the

sam

ple

for

thes

e re

gres

sion

s on

ly in

clud

ed o

bser

vatio

ns fo

r ch

ildre

n w

hose

mot

hers

res

pond

ed t

o th

e so

cial

cap

ital s

ectio

n in

the

You

ng l

ives

que

stio

nnai

re.


2�

Onl

ystr

uctu

rals

ocia

lcap

ital(

citiz

ensh

ip)w

asfo

und

tob

esig

nific

anta

tthe

indi

vidu

alle

vel(

atth

ete

npe

rcen

tsig

nific

ance

leve

l).T

he

effe

ctd

isapp

eare

dw

hen

the

com

mun

ityle

velw

asin

trodu

ced,

and

was

not

seen

ata

llw

hen

resp

onse

swer

eco

nfin

edto

mot

hers

.The

mod

els

acco

unte

dfo

rbet

wee

n34

and

37

perc

ento

fthe

var

ianc

ein

the

depe

nden

tvar

iabl

e.

Tabl

e17

pre

sent

sthe

resu

ltsfo

rthe

mod

ele

xpla

inin

gth

eod

dso

fac

hild

bei

ngo

n-ag

eat

scho

ol.N

ova

riabl

esw

ere

foun

dsig

nific

anta

tthe

in

divi

dual

leve

l,bu

tatt

hec

omm

unity

leve

lcog

nitiv

eso

cial

cap

italh

ada

pos

itive

rela

tions

hip

with

bei

ngo

n-ag

e,a

spre

dict

ed.T

hiss

eem

slik

ea

robu

stef

fect

,giv

enth

atit

app

ears

bot

hfo

rthe

who

lesa

mpl

ean

dfo

rthe

mot

hers

asr

espo

nden

tson

ly,a

ndis

sign

ifica

nta

tthe

five

per

ce

ntle

vel.

Tabl

e 17

: Log

-odd

s co

effic

ient

s fo

r so

cial

cap

ital

var

iabl

es in

logi

stic

reg

ress

ion

for

on-a

ge (

=1)

vers

us o

vera

ge (

=0)

child

rena

Who

le sa

mpl

e (in

divi

dual

leve

l onl

y)W

hole

sam

ple

(add

ing

com

mun

ity

vari

able

s)M

othe

rb(in

divi

dual

leve

l onl

y)M

othe

r(a

ddin

g co

mm

unit

y var

iabl

es)

Coe

f.C

IC

oef.

CI

Coe

f.C

IC

oef.

CI

Low

erU

pper

Low

erU

pper

Low

erU

pper

Low

erU

pper

Soci

al ca

pita

l var

iabl

es (i

ndiv

idua

l lev

el)

Fact

or 1

(cog

nitiv

e)-0

.039

-0.3

400.

261

-0.1

58-0

.503

0.18

7-0

.011

-0.3

150.

294

-0.1

18-0

.466

0.23

0

Fact

or 2

(str

uctu

ral –

citiz

ensh

ip)

0.10

6-0

.177

0.39

00.

173

-0.1

390.

484

0.11

9-0

.177

0.41

50.

193

-0.1

310.

517

Fact

or 3

(str

uctu

ral –

mem

bers

hip)

-0.1

54-0

.382

0.07

4-0

.161

-0.4

150.

093

-0.1

14-0

.367

0.14

0-0

.127

-0.3

990.

144

Soci

al ca

pita

l var

iabl

es (c

omm

unit

y lev

el)

Fact

or 1

(cog

nitiv

e)0.

682*

*0.

066

1.29

80.

683*

*0.

038

1.32

8

Fact

or 2

(str

uctu

ral –

citiz

ensh

ip)

-0.3

33-0

.824

0.15

8-0

.354

-0.8

850.

177

Fact

or 3

(str

uctu

ral –

mem

bers

hip)

-0.0

06-0

.466

0.45

30.

029

-0.4

830.

