An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-sig-teaching/attachments/20121003/ae97dfe0/attachment.pl>
Interpreting the results
5 messages · Ali Zanaty, Jeff Laux, Paul Johnson +1 more
An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-sig-teaching/attachments/20121003/66daece1/attachment.pl>
There appears to be a little mistake in your code
On Wed, Oct 3, 2012 at 3:57 PM, Ali Zanaty <zanaty2005 at yahoo.com> wrote:
Dear All: I have two variables: X (nominal categorical) has 9 categories. Y(ordinal categorical) has 4 categories. 1-I fit a logistic regression using Y as the response variable and X as the independent variable. The P VALUE was 0.2373. This tells me that the independent variable X is NOT significant predictor.
if X is nominal with 9 categories, the fitted regression should include an intercept and 8 categories. If it is only fitting one predictor, that means your X is not properly understood as a factor variable. If y is ordinal, then fitting the model requires use of some addon package, not something provided with R. You'd need polr, ordinal, or lrm from the rms package. So I think you don't have what you think you have, and you'd better double check and then post the full code for the model along with the output of str() on your data frame. pj
2- I run a
chi-square test to test if there is a relationship (association ship) between
both X and Y variables. The P VALUE was 0.0013. That is there is statistical
evidence of association between the two variables.
3- I collapsed the 4 categories of the Y-variable into 2 categories. I run a
chi-square test to test if there is a difference(s) among proportions of the X-variable (the equality of the 9 proportions). The P VALUE was 0.0037. That is there is statistical
evidence of differences among the 9 proportions.
How
I can explain results of one versustwo, and one versus three. I am not sure if there is a contradiction among these results.
many
thanks
Ali
[[alternative HTML version deleted]]
_______________________________________________ R-sig-teaching at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-teaching
Paul E. Johnson Professor, Political Science Assoc. Director 1541 Lilac Lane, Room 504 Center for Research Methods University of Kansas University of Kansas http://pj.freefaculty.org http://quant.ku.edu
An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-sig-teaching/attachments/20121004/aae1f1aa/attachment.pl>
An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-sig-teaching/attachments/20121004/ea5184a6/attachment.pl>