Date: Wed, 8 Sep 1999 17:19:35 -0400
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
Subject: Re: Comparing two chi-square tables
Content-Type: text/plain; charset=us-ascii
This is a followup to the reply to Michael Kruger's question on comparing
two chi square tables of the same size.
The response was this:
> It has been a while but I did what you want to do during my MSc thesis.
> I remember correctly, you use the chi-square test as you would for any
> 2x2 table except that instead of calculating an expected table to compare
> the observed, you compare the two observed tables. If the two tables are
> very similar the chi-square statistics will be small as is the case
> classically when the observed 2x2 does not differ from the expected.
I am not quite clear on what is being recommended. Subract one table from
the other on a cell by cell basis to make a new table and compute a chi
square on that basis (Would the df be found by the usual formula)? Subtract
the two chi squares (but what then is the df for the difference)?
Could someone elaborate on this reply a bit so that i could see how the
recommended procedure would be implemented?