542

Con

stan

t12

.183

***

7.42

616

.940

12.1

95**

*7.

556

16.8

3310

.959

***

6.41

715

.502

10.9

84**

*6.

530

15.4

37

Num

ber o

f obs

erva

tions

534

534

488

488

Pseu

do r–

squa

re0.

190.

200.

170.

18

Log

likel

ihoo

d-2

08.2

1-2

05.7

3-1

96.3

9-1

94.0

5

* si

gnifi

cant

at

10%

; **

sign

ifica

nt a

t 5%

; ***

sig

nific

ant

at 1

%

No

tes

a t

he r

egre

ssio

n co

ntro

lled

for

the

follo

win

g va

riab

les

at t

he in

divi

dual

leve

l: se

x, h

eigh

t-fo

r-ag

e, c

hild

wor

ks, c

row

ding

inde

x, m

othe

r’s e

duca

tion,

mot

her’s

lang

uage

, chi

ld’s

age

whe

n he

/she

firs

t w

ent

to s

choo

l, liv

ing

with

bio

logi

cal p

aren

ts, p

aren

ts h

elp

with

chi

ld’s

hom

ewor

k at

hom

e, a

tten

ded

pres

choo

l, ty

pe o

f sch

ool,

area

and

soc

io-e

cono

mic

sta

tus.

b t

he s

ampl

e fo

r th

ese

regr

essi

ons

only

incl

uded

obs

erva

tions

for

child

ren

who

se m

othe

rs r

espo

nded

to

the

soci

al c

apita

l sec

tion

in t

he Y

oung

liv

es q

uest

ionn

aire

.


28

4. discussionEducationinPeruisindeedasourceofgreatconcern.In2003,thePeruvianGovernmentdeclarededucationtobeinastateofemergency,duemostlytothefactthattheProgrammeforInternationalStudentAssessmentstudyinthatyearshowedaround80percentof15-year-oldsinschoolinPeruwereatthelowestornexttolowestreadinglevel(ietheyexperienceddifficultyinterpretingevenverysimpletexts).Thisresultwasinperfectagreementwiththenationalevaluationcarriedoutin2001andtheinternationalevaluationcarriedoutbyUNESCO(UMCandGRADE,2001)whichshowedverypoorachievementforstudentsatdifferentgradelevels.Yetthesuccessofthemeasuresderivedfromthisemergencyisstillunknown,morethantwoyearsafteritwasdeclared.Tobeginwith,theemergencydidnotresultinspecificactionsforoverayear.Later,2,500schoolsinimpoverishedareas(100ineachofthe25departmentsofPeru)weredeclaredasprioritiesforintervention.Theinterventionconsistedoftrainingteachersonreadingandmathematicstechniques.Itisunclear,though,whethertheprogrammehasindeedbeenimplementedinallschoolsorhashadanyimpact.

TheresultsofthisstudyshowpooreducationaloutcomesformanyPeruvianstudents,whocannotmastersimpletasksandarebehindthegradefortheirage.ManyPeruvianstudentsinsecond,thirdorfourthgradecannotsolveasimplemathematicscomputation,orreadorwriteasentence.Thisfindingisconsistentamongseveralstudies.MostchildreninPerudoattendschool,butitisobviousthattheydonotlearnwhatisprescribedinthecurriculumnoradvancethroughthegradelevelsatthepacetheyshould.Thecurrentstudyalsoshowsthatthereisaclearnegativeassociationbetweeneducationalachievementandpoverty.Overall,ruralstudentsarepoorer,andthusmorepronetolowachievementandfallingbehindtheirexpectedgrade.Withinurbanandruralareas,thereseemtobenostrongdifferencesbetweenboysandgirlswithregardtoenrolmentorachievementinliteracy,buttherearesomedifferenceswithnumeracy,infavourofboys.Nevertheless,themainproblemswithineducationinPeruarelowachievementbystudentsinstandardisedevaluationsandhighinequality,associatedwithsocio-economicindicators,forbothboysandgirls.

