complimentreallyworth?
ThisexperimentstudiesLASAseniorswillingnesstodoafavorbasedonwhetherthey
hadreceivedacomplimentornot.Todothis,20randomlyselectedLASAseniorswere
givenacomplimentandanother20werenotgivenacompliment.Thenbothgroups
wereaskedtofilloutasurvey.Basedontheirresponses,achisquaredtestof
homogeneitywasruntodetermineiftheproportionofresponseswerethesamefor
bothcategories.
InquiryPitch
AndreasToprac
StatisticsPeriod3
20152016
5/22/16
TableOfContents
ProjectReport
WorksCited
AppendixA(ProjectData)
AppendixB(RCode)
11
AppendixC(Signatures)
12
AppendixD(ProofofRandomness)
13
AppendixE(Graphs)
14
AppendixF(WebsiteLink)
15
AppendixG(RevisedInquiryPitch)
16
AppendixH(OriginalInquiryPitch)
19
AppendixI(ProjectRoughDraft)
22
AppendixJ(PeerReview)
26
HowMuchisaComplimentReallyWorth?
Astudyreportedby
Forbes
magazineconcludedthatreceivingacomplimenthas
thesamepositiveimpactasreceivingmoney.Totestthis,researcherstook48adults
andhadthemtypeoutaspecificpatternonakeyboardasfastaspossible.Theirtimes
wererecorded.Theythensplittheadultsintothreegroups.Onegroupreceiveda
compliment,onegroupwatchedanotherpersonreceiveacompliment,andthethird
groupreceivednocomplimentandevaluatedthemselvesonagraph.Theadultswere
thenbroughtinthenextdayandaskedtotypeoutthesamepatternonthekeyboardas
fastaspossiblefrommemory,andtheirtimeswererecordedonceagain.The
researchersfoundthatthegroupthatreceivedacomplimenthadsignificantlyfaster
timesthantheothertwogroups.Afterdoingmoreresearch,theyfoundthatthesame
partofthebrainthatwasstimulatedbythecomplimentwasalsostimulatedbyreceiving
acashreward(DiSalvo).Thesefindingsreinforcetheideaaboutgoodmodern
managementtechniquesthatpraisecostsnothing,butitproducesthesamesignificant
benefitsasmoneywould(WhyComplimentsImprovePerformance).
Becausetheseresearcherswereabletofindsignificantevidencethata
complimentcansignificantlyimproveperformanceonasimpletasksuchastyping,
therecouldalsobearelationshipbetweenreceivingacomplimentandotherthingslike
favors.Thequestionthatwillbetestedisifthereisacorrelationbetweenbeinggivena
complimentandwillingnesstodoafavor.Forthistest,thenullhypothesiswillbethat
theproportionofresponsesarethesamewhetherornotthesubjectwasgivena
compliment.Thealternativehypothesiswillbethattheproportionofresponsesarenot
thesamewhetherornotthesubjectwasgivenacompliment.Thepopulationthatthe
sampleisbeingdrawnfromisallLASAseniors.Basedontheresults,therecouldbe
significantevidencetosuggestthatgivingaLASAsenioracomplimentcaninfluence
themtodoafavorforyou.AssumingthataLASAseniorcanaccuratelyrepresentthe
averageperson,thiscoulddefinitelybeusedbytheaveragepersontotakeadvantage
ofthosearoundthemandgetfreefavorsoutofpeoplethattheycompliment.
Totestmyhypothesis,thepopulationbeingusedisallLASAHighSchool
seniors.ThesamplewasmadewasbyobtainingalistofalltheLASAseniorsthatwas
thennumberedfrom1252.Usinganonlinenumbergenerator,40numberswere
randomlygeneratedthatcorrelatedto40differentstudents.Thefirst20thatwere
generatedreceivedacomplimentandthesecond20thatweregenerateddidnot.This
createsasimplerandomsamplewhichwilleliminateasmuchbiasaspossible.Oneof
themajorbiasesthatcouldbepresentinthissampleisthatmanyoftherandomly
selectedstudentsknowme,sotheycouldbebiasedonhowtheyrespondtomeasking
forafavorbasedonhowwelltheyknowme.
