Matched analysis for ordinal data

[Peter Horby]

Hi folks,

I have data from a matched case-control study (4 controls per case). Matching is on age and gender.

Using Epi-Info and SPSS I am undertaking univariate analysis using a Mantal-Haenszel matched odds ratio for variables with a binary response (yes / no).

However,one variable is ordinal - highest level of education (primary, secondary, high school, university). How should I do a matched analysis using one category as the reference category (e.g. University)?

Peter

 

[JohnM]

Peter,

This sounds like a very specialized question, beyond most of our abilities to handle (I dealt with Mantal-Haenszel once back in grad school)

I'm going to refer you to UCLA's Statistics Department - they have a free statistics consulting site - the profs there probably have more experience dealing with these issues:

http://www.stat.ucla.edu/freeconsulting

They're pretty good about responding and should get back within a couple of days.

John

 

[Peter Horby]

Thanks John, I'll try them.

Peter