Thispaperanalysedtherelationshipbetweenstructuralandcognitivesocialcapitalandeducationaloutcomes.Theresultsdonotsupportthehypothesisthatthereisapositiveassociationbetweenthequalityandquantityofsocialcapitalavailabletothefamilyandeducationalresults,exceptfortheassociationbetweencognitivesocialcapital(atthecommunitylevel)andbeingon-ageinschool.Thismeansthatcommunitieswithmorequalityintheirsocialrelationships(egtrust)aremorelikelytohavechildrenwhoareon-ageinschool.Thisresultappliestohowchildrenareprogressingatschoolbutdoesnotextendtochildren’sachievement.

Wewillnowreturntothequestionofhowsocialcapitalvariablescouldaffecteducationaloutcomes.InPeru,parentsmaychoosetosendtheirchildrentoanyschool(publicorprivate),butprivateschoolsregardedas‘highquality’arealsoexpensive(chargingUS$300permonthormore).Publicschoolsaremostlyfree(exceptforafeefortheparents’association,aroundUS$14peryear,whichmaybewaived),butthereareothercostsinattendingschool,suchastransportation,uniforms,extrabooks,Englishlessons,etc.Thus,whenparentsdecidewheretosendtheirchildrentoschool,theytendtochooseaschoolthatfitsthefamily’ssocio-economiccharacteristics.Forinstance,inurbanareassomepublicschoolsmaydemandmoreworkoradditionalresourcesfromparents,andinruralareasusually


2�

thereisonlyoneschoolthechildmaygoto(thenextschoolbeingseveralkilometresaway,withnopublictransportationavailable).Overall,schoolsarerunbyteachers,withverylittleornoparticipationfromtheparentsorparents’representatives.Theretendstobelessintra-schoolvariabilitythaninter-schoolvariability,hencethestrongassociationbetweenthesocio-economicstatusofthestudentandhis/hereducationaloutcomes.(ForareviewofthePeruvianliteratureondeterminantsofachievementinPeruseeCuetoandRodriguez,2003.)Arecentstudy(AgüeroandCueto,2004)showedstrongpeereffectsinachievementinPeruvianclassrooms(peereffectsweredefinedastheachievementresultsofanygivenstudent’sclassmates).Thisstudysuggeststhattheaverageachievementcouldbeincreasedbymixingstudentsofhighandloweducationalachievementinthesameclassrooms.

InthePeruviansystemtherearenopubliclyprovidedprogrammestohelpstudentscomingfrompoorereducationalenvironments,norprogrammestargetedatstudentswho,forwhateverreason,fallbehindtheirpeers.Infact,studentswhofailagivencourseandwanttotakeremedialcourseworkoverthesummermustpayforit.Somestudies(egCueto,RamirezandLeon,inpress)suggestthattheopportunitiestolearninsidetheclassroomarealsomarkedbysocio-economicdifferences(forinstance,childrenfrompoorerfamiliessolvefewermathematicsexercisesandreceivelessfeedbackfromtheirteachers).Thesystemissetupsothatthepoorerstudentsaremorepronetolowachievementandeventuallytodroppingoutofschool.Ithasbeendubbeda‘Darwinian’system,inthatonlythe‘fittest’willsurvivebyadaptingtotheenvironmentintheschools,whicharemoreorlessinflexibleinstitutions(Cueto,2004).Thusitwouldseemthatthepossibilitiesofsocialcapitalvariableshavingamarkedinfluenceoneducationaloutcomesare,indeed,limitedbythecharacteristicsofaneducationsystemthatismarkedbylowqualityandhighinequality.Itmustalsoberememberedthatnoneofthecommunity-basedorganisations,describedearlierinthispaper,hasasoneofitsgoalstheimprovementofeducationaloutcomesforprimaryschoolagedchildren.TheemergencyineducationdeclaredbytheGovernmentin2003didnotincludetheparticipationofcommunity-basedorganisationsorfamiliesinanyway;itwasdesignedasapurelyeducationalintervention(iefocusingonteachertraining).Ontheotherhand,itmightbethatsocialcapitaldoesimpactonsomeoutcomes(egnutrition,localsecurity,daycare),butnotoneducation,whichmaybeperceivedasafamily,notasocial,concern.

