Panel data regressions

[Jane McGrath]

I am running regressions for 100 companies with 10 years of data each. I am wondering what are the panel data specific prompts I need to include to make sure the regressions are correct?

For example, I have been using xtreg and ,fe in each regression. Are there any other panel data specific regression tools I need to use? Such as i.timed? Or is this not necessary?

[Jane McGrath]

Any help would be greatly appreciated

[Nick Cox]

I'm not a panel expert, which is why I didn't answer. Immensely more important is why no panel expert answered this, and there are hints in
https://www.statalist.org/forums/help#noanswer

[Jane McGrath]

Hi Nick, I am struggling to read between the lines. Can you please help me understand your answer

[Nick Cox]

There isn't a way of spelling it out that isn't at risk of seeming patronizing or obnoxious. You have asked several questions in the last few days as someone new to Stata and Statalist. All the questions that are concrete and specific got quick, concrete, specific answers. So far, so good. This one didn't. Why not?

Statalist is a forum, not a helpline. The difference is that on a helpline there is an obligation that a question gets an answer. On a forum, no one is obliged to answer anything. Anything that looks like "What do I need to know and what should I do next in my project?" is at risk of no answer.

[Daniel Schaefer]

I'm not a panel expert, but based on what you've said here, xtreg seems like it should be an appropriate model. You might want to use a Hausman test to determine whether re or fe is more appropriate. I don't know what you mean by "panel data specific prompts." You might want to look through the [XT] documentation for this. I also have no idea what "i.timed" refers to in this context.

I considered trying to answer this shortly after you posted, but it wasn't clear to me what kind of help you needed, and I didn't want to risk rambling unhelpfully about panel data analysis broadly. Hope that helps.

[Carlo Lazzaro]

Jane:
first off, why are you writing about "regressions" instead of a singe panel data regression with N=100 and T=10?
In addition, and in my opinion, much more relevant to get positive replies, please, as per FAQ, share what you typed and what Stata gave you back. Thanks.