Hi all,
I'm doing some baseline regressions without adding any controls. Just try to see a bit and add elements step by step. I have a huge panel data, where my panel variable is job posting (each record）, and my time var is year-month.
Here is my regression result with and without different FEs:
I notice my F stats is huge. I consider different possibilities I can find on the forums:
1. "If a large F appears with skyrocketed R2, it suggests the model is over-fitting", here the adjusted R2 looks normal
2. "If a large F appears with huge t statistics of some independent variables, it means there are omitted variables which cause endogeneity", here the t-stats of the independent variable are small
3. Large F can also be caused by different unit measurements of different variables on both sides of the regression, but here they are both shares ([0,1])
So I still cannot figure out what might be the problem. Related to 3, I checked my independent variable distribution, it has a very high kurtosis (108.685), and the skewness is 8.879555 (not sure if this is fine). Could this be the problem?
Can anyone help me what might cause the huge F stats?
Thank you!
I'm doing some baseline regressions without adding any controls. Just try to see a bit and add elements step by step. I have a huge panel data, where my panel variable is job posting (each record）, and my time var is year-month.
Here is my regression result with and without different FEs:
I notice my F stats is huge. I consider different possibilities I can find on the forums:
1. "If a large F appears with skyrocketed R2, it suggests the model is over-fitting", here the adjusted R2 looks normal
2. "If a large F appears with huge t statistics of some independent variables, it means there are omitted variables which cause endogeneity", here the t-stats of the independent variable are small
3. Large F can also be caused by different unit measurements of different variables on both sides of the regression, but here they are both shares ([0,1])
So I still cannot figure out what might be the problem. Related to 3, I checked my independent variable distribution, it has a very high kurtosis (108.685), and the skewness is 8.879555 (not sure if this is fine). Could this be the problem?
Can anyone help me what might cause the huge F stats?
Thank you!