FromourreviewoftheliteratureofdeterminantsofeducationinPeruandotherLatinAmericancountries(egMurillo,2003),interventionsforimprovingthequalityofthesystemsthemselvesaremostlycomingfromwithintheeducationsystems,andthebiggestchallengesareaboutbreakingthestrongassociationbetweeneducationlevelandsocio-economicstatus.Indifferentcontexts,interventionsaimedatimprovingthequalityofteaching,educationalmaterials,activelearningtimeandschoolautonomy(undertheleadershipoftheschoolprincipal)haveproveneffective.Thisisnottosaythatinterventionsatthecommunitylevel,includingsocialcapital,couldnothaveanimpactonpooreducationaloutcomes.However,itishardtoimaginethatsocialcapitalper sewouldresultin,say,animprovementineducationalachievementamongthestudents,ifteachersdidnotalsoimprovetheirteachingmethodsorincreasethetimespentonactivelearningwithintheclassroom.Forexample,Cuetoet al(2004)haveshownthatstudentsfrompoorercontextshavefeweropportunitiestolearnmathematicsthanchildreninrichercontexts(bothinpublicschools).Alongitudinalanalysisshowedthattheachievementgapbetweenpoorerandricherstudentstendedtoincreaseovertime,andopportunitiestolearnwasanexplanatoryvariableforhigherachievement.


30

AnumberofissuestoconsiderregardingtheanalysisperformedhereandthepossibilitiesforfutureroundsoftheYoungLivesprojectarenowexplored.

The design in this first round was cross-sectional, not longitudinalThecovariates,therefore,mightnothavebeensufficienttocontrolforthepovertylevelofthefamily.Inotherwords,itmightbethatfamiliesareindeedgettingbettereducationaloutcomesbecauseoftheirsocialcapital,butthedesignandanalysiswasnotstrongenoughtocapturethis.

Validity of the social capital questions in the questionnaireThequestionsbelongtotheA-SCATinstrument,developedbytheWorldBank.However,theinstrumentwasnotusedinitscompleteform.DeSilvaet al(2004b)carriedoutavaliditystudyoftheinstrumentandfoundthat,ingeneral,itpresentedacceptablefacevalidity.However,someitemsmighthavebeenmisunderstood.Forinstance,DeSilvaet al foundthatwhenaskedabout‘theircommunity’respondentshaddifferentinterpretationsofwhatthismeant.Inaddition,groupmembershipmighthavebeenunder-reportedbecausepeoplewouldnotconsidersomeoftheirgroupstobelongtothespecificcategoriesposedbythequestionnaire.Socialsupportmayalsohavebeenunderestimatedbecause,whenasked,peopleapparentlythoughtmostlyofeconomicsupport,althoughitwasintendedthatemotionalandinstrumentalsupportwouldalsobeconsidered.Finallythequestionsregardingtrustweredifficultforpeopletoanswer,becausetheywerenotwillingtocommentonmembersoftheircommunitywhomtheydidnotknowpersonally.DeSilvaet almakeseveralsuggestionstobeconsideredinfutureroundsofthestudy.Itmightbethatobjectivequestions,suchasthoseposedbytheinstrumentused,arenotthebestwaytoexploresocialcapital–duetodifferencesininterpretations–oratleastthatmorequalitativemethodsshouldbeusedwiththequestionnairetoexploretheimportanceofsocialcapitalinPeru.Asmentionedintheliteraturereview,thestudiesthatdidfindassociationsbetweensocialcapitalandeducationintheUSAdidnotapplytheinstrumentusedinthecurrentstudy,butusedlimiteddefinitionsofsocialcapital(egpresenceofbothbiologicalparentsathome).

