I am trying to prove that SUM(y_i - yhat_i)^2 / (n-1) is an unbiased estimator of sigma^2 in simple linear regression without an intercept: Y = B1X
I have tried everything without luck. My last attempt looked fruitful (for a while....). I figured out that SSE can be written as SUM(y_i^2) - b1^2*SUM(X_1^2). I tried to work from there without luck - I did not see how it could resolve. Is there a *trick* for lack of a better word I am missing?