Hi! I am doing a research on the impact of dividend policy on the investors' rate of return. Now I have the following data set. Array I use Fama's market model: Rit=a+bRmt+e. (a is constant, b is coefficient, e is error). I need to test whether certain errors on certain date (the dividend announcement date) is significant from zero.Should I run the regression with 30 companies past 5 years' daily Rit (rate of return of company i, eg,. 2013-2018 daily data)and Rmt (rate of return of the market, 2013-2018 daily data), get the errors, then select the errors on the dividend announcement date, and test whether they are significant from zero? Or Should I run the regression only with the Rit and Rmt on the dividend announcement date (like Ri and Rm on 30/12, 31/03, 30/06, 30/09), Then get the errors and test whether they are significant from zero? And how to select the errors on the certain date?