The questionnaires did not include questions about social capital at schoolThereisinternationalevidenceshowingtheimportanceofschoolsinexplainingeducationaloutcomes,andthereissomeliteraturethatsuggeststhatsocialcapitalatschoolisimportant(Goddard,2003).Socialcapitalatschoolmaybemorerelevantthanfamilysocialcapital,especiallyinearliergradeswhenthebasicskillsofreading,writingandmathematicsarebeinglearned.

There is no exploration of the mechanisms by which social capital could explain educational outcomes

Isitadditionalresources,familyvalues,orthecollaborationofcommunitymembersintheeducationoflocalchildrenthatwouldexplaineducationaloutcomes?BasedonthetheoreticalmodelpresentedbyHarpham(2002)andothers,wesuggestperformingqualitativeinterviewstoexplorehowitisthatsocialcapital(especiallysocialcapitalatthecommunitylevel)mighthaveaninfluenceoneducationaloutcomes,beforeembarkingonfurtherresearchonthisissue.

Fromtheresultsofasinglestudyitwouldbecarelesstoaffirmthatsocialcapitalshouldnotbeamongthepotentiallybeneficialpolicyinterventionsineducation–especiallysincesocialcapitalappearstobe

1.

2.

3.

4.


3�

importantinimprovingthequalityoflifeofchildreninothercountries,andgiventheimportanceofnetworkingamongpoorcommunitiesinurbanandruralPeru.Wedosuggest,however,thatdifferentinstrumentsanddesignsbeused,inadditiontoanalysisbasedpurelyonobjectivedatacomingfromquestionnaires.


32

referencesAgüero,JandSCueto(2004)‘Dimeconquiénestudiasytedirécómorindes:Peer-effectscomodeterminantesdelrendimientoescolar’,Lima,unpublishedresearchreportpreparedforthe Economic and Social Research Consortium

Ansión,J,ALazarte,SMatos,JRodríguezandPVegaCenteno(1998)Educación: La Mejor Herencia,Lima,PontificiaUniversidadCatólicadelPerú

Benavides,M(2002)‘ParaexplicarlasdiferenciasenelrendimientoenmatemáticadecuartogradoenelPerúurbano:Análisisderesultadosapartirdeunmodelobásico’,inJRodríguezandSVargas(eds)Análisis de Resultados y Métodos de las Pruebas Crecer 98,MinisteriodeEducación,DocumentosTécnicos13,83–108,February

Blackwell,DLandDKMcLaughlin(1999)‘Doruralyouthattaintheireducationalgoals?’,Rural Development Perspectives,13(3),37–44

Blondet,C(1986)Muchas Vidas Construyendo una Identidad. Las mujeres pobladoras de un barrio limeño,Documentodetrabajo(workingpaper)Nº9,Lima,InstitutodeEstudiosPeruanos

Blondet,CandCMontero(1995)Hoy: Menú popular,Lima,InstitutodeEstudiosPeruanos;UNICEF

Bryk,AandSRaudenbush(1992)Hierarchical Linear Models,NewburyPark,California,Sage

Coleman,JS(1988) ‘Socialcapitalinthecreationofhumancapital’, American Journal of Sociology, 94SupplementS95-S-120

CVR(ComisióndelaVerdadylaReconciliación)(2003)Informe Final,Lima,consultedinMarch2004atwww.cverdad.org.pe/ifinal/index.php

Cortéz,R(2000)‘Acumulacióndecapitalhumano:EldesempeñoescolarenelPerú’,Lima,DepartamentodeEconomía,UniversidaddelPacífico,unpublished

