Teffect psmatch with sample weights

[John Matrix]

I am using propensity score matching on data that requires using sample weights. The data comes from the medical expenditure panel survey. I am using teffect psmat h code but it is not allowing an option for weights. Any suggestions on how to calculate this model with this type of data? Thank you.

[Andrew Musau]

The syntax definitely allows for weights.

teffects psmatch (ovar) (tvar tmvarlist [, tmodel]) [if] [in] [weight] [, stat options]

What did you try?

[John Matrix]

I tries it, and it did not work. I for the following error message:

. teffects psmatch ( ttlpx ) (time age female white) [weight=weight]

(frequency weights assumed)
may not use noninteger frequency weights

[Andrew Musau]

Sampling weights are denoted pweights. See

help weights

So you want:

teffects psmatch ( ttlpx ) (time age female white) [pweight=weight]

[John Matrix]

Thank you, but it still it is giving me an error message:

teffects psmatch ( ttlpx ) (time age female white) [pweight=weight]
pweights not allowed
r(101);


Any other suggestions, please?

[Andrew Musau]

See George Ford's suggestion to look at psmatch2's documentation which has a discussion on sampling weights and some recommendations: https://www.statalist.org/forums/for...sample-weights

ssc install psmatch2, replace
help psmatch2

[George Ford]

can ononly use fweight with teffects psmatch. says so in help.

[John Matrix]

OK. Thank you.