+ All Categories
Home > Documents > Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors...

Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors...

Date post: 15-Jul-2020
Category:
Upload: others
View: 1 times
Download: 0 times
Share this document with a friend
35
Friends Asking Friends for Charity: The Importance of Gifts and Audience Marco Castillo* Ragan Petrie* Clarence Wardell** August 2017 Abstract: We explore the underlying motivations for donors to ask their friends to give to charity and the effectiveness of alternative ways to ask for support. Using data from a field experiment with a global crowdfunding community and over 9,800 unique donors, we find that asking a friend in front of his friends is the most effective way to fundraise. Our estimates show that $1 spent to encourage donors to ask friends pays for itself by returning about $1.60. The social pressure that an audience provides is the primary driver for this result, rather than asking a larger group of individuals. Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for the friend, however, increases the willingness to ask. Both social context and donor incentives are equally important in explaining giving. Keywords: fundraising, peer-to-peer solicitation, field experiment, social media, online giving, social pressure, audience JEL codes: D64, C93 * Department of Economics, Texas A&M University, College Station, TX USA and Melbourne Institute, University of Melbourne, Australia. Castillo: [email protected], Petrie: [email protected] ** What Works Cities Initiative, Washington, DC USA. Wardell: [email protected]
Transcript
Page 1: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

FriendsAskingFriendsforCharity:TheImportanceofGiftsandAudienceMarcoCastillo*RaganPetrie*ClarenceWardell**August2017Abstract:Weexploretheunderlyingmotivationsfordonorstoasktheirfriendstogivetocharityandtheeffectivenessofalternativewaystoaskforsupport.Usingdatafromafieldexperimentwithaglobalcrowdfundingcommunityandover9,800uniquedonors,wefindthataskingafriendinfrontofhisfriendsisthemosteffectivewaytofundraise.Ourestimatesshowthat$1spenttoencouragedonorstoaskfriendspaysforitselfbyreturningabout$1.60.Thesocialpressurethatanaudienceprovidesistheprimarydriverforthisresult,ratherthanaskingalargergroupofindividuals.Socialpressureiscostly,anddonorsarereluctanttouseitwiththeirfriends.Providingdonorswitha“gift”forthefriend,however,increasesthewillingnesstoask.Bothsocialcontextanddonorincentivesareequallyimportantinexplaininggiving.Keywords:fundraising,peer-to-peersolicitation,fieldexperiment,socialmedia,onlinegiving,socialpressure,audienceJELcodes:D64,C93*DepartmentofEconomics,TexasA&MUniversity,CollegeStation,TXUSAandMelbourneInstitute,UniversityofMelbourne,Australia.Castillo:[email protected],Petrie:[email protected]**WhatWorksCitiesInitiative,Washington,DCUSA.Wardell:[email protected]

Page 2: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

1

1.Introduction

DonationstocharitableorganizationsamounttoUS$350billion(about2%ofGDP),and

7%camefromonlinegiving(GivingUSA2014,BlackbaudIndex).Therearenumerous

fundraisingwebsitesthatprovideinformationaboutcharitiesandfacilitatesecureonline

donations(e.g.Causes,Crowdwise,NetworkforGood,GiveGab).Mostsitesprovideaway

fordonorstoconvenientlyinformfriendsthattheyhavesupportedacharityandasktheir

friendstogivetothatcharityaswell.Whileaskingafriendtodonatemaybeaclickaway,it

isnotinconsequentialtodoso.Therearesocialcostsfortheindividualmakingtherequest

and,potentially,forthefriendshouldshechoosetosayno.Nonetheless,itmightbeworth

asking,especiallyifthefriendmakesadonation.Howthesebenefitsandcostsaffecta

donor’swillingnesstoask,afriend’swillingnesstocomplyandtheeffectivenessof

differentsolicitationapproaches(e.g.inprivate,infrontofothers,withagiftforthefriend)

arestillopenquestions.Wesystematicallyaddresstheseusingafieldexperimentin

partnershipwithaglobalonlinecrowdfundingcommunity.

Evidencesuggeststhatrequestsfromfriendscanbebeneficial(seeCastilloetal,

2014;Sandersetal,2014;Meer,2011).However,thereasonswhythisworksarelessclear.

Ourcontributionliesinidentifyinghowthemechanismofpeer-to-peersolicitation

operates.Wefindthatnotallmannersofaskingareusefulorused.Thecontextofthe

solicitationisimportant.Itcantriggersocialpressure,andinturngenerateadonation,but

itisnotwithoutsocialcosts.Wefindthatdonorsaremorewillingtobearthesecostswhen

theycanpairtherequestwithagifttotheirfriend.Subsidizingpeer-to-peersolicitationby

providingagiftforexchangeorfacilitatingrequestscouldbeanovelapproachforcharities

tofundraise.

Ourfieldexperimentisdesignedtoexaminetheperceivedcoststothedonorof

askingamemberofhisonlinesocialnetwork(onFacebook)tosupporthischarityandthe

mechanismsforwhyfriendsmightrespond.1Inourexperiment,werandomlyvarythe

1Socialmediahasobviousinformationaladvantagesoverotherwaysoffundraising,andourresultssuggestthatpartofitssuccessislikelyduetodonors’concernsforsocialimage.Businessesandmarketingprofessionalsstresstheimportanceofreferralsandwordofmouthcampaignstopromoteandsellproductsandservices,andourstudyexaminesthisphenomenoninthecontextofcharitablegiving.Kumaretal(2007)findthatwhileroughly70%offinancialcustomerssaytheywouldreferafriend,only30%do.Similarly,we

Page 3: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

2

availablesocialcontextandmannerinwhichadonorcanaskfriendsandcompare

solicitationrates.Weexplorethreecommonwaysofasking:1)ageneralsolicitationto

manyfriendstogiveviaapostonthedonor’sFacebookwall,2)adirectsolicitationtoone

friendinprivateviaaFacebookprivatemessage,and3)adirectsolicitationtoonefriendin

frontofhisfriendsviaapostonafriend’sFacebookwall.Tomeasuretheperceivedcostsof

asking,wealsorandomly“subsidize”thesolicitationbyofferingeithernoincentiveora

smallmonetaryincentive($1or$5)todonorstoasktheirfriends.Theincentiveisapplied

asanadd-ondonationmadetothedonor’schosencharitywhenthedonorcompletesthe

solicitation.Toavoidpotentialcrowdingoutofdonations,allthetreatmentsinour

experimentwereimplementedafteradonationhadbeencompletedsothattheinitial

donationdecisionisunaffected.

Wefurthervarythesocialcontextoftheadd-ondonationbyrandomlypositioning

theincentiveasa“gift”tothefriendwhenmakingthesolicitation.Thisisdoneby

informingthedonorthattheadd-ondonationwilleitherbemadein1)hisname,2)the

friend’snamebutkeptsecretfromthefriend,or3)thefriend’snameandmadeknownto

thefriend.Anadd-ondonationinthefriend’snameandannouncedtothefriendallowsthe

donortosignalhisgenerositybynotonlymakingadonationhimselfbutalsooneforhis

friend.Additionally,itcouldserveasawayforthedonortopayforwardforhisfriend’s

donation(i.e.thankingthefriendinadvance,ormitigatingfutureobligations)andcould

simultaneouslyplaceadditionalpressureonthefriendtogive.Inessence,the“gift”

providesthedonorafavorhecouldtradewithhisfriend,ifhesodesires.Also,the“gift”

mayhelptolessenareluctancetoaskafriendinfrontofhisfriends,whensocialpressure

orimagecostsarehighest.Ourtreatmentarmsallowustomeasurethis,andwefindthat

havingagiftinthiscontextincreasessolicitations.

Crucialtoourdesignandmeasureofeffectivenessisthatwhethertheadd-on

donationismadeinthedonor’snameorthefriend’sname,thecosttothecharityisexactly

thesame.Thisallowsustoexaminewhatisthemostcost-effectiveway,inthis

environment,foracharitytospendafundraisingdollar.Wethencalculateanestimated

findadifferenceintheproportionofdonorswhoactuallydosolicitafriendandthosewhomanifestinterestindoingso.

Page 4: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

3

rateofreturnofthevariousaskmethods,incentiveamounts,andabilitytofavortradeas

fundraisingstrategies.

Wehaveseveralkeyfindings.First,askingafriendinfrontofhisfriendsisthemost

effectiveatraisingadditionaldonations--thepercentofsolicitationsthatresultinnew

donationsis3.5%,roughly50%morethanaskingmanyfriendsatonce(2.3%).2Onnet,it

isthemosteffectivefundraisingmethodandreturns$1.63indonationsforevery$1spent

inadd-ondonations.Nonetheless,donorsaretheleastlikelytousethismethodwhen

offeredasitappearstobethemostsociallycostly.Sendingaprivatemessagetoafriend

generatednonewdonations.