Cueto,S(2004)‘FactoresPredictivosdelRendimientoEscolar,DesercióneIngresoaEducaciónSecundariaenunaMuestradeEstudiantesdeZonasRuralesdelPerú’,Education Policy Analysis Archives,12(35),28July,availableathttp://epaa.asu.edu/epaa/v12n35/

Cueto,S,CRamírezandJLeón(inpress)‘OpportunitiestolearnandachievementinasampleofsixthgradestudentsinLima,Peru’,Educational Studies in Mathematics

Cueto,S,CRamírez,JLeónandGGuerrero(2004)‘OportunidadesdeaprendizajeyrendimientoenmatemáticadelosestudiantesdetercerycuartogradosdeprimariaenLimayAyacucho’,inMBenavides(ed)Educación, Procesos Pedagógicos y Equidad(pp.15–68),Lima,GRADE

Cueto,SandJRodríguez(2003)‘EstudiosempíricossobredeterminantesdelrendimientoescolarenelPerú’,inJMurillo(ed)La Investigación sobre Eficacia Escolar en Iberoamérica(pp.419–50),Bogota,ConvenioAndrésBello&CentrodeInvestigaciónyDocumentaciónEducativa

Cueva,HandAMillán(2000)Las Organizaciones Femeninas para la Alimentación y su Relación con el Sector Gubernamental,Lima,CentrodeInvestigacióndelaUniversidaddelPacífico


33

DeSilva,MJ,THarpham,SHuttly,RBartoleni,MEPennyandJEscobal(2004a)‘UnderstandingSourcesandTypesofSocialCapitalinPeru’,Community Development Journal.AdvanceElectronicAccess

DeSilva,MJ,THarpham,SHuttly,RBartoleni,MEPennyandJEscobal(2004b)‘ValidationofaSocialCapitalMeasurementToolforUseinPeru’,Social Science and Medicine. AdvanceElectronicAccess

Escobal,J,CFLanata,SMadrid,MEPenny,JSaavedra,PSuárez,HVerástegui,EVillarandSHuttly(2003)Young Lives Preliminary Country Report: Peru,London,YoungLives

Escobar,AandSAlvarez(1992)The Making of Social Movements in Latin America,Colorado,WestviewPress

Furstenberg,FFJrandMEHughes(1995)‘Socialcapitalandsuccessfuldevelopmentamongat-riskyouth’,Journal of Marriage and the Family, 57,580–92

Goddard,RD(2003)‘Relationalnetworks,socialtrust,andnorms:Asocialcapitalperspectiveonstudents’chancesofacademicsuccess’,Educational Evaluation and Policy Analysis,25(1),59–74

Guadalupe,C(2002)La Educación Peruana a Inicios del Nuevo Siglo,ProgramaEspecialMejoramientodelaCalidaddelaEducaciónPeruana,DocumentodeTrabajo(workingpaper)12,Lima,MinisteriodeEducación

Harpham,T(2002)Measuring the Social Capital of Children, WorkingPaperNo.4,London,SouthBankUniversity

Harpham,T,EGrantandEThomas(2002)‘Measuringsocialcapitalwithinhealthsurveys:Keyissues’,Health Policy and Planning,17(1),106–11

Horvat,EM,EBWeiningerandALareau(2003)‘Fromsocialtiestosocialcapital:Classdifferencesintherelationsbetweenschoolsandparentnetworks’,American Educational Research Journal,40(2),319–51

INEI(1995)Atraso y Deserción Escolar en Niños y Adolescentes,Lima,INEIandWorldFoodProgramme

Israel,GD,LJBeaulieuandGHartless(2001)‘Theinfluenceoffamilyandcommunitysocialcapitaloneducationalachievement’,Rural Sociology,66(1),43–68

Jack,GandBJordan(1999)‘Socialcapitalandchildwelfare’, Children and Society,13,242–56

Lindstrom,M(2004)‘Socialcapital,theminiaturizationofcommunityandself-reportedglobalandpsychologicalhealth’,Social Science and Medicine,59,595–607

