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
Confidence Intervals Intro - Day 1 - Complete SupposewehaveaNormalpopulationwithanunknownmean,μ,and standarddeviationof20. Ingroups,Iwouldlikeyoutoestimatethemean,μ,ofthispopulation. Iwillgenerateasampleofsize16fromthispopulationandsincesample mean,,isanunbiasedestimatorofpopulationmean,Iwillgiveyouthe meanfromthissample,. YourjobistocomeupwithanintervalofREASONABLEvaluesthatwill containthetruevalueμ. Confidence Intervals Intro - Day 1 - Complete De#inition:ApointestimatorisaSTATISTICthat providesanestimateofapopulationparameter.The valueofthatstatisticfromasampleiscalleda point estimate.Ideally,apointestimateisour"bestguess"at thevalueofanunknownparameter. Thismeansourestimatorshouldbean _______________________andhaveasamplingdistribution with____________________________. Sowehaveapointestimator,,ofourunknownparameter-μ- andwewanttoknowhowclosethisestimateistotheactualμ.Itis VERYunlikelythatwegottheactualvalueofμwithourpoint estimator,. Inordertoanswerthisweneedtoknowhowthesamplemean,, wouldvaryifwetookmanySRSsofthesamesize(16inthiscase) fromthissamepopulation. SAMPLINGDISTRIBUTIONSanswerthisquestion. Confidence Intervals Intro - Day 1 - Complete SamplingDistributions: Shape:BecausewehaveaNormaldistributiontostartwithour samplingdistributionofisalsoNormal. Center:Themeanofthesamplingdistributionofisthesameas μ(theparameterwearetryingtoestimate ) Spread:Thestandarddeviationofthesamplingdistributionof foranSRSof n=16is =5 Sampling Distribution of Population Distribution =5 σ = 20 μ Values of X μ Values of 1.Toestimateμweusethesamplemean,,oftherandomsample.We don'texpecttobeexactlyequaltoμ,sowewanttosayhowaccurate theestimateis. 2.Inrepeatedsamples,thevaluesoffollowaNormaldistributionwith meanμandstandarddeviation,asinthe_igureabove. 3.The95partofthe68-95-99.7ruleforNormaldistributionssaysthat iswithin2standarddeviationsofthepopulationmeanμinabout95%of allsamplesofsizen. Confidence Intervals Intro - Day 1 - Complete 95% of all sample µ-2 µ s µ+2 4.Wheneveriswithin2ofμ,μiswithin2of.Thishappensinabout 95%ofallpossiblesamples.Sotheintervalfromμ-2toμ+2"captures" thepopulationmeanμinabout95%ofallsamplesofsize n. 5.Ifweestimatethatμliessomewherebetweenμ-2andμ+2,we'dbe calculatingthisintervalusingamethodthatcapturesthetrueμinabout95%ofall possiblesamplesofthissize. Allcon_idenceintervalswillhaveaformsimilartothis: estimate±marginoferror Keepinmindthis"error"hastodowithchancevariation duetorandomsamplingorrandomassignment-i.e. samplingvariability-notwhetherornotwemadean errorincalculatingorANYothertypeoferror. Confidence Intervals Intro - Day 1 - Complete De#initions: Acon$idenceintervalforaparameterhastwoparts: 1)The marginoferrortellshowclosetheestimate tendstobetotheunknownparameterinrepeated randomsampling. 2)Acon$idencelevel,C ,whichgivestheoverallsuccess rateoftheMETHODforcalculatingthecon_idence interval.ThatisinC%ofALLPOSSIBLESAMPLES,the methodwouldyieldanintervalthatcapturesthetrue parametervalue. InterpretingCon#idenceLevelsandCon#idenceIntervals 1.Gotowww.whfreeman.com/tps4eandlaunchtheapplet( Con$idenceInterval applet).Thedefaultsettingforthecon_idencelevelis95%.Changethisto90%. Wewillbeusingthisapplettoinvestigatetheideaofcon_idencelevel. 2.Click"Sample"tochooseanSRSanddisplaytheresultingcon_idenceinterval. Didtheintervalcapturethepopulationmeanμ(whattheappletcallsa"hit")? Dothisatotalof10times.Howmanyoftheintervalscapturedthepopulation meanμ? 