Second,whileaskingafriendinfrontofhisfriendsisusedtheleast,providingthe

donoranadd-ondonationhecanpubliclygivehisfriendincreasesthistypeofsolicitation

byalmost50%.Socialpressureappearstobeaneffectivemechanismandmaybemore

palatablewhensweetenedwitha“gift.”Or,donorsprefertheheightenedpressurethe

knowledgeoftheadd-ondonationprovides.Thisimpliesthatcharitiescanpartiallyoffset

thereluctancetoaskfriendsinpublicbypositioningadd-ondonationsinthedonor’s

friend’sname.

Finally,behaviorisnotconsistentwithdonorsattemptingtomaximizethesizeof

theaudienceexposedtothesolicitation.Donorsaresignificantlylesslikelytoaskafriend

infrontofhisfriendsthanaskthefriendinprivate(6.3%comparedto8.1%).Thisisnot

becausedonorsdonotcareaboutthesizeoftheaudience--themostpreferredmethodis

toaskallfriendsatonceinpublic(13.4%)--itisthatdonorspreferaskingafriendinfront

ofhisfriends’friendstheleast.Thissuggeststhatapotentialreasonforthesuccessof

onlinegivingcampaigns,apartfrominformationefficiency,isthemagnificationofaudience

effects.

Ourresultsmakeseveralcontributionstotheliterature.Peerpressurecanhave

strongeffectsonbehavior(Asch,1951;Manski,2000),andourresultssuggestthatdonors

understandthoseeffectswhenaskingfriendstogive.Havinga“gift”availabletoshowgood

2Donationratesconditionalonbeingaskedaretypicallyaround2%formailcampaigns(HuckandRasul,2011;Karlanetal,2011;KarlanandList,2007;EckelandGrossman,2008)

Page 5: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

4

willisimportant–asfriendsmightex-anteprefernottobeaskedtomakeadonation.3This

resultaddstotheliteratureonconcernsforsocialimage(KarlanandMcConnell,2014;

Dellavignaetal,2012;BekkersandWiepking,2011;AndreoniandBernheim,2009;

Charness,RigottiandRustichini,2007).4

Thereisalargeliteraturedocumentingthatmatchingandrebateincentivescan

affectgiving(forexamples,seeHuckandRasul,2011;Karlanetal,2011;KarlanandList,

2007;EckelandGrossman,2008).Ourresultsshowthatusingfundraisingdollarsinstead

tooffersmallsubsidiestodonorstoaskfriendstogiveiscosteffective,especiallywhen

donorscanaskafriendinfrontofhisfriends.Thisnewstrategycouldbeofinterestfor

charities.

Ourfindingscomplementpreviousstudiesshowingthataskingforadonation,in

general,hasastrongeffectongiving(AndreoniandRao,2012;Carmen,2003;Meer,2011;

Castilloetal,2014;SandersandSmith,2016;Smithetal,2015).5Distincttothispaperis

that,byexaminingthesolicitationmechanismitself,weshowthatnotallmethodsofasking

areeffectiveortakenup.Ourfindingshighlighttheimportanceofgifts,audienceeffects

andprestigemotiveswhensolicitingdonationsfromfriends.Thatis,socialaspectsof

givingexplainasmuchormoreofthevarianceasmonetaryincentives.Offeringmonetary

incentivestodonorsexplains1.2%ofthevarianceinthewillingnesstoaskafriend,but

audienceeffectsareasimportantastheyexplainanother1.4percentagepoints.Prestige

motivesandgiftsexplainamuchsmallerportion(anadditional0.18percentagepoints).

Ourexperimentsrevealarichtapestryofsocialinteractionswithinwhichdecisionsto

donatetakeplace.

Thepaperproceedsasfollows.Thenextsectiondescribesourresearch

environments.Section3elaboratestheexperimentaldesign.Section4discusseswhat

3Previousresearchhasshownthatnotalldonationsarenecessarilywelfareenhancing(DellaVignaetal,2012;Andreonietal,2017).4Socialinformation,networksandsolicitationeffectivenessarealsoimportant(Goeletal,2012;FowlerandChristakis,2011;Randetal,2011).Allowingindividualstotargetreciprocitycanbeeffectiveinraisingcontributionstoapublicgood(JacobsonandPetrie,2014).5Askingafriendtogivetoaparticularcharitycansignalleadershipandcharityqualityorneed,whichhavebeenshowntoaffectgiving(Scharf,2014;RondeauandList,2008;Vesterlund,2003).Inasocialmediacontext,broadcastingadonationpledgehasapositiveeffectonpledgesbytheuser’scontacts(Laceteraetal,2016).

Page 6: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

5

donorsmightbeexpectedtodointhevarioustreatments.Section5presentsresults,and

Section6concludes.

2.ResearchEnvironments:FacebookandtheOnlineCrowdfundingCommunity

Theresearchiscarriedoutintwoenvironmentsthatallowustoexploredonormotives

andtheeffectivenessofincentivizingfriendstoaskfriendstodonate.

2.1Facebook

Facebookisapopularsocialnetworkingsite,withover1.96billionmonthlyusers

worldwide,asofMarch2017.Almosteightinten(79%)onlineAmericanshaveaprofileon

Facebook(PewResearchCenter,2016).Theplatformprovidesaconvenientandlow-cost

wayforfriendstocommunicatewithfriendswithintheirsocialnetwork(e.g.bysharing

information,photosandvideos).6

TherearethreecommunicationmechanismsonFacebookthatlendthemselveswell

toourresearchdesign.First,eachFacebookuserhasa“wall.”Ontheuser’swall,heorhis

friendscanpostmessages,photos,videosorweblinksthatallhisfriendscansee.Second,

eachusercansendaprivatemessagetoafriend.Theprivatemessageisadirectmessage

toafriendthatonlythefriendcansee.Thisissimilartoaprivateemail.Third,eachuser

canaccesseachofhisfriend’swalls(ortimelines).Ausercanpostmessages,photos,videos

andweblinksonhisfriend’swallthatallofhisfriend’sfriendscansee.7Thesethree

mechanismsprovideusdifferentsocialcontextsinwhichadonormightaskafriendto

give.

6Adonor’sFacebooknetworkwillmostlikelynotbeanexhaustivelistofallfriendsinhislife,sincenoteveryonehasaFacebookaccountnorareallfriendconnectedonFacebook.Nonetheless,onlinesocialnetworksarealow-costmechanismtoexploremotivesandeffectivenessoffriendsaskingfriendsforcharity.7ThedefaultsettingsonFacebookarethatauser’spostsonhiswallareviewedbyallhisfriendsandhisfriendscanpostonhiswall,inwhichcaseallofauser’sfriendswouldseehisfriend’spost.Likewise,ausercanpostonafriend’swall.Auserwouldneedtotakeactiontochangethesesettings.BecauseofprivacyrulesonFacebook,wedonotknowifthesesettingswerealteredbyindividualsinourfieldexperiment.Changestosettingswouldnotimpactthemechanicsofourtreatments(e.g.afriendwhoblockedauserwouldstillappearontheuser’sfriendslist).Becausedonorswererandomlyassignedtotreatment,thepropensitytochangeFacebooksettingsshouldbeuncorrelatedwithestimatedeffects.

Page 7: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

6

2.2TheOnlineCrowdfundingCommunity

Weworkwithaglobalonlinecrowdfundingcommunity.Here,wesystematicallyvarythe

mannerinwhichdonorscanasktheirfriendsviaFacebookandtheavailabilityand

attributionofanadd-ondonation.

Theonlinecrowdfundingcommunityconnectsdonorstocharitiesbyhostingaweb

sitewithinformationonprojectsseekingfunding.Theorganizationhasanexcellent

reputationinthegivingcommunityforhavinglowadministrativecostsandconducting

properprojectscreeningtomakesurecharitiesandprojectsarelegitimate.Adonorcan

easilysearchtheorganization’swebsiteforprojects(e.g.hurricanerelief,girl

empowerment,foodassistance,etc.)andmakeadonationthroughasecurecheckout

procedure.

Aftercheckout,thedonorseesapagewithclickablebuttonstoshareinformation,

viaeitheranemailmessageorbypostingonFacebook,aboutthecharitythatreceivedthe

donation.Becausedonorsarenormallyofferedtheoptionofsharinginformationaboutthe

charitywiththeirfriendsonFacebook,thisisanaturalenvironmentinwhichtoexplore

ourresearchquestions.