Murillo,J(ed)(2003)La Investigación sobre Eficacia Escolar en Iberoamérica,Bogota,ConvenioAndrésBello&CentrodeInvestigaciónyDocumentaciónEducativa

Narayan,D(1999) Bonds and Bridges: Social Capital and Poverty, Washington,DC,WorldBank

PISA(2003)Literacy Skills for the World of Tomorrow. Further Results from PISA 2000,ProgrammeforInternationalStudentAssessment,OECD-UNESCO,June


34

Pong,S(1997)‘Familystructure,schoolcontext,andeighth-grademathandreadingachievement’,Journal of Marriage and the Family,59, 734–46

Portes,A(1998)‘Socialcapital:Itsoriginsandapplicationsinmodernsociology’,Annual Review of Sociology,24,1–24

Pozzi-Escot,I(1998)El Multilingüismo en el Perú,Cusco,PROEIBAndes,CentrodeEstudiosRegionalesAndinos‘BartolomédelasCasas’

Putnam,RD(1995)‘Bowlingalone:America’sdecliningsocialcapital’, Journal of Democracy,6(1),65–78

Ream,RK(2003)‘CounterfeitsocialcapitalandMexican-Americanunderachievement’,Educational Evaluation and Policy Analysis,25(3),237–62

Runyan,DK,WMHunter,RRSSocolar,LAmaya-Jackson,DEnglish,JLandsverk,HDubowitz,DBrowne,SIBangdiwalaandRMMathew(1998)‘Childrenwhoprosperinunfavorableenvironments:Therelationshiptosocialcapital’,Pediatrics,101(1),12–18

Teachman,JD,KPaaschandKCarver(1996)‘Socialcapitalanddroppingoutofschoolearly’,Journal of Marriage and Family,58(3),773–83

UMC(UnidaddeMedicióndelaCalidadEducativa)andGRADE(2001)‘ElPerúenelprimerestudiointernacionalcomparativodelaUNESCOsobrelenguaje,matemáticayfactoresasociadosentercerycuartogrado’,Boletín UMC 9,Lima,MinisteriodeEducación


35

appendix: Factor analysis for derivation of summary social capital variables

Thefactoranalysisdescribesthecovariancerelationshipamongallthesocialcapitalquestions.Theresultsofthefactoranalysisshowthattherearethreefactorsdescribingdifferentdimensionsofsocialcapital.Thefirstfactordescribescognitivesocialcapital;themajorityofloadingsarepositiveexceptforthevariablethatisanegativeitemanditexplainsalmost23percentofthevariance.Thesecondfactorgroupstogethervariablesrelatedtocitizenship,andalltheloadingsarepositive.Thelastfactorbringstogethervariablesrelatedtothenumberofgroupsorinstitutionsofwhichthecaregiverisamemberandthesupportthecaregiverreceivesfromindividualsorinstitutions.Alltheloadingsarepositive.Fortherotationofthefactorsweusedthevarimaxrotation,andthemethodofextractionusedwasprincipalcomponents.Thevarianceaccountedforbythethreefactorswas57.42percentofthetotalvariance.TableA1presentstherotatedloadingfactorshigherthan0.4orlowerthan-0.4fromthefactoranalysisandthevarianceaccountedforbyeachfactor.