3.Resettheapplet.Click"Sample50"tochoose50SRSsanddisplaythe con_idenceintervalsbasedonthosesamples.Howmanycapturedtheparameter μ? Keepclicking"Sample50"andobservethevalueof"Percenthit".Whatdoyou notice? Confidence Intervals Intro - Day 1 - Complete Astheappletcon_irmed,thecon(idencelevelisthe overall"capturerate"ifthemethodisusedanytimes. Fora95%con_idencelevelimpliesthat inrepeated sampling,95%oftheintervalsconstructedinthis waywillcontaintheunknownpopulationmeanμ . Con(idenceLevel:Tosaythatweare 95%con(identis shorthandfor"95%ofallpossiblesamplesofagiven sizefromthispopulationwillresultinanintervalthat capturestheunknownparameter." Con$idenceInterval:Tointerpreta C%con_idence intervalforanunknownparameter,wesay,"Weare C% con_identthattheintervalfrom_________to_______ containstheactualvalueofthe[populationparameter INCONTEXT]." Confidence Intervals Intro - Day 1 - Complete What'stheprobabilitythatour95%con6idence intervalcapturestheparameter? NOT95%!!! Thislevelonlytalksaboutthecon_idenceintheMETHOD,notONEsample! BEFOREwesample,wehavea95%chanceofgettingasamplethatwill producea95%con_idenceintervalthatcapturesthetrueparameter. However,oncewesampleourresultingcon_idenceintervaleitherwill containtheparameterorwon't-i.e.theprobabilitythatitcontainsthe parameteriseither1(itdid)or0(itdidn't). Thecon6idencelevelDOESNOTtellusthechancethatAPARTICULAR con6idenceintervalcontainsthepopulationparameter. Somethingtokeepinmind: Acon_idenceintervalisastatementaboutaparameter- notasamplestatistic. Don'tsay"Weare95%con_identthattheintervalfrom _______to_____containsthesampleproportion/meanof all_______" Weknowitcontainsthesamplestatisticsothismakes NOsense! Confidence Intervals Intro - Day 1 - Complete Checkyourunderstanding: HowmuchdoesthefatcontentofBrandXhotdogsvary?To_indout, researchersmeasuredthefatcontent(ingrams)ofarandomsample of10BrandXhotdogs.A95%con_idenceintervalforthepopulation standarddeviationσis2.84to7.55. 1.Interpretthiscon_idenceinterval. 2.Interpretthecon_idencelevel. 3.Trueorfalse:Theintervalfrom2.84to7.55hasa95%chanceof containingtheactualpopulationstandarddeviationσ.Justifyyour answer. Choosingacon#idencelevel ReturntotheCon$idenceIntervalapplet. Wearegoingtoexploretherelationshipbetweenthecon_idencelevelandthe con_idenceinterval. 1.Setthecon_idencelevelat95%andclick"Sample50".Observethepercenthitand thelengthofthecon_idenceinterval. 2.Changethecon_idencelevelto99%.Whathappenstothelengthofthecon_idence interval?Tothepercenthit? 3.Nowchangethecon_idencelevelto90%.Whathappenstothelengthofthe con_idenceinterval?Tothepercenthit? 4.Finally,changethecon_idencelevelto80%.Whathappenstothelengthofthe con_idenceinterval?Tothepercenthit? Confidence Intervals Intro - Day 1 - Complete Asweincreasecon_idencelevel,weincreasethewidth ofourinterval... Whatelsemighteffectthewidthofourcon_idenceinterval? Let'slookattheformofthecon_idenceintervalmoreclosely: estimate±(criticalvalue)(standarddeviationofthestatistic) Clearlythesmallerthemarginoferror,thenarrowerthe con_idenceinterval-themorepreciseourestimateofthe populationparameter. Themarginoferrordependsontwothings: · Thecriticalvalue -thisdependsonthesampling distribution(moreonthis)andonthecon_idencelevel- thegreaterthecon_idencethegreaterthecriticalvalue · · Thestandarddeviationofthestatistic -thisdepends onthesamplesizen-thebiggerthesample,thesmaller thestandarddeviationofthestatistic,andthenarrower theinterval. Confidence Intervals Intro - Day 1 - Complete CONDITIONS: 1.Random:Thedatashouldcomefromawell-designedrandomsampleora randomizedexperiment. 2.Normal:Themethodweusetoconstructcon_idenceintervalsforμdependson thefactthatthesamplingdistributionofthestatisticisatleastapproximately Normal 3.Independent:Theproceduresforcalculatingcon_idenceintervalsassumethat individualobservationsareindependent.Randomsamplinghelpsensurethisso longasourpopulationisin_inite.IfNOTandwearesamplingwithoutreplacement weshouldcheckthe10%condition.