A Regression Paradox for Linear Models: Sufficient Conditions and Relation to Simpson's Paradox.
A study was conducted to examine sufficient conditions wherein the regression paradox will occur in a linear, Gaussian system are described. Regression paradox occurs when group effects are compared from direct and reverse regressions, causing seemingly contradictory group effects when the predicto...
| Published in: | American Statistician Vol. 63; no. 3; pp. 218 - 226 |
|---|---|
| Main Authors: | , , |
| Format: | Article |
| Published: |
American Statistical Association
August 2009
|
| Subjects: | |
| Online Access: | View this record in EBSCOhost |
| fields | @attributes: recordID: 1 pdfLink: plink: https://search.ebscohost.com/login.aspx?direct=true&db=ssf&AN=508081608&site=ehost-live header: @attributes: shortDbName: ssf uiTerm: 508081608 longDbName: Social Sciences Full Text (H.W. Wilson) uiTag: AN controlInfo: bkinfo: jinfo: jid: 00031305 STT jtl: American Statistician issn: 00031305 maglogo: N pubinfo: dt: August 2009 vid: 63 iid: 3 pid: 543 pub: American Statistical Association artinfo: ui: 508081608 10.1198/tast.2009.08220 ppf: 218 ppct: 8 formats: tig: atl: A Regression Paradox for Linear Models: Sufficient Conditions and Relation to Simpson's Paradox. aug: au: Chen, Aiyou Bengtsson, Thomas Ho, Tin Kam su: Paradox Statistics Linear statistical models Regression analysis Mathematical models of customer satisfaction sug: subj: Paradox Statistics Linear statistical models Regression analysis Mathematical models of customer satisfaction ab: A study was conducted to examine sufficient conditions wherein the regression paradox will occur in a linear, Gaussian system are described. Regression paradox occurs when group effects are compared from direct and reverse regressions, causing seemingly contradictory group effects when the predictor and regressand are interchanged. Data were obtained from a recent investigation of customer satisfaction data, in which the regression paradox was rediscovered, are discussed. Findings revealed that the paradox can arise naturally in some scenarios and is not necessarily the result of sampling error, collinearity, or misspecified models, as has been previously implied. Findings also revealed that the phenomenon is possible in more general, non-Gaussian settings. Findings are discussed in detail. pubtype: Academic Journal doctype: Article src: R language: English refInfo: copyright: @attributes: flag: N holdings: @attributes: islocal: N |
|---|