3.TheFieldExperiment

3.1Design

Attheonlinecrowdfundingcommunity’swebsite,aftercheckout,adonorispresented

withtheopportunitytosharethefactthathedonatedtoaparticularcharitywithfriends

andaskthemtodonateaswell.Sometimesthedonorisofferedamonetaryincentiveto

ask,intheformofanadd-ondonationtothecharityhejustgaveto.Also,ifanadd-on

donationisoffered,itismadeeitherinthedonor’snameorinthefriend’sname.8

Aftercheckout,adonorisnormallypresentedwithawebpageconfirmingthe

donation(seeFigure1).Onthepage,heisalsogiventheopportunitytotellfriendsabout

8TheexperimentalprotocolissimilartoCastilloetal(2014),however,itincludesoneadditionalsolicitationmethod(postingonafriend’swall)andthreedifferentattributionsfortheadd-ondonation.

Page 8: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

7

thecharity,eitherbyclickingabuttontosendapre-filledemailoronetosendapre-filled

Facebookpostforhisownwall.Bothoptionsincludehyperlinkstothelandingpageforthe

charity.Thedonorcanaddadditionaltextbutcannoteditthepre-filledFacebookpost.9

Adonorisrandomlyassignedtoeitherseethenormalpost-checkoutpageoroneof

17treatmentpages.Thetreatmentsvaryonthreeimportantdimensions(seeTable1).The

firstisthemethodbywhichthedonorcanaskfriendstodonate.Thedonorcaneither

makeageneralasktomanyfriendsatonce(bypostingonhisownwall),aprivatedirect

asktoonefriend(bysendingaprivatemessage)oradirectasktoonefriendinfrontofan

audience(bypostingonhisfriend’swall).Adonorisrandomlypresentedoneofthethree

askmethodsandcanvoluntarilychoosetoaskornot.ThepanelsinFigure2showthe

screensadonorwouldseeforeachofthethreemethodsofaskingfriendstodonate.

Theseconddimensionofthetreatmentiswhetherornotanadd-ondonationtothe

charityisofferedtothedonorifhecompletestheposttoFacebook.Theamountoftheadd-

ondonation,ifoffered,is$1or$5.Thedonorisrandomlypresentedoneofthethree(none,

$1,or$5).

Thefinaldimensionofthetreatmentsistowhomtheadd-ondonationisattributed.

Thedonationiseitherinthedonor’sname,inthefriend’snamebutisnotannouncedtothe

friend,orinthefriend’snameandthefriendisinformedofthedonationbutnotthe

amount.ThethreepanelsofFigure3showexamplescreens.Thedonorisinformedthatif

hepostseither“we’llgiveanextra$1/$5to[charityname]onyourbehalf,”“we’llgivean

extra$1/$5privatelyto[charityname]onyourfriend’sbehalf,”or“we’llgiveanextra

$1/$5to[charityname]onyourfriend’sbehalf,whichwe’llletyourfriendknowcame

fromyou.”Ifanadd-ondonationisoffered,thedonorisrandomlypresentedoneofthese

threemessages.

Intreatmentsinwhichadonorisaskedtopostaprivatemessageoronafriend’s

wall,apop-upwindowappearsaskinghimtochooseafriend.10Oncehechooses,another

pop-upwindowappearsshowingthedonorthepre-filledinmessagethatwillbepostedon

9BecauseofprivacyrulesonFacebook,wedonotknowifadonoraddedtextorwhatthecontentofthetextmighthavebeen.10Thepop-upwindowpopulatesadrop-downmenuwithalistofthedonor’sfriendsfromhisFacebookaccount.Thelistoffriendsisdisplayedinrandomorder.BecauseofprivacyrulesonFacebook,wecannotanddidnotrecordthenamesofthedonor’sfriends.

Page 9: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

8

Facebook.Ifthedonorisinatreatmentwherethemessageispostedonhisownwall,the

firstpop-upwindowtochooseafrienddoesnotappear.11Ifthedonorisnotalreadylogged

intoFacebookordoesnothaveaFacebookaccount,heispromptedtologinorsignupfor

one.12

ThereareonlytwomessagesthatultimatelygetpostedonFacebook(seepanelsin

Figure4).Thetexteitherreads“Imadeadonationto[charityname]on[online

crowdfundingcommunity].Joinmeindonatingtoo.”Orinthecasetheattributionofthe

add-ondonationisinthefriend’snameandknowntothefriend,“Imadeadonationto

[charityname]on[onlinecrowdfundingcommunity],alongwithanadditionaldonationin

yourname.Joinmeindonatingtoo.”IfafriendclicksonthemessageinFacebook,hewill

betakentothelandingpageforthecharity.Noneofthepre-filledmessagetextcanbe

editedbythedonor.

Onceadonorclicksthefinalpop-upwindowwithapreviewoftheFacebook

message,themessageisactuallypostedonFacebook.Thedonoristhankedforsharing,

andifanadd-ondonationismade,aconfirmationmessageisdisplayed.Ifadonorclicksto

takeupthepostbutdoesnotcompleteit,noadd-ondonationismade.

Thedonorhasseveralpiecesofinformationavailablepriortomakinghisdecisions.

Beforeadonorclickstotakeuppostingthemessage,heisfullyinformedoftheask

method,iftherewillbeanadd-ondonationandinwhosename.Beforethedonorclicksto

completethepost,heisfullyinformedofthecontentofthemessagethatwillbepostedon

Facebook.Thus,weobservetwodecisionsmadebydonors:take-upoftheoffertopostto

Facebookandcompletionofthepost.

Inaddition,wetrackdonationsspurredbytheFacebookmessageposts.Anyone

whoclicksontheFacebookmessagewillarriveatthecharity’slandingpage.Iftheperson

thenmakesadonation,wecanlinkthedonationtothedonorwhopostedthemessageand

11Thepop-upwindowsarenotaffectedbyInternetpop-upblockerssetbyadonor.Thewindowsareinternaltotheonlinecrowdfundingcommunity’swebsite.So,ifadonorclicksthebuttontopostthemessage,heseesthepop-upwindow.12RoughlyhalfofalldonorswerealreadyloggedintoFacebook,andthisproportionisconstantacrossthe17treatments.Thereisarolloffbetweentake-uptopostamessageandcompletionofthepost(roughlyone-third).ThismightreflectthenuisanceofhavingtologintoFacebookornothavinganaccount,howeverthisrolloffisconstantacrosstreatments.Sincewedonotknowifnon-loggedindonorshaveaFacebookaccountornot,ourmainresultsmeasureintent-to-treateffects.OurfindingsholdifthesampleisrestrictedtothoseloggedintoFacebookattreatment(TableA1intheAppendix).

Page 10: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

9

thetreatmentconditionthedonorwasplacedin.Ourdonationsspurredmeasureisalower

boundonthenumberofdonationsgeneratedasafunctionofadonoraskingfriends.Ifa

friendseesthemessagepostandcomestotheonlinecrowdfundingcommunitythrough

anyothermeansthanclickingthemessage(e.g.bytypinginthewebsiteaddressdirectly)

andmakesadonation,wecannotlinkthatdonationtothemessagepost.

Itisimportanttonotethatbecausethisisafieldexperiment,atnotimewere

donorstoldtheywereinanexperiment.Thetreatmentpagesweredesignedtobe

consistentwiththeonlinecrowdfundingcommunity’simageandnormalpost-checkout

procedures.Choicesreflectadonor’sreactiontodifferentinformationinanaturally-

occurringenvironment.BecauseofprivacyrulesonFacebook,wehavenodataonthesex

ofthedonor,towhichfriendamessagewassent,thenamesornumberoffriends,the

donor’slocation,thetextofthemessagessent,etc.Theonlydatawehaveondonorsare

howmuchtheygaveatcheckout,andtheonlinecrowdfundingcommunityprovidedthese.

