Population Assessment of Tobacco and Health

Cross-sectional weights for analyses on trans/cis populations Tips for Using the PATH Study Data User Forum

Return to the Population Assessment of Tobacco and Health (PATH) Study Series page.

2 posts / 0 new
Last post
sawyeran
Cross-sectional weights for analyses on trans/cis populations

Hey everyone,

I have a question about using weights for analyses looking at differences between transgender and cisgender individuals. I know that it is suggested that those analyzing PATH data use the weights in analyses, but I have a concern because the weights include "sex." When looking at "sex" vs "gender identity" variables, people's responses aren't matching up in a clear way. It's about a 75/25% split between how people answer for "sex" (based only on those with binary transgender identities: MTF, FTM).

I've run the analyses weighted and unweighted and the estimates show quite a bit of difference, so I'm trying to figure out the best course of action for reporting the results - Should I use weighted, unweighted, or both in reporting? I don't have a ton of experience with weights, so I don't have a great idea of which things are most important to consider here. Any help is appreciated. 

Thank you!

Ash

Joy Jang
Hi Ash,

Hi Ash,

This is an interesting question - to what extent the inclusion of ‘sex’ variable in the weighting would affect the results of comparison by sexual orientation.

First of all, PATH Study recommends (actually requires) the use of weights to compensate for selection, nonresponse rates, and possible sampling frame deficiencies (p.37 on PATH Study Public Use Files User Guide). So, (if I may jump to the conclusions) I would report weighted estimates.

I have experiences in which weighted and unweighted estimates were quite different. That’s because of the complex sampling design, probably not because of the inclusion of a particular variable (in your case, sex) in the weighting. I think using weights with and without sex would not make a huge difference. I once created my own weighting variable (i.e., attrition, sample selection) with MTF, and adding one variable or dropping one in weighting did not particularly affect results. ** But it’s possible that your case is different. If you want to check, I would contact Westat who originally creates the PATH weighting variable to get some advice to create your own weighting variable (without sex). That way, you may be able to better understand what’s going on with your data/analyses.

One more thing I wanted to highlight is that as a journal reviewer, I sometimes see researchers do not weight their data and say that their findings are from national data and thus representative/applied to general population. Unless you weight the data appropriately, that is not correct - you are just using another selective/convenience sample.

I’m relatively new to PATH Study (recently start to look at the data). Others may have better insights on your case. Hope to see how others think..

 

Thanks,

Joy  

Log in or register to post comments