Central limit theorem, panel study

In summary, the conversation discusses the use of multiple linear regression in a panel study with a time dimension. The question is raised about the normal distribution of residuals, which is a requirement for the regression model. The expert notes that the central limit theorem cannot guarantee this and that the errors may or may not be normally distributed. Additionally, the fact that the data is measured repeatedly over time poses a potential issue with correlated errors, making it more of a time series problem than a simple regression problem.
  • #1
monsmatglad
76
0
I am doing a panel study with multiple linear regression.
When I want to make sure that the residuals are normally distributed, as is a requirement for the regression model, can I assume so due the Central limit theorem (given the size is sufficient)? Or does it not apply when there is a time dimension?
The study is based on 73 companies with variable values once a year for 6 or 7 years.

Mons
 
Physics news on Phys.org
  • #2
The central limit theorem will not guarantee this; nor will anything, really. Given an appropriately chosen regression model, the errors may or may not be normally distributed. If they're not, then you'll have to either transform the data in some way, or fit a regression model with a different error distribution (which takes a fair bit of expertise).

A bigger problem is the fact that your errors will almost certainly be correlated, since companies are being measured repeatedly. This is really a time series problem, not a straightforward regression problem.
 
Back
Top