Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Click and Learn Sampling and Normal Distribution Educator Materials OVERVIEW Normaldistribution,sometimescalledthebellcurve,isacommonwaytodescribeacontinuous distributioninprobabilitytheoryandstatistics.Inthenaturalsciences,scientiststypicallyassumethata seriesofmeasurementsofapopulationwillbenormallydistributed,eventhoughtheactualdistribution maybeunknown.Butevenifyouassumethatmeasurementsofapopulationshouldbenormally distributed,asampletakenfromthatpopulationwillnotnecessarilybenormallydistributed.Whyis that? InthisClickandLearn,youwillexplorewhatsampledistributionlookslikewhensamplesaretakenfrom anidealizedpopulationofadefinedmeanandstandarddeviation.Studentswillexplorehowstandard deviationaffectsthedistributionofmeasurementsinapopulation.Next,theywillexplorehowsample sizeaffectsthedistributionofmeasurementsandthereforethesamplemean.Throughthisexploration, studentswilldevelopanunderstandingofhowsamplesizeaffectsthedistributionofsamplemeans drawnfromthesamepopulationandhowthisphenomenonismodeledinanequationforcalculating thestandarderrorofthemean. KEYCONCEPTSANDLEARNINGOBJECTIVES • Theappearanceofahistogramofmeasurementsinasampledependsonthepopulationfrom whichthesamplecame. • Theappearanceofthehistogramalsodependsonthesamplesize. • Smallsamplestakenfromanormallydistributedpopulationmaynotappeartobenormally distributed.Largersamplesstarttoapproximateanormaldistribution. • Whenapopulationissampledrepeatedly,ameancanbecalculatedforeachsample,toobtain manydifferentmeans.Ifthosemeansareplottedasahistogram,theywillbeapproximately normallydistributed. • Thestandarddeviationofsuchadistributionofmeansiscalledthestandarderrorofthemean. Studentswillbeableto • explainthatstandarddeviationisameasureofthevariationofthespreadofthedataaround themean. • explainthatlargersamplesizesaredesirablewhencollectingdataaboutapopulationbecause theyaremorelikelytoreflectthedistributionofmeasurementsinapopulation. • calculateStandardErroroftheMean(SE)usingtheequationSE𝑥 = • explainthatSEofthemeanisameasureofthereliabilityofthemeanofasampleasareflection ofthemeanofthepopulationfromwhichthesamplewasdrawn. www.BioInteractive.org $ % . PublishedMarch2017 Page1of8 Click and Learn Sampling and Normal Distribution • Educator Materials useSEtodeterminethe95%ConfidenceIntervaltoadderrorbarstoagraphandusethese errorbarstodetermineifthereisadifferencebetweenthepopulationsfromwhichthesample came. CURRICULUMCONNECTIONS APBiology(2012-2013)SP2,SP5 KEYTERMS NGSS(2013)SEP4 measurement,sample,population,normaldistribution,randomsampling,mean,standarddeviation, standarderrorofthemean,95%ConfidenceInterval,errorbar TIMEREQUIREMENTS Completingallpartsofthislessonwillrequireuptothree50-minuteclassperiods.However,some portionscanbeassignedforhomework. SUGGESTEDAUDIENCE Part1ofthisactivityisappropriateforafirstyearandanadvanced(honors,AP,orIB)highschool biologycourse.Parts2and3areappropriateforanadvanced(honors,AP,orIB)highschoolor introductorycollegebiologycourse. PRIORKNOWLEDGE Studentsshouldbefamiliarwith • statisticalconceptofmeanasanaverageofasample’smeasurements. • histogramsasadisplayofthefrequencyofmeasurementsinasample. MATERIALS • SamplingandNormalDistributionClickandLearnat http://www.hhmi.org/biointeractive/sampling-and-normal-distribution • DistributionofMeansgrid(lastpageofthisdocument;thesecanbelaminatedtobereusedby multipleclasses) TEACHINGTIPS • ThisactivityassumesnopriorknowledgeofStandardDeviationorStandardErroroftheMean. Therefore,itcanbeusedtointroducetheuseofstatisticstodescribeadataset.Itisimportant thatstudentscandistinguishbetweenthetermsmeasurement,sample,andpopulation.A sampleisacollectionofindividualmeasurementsdrawnfromapopulation.Priortostarting Part1,studentsshouldunderstandthatitistypicallynotpossibletomeasureeveryindividualin