Forthisexperimentthecomplimentwasaboutsomethingthatthestudentwas
wearing(whetheritbetheirshoes,shirt,etc.).Forexample,Heythatshirtlooksgood
onyou.Thefavoraskedwas,Wouldyoubewillingtofilloutthissurveyforme?The
surveywasbeapagefullofrandomquestionsabouttheirsex,age,haircolor,etc.The
questionsonthesurveydonotreallymatter,becausethestudentsresponsetothe
favoristheactualdata.Thestudentswerenotactuallyrequiredtofilloutthesurvey,
becausetheresponsetothefavoristheonlynecessarymeasurementforthe
experiment.Tocollectthedata,eachstudentwasfoundindividuallyandaskedtofill
outthesurvey.Theirresponsewererecordedandasignaturewasacquiredas
verificationofparticipationintheexperiment.
Beforerunningthechisquaredtestofhomogeneity,first,theassumptionsforthe
testneedtobechecked.Therandomassumptionismetbecausethesampleisa
simplerandomsample,whichwasdescribedearlier.Theindependentassumptionthat
eachsubjectmustonlybeobservedoncewasmetbecausetestsubjectscouldnot
providemultipleresponsestotherequestforafavor.Furthermore,studentscouldnot
becomplimentedandnotcomplimentedatthesametime.Thefirstexpectedcount
assumptionthateveryexpectedcountmustbegreaterthanonewasmetbecauseall
theexpectedvaluesweregreaterthanone,withthesmallestexpectedcountbeing3.5.
Butthesecondexpectedcountsassumptionthatnomorethan20%oftheexpected
countscellcanbelessthan5wasnotmet,becausetwooutofthefourexpectedcount
cellswere3.5.Because50%oftheexpectedcountswerelessthanfive,thismeans
thatthedatacouldbesubjecttoquestion.
AchisquaredtestofhomogeneitywasthenruninRstudiotoanalyzethedata.
Runningachisquaredtestofhomogeneityproducedachisquaredvalueof1.5584,a
chisquaredstatisticof3.84,1degreeoffreedom,andapvalueof0.2119.Thepvalue
saysthatthereisa0.2119probabilityofgettingtheseresults,ormoreextreme,given
thatthenullhypothesisistrue.ForthistestIusedanalphavalueof0.05,whichmeans
thatthereisa0.05probabilityofrejectingthenullhypothesiswhenthenullhypothesis
istrue.IusedthisalphavaluebecauseitmakestheprobabilityofaType1errorvery
small.Becausethepvalueof0.2119isgreaterthanthealphavalueof0.05andthe
chisquaredvalueof1.5584islessthanthechisquaredstatisticof3.84,wefailtoreject
thenullhypothesisthattheproportionofresponsesarethesamewhetherornotthe
studentreceivedacompliment.Therefore,wedonothavesignificantevidenceto
supporttheclaimthattheproportionofresponsesarenotthesamewhetherornotthe
studentreceivedacompliment.
TheresultsindicatethatforthepopulationofLASAseniors,receivinga
complimentdidnothaveasignificantinfluenceontheirwillingnesstodoafavor.Ofthe
40students,fivewhodidnotreceiveacomplimentdidnotwanttofilloutthesurveyand
onlytwowhodidreceiveacomplimentdidnotwanttofillitout.Thismeansthatfewer
studentswhoreceivedacomplimentdidnotwanttofilloutthesurvey,butitwasnota
significantdifference.IfitisassumedthataLASAseniorisrepresentativeofthe
averageperson,itwouldmeanthatobtainingfreefavorsfrompeoplebygivingthem
complimentswouldnotactuallywork.
