Springer.J.K.Lindsey.Applying.Generalized.Linear.Models

《Springer.J.K.Lindsey.Applying.Generalized.Linear.Models 》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、ApplyingGeneralizedLinearModelsJamesK.LindseySpringerPrefaceGeneralizedlinearmodelsprovideaunifiedapproachtomanyofthemostcommonstatisticalproceduresusedinappliedstatistics.Theyhaveap-plicationsindisciplinesaswidelyvariedasagriculture,demography,eco-logy,economics,education,engineering,environmentalst

2、udiesandpollu-tion,geography,geology,history,medicine,politicalscience,psychology,andsociology,allofwhicharerepresentedinthistext.IntheyearssincethetermwasfirstintroducedbyNelderandWedder-burnin1972,generalizedlinearmodelshaveslowlybecomewellknownandwidelyused.Nevertheless,introductorystatisticstextb

3、ooks,andcourses,stillmostoftenconcentrateonthenormallinearmodel,justastheydidinthe1950s,asifnothinghadhappenedinstatisticsinbetween.Forstudentswhowillonlyreceiveonestatisticscourseintheircareer,thisisespeciallydisastrous,becausetheywillhaveaveryrestrictedviewofthepossibleutilityofstatisticsintheirch

4、osenfieldofwork.Thepresenttext,beingfairlyadvanced,isnotmeanttofillthatgap;see,rather,Lindsey(1995a).Thus,throughoutmuchofthehistoryofstatistics,statisticalmodellingcentredaroundthisnormallinearmodel.Booksonthissubjectabound.Morerecently,loglinearandlogisticmodelsfordiscrete,categoricaldatahavebecomec

5、ommonundertheimpetusofapplicationsinthesocialsci-encesandmedicine.Athirdarea,modelsforsurvivaldata,alsobecameagrowthindustry,althoughnotalwayssocloselyrelatedtogeneralizedlinearmodels.Incontrast,relativelyfewbooksongeneralizedlinearmodels,assuch,areavailable.Perhapstheexplanationisthatnormalanddiscr

6、ete,aswellassurvival,datacontinuetobethemajorfieldsofapplication.Thus,manystudents,eveninrelativelyadvancedstatisticscourses,donothavevianoverviewwherebytheycanseethatthesethreeareas,linearnormal,categorical,andsurvivalmodels,havemuchincommon.Fillingthisgapisonegoalofthisbook.Theintroductionoftheidea

7、ofgeneralizedlinearmodelsintheearly1970shadamajorimpactonthewayappliedstatisticsiscarriedout.Inthebeginning,theirusewasprimarilyrestrictedtofairlyadvancedstatisticiansbecausetheonlyexplanatorymaterial