资源描述:
《don t let the perfect be the enemy of the good is bad data really bad》由会员上传分享,免费在线阅读,更多相关内容在工程资料-天天文库。
1、CHAPTER11Don’tLetthePerfectBetheEnemyoftheGood:IsBadDataReallyBad?BrettGoldsteinAsadatascientist,yourgutandyourtrainingtellyoutouseperfectdataforananalysis.Thisisoftenafunctionofclassicalstatisticseducation,withanintenttosubmitresearchandanalysisforpublication.Thisisfineandnoble,butuponencounteringr
2、eal-worlddata,thecoldrealityofdirtydatabecomesprominentandonemustlearntoabandonhopeofperfectionorfaceanendlessloopoffrustration.MywifeSarah,whodidhergraduateworkinpublichealth,hasoftenusedthephrase:“Don’tlettheperfectbetheenemyofthegood.”1Whenconfrontedwithimperfectdata,myclassicaltrainingwouldsayth
3、atthisdataisbeyondhope,thatitiscouldneverbecleanedsufficiently,andthatwewouldbeunabletoobtainanythingthatwastrulymeaningful.However,thisiswherewegettothekeyprinciplethatthisshouldnothavetobeazero-sumdecision.Itisnotgood,norisitbad,butitcertainlyisviable.Howcanweimproveourpoliciesandstrategiesinabsen
4、ceofperfectdata?Ifitdoesn’tmeetthepristinestandardsoftheclassicalapproach,wemustfindwaystomakethedataworksothatitcaninformthecriticaldecisionsthatarenecessarytomoveahead.ButFirst,Let’sReflectonGraduateSchool…Toexplain,Ineedtostepbackintime.Ingraduateschool,Ihadthesameprofessorforallofmystatisticscla
5、sses.Shewasthorough,excellent,andmeticulous.Whileitmaybeeasiertohavesuchkindmemorieslongafterthecoursesended,shetaughtmeagreatdealaboutclassicalstatistics.1.Thisphrase,Ilaterlearned,isanadaptationofaVoltairequote:Lemieuxestl’ennemidubien.143Theprofessortaughteachofthetechniquesinaveryspecificway.The
6、majorityofthetimewouldbespentonpreparingandcleaningthedata.Therewasaremarkablyrig‐orousprocedureandseriesofstepsdependingontheapplicationofthetest.Inordertomaximizethepowerofthegiventechnique,eachofthetestshadarequiredsetofassumptions.Asweworkedthrougheachandeveryoneofthesemethods,shehadusmeticulous
7、lydocumenteachstepofourwork(which,inthiscase,wasinSPSSsyntax).Wewouldsubmitahugeamountofwork,whichshewouldtirelesslycritique.Eachtimewetookapathormadeadecisionthatwasincorrect,shewouldreturnourpileofs