3.2Implementation

Thefieldexperimentwasconductedduringtwoperiodsin2013,January-Februaryand

June-October,withourpartneronlinecrowdfundingcommunity.13

Afteradonorpaidforhisdonation,hewasscreenedtoseeifhequalifiedtobeinthe

experiment.Donorswhohadpreviouslyparticipatedwereexcludedfromseeingthe

treatmentagain.14Thesedonorssawthenormalcheckout.Additionally,anydonorwho

madeadonationwithagiftcard,mademultipledonations,ormadeadonationtoacharity

participatinginaspecialpromotioncampaignwerealsoexcluded.Thesenon-qualified

donorssawthenormalcheckoutscreen.15

13Thedatawerecollectedintwophasestoaccommodatetheonlinecrowdfundingcommunity’sinternalconstraintsandprogrammingtime.14Theonlinecrowdfundingcommunityassignsauniqueuseridtoeachdonor.Onceadonorseesatreatment,heistaggedashavingparticipated.Ifhereturnstomakeanotherdonationduringthedatacollectionperiod,heisthenexcluded.15Thesedonorswereexcludedbasedontheadviceoftheonlinecrowdfundingcommunity.Forpracticalreasons,weexcludeddonorswhogavetomultipleprojectsbecausethenthedonorwouldneedtochoosewhichproject/charitytosharewithfriends.Ofthenon-qualifieddonors,70.1%gavetoaprojectinaspecialcampaign,21.6%weredisqualifiedformultiplereasons(typicallypreviousparticipantandspecial

Page 11: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

10

Table1showsthat,ofthequalifieddonors,450-500wereassignedtoeach

treatmentorthenormalpost-checkoutpage,givingatotalof9,814uniquedonor

observations.Giventheratesofcompletedpostsacrossthetreatmentconditions,atotalof

$2,311ofadd-ondonationswerepaidouttocharities.

4.Whatwemightexpectdonorsandfriendstodo

Wediscussthedifferentmotivationsdonorsmighthavetoaskfriendsfordonationsand

forfriendstorespondtosuchsolicitations.Weillustratehowreputationconsiderationson

thesideofthedonorandthefriendcanexplainthepatternsandeffectivenessofthe

differentmethodsofasking.Thekeyelementsofthetreatmentsandhypothesesare

summarizedinTable2.16

First,donorswhoaremotivatedtomaximizethecharity’searningsmightwantto

reachthelargestaudiencepossibletoincreasethenumberoffriendswholearnofthe

opportunitytogive.Thisproducesastrongpredictiononthetake-upratesbytreatments.

Hypothesis1:Ifdonorsareonlyconcernedwithmaximizingaudiencesize,take-up

rateswillbehigheronownwallandfriend’swallrelativetosendingaprivate

message.

Wallpostingsreachalargeraudienceandwouldpresumablygeneratemore

donations.17Becausewedonotknowwhetherthedonorhasmorefriendsthanhisfriend,

wecannotrankpostingonthedonor’sownwallrelativetopostingonafriend’swall.

Second,ifdonatingisprestigious,anindividualmightpreferadd-ondonationsmade

inhernameandtorevealadd-ondonationsmadeinafriend’sname.18Thiscreates

campaign),4.0%haddonatedtomultipleprojects,2.3%werepreviousparticipantsonly,2.0%usedagiftcard.16AfulldescriptionofthehypothesesisincludedinTableA2intheAppendix.17Assumingthateachfriendisequallylikelytodonate,havingmorefriendswillgeneratemoremoneyforthecharity.18Harbaugh(1998)discussesprestigemotivestogiving.BenabouandTirole’s(2006)modelofgivingbehavioralsohighlightshowpublicityofdonationsmightencouragemoreprosocialbehavior.Thecharityitselfcouldbeanaudience.Adonormightpreferdonationsmadeinhernamebecausetheyarerevealedtothecharityaswellasothers.

Page 12: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

11

additionalpatternsofbehavior.Inparticular,Hypothesis2addsthat,givenanaskmethod,

take-uprateswillbehigherwhentheadd-ondonationisinthedonor’snameorinthe

friend’snameandmadeknowntothefriendcomparedtowhentheadd-ondonationisin

thefriend’snamebutkeptsecretfromthefriend.Prestigemotiveshaveanambiguous

predictiononthenumberandsizeofgenerateddonationssinceprestigeseekersmaynot

necessarilycareifthefrienddonatesornot.Prestigecouldbeconferredsimplyfrom

receivingtheadd-ondonationbyasking.

Hypothesis2:Ifdonorscareaboutprestige,take-uprateswouldbehigherwhenadd-

ondonationsareinthedonor’snameorinthefriend’snameandmadeknowntothe

friend.

Third,askingafriendtomakeadonationmaynotbeinnocuous.Bydoingso,donors

mightimposenon-pecuniarycostsonthefriendbymakinghimfeelobligatedtodonateor

raisingimageconcernsthatifhedoesnotdonatehisreputationcouldbedamaged.Ifthisis

thecase,thenthefriendmightprefernottobeaskedinthefirstplace.19Notethatthese

costswouldnotexistforaltruists.Donationsarevoluntaryandaskingprovidesthefriend

anopportunitytoengageinaltruism,soaskingforavoluntarydonationcanonlybeweakly

welfareenhancing.

Ifnon-pecuniarycostsexist,thiscanexplainpatternsofbehaviorthatareatodds

withaudiencesizeanddonorprestige.Adonormightprefertoaskafriendforadonation

inprivateratherthaninpublicifbyaskinginpublicthedonorrisksshamingthefriend.

However,ifthefriendhimselfcaresforprestige,hemaybemoreresponsivetosolicitations

madeinpublicthaninprivate.Thispossibletensionfromtheunderlyingmotivesofdonors

andfriendsimpliesthatdonorswillhavetobalancethebenefitsfromsolicitingafriendin

publicwiththepotentialcostsimposedonthefriend.Ifadonorestimatesthatthecoststo

askingafriendinpublictobeonnettoohigh,take-upratesshouldbelowerinpublicthan

inprivate.

19DellaVignaetal(2012)empiricallyestimatethedisutilitycoststoadonorofbeingaskedbyadoor-to-doorfundraiserandfindasignificantreductioninutility.

Page 13: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

12

Becauseofthepossibilityofnon-pecuniarycoststothefriendoverwhelmingthe

benefitsofaskingforadonation,havingawaytoreducethosecostswouldbepreferred

(i.e.byprovidingthedonora“gift”forthefriendthatispairedwiththeask).Hypothesis3

predictsthatwhenaskingafrienddirectly,eitherinprivateorinfrontofhisfriends,take-

upratesarehigherwhentheadd-ondonationismadeinthefriend’snameandmade

knowntothefriend.This,ofcourse,assumesthatmakingadonationinafriend’snameis

consideredtobeapositiveact.Favortradingwouldalsobeconsistentwithhighertake-up

rateswhentheadd-ondonationtothefriendisknown.

Hypothesis3:Iftherearenon-pecuniarycoststothedonorofasking,take-uprates

willbehigherwhentheadd-ondonationisinthefriend’snameandmadeknownto

thefriend(i.e.a“gift”forthefriend).Add-ondonationsinafriend’snamemade

knowntothefriendaremoreeffectiveatgeneratingdonationswhenthecoststothe

friendofignoringtherequestarelargest(i.e.whentheaskismadeinpublic).

Totesttheneteffectofthepresenceofanaudienceontake-uprates,welookatthe

differencebetweensecretandknownadd-ondonationsinafriend’snameacrossthe

privatedirectaskandthepublicdirectasktreatments.Thedifferenceintake-uprates

capturestheneteffectofthefriend’sfriendsknowingofthedonation.Inthiscase,we

wouldexpectthatthedifferenceshouldbelargerwhenthefriendisinformedoftheadd-on

donationinfrontofhisfriendsthanwhenaskedprivately.

5.Results

5.1Take-upandCompletedPostRates

Westartbypresentingsomedescriptivestatisticsonthedonors.Themeandonationat

checkoutforoursampleofdonorsis$64.91andissimilaracrossthe17treatmentsandfor

Page 14: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

13

thoseinthecontroltreatmentwhosawthenormalcheckout.20Mostdonorsaredonating

forthefirsttime(69%).

Themaineffectsforthesubsidy,theaskmethodandtheadd-ondonation

attributionaresummarizedinTable3andshowninFigure5.PanelAinFigure5showsthe

take-upandcompletedpostratesacrossthethreeincentives(none,$1,$5).Thereisabout

aone-thirdrolloffbetweentake-upandcompletedpostrates,butthegeneralpatterns

acrosstreatmentconditionsalwaysremain.21Justbeingaskedtopostamessagetofriends

yieldsa6.1%take-upanda3.5%completionrate.22Offeringa$1incentivetopostmore

thandoublesthetake-uprate(to13.7%)andalmosttriplesthecompletionrate(to8.9%).

Offeringanadditional$4(increasingtheincentivefrom$1to$5)yieldsa31%increase(to

17.9%)fortake-upand30%increase(to11.6%)forcompletion.23So,asmalladd-on

donationof$1hasalargeandsignificanteffectonincreasingsolicitationsrelativetono

subsidy,andaddinganextra$4doessignificantlyraisesolicitationsfurther.However,the

increaseinsolicitationswiththeextra$4ismuchsmallerthantheinitial$1.Also,while

donorsdorespondtothesizeoftheincentive,itisinterestingtonotethatitisaminorityof

donorswhoarewillingtoaskfriends,evenforanadditional$5totheircharity.Almost

90%donotask.

