Two wave panel data analysis

[Antonia Diaz-Valdes]

Hello all,

It is my understanding that I can only do lagged models or change score models when using two waves (or depending on the type of analysis, t-test, ancova, and so on). So my question is, would it be appropiate to run a fixed or random effect model for two waves?

Thanks you!

[Carlo Lazzaro]

Antonia:
I beg your pardon, but I find your query a bit unclear.
If you have a panel dataset you should consider -xt-prefixed commands first and if there's no evidence of a panel-wise effect, you should switch to a pooled regression (t-test and ancova seems off target with panel data).
Hence, if your concern is about performing a panel data regression with two waves of data, this is pefectly legal.
Which specification (fixed or random effect, if there's evidence of panel-wise effect) fits your data better than the alternate option, should be investigated (eg, via -hausman-).
Without knowing further details about your data, I find difficult to reply more positively.

[Antonia Diaz-Valdes]

Thank you Carlo,

Yes, I did not express myself correctly. But you got it and get me the answer I was looking for. Thanks for your answer.