You are on page 1of 5

AStatisticalStudyofBottleRattle

MorganOlson
SydneyBrooks
ChanelleLogan
EmilyMajewski

Statistics
Mr.Warneke
January6th,2014

ABSTRACT
Thispaperdescribesapreliminarystudyoftheminutetowinitgamebottle
rattleusingagroupofstudentsinastatisticsclass.Wetestedplayersonhowtheleft
orrighthandaffectedtheirskillsusingvolunteersinarandomizedmatchedpairs
experiment.WedrewourvolunteersfromByronHighSchoolstatisticsclass.Using
a=.05,thisstudysupportsthehypothesisthatusingyourlefthandinthegameofbottle
rattledecreasesaplayersscore.Thepaperalsoincludesanobservationalstudythat
showshowplayersscorebetterintheirsecondtrialroundofbottlerattleregardlessof
whichhandtheyused.

INTRODUCTION
Bottlerattleisapopularminutetowinitgame,wewantedtodoadeeper
statisticalanalysisofthegame.Inparticular,wewantedtotesttheideathatusingthe
lefthandwilldecreaseyourscore.Weperformedthestudyonasampleconsistingof
17studentsinMr.WarnekesthirdblockFall2014Statisticsclass.Underthe
assumptionthatstudentsinthisclassinByron,MNhavebottlerattleskillsthatcan
representanySoutheastMinnesotahighschoolstudent,webelievethattheresultsof
ourexperimentapplybeyondourpopulation.Becauseofrandomorderingwithina
matchedpairsexperiment,wecansafelypredictwhetherornotusingthelefthandwill
affectthestudentscores.

EXPERIMENTALDESIGNANDPROCEDURE
Bottlerattle:Weusedtwoemptyplastic1litersodabottles,andtiedthemneck
toneckusingducttape.Beforetapingthemtogetherweput116oz.bagoflargejelly
beansinside.Wemarkedthe1literbottleswith5levelsthatwere1inchapart,starting
fromthebottomofthebottle,measuringtheamountofjellybeanstheygotfromone
bottletoanother.Subjectshadtouseonehandtoshakethejellybeansfromone1liter
bottletotheotherempty1literbottle,inoneminute.Youcannotswitchhands.Each
subjecthastouseboththeirrighthandandlefthand,buttheonethattheyusedfirstis
chosenrandomlybyarandomnumbergenerator.Inthisstudyweinvited17students
fromourclasstodothisexperiment.Eachpersonplayedthegametwice.Onewiththeir
righthand,andonewiththeirlefthand.Allsubjectswererequiredtoremainstandingto
removepotentialbias.Thefirstpartofthestudytestedthehypothesisthatusingtheir
lefthanddecreasesthelevelofjellybeansmakingitintotheotheremptybottle.After
thedatawascollectedfromthe17playersthedifferenceofthetwoscoresforeach
playerwascomputed.
Fortheexperimenttestingtheeffectusingthelefthand,thedifferencewas:
scorewithleftscorewithright
Forobservationalstudycomparingthefirstandsecondtrials.
scoreforfirstscoreforsecond

EXPERIMENTALRESULTS
Eachofthe17studentsinthestudycompletedthegameofbottlerattletwice
onewiththeirlefthandandonewiththeirrighthand.Werandomizedwhichhandthe
studentusedfirsttocontrolfortheconfoundingfactorthatstudentsmaydobetteron
theirsecondgame.Thedatacollectedfromoursubjectsisbelow.
SubjectNumber

ScorewithRight

ScorewithLeft

Difference

10

11

12

13

14

15

16

17

Wetesteddifferencesinscoresforeachplayerusingabootstraptestofasinglemean.
Thistestisusedtoanalyzealistofdifferencesfrommatchedpairsdata.Weexpected
scorestobelowerwhentheplayersusedtheirlefthands.Weranthetestwiththe
followinghypothesis:
H0:nodifferencebetweentreatmentgroups(leftright=0)
Ha:studentsperformedworsewithlefthand(leftright<0)
Wedecidedtotestagainstthealphalevelofa=.05.

