Interpretation of omitted variable bias

[Torben Lieber]

Hey everyone,
I did a few regressions between different subgroups to check the significance of one exogenous variable, including some control variables. 3 of my outcomes for the exogenous variable were statistically significant.
After I did the regression again without my control variables just two of my outcomes were significant. Both of them were a part from the 3 significant outcomes earlier.
Now I'm having trouble with the interpretation. I'm guessing it is a hint for O.V.B. but in what way and how can I test it?
I used the "e(sample)" command to control for the number of people observed.

Thank you in advance,
Torben Lieber

[Carlo Lazzaro]

Torben:
you may want to take a look at -estat ovtest-

[Torben Lieber]

Hey Carlo,

unfortunately it is not possible without my control variables, because all my explanatory variables are indicator variables now.
When I'm including my control variables no sign of O.V.B is found.
I'm having troubles mostly with my interpretation.
Why do just two samples have a significant outcome and not three?
Are the variables correlated with my endogenous variable or with the observed exogenous variable and how can I test this?

Thanks again,
Torben

[Rich Goldstein]

your question is completely unclear to me; please read the FAQ, especially the sections on how to ask questions and how to post both commands and results, and then re-post