home *** CD-ROM | disk | FTP | other *** search
/ NetNews Usenet Archive 1992 #16 / NN_1992_16.iso / spool / bit / listserv / statl / 1222 < prev    next >
Encoding:
Text File  |  1992-07-30  |  884 b   |  22 lines
  1. Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
  2. Path: sparky!uunet!paladin.american.edu!auvm!LINC.CIS.UPENN.EDU!MATSUDA
  3. Posted-Date: Thu, 30 Jul 92 13:06:56 -0400
  4. Message-ID: <9207301706.AA25813@linc.cis.upenn.edu>
  5. Newsgroups: bit.listserv.stat-l
  6. Date:         Thu, 30 Jul 1992 13:06:56 -0400
  7. Sender:       "STATISTICAL CONSULTING" <STAT-L@MCGILL1.BITNET>
  8. From:         N o s t a l g i a <matsuda@LINC.CIS.UPENN.EDU>
  9. Subject: Re: interaction effects in regression models
  10. Lines: 10
  11.  
  12. Supposing that there are some grounds in putting the non-significant variable
  13. which turned out to interact with some other variable(s) in the model together
  14. with the interaction term, how would you deal with the multicollinearity that
  15. would surely ensue between that non-significant variable and the interaction
  16. term?
  17.  
  18. Kenjiro Matsuda
  19. Dept. of Linguistics
  20. Univ. of Pennsylvania
  21. matsuda@linc.cis.upenn.edu
  22.