PanelBinFigure5showsthetake-upandcompletionratesforthethreeask

methods.Take-upandcompletionofageneralasktomanyfriends(postingonadonor’s

ownwall)ismostpopular,with19.7%ofdonorstakingupand13.4%completingthepost.

Makingaprivate,directask(sendingaprivatemessage)isthesecondmostpopular,with

15.0%take-upand8.1%completion.Askingafrienddirectlyinfrontofhisfriends(posting

onafriend’swall)istheusedtheleast,with8.1%take-upand6.3%completed.The

differencesacrossaskmethodsaresignificantforbothtake-upandcompletionrates

(differenceinmeansp-values<0.01).

20Themediandonationis$25across11ofthe17treatments,$50in5ofthem,and$40inone.21Take-upisdefinedaswhetherornotthedonorclickedthefirstbuttontopostamessagetoFacebook.Atthatpoint,thedonorknowstheaskmethod,thesubsidyamountandtheadd-ondonationattribution.CompletionisdefinedaswhetherornotthedonorclickedthefinalbuttontocompletetheposttoFacebook.Atthatpoint,thedonorseesthetextofthemessagethatwouldultimatelybepostedonFacebook.22ThecompletedpostratefortheNormalCheckoutis4.22%,andthisincludeseitherpostingtothedonor’sownwallonFacebookorsendinganemailmessage.23Allincreasesintake-upandcompletionratesaresignificantacrossthethreeincentives(none,$1,$5)withp-values<0.01.

Page 15: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

14

Askingafriendinfrontofhisfriendsappearstobethemostsociallycostly,since

donorsaretheleastlikelytovoluntarilytake-upthatmethodofasking,andmakinga

generalaskistheeasiesttodo.ThesedatapatternsareonlypartiallyconsistentwithH1

(maximizingaudiencesize).Takeupratesarelargerforownwallcomparedtoaprivate

message,however,ratesarethelowestforafriend’swall.Absentinformationonthe

effectivenessofanyofthethreeaskmethods,thesedatapatternscouldimplythataskinga

friendinfrontofhisfriendsistheleasteffectivefundraisingmethod(orthemostsocially

costly).Wereturntothispointinthenextsection.

ThefinalpanelinFigure5examinestheeffectsoftheattributionoftheadd-on

donation.Adonorislesslikelytotake-upandcompletethepostwhentheadd-ondonation

isinthefriend’snamecomparedtoinhisname--15.0%wheninthedonor’snameand

11.8%wheninthefriend’sname(9.7%and7.4%completionrates).Thesedifferencesare

significant(differenceinmeanstestp-value<0.01).Also,donorsaremorelikelytotake-up

andcompletethepostwhentheadd-ondonationinthefriend’snameismadeknowntothe

friend.Take-upratesare12.8%whenknownand10.7%whensecret(8.4%and6.5%

completionrates).Thesedifferencesarealsosignificant(p-values<0.01).Donorsprefer

theadd-ondonationtobeintheirname,butwhenitisnot,theypreferthefriendknows.

ThishintsthatdonorscareabouttheirsocialimageandisconsistentwithH2(prestige

motives).Adonorwhoonlycaresaboutsignalinggenerositytoherselfshouldbe

indifferentastowhogetstheattributionoftheadd-ondonation.24

Asimpledecompositionofvarianceillustratestherelativeimportanceofmonetary

incentives,thepresenceofanaudiencewhenaskingafriend,prestigemotivesandgiftson

thewillingnesstoaskafriendtodonate.Monetaryincentivesexplain1.2%ofthevariance,

andaudienceeffects(eitherpostingonone’sownwallorafriend’swall)explainan

addition1.4percentagepoints.Prestigeandgiftscontributeasmallerexplanationandadd

anaddition0.18percentagepoints.25

Turningtotheinteractioneffectsofmonetaryincentivesandtheaskmethod,wesee

similarpatternstothemaineffects.Table4showsregressionsoftake-upandcompletion24Foradiscussionofself-signalingandsocial-signalingingiving,seeGrossman(2015).25WerunOLSregressionsandexaminehowtheR-squaredchangesasexplanatoryvariablesareadded:incentivestopost(e.g.$1,$5add-on),audience(e.g.ownwall,friend’swall),prestige(e.g.ownname,friend’sname),gift(e.g.add-oninfriend’snameisknowntofriend).

Page 16: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

15

ratesbyaskmethods,monetaryincentivesandinteractiontermsfortheadd-ondonation

inthedonor’sname.PatternsseeninthefirsttwopanelsofFigure5emergeinthe

regressionsaswell.Donorsaremorelikelytoaskafriendwhenthesizeoftheadd-on

donationincreases,aremostlikelytoaskmanyfriendsatonceandareleastlikelytoaska

friendinfrontofhisfriends.Inaddition,wecanrejectthattheincreaseincompletedposts

fora$1subsidyandthemarginalincreasewhenincreasingthesubsidyto$5areequal

whenaskingmanyfriendsatonce(p-values=0.0118).However,thiscannotberejectedfor

askingafriendprivatelyorinfrontofhisfriends(p-valuesare0.1265and0.1742).This

suggeststhat,ingeneral,spendinganadditional$4foranadd-ondonationdoesnot

substantiallyincreaseadonor’swillingnesstoaskandmaynotbeworththeextracost.To

testtherobustnessoftheresultsinTable4,wesplitthesampleintodonorswhowere

alreadyloggedintoFacebookatthetimeoftreatmentandthosewhowerenot.Allthemain

treatmenteffectsstillhold,andthecoefficientsarelargerforthosealreadyloggedinto

Facebook.26

Next,weexaminewhetherdonorsaremotivatedtoprovidethemostdonation

moneypossibletothecharity.Whenanadd-ondonationisofferedinthedonor’snameor

inthefriend’snamebutnotannouncedtothefriend,boththemessagepostedonFacebook

(seeFigure4,panel(a))andtheamountofmoneygoingtothecharitythroughtheadd-on

donationareidentical.Absentothermotivations,behaviorshouldbethesameinboth

treatments.PanelAinFigure6showscompletedpostratesforthesetwotreatments.

Whenaskingafrienddirectly(privatemessage),11.1%ofpostsarecompletedwhenthe

add-onisinthedonor’sname,and7.7%arecompletedwhenitisinthefriend’snameand

secrettothefriend.Theseratesaresignificantlydifferent(differenceinmeansp-value<

0.01).Whenaskingafriendinfrontofhisfriends(friend’swall),theratesare8.0%and

5.2%,respectively(differenceinmeansp-value<0.10).Thisdifferenceinwillingnessto

askshowsthatdonorsarenotmerelymotivatedbymaximizingrevenueforthecharity.

Favortradingcouldalsobeimportanttodonors,especiallysincetheyareasking

friendstohelpthemoutbydonatingtotheircharityandthismaybecostlyforthefriend

26SeeTableA1theAppendix.NotethatwecannottellifthosenotalreadyloggedintoFacebookhaveaFacebookaccountornot,sowecannotattributethedifferenceinmagnitudeofcoefficientstonothavinganaccountortothenuisanceofsigningin.

Page 17: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

16

(asoutlinedinH3).Toexaminethis,welookatcompletedpostrateswhentheadd-on

donationisinthefriend’snameandsecrettothefriendcomparedtowhenitismade

knowntothefriend.Inthiscase,themessagepoststoFacebookwilldiffer.Figure4shows

examplesofthetwoposts.Panel(a)showsthepostwhentheadd-onissecret,andpanel

(b)showsthepostwhentheadd-onismadeknown.Favorscanonlybetradedwhenthey

areknown,andinthelattercase,thefriendistoldaboutthis“gift”ofadonationinhis

name.

