home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!paladin.american.edu!gatech!usenet.ins.cwru.edu!news.ysu.edu!psuvm!auvm!BINGVMB.BITNET!LJONES
- Message-ID: <STAT-L%93010517091554@VM1.MCGILL.CA>
- Newsgroups: bit.listserv.stat-l
- Date: Tue, 5 Jan 1993 16:32:14 ECT
- Sender: STATISTICAL CONSULTING <STAT-L@MCGILL1.BITNET>
- From: Lary Jones <LJONES@BINGVMB.BITNET>
- Subject: Re: X^2 interaction
- In-Reply-To: Message of Tue, 5 Jan 1993 11:36:29 CST from <EJOHNSO3@UA1VM>
- Lines: 45
-
- On Tue, 5 Jan 1993 11:36:29 CST <EJOHNSO3@UA1VM> said:
- >I'd appreciate your insight in interpreting the results of a 2X3 contingency
- >table.
- >
- >I've run the standard chi-square test with significant results (p<.01), but I
- >see this as analogous to an omnibus ANOVA--it tells me something is happening
- >among these six cells, but it doesn't tell me what; therefore, my question is
- >whether there is anything like a test for main effects or interaction for a
- >chi-square? ...
-
- Chi-Square for a contingency table is an omnibus statistic, similar to
- the "Explained" term which is sometimes displayed in ANOVA tables. I
- remember seeing procedures for computing the "main effects" -- which
- is really testing the goodness of fit of the marginal (one way)
- frequencies. I am going to guess that this may not be of interest.
-
- There is a procedure for multiple comparisons. The only reference
- that I have at hand is an old one:
-
- Ryan, T. A. Significance tests for multiple comparison of proportions,
- variances and other statistics. _Psychological Bulletin_, 1960, _57_,
- 318-328.
-
- You should get additional verification for the appropriateness of this
- procedure.
-
- The basic idea follows that for SNK and other procedures working with
- tables of means. One decomposes the 2 by N contincency table into a
- series of 2 x 2 comparisons. This makes the most sense when the two
- category dimension is something like Agree/Disagree, and the other
- dimension represents "groups".
-
- The entries in the table are the Chi-Square values for the 2 x 2
- comparisons. The rows and columns are the categories of the larger
- dimension ordered by the proportion falling into a target category of
- the 2-level dimension (i.e., "agree" in the present example). These
- Chi-Square values are then compared to Ryan's table of critical
- values. For the problem you describe, the critical values are
- X(2)=5.76, X(1)=4.54 . (The value in parentheses is the range tested
- -- largest-smallest is 2, adjacent levels are 1).
-
- Hope this helps,
- __________________________________________________________________
- Lary Jones...Computing Services...Binghamton Univ...(607) 777-2879
- School of Management Research Support ...... LJones@BingVMB.Bitnet
-