Afterrunningaonesamplebootstraphypothesistestonthedata,wefoundap
valueof0.0016.Sincethisislessthana,werejectedthenullhypothesisandsidedwith
thealternativehypothesis.Giventhisresultandthefactthatthiswasarandomizedand
controlledexperiment,weconcludedthatusingthelefthandcausesstudentsscoresto
drop.Toestimatetherangeofthedifferencewefounda95%confidenceintervalofthe
listofdifferences.Weare95%confidentthattheplayersscoreanaverageof.65.409
lessusingtheirlefthandcomparedtousingtheirrighthands.Inordertoreducethis
marginoferrorwewouldneedtoincreasethesamplesize.

OBSERVATIONALSTUDYRESULTS
Asanadditionalstudyweobservedthedifferencebetweeneachplayersfirstand
secondtrialtoseeifitmadeadifference.Sinceitisimpossibletorandomlyassignthe
firstandsecondtrial,itwassimplyobservedandnotapartofthecontrolexperiment.
Wetesteddifferencesinthefirstandsecondscoresforeachplayerusingone
samplebootstraphypothesistest.Weexpectthesecondtrialtobehigherbecausethey
mightfindoutamoreefficientstrategytogetthejellybeansfromonebottletotheother.
Sinceweexpectedthesecondtrialtobehigherweusedthefollowinghypothesis:

H0:nodifferencebetweenfirstandsecondtrial(2nd1st=0)
Ha:studentsperformedbetteronthesecondtrial(2nd1st>0)
Wedecidedtotestthealphalevelofa=.05.
Afterrunningabootstraptestonthedatawefoundapvalueof0.012.Sincethis
islessthanthealphavalueof.05wedecidedtorejectthenullhypothesisandsidewith
thealternativehypothesis.Sincethispartisnotacontrolledexperiment,wecannot
determinewhysubjectsdidbetterintheirsecondgame,justthatthesecondgamewent
better.Ourassumptionisthatpracticecausesplayerstogetbetter.Thiscouldbetested
inafutureexperimentthatvariestheamountofpracticeplayersaregivenbeforethey
arescored.
Wefounda95%confidenceintervalofthedifferenceofthetwotrials.Weare
95%confidencethattheplayersscoreanaverageof.531.41morepointsonthe
secondtrialthanthefirst.Inordertoreducethemarginorerror,wewouldneedalarger
samplesize.Tomaketheintervalasvalidaspossible,itwouldbeimportanttoselect
playersusinganSRSofthetargetpopulation.

INCONSISTENCIESANDERRORS
Throughouttheexperiment,playersmighthavewatchedotherplayersperform
thegameandfigureoutawaytogetthejellybeansdownfaster.Thetapingofthe
bottlesmighthaveweakenedasmoreplayerswereusingthem,makingtheliterbottles
floppyandmorechallengingtogetthejellybeanstotheotheremptybottle.Another
factorcouldalsobethatthebottleswereslightlywetontheinsidemakingthejelly
beanssticktoeachothermakingitharderforthejellybeanstotransferfromonebottle
totheother.Therealsocouldhavebeendistractionsoccurringduringthegamecausing

playerstoperformworse.

CONCLUSION
Withapvalueof0.0016inacomparisonstudyoftheeffectofusingthelefthand
onthebottlerattlescores,itseemsthatthelefthandcausesadecreaseinstudents
abilityatthebottlerattle.Toverifythisresultandconfirmitwithastrongerpvalue,it
shouldberepeatedinotherschools.Abettersamplesizeforafutureexperimentwould
bearound30or40ratherthanour17.Theobservationalstudycomparingthesecond
andfirsttrialsresultedinapvalueof0.012.Thisresultconfidentlysuggeststhatthe
orderofthetrialsdoesmatterandthatiswhywerandomlydeterminedwhichtreatment
grouptheyperformedfirst.Howeveritdoesnotexplainwhytheordermatters.
Experimentsinthefuturecouldlookintothisobservationtoexplainwhyithappened.

You might also like