Howeverthereareseveralproblemswiththedatathatcouldleadtoa
questionableconclusion.Firstofall,oneoftheexpectedcountsassumptionswasnot
metwhichcouldmeanthatachisquaredtestofhomogeneitymaynotbeappropriate.
Also,mostofthestudentswhodidnotwanttofilloutthesurveyknowme,possibly
becausetheyweremorecomfortabletellingmethattheydidnotwanttofillitoutthan
thosewhodidknowme.Ontheotherhand,thosewhodidnotknowmethatwellmay
havejustactedpoliteandfeltawkwardrefusingtodoafavorforme.Thisisdefinitelya
majorbiasthatcouldleadtoquestionableresults.Inordertoperformthisexperimentin
alessbiasedway,Iwoulduseasamplefromalargegroupofrandomstrangersina
placesuchasthemall.Randompeoplewalkingaroundthemallcouldbe
complimentedandthenaskedtofilloutthesurvey.Mostlikelythesetestsubjects
wouldnotknowme,sotheywouldnotbebiasedinanswering.Also,alargersample
wouldhelpmeettheexpectedcountsassumption.Howevertheexperimentwouldno
longerbeasimplerandomsample,soconveniencebiaswouldbepresent.Thisis
becauseifthesamplewastakeninasinglelocationlikethemall,itwouldonlyuse
subjectsthatwereconvenienttobepartoftheexperiment.Butthissamplewould
probablybemorerepresentativeoftheaverageperson.
WorksCited
DiSalvo,David."Study:ReceivingaComplimentHasSamePositiveEffectasReceiving
Cash."
Forbes
.ForbesMagazine,9Nov.2012.Web.16May2016.
"WhyComplimentsImprovePerformance."
TheBritishPsychologicalSociety
.BPS.org,
13Nov.2012.Web.16May2016.
A.ProjectData
Compliment
NoCompliment
Total
Yes
18
15
33
No
20
20
40
Total
Name
Compliment
Answer
Amstutz
Yes
Yes
Asdar
No
Yes
Asghar
Yes
Yes
Beckett
Yes
Yes
Bhakta
Yes
Yes
Bixby
No
Yes
Brodkin
No
Yes
Dennis
Yes
Yes
Devany
Yes
Yes
DuBose
No
No
Figueroa
No
Yes
Flores
Yes
Yes
GardunoSalazar
No
Yes
Guzman
Yes
Yes
HumphreysLucas
No
No
Johnson
No
Yes
Keel
Yes
Yes
Kemp
Yes
Yes
Larsen
No
Yes
Li
Yes
No
Liu
Yes
Yes
Martinez
No
Yes
McHaney
No
No
Moreno
Yes
Yes
Newman
Yes
Yes
Odell
No
Yes
Pace
No
Yes
PfundPulliam
No
No
Ploeger
Yes
Yes
Ramirez
Yes
Yes
Randall
Yes
No
Read
Yes
Yes
ReesEissler
Yes
Yes
Sajbel
Yes
Yes
Seiler
No
Yes
Sheblak
No
Yes
Stankiewicz
No
Yes
10
Tanaka
No
Yes
TorresSkillicorn
No
Yes
Zachgo
No
No
11
B.RCode
>tab<table(raw$Type,raw$Answer)
>tab
NoYes
Compliment218
NoCompliment515
>chisq.test(tab,correct=FALSE)$expected
NoYes
Compliment3.516.5
NoCompliment3.516.5
Warningmessage:
Inchisq.test(tab,correct=FALSE):
Chisquaredapproximationmaybeincorrect
>chisq.test(tab,correct=FALSE)
Pearson'sChisquaredtest
data:tab
Xsquared=1.5584,df=1,pvalue=0.2119
Warningmessage:
Inchisq.test(tab,correct=FALSE):
Chisquaredapproximationmaybeincorrect
>barplot(tab,main="Willyoudoafavorforme?",xlab="Response",legend=T)
>expect<chisq.test(tab,correct=FALSE)$expected
Warningmessage:
Inchisq.test(tab,correct=FALSE):
Chisquaredapproximationmaybeincorrect
>barplot(expect,main="Expectedresponsestoafavor",xlab="Expected
Responses",legend=T)
12
C.Signatures
13
D.ProofofRandomness
14
E.Graphs
15
F.WebsiteLink
http://statsfinal.weebly.com/
16
G.RevisedInquiryPitch
Astudyreportedby
Forbes