PanelBinFigure6showscompletedpostratesacrossthesetwotreatments.Both

forsendingaprivatemessageandpostingonafriend’swall,completedpostratesare

higherwhentheadd-ondonationismadeknowntothefriend.Ratesgofrom7.7%to8.4%

(a0.7percentagepointincrease)forprivatemessagesbutthedifferenceisnotsignificant

(differenceinmeansp-value=0.5455).Forafriend’swall,ratesgofrom5.2%to8.3%(a3.1

percentagepointincrease)andthedifferenceissignificant(differenceinmeansp-

value=0.0095).Thisillustratesthatdonorsaremoremotivatedtoaskafriendwhenthey

cancrediblyinformthefriendthattheadd-ondonationinhisnameandprovidessome

evidenceoffavortrading.27

ThelastpartofH3predictsthatifaudienceeffectsareimportantinmotivatinga

donortoaskafriendthenthedifferentialincreaseinpostingwhentheadd-ondonationin

thefriend’snameandknownwillbelargerwithanaudience(friend’swall)thanwhenitis

private(privatemessage).The3.1percentagepointincreaseforafriend’swallislarger

thanthe0.7percentagepointincreaseforaprivatemessageaspredictedbyaudience

effects,butthisisnotsignificant(theone-sidedp-value=0.097).28

Insum,wefindthatdonorsarewillingtoaskfriendstodonate,however,notalldo,

evenwhentheircharitywouldbenefitwithanadditionaldonation.Providingthedonora

“gift”toofferthefriendwhenasking,especiallyinfrontofhisfriends,isimportanttoa

donor’swillingnesstomakethesolicitation.Ourresultsaremostconsistentwiththe

existenceofnon-pecuniarycoststoaskingfriends,prestigemotivesandsomeevidenceof27Anon-parametrictestofwhetherthedistributionofcompletedpostsisdifferentbyaskmethod(privatemessage,postingonafriend’swall)andtheadd-ondonationisknowntothefriend(unknowntofriend,knowntofriend)yieldsaFischerexacttestp-valueof0.080.28Thisistestedwitharegressionofcompletedpostratesintreatmentswheretheadd-ondonationisinthefriend’snameonadummyvariableforfriend’swall,adummyvariableforapublicadd-ondonationandaninteractiontermofthetwo.Thep-valueontheinteractiontermisusedtotestforsignificance.

Page 18: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

17

favortrading.Donorsarenotmotivatedtomaximizeadditionalmoneyforthecharityifit

meanstheydonotgetthecreditforadd-ondonation,noraretheyseekingtomaximizethe

numberoffriendsasked.

5.2Efficiency

Wenowexaminetheefficiencyofthevariousaskmethods,incentivesandattributionsin

termsofgenerateddonations.

IfafriendclickedontheFacebookmessagepost(viathehyperlink)andmadea

donation,wecantrackthisandtieitbacktothedonorsolicitationandexperimental

treatment.However,ifafriendwasinspiredbythesolicitationbutwenttotheonline

crowdfundingcommunitydirectly(notthroughtheFacebookpost)andmadeadonation,

wecannottiethistoourtreatments.So,ourmeasureofeffectivenessisalowerbound

sincedonationsmayhavebeenspurredbyourtreatmentsbutwecannotmeasurethem.

Ofallthecompletedposts,1.70%resultedinnewdonations.Thisissignificantly

differentthanzero(p-value<0.01),soaskingdonorstoasktheirfriendsincreases

donationsontheextensivemargin.Thispercentreturnfromaskingpeopletogiveisinline

withresearchusingmailcampaigns(HuckandRasul,2011;EckelandGrossman,2008;

KarlanandList,2007).Therewere14donationstrackedthroughFacebookposts,with7

frompoststothedonor’sownwall,7frompoststoafriend’swall,and0fromprivate

messages.Thus,thepercentofasksresultinginadonationfromaposttothedonor’sown

wallis2.28%and3.50%fromaposttoafriend’swall.Thesearenotsignificantlydifferent

fromeachother(p-value=0.4127),however,poststoownwallorafriend’swallyield

significantlymoredonationsthansendingaprivatemessage.29Theaveragedonationofthe

friendwasslightlylower($46.43)thanthedonorwhomadethesolicitation($64.91).The

totalamountofmoneyraisedthroughtrackeddonationsis$650.

29Similarresultsholdifweinsteadlookatefficiencyusingintenttotreat.Ofalltreateddonors,0.14%ofthosetreatedgeneratednewdonations(0.31%inownwall,0.22%infriend’swall).Thesepercentagesaresignificantlydifferentfromzero(p-value<0.01).Thepercentofdonationsspurredfromthoseintheownwalltreatmentarenotsignificantlydifferentfromthepercentinthefriend’swalltreatment(p-value=0.5415),buttheyaresignificantlyhigherthanthoseintheprivatemessagetreatment.

Page 19: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

18

Howeffectivewasprovidingdonorsadd-ondonationstoasktheirfriendstogive?If

welookatofferingdonorsa$1add-ondonation(sincethe$5add-ondonationyieldeda

smallextrareturnforadditional$4),acharitywouldspend$100inadd-ondonationsand

get100posts.Lookingfirstatownwallposts,ifweassumethat2.28%ofthoseposts

yieldedadonationandthedonationwasonaverage$46.43,thenthecharitywouldget

$105.86innewdonations(100postsx2.28%x$46.43).Lookingatpostsonafriend’swall,

thecharitywouldget$162.51innewdonations(100postsx3.50%x$46.43).Thissays

thatthereturnfromspending$1foranadd-ondonationwouldyielda$1.06returnif

donorswereaskedtoposttotheirownwalland$1.63toposttoafriend’swall.

Ofcourse,acharitycouldchoosetonotofferadd-ondonationsandmerelyask

donorstopostontheirownwalloronafriend’swall.Thiswouldcostthecharitynothing.

Welookfirstatanestimateofthedonationsgeneratedbyaskingdonorstopostontheir

ownwallwithoutanadd-ondonation.Foracharitytospend$100in$1add-ondonations

toget100posts,theywouldhavetoask631donors.Thisisbecause15.85%ofdonors

completedthepostinthe$1add-ondonationtreatments(100posts/0.1585=631

donors).Ifthecharityinsteadaskedthese631donorstopostontheirownwallwithout

anyincentive,thiswouldyield40posts(631donorsx6.3%whopostwhennotgivenan

add-ondonation).Givenourassumptionsoutlinedabove,thiswouldyield$42.34innew

donations(40postsx2.28%x$46.43).Ifinsteadthecharityaskeddonorstopostontheir

friend’swallwithoutanadd-ondonation,thecharitywouldraise$24.38innew

donations.30

Insum,usingourexperimentalresultsofthepercentofpoststhatyieldedanew

donationandassumingthedonationwouldbetheaveragenewdonationinourdata,

offeringincentivestodonorstoaskfriendsbypostingontheirownwallisnotcost

effective(thereturnis$1.06onthedollar).Thecharitywoulddobettertosimplyask

donorstoasktheirfriendsandnotofferamonetaryincentive.However,askingdonorsto

askafriendbypostingonthefriend’swalldoesyieldapositivereturn(thereturnis$1.63

onthedollar).30Thecharitywouldneedtoask1,742donorstopostonafriend’swallwitha$1add-onincentivetospend$100(1,742x5.74%=100posts).Ifinsteadthecharitymerelyaskedthosedonorstopostwithoutamonetaryincentive,itwouldget15posts(1,742x0.87%).Givenourassumptions,thiswouldyield$24.38innewdonations(15postsx3.50%x$46.43).

Page 20: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

19

6.Conclusion

Usingafieldexperimentwithaglobalonlinecrowdfundingcommunityandover9,800

uniquedonorobservations,weexamineadonor’swillingnessandmotivationtoaskfriends

intheirsocialnetworkonFacebooktodonatetocharity.Ourresearchdesignhasthe

advantageofobservingdonorbehaviorinanaturalsetting,sowecaninterpretour

findingsasexternallyvalid.Byrandomlyassigningdonorstotreatments,wecanidentify

theeffectsoftheprivateandpublicnatureofaskmethods,theimpactofanaudienceand

theimportanceofgiftswhenaskingfriendstodonate.Ourresultssystematicallydocument

thecostsandbenefitsofonlinepeer-to-peersolicitationviasocialmediaforbothdonors

andfriends.Wefindthatvariationinsolicitationisexplainedasmuchbyaudienceeffects

asitisbymonetaryincentives.

Wehaveseveralkeyfindings.First,askingmanyfriendsatonceisthemostpopular

solicitationmethod,butaskingonefriendinfrontofhisfriendsisusedtheleastand

appearstobethemostsociallycostly.Bothmethodsgeneratedonations,butaskingone

friendinfrontofhisfriendsincreasesthedonationsresultingfromasolicitationby50%,

despiteitssparinguse.Second,friendsaskingfriendsviaonlinesocialmediamaybeacost-

effectiveandviablemethodforcharitiestoconsiderwhenseekingtoraisedonationsand

findnewdonors.Ourestimatesshowthatofferingadonor$1inanadd-ondonationtohis

charitytoaskonefriendinfrontofhisfriendstodonateyields$1.63innewdonations.

Third,themannerinwhichdonorincentivesareimplementedisimportant.When

theattributionoftheadd-ondonationisinthedonor’sname,ratherthanthefriend’s,

donorsaretwiceaslikelytoaskfriendstodonate.Thisisimportantforcharitiesbecause

thefundraisingcosttothecharityfortheadd-ondonationisthesame,butsolicitationsare

higher.Also,providingthedonoragifttoofferafriendcandoublesolicitationrateswhen

askingafriendinfrontofhisfriends.

