I'm currently doing a project in using Econometrics, I am using Stata to regress my figures. I currently have 300 panel data observations over 15 periods, ie 20 observations per period. I have estimated my model using the following methods:
- Pooled OLS
- Fixed Effects
- Random Effects
- First Differences
However in all my models my P value is no where near even the 10% significance level. The smallest I can get the value is when using a Fixed effects model with a p-value of around 0.3.
I am trying to test this significance, so yielding an insignificant variable won't really help my project. What can I do to resolve this?
My project is a study on team performance in English Premier League Soccer and I am specifically trying to study whether English players(measured using the ratio of English player values to total club value) have an impact on team performance(measured by points per season)
Econometrics - Insignificant Variable Problem Watch
- Thread Starter
- 05-02-2016 20:01
- 16-02-2016 10:50
To improve the significance you can either include more variables (i.e. more data) so that you are overcoming the omitted variable bias issue, or increase your sample size (i.e. more data) so that the regression can pick out more of the variance. Unfortunately it's probably not going to be an easy task to get additional data (otherwise I presume you would use it in the first place).