magazineconcludedthatreceivingacomplimenthas
thesamepositiveimpactasreceivingmoney.Totestthis,researcherstook48adults
andhadthemtypeoutaspecificpatternonakeyboardasfastaspossible.Theirtimes
werethenrecorded.Theythensplittheadultsintothreegroups.Onegroupreceiveda
compliment,onegroupwatchedanotherpersonreceiveacompliment,andthethird
groupreceivednocomplimentandevaluatedthemselvesonagraph.Theadultswere
thenbroughtinthenextdayandaskedtotypeoutthesamepatternonthekeyboardas
fastaspossible,andtheirtimeswererecordedonceagain.Theresearchersfoundthat
thegroupthatreceivedacomplimenthadsignificantlyfastertimesthantheothertwo
groups.Afterdoingmoreresearch,theyfoundthatthesamepartofthebrainthatwas
stimulatedbythecomplimentisalsostimulatedbyreceivingacashreward(DiSalvo).
Becausetheseresearcherswereabletofindsignificantevidencethatacomplimentcan
improveperformanceonasimpletask,Iwonderedifthereisasimilarresultwhen
askedtodoafavor.ThequestionIwillbetestingisifthereisacorrelationbetween
givingacomplimentandwillingnesstodoafavor.
TotestthisquestionIwillbedrawingfromthepopulationofLASAHighSchool
seniors.IamusingonlyseniorsbecauseIknowhowtotrackdownmostpeopleinthe
seniorclassanditwouldntbeweirdformetocasuallygivethemacompliment.Todo
this,Iwillassigneverysenioranumber.Iwillthenrandomlypick40numbersthatwill
correlatetothestudentswhowillbepartofmysample.Theonlybiasthatcouldbe
presentinthissampleisthatmanyofthestudentsthatarerandomlychosenprobably
knowme,sotheycouldbebiasedonhowtheyrespondtomeaskingforafavor.After
17
creatingmysample,Iwillgive20ofthestudentsacomplimentandtheother20no
complimentbeforeaskingforafavor.Forthisexperimentmycomplimentwillsimplybe
acomplimenttosomethingthattheyarewearing(whetheritbetheirshoes,shirt,etc.).
MyfavorwillbeWouldyoubewillingtofilloutthissurveyforme.Thesurveywill
simplybeapagefullofquestionsaboutthemselveslikesex,age,haircolor,etc.To
collectthisdataIwillfindeachstudentindividually,complimentthemornot,andthen
askthemtofilloutthesurvey.Iwillthenrecordtheirresponseandgettheirsignature
toshowthattheywerepartoftheexperiment.
Mydatawillbesplitintotwogroups,thosewhoreceivedacomplimentandthose
whodidnot.Theirresponsetotherequestforafavorwillberecordedaseitheryesor
no.BecauseIamevaluatingtwocategoricalvariablesthatweretakenfromtwo
differentgroups,Iwillbeusingachisquaredtestofhomogeneity.Forthistest,mynull
hypothesiswillbethattheproportionofresponsesarethesamewhetherornotIgave
themacompliment.Myalternativehypothesiswillbethattheproportionofresponses
arenotthesameacrossallcategories.BeforeIcanrunthistestIneedtomakesure
thatmyassumptionsaremet.BasedonmydatacollectionIalreadyknowthatmydata
willberandom.Also,Iknowthatasinglesubjectcantbeinboththeyesandno
categoryforrespondingtoafavor,andalsocantbothreceiveandnotreceivea
compliment.AfterIcollectmydataIwillbeabletotestmylasttwoassumptionsabout
expectedcounts.TofindmyconclusionforthisquestionIwillputmydataintoa
calculatorandrunachisquaredtestofindependencesothatIcanfindthepvalueand
compareittomyalphaof0.05.