Finally,donorsdoweighthesocialcostsandbenefitsofaskingfriends.The

willingnesstoaskfriendsseemstobedrivenbythedifferentcostsimposedonfriends

whenaskingprivatelyandinpublicandbythesocialpressureofaskingfriendsinfrontof

others.

Page 21: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

20

Givingmoneytocharitycanbeasocialactivity,andindividualsmayenjoydonating

withandamongfriends.Friendscanprovideimportantinformationontheexistenceand

qualityofcharitiesthatmightbeofcommoninterest.Indeed,friendsmayhaveastrong

influenceondonationbehavior,andsurveysonwhyindividualsgivesuggestthatbeing

askedandaskedbyafriendareimportant(VanSlykeandBrooks,2005).Ourresearch

confirmsthisandprovidesinsightsintomotivationsforwhyfriendsmightbewillingtoask

friendsandwhyfriendsmightthengive.Wefindthatitisnotmerelyaskingfriendsthat

generatesnewdonations.Therearewaysofaskingthataremoreeffectivethanothers.The

existenceofsocialpressure,audienceandgiftsareimportanttospurfriendstoaskfriends

andtoinducefriendstogive.

Page 22: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

21

ReferencesAndreoni,J.(1989)"GivingwithImpureAltruism:ApplicationstoCharityandRicardian

Equivalence,"JournalofPoliticalEconomy,1447-1458.Andreoni,JamesandDouglasBernheim(2009).“Socialimageandthe50–50norm:A

theoreticalandexperimentalanalysisofaudienceeffects.”Econometrica,77(5),1607-1636.

Andreoni,JamesandJustinRao(2011).“Thepowerofasking:Howcommunicationaffectsselfishness,empathy,andaltruism.”JournalofPublicEconomics,95(7-8),513-520.

Andreoni,James,JustinRaoandHannahTrachtman(2017).“AvoidingtheAsk:AFieldExperimentonAltruism,EmpathyandCharitableGiving.”JournalofPoliticalEconomy,125(3),625-653.

Asch,Solomon(1951).“EffectsofGroupPressureUpontheModificationandDistortionofJudgment.”InH.Guetzkow(ed.)Groups,leadershipandmen.Pittsburgh,PA:CarnegiePress

Bekkers,R.,&Wiepking,P.(2011).ALiteratureReviewofEmpiricalStudiesofPhilanthropy:EightMechanismsThatDriveCharitableGiving.NonprofitandVoluntarySectorQuarterly,40(5),924-973.

Bénabou,Roland,andJeanTirole.2006."IncentivesandProsocialBehavior."AmericanEconomicReview,96(5):1652-1678.

Carman,K.(2003).“Socialinfluencesandtheprivateprovisionofpublicgoods:Evidencefromcharitablecontributionsintheworkplace.”Manuscript,StanfordUniversity.

Castillo,Marco,RaganPetrieandClarenceWardell(2014),“FundraisingThroughOnlineSocialNetworks:AFieldExperimentonPeer-to-PeerSolicitation”JournalofPublicEconomics,114,29-35.

Charness,Gary,LucaRigottiandAldoRustichini,(2007),“IndividualBehaviorandGroupMembership,”AmericanEconomicReview,97(4),1340-1352

Dellavigna,Stephano,List,JohnandUlrikeMalmendier(2012).“TestingforAltruismandSocialPressureinCharitableGiving.”TheQuarterlyJournalofEconomics,127(1),1-56.

Eckel,CatherineandPhillipGrossman(2008).“Subsidizingcharitablecontributions:anaturalfieldexperimentcomparingmatchingandrebatesubsidies.”ExperimentalEconomics,11(3),234-252.

Fowler,J.,&Christakis,N.(2010).“Cooperativebehaviorcascadesinhumansocialnetworks.”ProceedingsoftheNationalAcademyofSciences,107(12),5334-5338.

Goel,S.,Watts,D.,&Goldstein,D.(2012).“Thestructureofonlinediffusionnetworks.”EC'12Proceedingsofthe13thACMConferenceonElectronicCommerc,623-638.

Grossman,Zachary(2015).“Self-SignalingandSocial-SignalinginGiving,”JournalofEconomicBehaviorandOrganization,117,26-39.

Huck,SteffenandImranRasul(2011).“Matchedfundraising:Evidencefromanaturalfieldexperiment.”JournalofPublicEconomics,95(5-6),351-362.

Jacobson,SarahandRaganPetrie(2014),“FavorTradinginPublicGoodProvision,”ExperimentalEconomics,17,439-60.

Karlan,DeanandJohnList(2007).“Doespricematterincharitablegiving?Evidencefromalarge-scalenaturalfieldexperiment.”AmericanEconomicReview,1774-1793.

Page 23: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

22

Karlan,Dean,List,JohnandEldarShafir(2011).“Smallmatchesandcharitablegiving:Evidencefromanaturalfieldexperiment.”JournalofPublicEconomics,95(5),344-350.

Karlan,DeanandMargaretMcConnell(2014).“HeyLookatMe:TheEffectofGivingCirclesonGiving.”JournalofEconomicBehaviorandOrganization,106:402-412.

Kumar,V,J.AndrewPetersenandRobertP.Leone(2007),“HowValuableisWordofMouth?”HarvardBusinessReview,October.

Lacetera,Nicola,MarioMacisandAngeloMele(2016),“ViralAltruism?CharitableGivingandSocialContagioninOnlineNetworks,”SociologicalScience.

Manski,Charles(2000).“EconomicAnalysisofSocialInteractions,”JournalofEconomicPerspectives,14(3),115-136.

Meer,Jonathon(2011).“Brother,canyouspareadime?Peerpressureincharitablesolicitation.”JournalofPublicEconomics,95(7),926-941.

PewResearchCenter(2016),“SocialMediaUpdate,”November.Rand,D.,Arbesman,S.,&Christakis,N.(2011).“Dynamicsocialnetworkspromote

cooperationinexperimentswithhumans.”ProceedingsoftheNationalAcademyofSciences,108(48),19193-19198.

Rondeau,DanielandJohnList(2008),“MatchingandChallengeGiftstoCharity:EvidencefromLaboratoryandNaturalFieldExperiments,”ExperimentalEconomics,11(3),253-267.

Sanders,MichaelandSarahSmith(2016),“CanSimplePromptsIncreaseBequestGiving?FieldEvidencefromaLegalCallCentre,"JournalofEconomicBehaviorandOrganization,125,179-191.

Sanders,Michaeletal(2014),“NetworkNudges:SecondandThirdOrderSocialEffectsinCharitableGivingFieldExperiments”,Presentedat2014ScienceofPhilanthropy(SPI)Conference,UniversityofChicago.

Scharf,Kimberly(2014),“PrivateProvisionofPublicGoodsandInformationDiffusioninSocialGroups,”InternationalEconomicReview,55(4),1019-1042.

Smith,Sarah,Windmeijer,FrankandEdmundWright(2015)“PeerEffectsinCharitableGiving:Evidencefromthe(Running)Field”,EconomicJournal,125(585), 1053–1071

VanSlyke,DavidandArthurBrooks(2005),“WhyDoPeopleGive?NewEvidenceandStrategiesforNonprofitManagers,”AmericanReviewofPublicAdministration,35(3),199-222.

Vesterlund,Lise(2003),“TheInformationalValueofSequentialFundraising,”JournalofPublicEconomics,87(3),627-657.

Page 24: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

23

FiguresandTables

Figure1Normalpost-checkoutscreen–afterdonorhaspaidforhisdonation

Page 25: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

24

(a)

(b)

(c)

Figure2Threeaskmethods–(a)generalasktomanyfriendsatonce(posttoownwall),(b)private

directasktoonefriend(sendaprivatemessage),(c)directasktoonefriendwithanaudience(posttofriend’swall)

Page 26: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

25

(a)

(b)

(c)

Figure3Threeattributionsofa$1add-ondonationifadonorpoststoonafriend’swall(theonlydifferenceforthe$5add-ondonationistheamount)–(a)add-ondonationindonor’sname,(b)add-ondonationinfriend’snamebutnotannouncedtofriend,(c)add-on

donationinfriend’snameandmadeknowntofriend

Page 27: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

26

(a)

(b)

Figure4Facebookposts

Page 28: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

27

Take-uprates

Completedpostrates

(a) byAdd-ondonationamount

(b)byAskmethod

(c)byAttributionofadd-ondonation

Figure5MainEffectsoftake-upandcompletedpostratesbyadd-ondonationamount(panela),ask

method(panelb)andattribution(panelc).Errorbarsarestandarderrorofthemean.

