Statistics and Time Series.

Some criteria in evaluating results of a Regression Analysis.

Once you are done with your statistical modeling process, it is time to evaluate the overall soundness of your work.
• First, do the signs of the estimated coefficients correspond to what you expected them to be? In other words, do the signs of the coefficients match the theoretical postulation of your research?
• Second, can you claim that the relationship you established in your theory and also observed in the data is statistically significant?
• Third, to what extent the model explain variation in your dependent variable?
• Fourth, does your model comply with the 7 assumptions of the linear regression model?
• Fifth, test for normality of the residuals (Histogram of residuals, Jarque-Bera Test, Normal Probability Plot, you name it).
At we can counsel you and always check your work by running your regressions once again, and again. If we both (you and I) can get the same results anytime we do it, you might be more than “alright” with the statistical part of your research.

Leave a Reply