would you employ a regression model if the intercept was statistically insignificant? Can it be ignored? F-stat and X1 are significant.
All kinds of places where the intercept ought to be 0. Drop a ball and measure distance fallen vs time. If I fit the regression Distance = a0 + a1*t + a2*t^2, I will bet that: a) My F-stat is significant b) My t-stats for a1, a0 are not significant. c) My t-stat for a2 is significant if I don’t do a really terrible job of measuring and get enough data.
Thanks Joey
If you decide not to use the intercept because it is insignificant, that’s fine, provided that it is consistent with the theory your model is based on. However, you should reestimate your other coefficients with the intercept set to zero; don’t just use the original coefficients and just ignore the intercept. If you do that, it will throw off your standard errors.
Good point, as well, Chad - thanks.