Using inverse probability weights to increase balance among covariates

[Karishma DSouza]

Hello. I'm running a difference-in-difference model, and so have two groups. I'm using inverse probability weights (ipw) to increase balance amongst covariates. I have to create summary statistics and want to show better balance between the two groups after using ipw by showing that the difference between the two groups reduces and/or is not statistically significant after weighting. What tests can I run to show this?

[FernandoRios]

Perhaps you can use -teffects-post estimation command -tebalance-

webuse cattaneo2,clear
teffects ipw (bweight) (mbsmoke mmarried c.mage##c.mage fbaby medu, probit)
tebalance summarize
*or 
tebalance density mage
tebalance density medu

you could also use -pstest- which comes when you install -psmatch2- (a community contributed command).
Fernando

[Karishma DSouza]

Thanks, Fernando. I see I wasn't clear in my earlier question. I'm using a GEE model over here and have created the weights myself. I want to show balance using the weights at the summary statistics level using only the pre-intervention period data. I'm not sure if teffects which gives me balance on average treatment effects is the way to do that. Or maybe there's something I'm missing here?

[FernandoRios]

Hi Karishma
In that case, perhaps your best bet is to use pstest. Since you can add information of the previously estimated IPW.
I don't think teffects (or more precisely tebalance) will allow you to do that.
Fernando