alargepopulation.Therefore,arandomlyselectedsampleofthepopulationismeasuredand thedataisusedtorepresentthewholepopulation. www.BioInteractive.org PublishedMarch2017 Page2of8 Click and Learn Sampling and Normal Distribution Educator Materials • Studentscanoftenrecognizethatsmallsamplesizesarenotrecommendedwhencollecting datafromapopulation.Doingasimpledemonstrationsuchasdrawingonlyafewcoloredbeads fromabagtodeterminethedistributionofcolorsinthebagormeasuringtheheightofonlya fewstudentstodeterminethemeanheightoftheclasscanshowstudentsthatsmallsample sizescanoftenleadtoamisrepresentationofthepopulationknownassamplingerror. • ThesimulationintheClickandLearnisrunbyaprogramthatcalculatesarandomsamplevalue fromanormallydistributedpopulationofinfinitesize.InPart1,thestudentcanmanipulate samplesizeaswellaspopulationmeanandstandarddeviation.InPart2,thestudentcan manipulatesamplesize.Dependingonthespeedofyourcomputer,resamplinginPart2can takeafewsecondsandthecalculationsoccurringinthebackgroundarecomplicated.(Forthose ofyouwhoaremathematicallyorstatisticallyinclined,theprogramusesBox-Mullertransform.) • AttheconclusionofPart1oftheactivity,studentsshouldbeabletoexplainwhatstandard deviationshowsaboutthedistributionofmeasurementsinapopulation.Theywillalsobeable toexplain,usingevidencecollectedintheactivity,whythemeansoflargersamplesizesare morelikelytoberepresentativeofapopulation’struemean. • AttheconclusionofPart2oftheactivity,studentswillunderstandwhytheequationforSEM givesanestimateofthestandarderrorofthemeanbasedonasample’ssizeandstandard deviation.TheywillalsobeabletousetheequationtocalculateSEM,95%confidenceintervals, andusethe95%CItogenerateerrorbarsonabargraph.Whilethisactivityfocusesonthe effectofsamplesize(assamplesizeincreases,SEMdecreases),studentsshouldbeableto predictfromtheequationthatthereisadirectrelationshipbetweenthestandarddeviationand SEM. • Remindstudentsthatonpage1oftheClickandLearn“SamplingfromaNormallyDistributed Population,”clicking“resample”issimulatingcollectinganewrandomlyselectedsetof measurementsfromthepopulation.Therefore,samplemeansandstandarddeviationswill likelybedifferentfromstudenttostudent.Thiswillnotaffectthefinaloutcomeoftheactivity. Studentsshouldalsoberemindedthatonpage2oftheClickandLearn“StandardErrorofthe Mean,”“resample”representsrepeatingthesamplecollection500times,andeachsample consistsofanumberofmeasurementsequaltothesamplesize.Thismeansthatforasample sizeof100,thesimulationtook50,000measurements. • InPart2,theteachermayneedtopointoutanddiscussthedifferencebetweensamplemean andstandarddeviationandthemeanandstandarddeviationof500means.Samplemeanand standarddeviationisdescribingthedatainthetopgraph,whilethemeanandstandard deviationof500meansisdescribingthedatainthebottomgraph. SUGGESTEDPROCEDURE Dependingontheskilllevelofthestudentsinthecourse,thisactivitycanbedoneindependentlyor guidedbytheteacher.Theprocedurebelowisforaguidedprocess,duringwhichtheinstructor checksforstudentunderstandingatkeypointsintheactivity. www.BioInteractive.org PublishedMarch2017 Page3of8 Click and Learn Sampling and Normal Distribution Educator Materials Introduction 1.Showstudentsthegraphbelowandaskthemtointerpretit.Askthemwhattheerrorbarsmean. Whileitdependsonanindividualstudent’spriorlearning,moststudentswillnotbeabletoexplain whattheerrorbarsmean.Ifthisisthecase,askthemtodescribetheerrorbar.Guidestudentsto observationssuchas • Theerrorbarfordarkdoesnotoverlaptheerrorbarforlight. • Thedarkerrorbarislongerthanthelighterrorbar. • Thelengthsoftheerrorbaraboveandbelowthetopofthebarareequal. Figure1.MeanLengthofCroftonSeedlingsafterOneWeekintheDarkorintheLight.