18
MyreasonforthisquestionisthatIwanttoseeifasimplecomplimentcan
changethewayapersonrespondstoyou.Byaddingthefavorofaskingthemtofillout
asurvey,IcanseeifpeoplearemorewillingtodoafavorthatIwouldnotnormallyask
ofthemafterIgivethemacompliment.Mymainissuewiththisexperimentwill
probablybetrackingdownallthepeoplewhoIrandomlychooseformysample.AlsoI
willhavetomakethecomplimentlooknaturalandnotforcedinanywaysothattheydo
notthinkitispartofanexperiment.ButIthinkIwillprobablybeabletoaccomplish
thesethings.
19
H.OriginalInquiryPitch
Astudyreportedby
Forbes
magazineconcludedthatreceivingacomplimenthas
thesamepositiveimpactasreceivingmoney.Totestthis,researcherstook48adults
andhadthemtypeoutaspecificpatternonakeyboardasfastaspossible.Theythen
splittheadultsintothreegroups.Onegroupreceivedacompliment,onegroup
watchedanotherpersonreceiveacompliment,andthethirdgroupreceivedno
complimentandevaluatedthemselvesonagraph.Theadultswerethenbroughtinthe
nextdayandaskedtotypeoutthesamepatternonthekeyboardasfastaspossible.
Theresearchersfoundthatthegroupthatreceivedacomplimentdidsignificantlybetter
thantheothertwogroups.Afterdoingmoreresearch,theyfoundthatthesamepartof
thebrainthatwasstimulatedbythecomplimentisalsostimulatedbyreceivingacash
reward(DiSalvo).Becausetheseresearcherswereabletofindsignificantevidence
thatacomplimentcanimproveperformanceonasimpletask,Iwonderedifthereisa
similarresultwhenaskedtodoafavor.ThequestionIwillbetestingisifthereisa
correlationbetweengivingacomplimentandwillingnesstodoafavor.
TotestthisquestionIwillbedrawingfromthepopulationofLASAHighSchool
seniors.Todothis,Iwillassigneverysenioranumber.Iwillthenrandomlypick40
numbersthatwillcorrelatetothestudentswhowillbepartofmysample.Theonlybias
thatcouldbepresentinthissampleisthatmanyofthestudentsthatarerandomly
chosenprobablyknowme,sotheycouldbebiasedonhowtheyrespondtomeasking
forafavor.Aftercreatingmysample,Iwillgive20ofthestudentsacomplimentand
theother20nocomplimentbeforeaskingforafavor.Forthisexperimentmy
20
complimentwillsimplybeIlikeyourshirt.,andmyfavorwillbeCanIborrowa
dollar?.TocollectthisdataIwillfindeachstudentindividually,complimentthemor
not,andthenasktoborrowadollar.Iwillthenrecordtheirresponse,returnthedollarif
theygavemeone,andgettheirsignaturetoshowthattheywerepartofthe
experiment.
Mydatawillbesplitintotwogroups,thosewhoreceivedacomplimentandthose
whodidnot.Theirresponsetotherequestforafavorwillberecordedaseitheryesor
no.BecauseIamevaluatingtwocategoricalvariablesthatweretakenfromasingle
sample,Iwillbeusingachisquaredtestofindependence.Forthistest,mynull
hypothesiswillbethatthereisnoassociationbetweenreceivingacomplimentand
willingnesstodoafavor.Myalternativehypothesiswillbethatthereisanassociation
betweenreceivingacomplimentandwillingnesstodoafavor.BeforeIcanrunthistest
Ineedtomakesurethatmyassumptionsaremet.BasedonmydatacollectionI
alreadyknowthatmydatawillberandom.Also,Iknowthatasinglesubjectcantbein
boththeyesandnocategoryforrespondingtoafavor,andalsocantbothreceive
andnotreceiveacompliment.AfterIcollectmydataIwillbeabletotestmylasttwo
assumptionsaboutexpectedcounts.TofindmyconclusionforthisquestionIwillput
mydataintoacalculatorandrunachisquaredtestofindependencesothatIcanfind
thepvalueandcompareittomyalphaof0.05.
