Michael Wood wrote (excerpted):
Joseph W. may want to consider dropping column 5 which has only one, unique
observation, which will allow the use of Stata's -symmetry- command.
[output edited]
The symmetry model fits, albeit poorly; the marginals are different. The main
source of difference for the symmetry chi sq is respondents (n=5) who reported
3 at time0 and 4 at time1, versus those (n=18) who responded 4 at time0 and 3
at time1.
--------------------------------------------------------------------------------
I believe that you can pad with zeroes in order to avoid throwing out data (see
below). Bowker's test doesn't account for the natural ordering of the scores.
When you do (by specifying the -trend- option), you get the same result as for
-mvtest- or -ttest-.
There might be something in the asymmetry of the self-reported health status
scores that Michael describes; knowledge of content would be crucial to any
interpretation here. I naively would be more inclined to be guided by
-ordplot-, which showed the two plots as essentially superimposable.
Joseph Coveney
clear
set more off
input byte sco0 byte cou1 byte cou2 byte cou3 byte cou4 byte cou5
1 28 18 6 0 0
2 21 78 44 1 0
3 7 34 96 5 1
4 0 3 18 16 0
5 0 0 0 0 0
end
reshape long cou, i(sco0) j(sco1)
symmetry sco0 sco1 [fweight= cou], trend
exit
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
