Casey, are you sure you need the weights? If they are survey weights, you're estimating a causal effect, and you're using one of these models and so aren't allowing the causal effect to vary, then you can use the weights as a control (if they are pre-treatment), but in most cases they aren't necessary for eliminating bias at least. If instead you're estimating some descriptive quantity (average of the variable), then weights can be crucial of course; but you're using ordinal logit, so I'm guessing that this isn't what you're after.

Gary
---
http://gking.harvard.edu

On 03/31/2009 11:17 AM, Casey Klofstad wrote:

Hay all-

I've see this issue addressed in a few previous posts, but I'm still unclear.

The long and short of my situation is that I did full matching in R. I
need to use ordered logit for my analysis, but as of right now Zelig
does not support the use of weights. So, I performed the analysis in
Stata with Clarify. The models come out fine, but I can't generate
substantive interpretations of the coefficients becuase simqi does not
appear to support the use of weights when generating predicted
probabilities.

Thoughts?

Best,

-c

-- Clarify mailing list served by Harvard-MIT Data Center [Un]Subscribe/View Archive: http://lists.gking.harvard.edu/?info=clarify