MyreasonforthisquestionisthatIwanttoseeifasimplecomplimentcan
changethewayapersonrespondstoyou.Byaddingthefavorofaskingforadollar,I
canseeifpeoplearemorewillingtodoafavorthatIwouldnotnormallyaskofthem
21
afterIgivethemacompliment.Mymainissuewiththisexperimentwillprobablybe
trackingdownallthepeoplewhoIrandomlychooseformysample.AlsoIwillhaveto
makethecomplimentlooknaturalandnotforcedinanywaysothattheydonotthinkit
ispartofanexperiment.Alsowhenaskingforadollar,itcantbeobviousthatthe
complimentwasjustanopenerforthefavor.ButIthinkIwillprobablybeableto
accomplishthesethings.Someothersmallproblemsthatmaycomewiththequestion
ismanypeoplemaybeunwillingtogiveupadollarnomatterwhatcircumstancethey
arein.Also,somepeoplemaynotactuallyhaveadollaronthem.Ifthishappens,Iwill
moveontothenextrandomlygeneratednumberandusethatstudentinstead.But
overall,IthinkthisisasolidexperimentandIminterestedtoseetheresults
.
22
I.ProjectRoughDraft
Astudyreportedby
Forbes
magazineconcludedthatreceivingacomplimenthas
thesamepositiveimpactasreceivingmoney.Totestthis,researcherstook48adults
andhadthemtypeoutaspecificpatternonakeyboardasfastaspossible.Theirtimes
were
then
recorded.Theythensplittheadultsintothreegroups.Onegroupreceiveda
compliment,onegroupwatchedanotherpersonreceiveacompliment,andthethird
groupreceivednocomplimentandevaluatedthemselvesonagraph.Theadultswere
thenbroughtinthenextdayandaskedtotypeoutthesamepatternonthekeyboardas
fastaspossible,andtheirtimeswererecordedonceagain.Theresearchersfoundthat
thegroupthatreceivedacomplimenthadsignificantlyfastertimesthantheothertwo
groups.Afterdoingmoreresearch,theyfoundthatthesamepartofthebrainthatwas
stimulatedbythecomplimentwasalsostimulatedbyreceivingacashreward(DiSalvo).
Becausetheseresearcherswereabletofindsignificantevidencethata
complimentcanimproveperformanceonasimpletask,therecouldalsobea
relationshipbetweenreceivingacomplimentandotherthingslikefavors.Thequestion
thatwillbetestedisifthereisacorrelationbetweenbeinggivenacomplimentand
willingnesstodoafavor.Forthistest,thenullhypothesiswillbethattheproportionof
responsesarethesamewhetherornotIgavethemacompliment.Thealternative
hypothesiswillbethattheproportionofresponsesarenotthesameacrossall
categories.ThepopulationthatthesampleisbeingdrawnfromisallLASAseniors.
Basedontheresults,therecouldbesignificantevidencetosuggestthatgivingaLASA
senioracomplimentcaninfluencethemtodoafavorforyou.Ifitisassumedthata
23
LASAseniorrepresentstheaverageperson,thiscoulddefinitelybeusedbythe
averagepersontotakeadvantageofthosearoundthemandgetfreefavorsoutof
peopletheycompliment.
Thewaythesamplewasmadewasbyobtaining
toget
alistofalltheLASA
seniorsandthenrandomlyselecting40ofthemwithanonlinerandomlistgenerator.
