Hello,
I'm conducting OLS regression in order to find value drivers for abnormal returns. My model includes 6 different independent variables, and i'm forming 20 subsamples (observations). My null hypothesis is that X variable does not significantly impact dependent variable.
How should I test hypothesis when i get results that X variable was significant in 5 samples and not significant in the rest of observations. Is there some threshold for null rejection?
P.S. an article with some suggestions would be perfect.
Thank you in advance!