computing age-specific fertility rates using tfr2 command and matching weights

[Ece Kafali]

Hi !

I am using a DHS (birth history) dataset and computing age-specific fertility rates using tfr2 command. This command uses DHS sampling weights by default.

 tfr2  [pweight=v005], len(3) ageg(5) bvar(b3_*) dates(v008) wbirth(v011)

I am also using propensity score matching and computing matching weights using this command:

 psmatch2 depvar [indepvars], odds logit

I want to recompute age specific fertility rates using the matched sample. To do so I used the same command (tfr2) with matching weights.
I'm not sure if using matching weights instead of sampling weights is okay or if it is possible to somehow employ both weights.
Any recommendation would be greatly appreciated.

Thank you in advance.
Regards,
Ece

[Steve Samuels]

Your question is addressed in this thread. See also this web page, which suggests that propensity score matching is a bad idea. Instead, the authors recommend matching on Mahalanobis distance or coarsened exact matching (cem). Follow by a tfr2 analysis that ignores the matching.