(FromUsingBioInteractive ResourcestoTeachMathematicsandStatisticsinBiologyhttp://www.hhmi.org/biointeractive/teacher-guidemath-and-statistics) 2.InstructstudentstocompletethePre-assessmentQuestion(whichcouldbecollectedonnotecards asaformativeassessment).TheninstructstudentstoaccesstheClickandLearnat http://www.hhmi.org/biointeractive/sampling-and-normal-distributionandcompleteitems2through5. Itisimportantatthispointtoensurethatstudentsunderstandthatanindividualmeasurementispart ofasampletakenfromalargerpopulation.Pointouttostudentsthecharacteristicsofanormally distributedpopulationbyreferencingtheredlineonthegraphandthatnumberofindividualmass measurementsarerepresentedbythebarsinthehistogram.Note:ThispartalongwithPart1items6 and7couldbeassignedtostudentsforhomeworkpriortocompletingtherestofPart1oftheactivityin class. PART1:SAMPLINGFROMANORMALLYDISTRIBUTEDPOPULATION 1. Studentsworkthroughthetaskandcompleteitems6and7toexplorehowmodifyingthestandard deviationaffectsthedistributionofmeasurementsinthepopulation.Itshouldbepointedoutto studentsthatchangingtheparameterschangesthesimulationprogram.Inarealdataset,the standarddeviationisdeterminedbytheactualmeasurementsinthepopulationorsample. www.BioInteractive.org PublishedMarch2017 Page4of8 Click and Learn Sampling and Normal Distribution Educator Materials 2.Havestudentsreadthesummarydescriptionofstandarddeviationanddiscussanyquestionsthey haveaboutstandarddeviationandnormaldistribution. 3.IntherestofPart1,studentsexploretheeffectofsamplesizeonthesamplemeancomparedtothe truemeanofthepopulation.Remindstudentsthattheyaresettingparametersfortheprogramrunning thesimulation(populationmean=50kgandstandarddeviation=10kg). 4.Item8canbeusedasaformativeassessmenttomonitorstudentunderstandingofstandard deviation.Acorrectstudentresponsewouldbe:“Forthispopulation,68%ofthemassesshouldbe between40and60kg(1standarddeviation),while95%ofthemassesshouldfallbetween30and70kg (2standarddeviations).” 5.Studentscompleteitems9and10.Aftercompletingthistask,studentsshouldrecognizethata samplesizeof1000ismorelikelytogiveyouasamplemeanthatreflectsthetruemeanofthe populationbecausethelargernumberofmeasurementswillreflectthenormaldistributionofthe population.Theyshouldalsorecognizethatcollectingmeasurementsfromasampleof1000individuals couldbetime-consuming,expensive,orsimplynotpractical. 6.Studentscomplete“Selectingtheappropriatesamplesize”bycompletingthetaskanditems11 through13.Providestudentswiththe“DistributionofMeans”grid.Thistaskcanbecompletedinpairs orasmallgroup.Thereshouldbeatleastonegraphintheclassforeachsamplesizeinthesimulation (4,9,16,25,100,400,1000).Laminatingthegridswillallowthemtobereusedbyseveralclasses. Discussitem13asawholeclass.Askstudentstojustifytheiranswertothequestionwithevidencefrom thegraphs.Studentstypicallyselect100asanappropriatesamplesize.Anexampleofthedata generatedfromthistaskisshownbelow. www.BioInteractive.org PublishedMarch2017 Page5of8 Click and Learn Sampling and Normal Distribution Educator Materials PART2:STANDARDERROROFTHEMEAN 1. Items1through8canbecompletedasahomeworkassignment.Studentsusethenextpageinthe ClickandLearntoextendtheirexplorationoftheeffectofsamplesizeonthedistributionofsample means.InPart2,resamplingwillgenerateahistogramofthemeansof500samplesshowinga normaldistribution.Studentsshouldcometotheconclusionthatwhilethesamplesizedoesnot affectthemeanof500means,itdoesaffectthestandarddeviationofthemeans.Forsmaller samplesizes,thesamplemeancouldbequitedifferentfromthepopulationmean,andthisis reflectedinthelargerstandarddeviationofthemeans.Thisshouldreinforcetheconclusionthey cametoattheendofPart1thatlargersamplesizeswillprovideabetterrepresentationofthe populationfromwhichthesamplewasdrawn. $ 2. Item9introducesstudentstotheequationforstandarderrorofthemean,SE𝑥 = .Items10and % 11havestudentscalculatetheSEusingtheequation.Theyarethenaskedtocomparethe empiricallymeasuredSEM(standarddeviationof500means)tothecalculatedestimationofthe SEM.StudentsshouldberemindedthatthemathematicalformulaforSEMallowsthemtoestimate