Thefirst20thatweregeneratedreceivedacomplimentandthesecond20didnot.To
testmyhypothesis,thepopulationbeingusedisallLASAHighSchoolseniors.Theonly
biasthatcouldbepresentinthissampleisthatmanyofthestudentsthatarerandomly
chosenknowme,sotheycouldbebiasedonhowtheyrespondtomeaskingfora
favor.Forthisexperimentthecomplimentwillsimplybeacomplimenton
to
something
thatthestudentiswearing(whetheritbetheirshoes,shirt,etc.).Thefavorwillbe,
Wouldyoubewillingtofilloutthissurveyforme?
.
Thesurveywillsimplybeapage
fullofrandomquestionsabouttheirsex,age,haircolor,etc.Thequestionsonthe
surveydonotreallymatter,becausethestudentsresponsetothefavoristheactual
data.Tocollectthedata,eachstudentwillbefoundindividuallyandthenaskedtofill
outthesurvey.Theirresponsewillberecordedandasignaturewillbeacquiredas
proofthatthestudentwasspokento.
Beforerunningthechisquaredtestofhomogeneity,firsttheassumptionshave
tobechecked.Therandomassumptionismetbecausethesampleisasimplerandom
sample.Theindependentassumptionismetbecauseastudentcannothavemultiple
responsestothefavorandtheycantbecomplimentedandnotcomplimentedatthe
sametime.Thefirstexpectedcountassumptionismetbecausealltheexpectedvalues
24
aregreaterthanone,thesmallestonebeing3.5.Butthesecondexpectedcounts
assumptionisnotmet,becausemorethan20%ofthevaluesarelessthanfive.Two
outofthefourvaluesare3.5.Thismeansthatthedatacouldbesubjecttoquestion.
Runningachisquaredtestofhomogeneityproducedachisquaredvalueof
1.5584,1degree
s
offreedom,andapvalueof0.2119.Becausethepvalueof0.2119
isgreaterthanthealphavalueof0.05,wefailtorejectthenullhypothesisthatthe
proportionofresponsesarethesamewhetherornotthestudentreceiveda
compliment.Therefore,wedonothavesignificantevidencetosupporttheclaimthat
theproportionofresponsesarenotthesamewhetherornotthestudentreceiveda
compliment.
ThismeansthatforthepopulationofLASAseniors,receivingacomplimentdid
nothaveasignificantinfluenceontheirwillingnesstodoafavor.Ofthe40students,
fivewhodidnotreceiveacomplimentdidnotwanttofilloutthesurveyandonlytwo
whodidreceiveacomplimentdidnotwanttofillitout.Thismeansthatfewerstudents
whoreceivedacomplimentdidnotwanttofilloutthesurvey,butitwasnotsignificant
enoughofadifference.I
f
thisdataisappliedtotheaverageperson,itwouldmean
s
thatgettingfreefavorsfrompeoplebygivingthemcomplimentswouldnotactually
work.
Butthereareseveralproblemswiththedatathatcouldleadtoaquestionable
conclusion.Firstofall,oneoftheassumptionswasnotmetwhichcouldmeanthata
chisquaredtestofhomogeneitywasnotappropriate.Also,mostofthestudentswho
didnotwanttofilloutthesurveyknowme,sotheyweremorecomfortablewithtelling
25
methattheydidnotwanttofillitout.Thisisdefinitelyabiasthatcouldleadto
questionableresults.Thebestwaythatthisexperimentcouldberepeatedtomakeit
lessbiasedwouldbetoperformitwithlargegroupofrandomstrangers.Thiswaythe
peoplewouldnotknowmesotheycouldnotbebiasedinansweringandalsoalarger
samplecouldbetakentohelpmeettoexpectedcountsassumption.Itwouldnolonger
beasimplerandomsample,soconveniencebiaswouldbepresent.Butitwould
definitelybemorerepresentativeoftheaverageperson.
26
J.PeerReview