0%

5%

10%

15%

20%

$0 $1 $50%

5%

10%

15%

20%

$0 $1 $5

0%

5%

10%

15%

20%

25%

Ownwall Privatemessage

Friend'swall0%

5%

10%

15%

20%

25%

Ownwall Privatemessage

Friend'swall

0%

5%

10%

15%

20%

Ownname Friend'sname Friend'sname- add-onsecret

Friend'sname- add-onknown

0%

5%

10%

15%

20%

Ownname Friend'sname Friend'sname- add-onsecret

Friend'sname- add-onknown

Page 29: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

28

(a)Maximizingmoneyforthecharity

(b)FavorTrading

Figure6

MaximizingmoneyforcharityandFavorTradingCompletedpostratesbyadd-ondonationattribution.

Errorbarsarestandarderrorofthemean.

0%

5%

10%

15%

PrivateMessage Friend'sWall

Ownname

Friend'sname:add-onsecret

0%

5%

10%

PrivateMessage Friend'sWall

Friend'sname:add-onsecret

Friend'sname:add-onknown

Page 30: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

29

Table1

ExperimentalTreatments-Numberofobservations

AskMethod Add-ondonationamountAdd-onindonor’sname

None $1 $5 Generalask(ownwall) 796 896 601Directask–private(privatemessage) 583 603 610Directask–audience(friend’swall) 461 435 468

Add-oninfriend’sname–secrettofriend

$1 $5 Directask–private(privatemessage) 435 427Directask–audience(friend’swall) 421 437

Add-oninfriend’sname–knowntofriend

$1 $5

Directask–private(privatemessage) 437 427Directask–audience(friend’swall) 496 452

Normalcheckout 829

Totalnumberofobservations 9,814

Page 31: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

30

Table2KeyElementsofExperimentalTreatments

Add-onindonor’sname

Generalask(ownwall)General-Own

Directask–private(privatemessage)Private-Own

Directask–public(friend’swall)Public-Own

Add-oninfriend’sname–secretDirectask–private(privatemessage)

Private-SecretDirectask–public(friend’swall)

Public-Secret Add-oninfriend’sname–known

Directask–private(privatemessage)Private-Known

Directask–public(friend’swall)Public-Known

Hypothesesontake-upratestoaskafriend:

H1(maxaudiencesize): Public>Private,General>Private H2(prestigemotives): Own>Known>Secret H3(non-pecuniarycoststoasking): Public-Known>Public-Secret Private-Known>Private-Secret

Page 32: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

31

Table3PercentofCompletePostsbyExperimentalTreatment

(numberofcompletedposts/numberofobservationsinparentheses)

AskMethod Add-ondonationamount Add-onindonor’sname

None $1 $5 Total Generalask(ownwall) 6.28

(50/796)15.85

(142/896)19.13

(115/601)13.39

(307/2294)Directask–private(privatemessage)

1.72(10/583)

8.13(49/603)

14.10(86/610)

8.07(145/1796)

Directask–audience(friend’swall)

0.87(4/461)

5.74(25/435)

10.04(47/468)

5.57(76/1364)

Total 3.48

(64/1840)11.17

(216/1934)14.77

(248/1679)9.68

(528/5553)

Add-oninfriend’sname–secrettofriend

$1 $5 Total Directask–private(privatemessage)

6.21(27/435)

9.13(39/427)

7.66(66/862)

Directask–audience(friend’swall)

4.04(17/421)

6.41(28/437)

5.24(45/858)

Total 5.14

(44/856)7.75

(67/864)6.45

(111/1720)

Add-oninfriend’sname–knowntofriend $1 $5 Total

Directask–private(privatemessage)

7.32(32/437)

9.60(41/427)

8.45(73/864)

Directask–audience(friend’swall)

7.86(39/496)

8.85(40/452)

8.33(79/948)

Total 7.61

(71/933)9.21

(81/879)8.39

(152/1812)

Totalnumberofobservations

9,814

Page 33: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

32

Table4OrdinaryLeastSquaresRegressionsontake-upandcompletedpostrateswhenadd-on

donationisindonor’snamebyadd-ondonationamountandaskmethod

(1) (2)VARIABLES Take-uprate Competedpostrate $1add-on 0.131*** 0.096*** (0.017) (0.014)$5add-on 0.179*** 0.129*** (0.019) (0.016)Privatemessage -0.055*** -0.046*** (0.019) (0.016)Friend’swall -0.082*** -0.054*** (0.020) (0.017)$1add-on*PM -0.014 -0.032 (0.026) (0.022)$1add-on*FW -0.070** -0.047* (0.029) (0.024)$5add-on*PM 0.011 -0.005 (0.028) (0.023)$5add-on*FW -0.070** -0.037 (0.030) (0.025)Constant 0.099*** 0.063*** (0.012) (0.010) Observations 5,453 5,453R-squared 0.056 0.040

Standarderrorsinparentheses***p<0.01,**p<0.05,*p<0.10

Page 34: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

33

Appendix

TableA1Table4rerunonsubsamplesofdonorswhowerealreadyloggedintoFacebookatthetime

oftreatmentorwerenotloggedin.OrdinaryLeastSquaresRegressionsontake-upandcompletedpostrateswhenadd-on

donationisindonor’snamebyadd-ondonationamountandaskmethod. (1) (2) (3) (4)VARIABLES Take-uprate–

AlreadyloggedintoFacebookat

timeoftreatment

Competedpostrate–Alreadyloggedinto

Facebookattimeoftreatment

Take-uprate-NOTalreadyloggedintoFacebookattimeof

treatment

Competedpostrate-NOTalready

loggedintoFacebookattimeoftreatment

$1add-on 0.211*** 0.170*** 0.060*** 0.037** (0.030) (0.027) (0.019) (0.014)$5add-on 0.245*** 0.202*** 0.111*** 0.077*** (0.033) (0.030) (0.022) (0.016)Privatemessage -0.081** -0.060** -0.042** -0.043*** (0.034) (0.030) (0.021) (0.016)Friend’swall -0.108*** -0.078** -0.066*** -0.043** (0.037) (0.033) (0.023) (0.017)$1add-on*PM -0.040 -0.069 0.019 0.012 (0.047) (0.043) (0.030) (0.022)$1add-on*FW -0.088* -0.068 -0.050 -0.032 (0.051) (0.046) (0.032) (0.024)$5add-on*PM 0.045 0.020 0.011 -0.017 (0.049) (0.044) (0.031) (0.024)$5add-on*FW -0.039 -0.035 -0.089*** -0.051** (0.052) (0.047) (0.033) (0.025)Constant 0.145*** 0.099*** 0.071*** 0.043*** (0.022) (0.020) (0.014) (0.010) Observations 2,357 2,357 2,704 2,704R-squared 0.082 0.065 0.044 0.030

Standarderrorsinparentheses***p<0.01,**p<0.05,*p<0.10

Page 35: Friends Asking Friends for Charity: The Importance of ......Social pressure is costly, and donors are reluctant to use it with their friends. Providing donors with a “gift” for

34

TableA2Hypothesesoftake-upratesacrossthemainelementsofthetreatments.

AskmethodsareGeneral(generalasktomanyfriendsatonce–postonownFBwall),Private(adirectasktoonefriendinprivate–sendaprivatemessageonFB)andPublic(adirectasktoonefriendinfrontofhis

friends–postonafriend’sFBwall).Add-ondonationattributionsareOwn(add-onindonor’sname),Secret(add-ondonationinfriend’snamebutkeptsecretfromthefriend)andKnown(add-ondonationinfriend’snameandmadeknowntothefriend).Table2describesthesekeyelements.A→BtodenotesthatoptionAis

preferredtooptionB,A↔BdenotesthatoptionAandBareequallydesirable,andA?Bdenotesnoprediction.

H1:Donorsarealtruisticandcareaboutmaximizingthesizeoftheaudiencewhenasking

Public–Own ↔ Public–Known ↔ Public–Secret

↓ ↓ ↓

Private–Own ↔ Private–Known ↔ Private–Secret

General–Own

H2:DonorscareaboutprestigePublic–Own → Public–Known → Public–Secret

↓ ↓ ↓

Private–Own → Private–Known → Private–Secret

General–Own

H3:Therearenon-pecuniarycoststoasking.Thefriendmightfeelobligatedtodonateorthinkhisreputationmightbedamagedbynotdonating.Public–Own ? Public–Known → Public–Secret

? ↕ ↕

Private–Own ? Private–Known → Private–Secret

?

General–Own


Recommended