Table A1: Estimated rotated factor loadings Components

Factor 1 Factor 2 Factor 3

You have joined with other community members to address a problem of common concern 0.82

You have talked with a local authority or governmental organisation about problems in this community 0.85

Can the majority of people in this community be trusted? 0.71

Do the majority of people in this community generally get along with each other? 0.76

Do you feel as though you are part of this community? 0.61

Would the majority of people in this community try to take advantage of you if they got the chance? -0.61

Number of groups the respondent/primary caregiver participated in 0.71

Number of persons or institutions that give support to the household 0.83

Total variance explained 22.92 18.97 15.53

The Young Lives Partners Centre for Economic and Social Studies (CESS), India

Department of Economics, University of Addis Ababa, Ethiopia

Ethiopian Development Research Institute, Addis Ababa, Ethiopia

General Statistical Office, Government of Vietnam

Grupo de Análisis Para El Desarrollo (GRADE), Peru

Institute of Development Studies, University of Sussex, UK

Instituto de Investigación Nutricional (IIN), Peru

London School of Hygiene and Tropical Medicine, UK

Medical Research Council of South Africa

Queen Elizabeth House, University of Oxford

RAU University, Johannesburg, South Africa

Research and Training Centre for Community Development, Vietnam

Save the Children UK

South Bank University, UK

Statistical Services Centre, University of Reading, UK

W O R K I N G P A P E R N O . 2 8

Social Capital and Education Outcomes in Urban and Rural PeruSantiago CuetoGabriela GuerreroJuan LeonMary De SilvaSharon HuttlyMary E PennyClaudio F LanataEliana Villar

Although enrolment in primary schools in Peru is very high, more than half of primary school children are one or more grades below the norm for their age. Furthermore, evaluations have shown that, when tested, Peruvian school children score well below the norms expected for their age. Their scores are also below the average levels of countries with similar socio-economic circumstances. The role of social capital in explaining these findings has not been studied, although research in the USA has suggested positive associations between social capital and educational achievement. Social organisations that focus on childcare are one example of strong community networking resources in Peru. The Young Lives study offers an opportunity to investigate whether social capital is associated with educational progress and achievement.

The results of the study confirmed poor educational outcomes for many Peruvian school children. High proportions were unable to master simple tasks and were in a lower school grade than they should have been for their age group. There is a clear negative association between educational achievement and poverty. Overall, rural students are poorer and are thus more prone to low achievement (lower results in tests) and falling behind their expected grade. However, there seem to be no significant differences between boys and girls in these outcomes.

The results do not support the hypothesis of a positive association between the social capital available to the family and the educational outcomes of their children, except for the association between cognitive social capital (at the community level) and children being in the correct grade for their age. This means that communities which experience more quality in their social relationships (e.g. trust) are more likely to have children who are in the correct school grade for their age. This result applies to how children are progressing at school (whether they are in the correct grade for age) but does not extend to children’s achievement (results in tests).

Perhaps the main interventions to improve the quality of the Peruvian education system are not to be found in the quality and quantity of social relationships within communities, but in improving educational inputs and processes within schools and breaking the strong association found between socio-economic status and educational performance. On the other hand, it might be that the types of instruments (objective questionnaires) and analysis (quantitative) used in this study do not offer the best way to capture the importance of social capital. If this is the case, then qualitative designs and analyses should be used to explore these issues.

Published by

Young Lives Save the Children UK 1 St John's Lane London EC1M 4AR

Tel: 44 (0) 20 7012 6796 Fax: 44 (0) 20 7012 6963 Web: www.younglives.org.uk

ISBN 1-904427-29-4 First Published: 2005

All rights reserved. This publication is copyright, but may be reproduced by any method without fee or prior permission for teaching purposes, though not for resale, providing the usual acknowledgement of source is recognised in terms of citation. For copying in other circumstances, prior written permission must be obtained from the publisher and a fee may be payable.

Designed and typeset by Copyprint UK Limited

Young Lives is an international longitudinal study of childhood poverty, taking place in Ethiopia, India, Peru and Vietnam, and funded by DFID. The project aims to improve our understanding of the causes and consequences of childhood poverty in the developing world by following the lives of a group of 8,000 children and their families over a 15-year period. Through the involvement of academic, government and NGO partners in the aforementioned countries, South Africa and the UK, the Young Lives project will highlight ways in which policy can be improved to more effectively tackle child poverty.

social capital and education outcomes in urban and rural peru · social capital and education...

Documents