Bivariate and Multiple linear regression

[Margarida Alves]

Hello guys,

I have cross-sectional data and I intend to perform a bivariate (linear) regression and multiple linear regressions with OLS with robust standard errors. I also want to do some main and specific tests, such as t-test, F-test, RESER, Breusch-Pagan and Chow test. In the multiple linear regression, I am aware that it makes sense to perform the statistic tests, but it makes sense to apply them, specially the specific tests, in the bivariate regression?

Thank you in advance.

[Carlo Lazzaro]

Margarida:
the issue is that bivariate regressions (ie, a set of simple OLS with the same regressand and one different predictor at a time) do not make much sense (even though they're often reported in technical journals).
That said, you can run the post-estimation tests you mention after bivariate regression; my foreboding is that many of them will let you down, as an OLS with one predictor only is, in all likelihood, misspecified.
As an aside, I would recommend -linktest- instead of -estat ovtest- (ie, the Ramsey's test).

[Margarida Alves]

Thank you so much for your help, Carlo!