In a regression analysis, suppose there is, in fact, no seasonal pattern to sales, and the trend line is estimated using dummy variables to account for seasonality. What effect would this have on the estimation?

Question

My guess is that it wouldn't affect the estimation....am I correct or missing something?

Answer 1

If there is, in fact, no seasonal pattern, then the parameters on the dummies ought to be zero (or very close) and insignificant. So, yes, they should not affect the estimation.

However, the dummies might be picking up something else besides seasonal variations. Unless you have boatloads of data, a few odd observations could very well trigger significant values on the dummies. So you would be attributing a seasonal pattern to some other unrelated phenomena. So, be careful.

I hope this helps