Hi,
With a cross-sectional regression, there are several diagnostic tests that I know how to conduct. For example:
- To identify heteroskedasticity, I use estat hettest
- To identify whether I have the correct functional form, I use ovtest
- To check the weakness of my instruments, I can run an F-test on my instruments
- To run the overidentification test for instruments, I use overid
- To check for normality in errors, I can look at a graph of the errors
I am currently using fixed effects estimation on panel data. I have 2 time periods, 3 years apart (so I don't think time series tests are necessary), with over 5000 observations. How would I run the above tests in this context? Can someone direct me to the correct commands and tests?
- Heteroskedasticity
- Functional form
- Instrument weakness
- Instrument exogeneity
- Normality of errors
With a cross-sectional regression, there are several diagnostic tests that I know how to conduct. For example:
- To identify heteroskedasticity, I use estat hettest
- To identify whether I have the correct functional form, I use ovtest
- To check the weakness of my instruments, I can run an F-test on my instruments
- To run the overidentification test for instruments, I use overid
- To check for normality in errors, I can look at a graph of the errors
I am currently using fixed effects estimation on panel data. I have 2 time periods, 3 years apart (so I don't think time series tests are necessary), with over 5000 observations. How would I run the above tests in this context? Can someone direct me to the correct commands and tests?
- Heteroskedasticity
- Functional form
- Instrument weakness
- Instrument exogeneity
- Normality of errors