counterfactual causal analysis and nonlinear probability models

counterfactual causal analysis and nonlinear probability models

ID:7298960

大小:336.92 KB

页数:21页

时间:2018-02-10

counterfactual causal analysis and nonlinear probability models_第1页
counterfactual causal analysis and nonlinear probability models_第2页
counterfactual causal analysis and nonlinear probability models_第3页
counterfactual causal analysis and nonlinear probability models_第4页
counterfactual causal analysis and nonlinear probability models_第5页
资源描述:

《counterfactual causal analysis and nonlinear probability models》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、Chapter10CounterfactualCausalAnalysisandNonlinearProbabilityModelsRichardBreenandKristianBerntKarlsonAbstractNonlinearprobabilitymodels,suchaslogitsandprobitsforbinarydependentvariables,theorderedlogitandorderedprobitforordinaldependentvariablesandthemultinomiallogit,

2、togetherwithlog-linearmodelsforcontingencytables,havebecomewidelyusedbysocialscientistsinthepast30years.Inthischapter,weshowthattheidentiÞcationandestimationofcausaleffectsusingthesemodelspresentseverechallenges,overandabovethoseusuallyencounteredinidentifyingcausalef

3、fectsinalinearsetting.ThesechallengesarederivedfromthelackofseparateidentiÞcationofthemeanandvarianceinthesemodels.Weshowtheirimpactinexperimentalandobservationalstudies,andweinvestigatetheproblemsthatariseintheuseofstandardapproachestothecausalanalysisofnonexperiment

4、aldata,suchaspropensityscores,instrumentalvariables,andcontrolfunctions.Naiveuseoftheseapproacheswithnonlinearprobabilitymodelswillyieldbiasedestimatesofcausaleffects,thoughtheestimateswillbealowerboundofthetruecausaleffectandwillhavethecorrectsign.Weshowthatthetechni

5、queofY-standardizationbringstheparametersofnonlinearprobabilitymodelsonascalethatwecanmeaningfullyinterpretbutcannotmeasure.Othertechniques,suchasaveragepartialeffects,canyieldcausaleffectsontheprobabilityscale,but,inthiscase,thelinearprobabilitymodelprovidesasimplean

6、deffectivealternative.IntroductionManyoftheoutcomesencounteredinsocialresearcharecategorical,ordinal,orcounts.Severalmodelshavebeendevelopedtodealwithdataofthesesorts:theyincludelog-linearmodelsforcontingencytables;logitandprobitmodelsforbinarydependentvariables;cumul

7、ativeprobabilitymodels,suchastheorderedlogitandorderedprobit,forordinaldependentvariables;andthemultinomiallogitforunorderedpolytomouscategoricaloutcomes.AlthoughmanysocialscientistsbelievethatcoefÞcientsfromthesemodelscanbeinterpretedinthesamewayascoefÞcientsfromline

8、armodels,thisisnotso.Innonlinearprobabilitymodelslikethelogit,probit,orderedlogit,R.Breen()DepartmentofSociology,YaleUniver

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。