xtprobit panel data prediction

[Kasper Nygaard]

Hi,

I'm currently working on a project with panel data. I need to calculate propensity scores by using a probit model.

When using "xtprobit" for my y and covarietes, I need to "predict", the values after running the probit model. However, when using "predict", the values are highly negative due to linear prediction. Which STATA command can be used to predict values from panel data (after xtprobit)?

I found this link, but it seems not to work anymore: https://www.statalist.org/forums/for...-with-xtprobit

Thanks in advance.

Best regards

[Phil Bromiley]

You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

There are subtle issues in predictions with panel logit and probit. If you don't want highly negative values, then look at the other options (see xtprobit postestimation and then look under Options for prediction). I don't see that xb is very interpret ble - pr or pu0 seem more likely to be what you want. These will be non-linear transformations of xb and so are likely to give you different match ups if you're using them to match data.