therealSEMgivenasmallsample,whilerepeatingsamplesmanytimesallowsthemtoempirically measuretheactualSEM.Studentsshouldfindthat,withtheexceptionofverysmallsamplesizes, $ usingtheequationSE𝑥 = isareasonablyaccuratewaytoestimatetheStandardErrorofthe % Meanfromthestandarddeviationofasampleandthesamplesize.Studentscanbenefitfroma discussionregardinghowtheequationmodelsthedistributionofmeansthattheyobserved. Encouragestudentstodiscusswhysamplesizeisinthedenominator.Asseeninthegraphbelow, standarddeviationofthemeansdecreasesassamplesizeincreases.Theyshouldrecallfromtheir observationsduringPart1oftheactivitythatlargersamplesizesaremorelikelytobeatruer reflectionofthepopulationfromwhichthemeasurementsaredrawnandthatverysmallsample sizesoftenresultininaccuraterepresentationsofthepopulation. Itishelpfultoreferstudentsbacktothedistributionof10meanstheyplottedinPart1ofthe activity.EmphasizethattheSEequationallowsonetoestimatethespreadofthemeansthatwould beexpectedfrommanysamplesdrawnfromthesamepopulationfromthestandarddeviationand samplesizeofasingleobservedsample. Standard Deviation of Mean 6 5 4 3 2 1 0 0 100 200 300 400 500 Sample Size (n) www.BioInteractive.org PublishedMarch2017 Page6of8 Click and Learn Sampling and Normal Distribution Educator Materials TheeffectofsamplestandarddeviationontheSEMisnotexploredinthesimulation.Thiswasdoneto avoidthemisconceptionthatsamplesizeaffectssamplestandarddeviation.Itstillmaybehelpfulto discusstheeffectstandarddeviationofthesamplehasonthestandarderrorofthemean.Thestandard deviationofthesampleisinthenumeratoroftheequationbecauseamorevariedpopulation(larger samplestandarddeviation)willincreasethelikelihoodthatthesamplemeasurementswillnotbeagood representationofthepopulationfromwhichtheyaretaken. 3. Havestudentsreadthesummaryandthencompleteitems12and13tolearnhowtousethe standarderrorofthemeantogenerate95%ConfidenceIntervalerrorbars.Readingthesummary willshowstudentshowtointerprettheseerrorbars. PART3:APPLYWHATYOUHAVELEARNED Thedatapresentedisauthenticdatacollectedbystudentsconductinganexperimenttotesttheeffect ofpectinaseandcellulaseonturningapplesauceintoapplejuice. Thispartoftheactivitycanbegivenasanassessmenttodeterminestudents’levelofunderstandingof theconceptofstandarderrorofthemeanandhowtousethestatistictoanalyzetheexperimentaldata. RECOMMENDEDFOLLOW-UPACTIVITIES EvolutioninAction:DataAnalysis(http://www.hhmi.org/biointeractive/evolution-action-data-analysis) RosemaryandPeterGranthaveprovidedmorphologicalmeasurements,includingwinglength,body mass,andbeakdepth,takenfromasampleof100mediumgroundfinches(Geospizafortis)livingonthe islandofDaphneMajorintheGalápagosarchipelago.Thecompletedatasetisavailableinthe accompanyingExcelspreadsheet. Inoneactivity,entitled“EvolutioninAction:GraphingandStatistics,”studentsareguidedthroughthe analysisofthissampleoftheGrants’databyconstructingandinterpretinggraphs,andcalculatingand interpretingdescriptivestatistics.Thesecondactivity,“EvolutioninAction:StatisticalAnalysis,”provides anexampleofhowthedatasetcanbeanalyzedusingstatisticaltests,inparticulartheStudent’st-test forindependentsamples,tohelpdrawconclusionsabouttheroleofnaturalselectiononmorphological traitsbasedonmeasurements. LizardEvolutionVirtualLab(http://www.hhmi.org/biointeractive/lizard-evolution-virtual-lab) IntheLizardEvolutionVirtualLab,studentsexploretheevolutionoftheanolelizardsintheCaribbean bycollectingandanalyzingtheirowndata. Thevirtuallabincludesfourmodulesthatinvestigatedifferentconceptsinevolutionarybiology, includingadaptation,convergentevolution,phylogeneticanalysis,reproductiveisolation,and speciation.Eachmoduleinvolvesdatacollection,calculations,analysis,andansweringquestions. www.BioInteractive.org PublishedMarch2017 Page7of8 Click and Learn Sampling and Normal Distribution Educator Materials www.BioInteractive.org PublishedMarch2017 Page8of8