Dear Stata colleges,

I have a problem with the comand teffects overlap. I have run two propensity score matching. They yield different coefficients but they give me the same predicted probabilities (using teffects overlap). I have 534 observations

Both of them are very easy. In the first model, I want to know if a type of course (TIPO) has an impact on Employment Quality (CALIDAD_EMP). Age sex education are using as tmvarlist. The second model is exactly the same but I want to know if the course has an impact on digital employment (EMP_DIGITAL). The models are the following:

teffects psmatch (CALIDAD_EMP) (TIPO age sex edu, logit)
teffects overlap
teffects psmatch (EMP_DIGITAL) (TIPO age sex edu, logit)
teffects overlap



COEFFICIENTS ARE DIFFERENT BUT THE PREDICTED PROBABILITIES ARE THE SAME

Thank you