Need help with Q's on Linear Models

chrishello · **Joined:** Sat, 26 Mar 2011 09:44:20 UTC **Posts:** 50

I have attempted the questions above, I just need some clarification on them. The question states that Bo is a known intercept parameter. Does it mean it could be any intercept parameter or an intercept parameter attained by the least squares method ( that I learned in my lectures.)

outermeasure · **Posted:** Sat, 11 Aug 2012 15:57:37 UTC

chrishello wrote:

I have attempted the questions above, I just need some clarification on them. The question states that Bo is a known intercept parameter. Does it mean it could be any intercept parameter or an intercept parameter attained by the least squares method ( that I learned in my lectures.)

Obviously $\epsilon_i$ being the symmetric 2-valued distribution $\epsilon_i=\pm\sigma$ gives a different likelihood distribution from the Gaussian distributed $\epsilon_i$ , so there isn't such a thing called "the maximum likelihood estimator of $\beta_1$ " and thus question (a) is nonsense.

$\beta_0$ being a known constant means you can't fiddle with it to fit the data, so it isn't really a parameter in the regression.

chrishello · **Joined:** Sat, 26 Mar 2011 09:44:20 UTC **Posts:** 50

Even a simple clarification of the q's can be of sooo much help.

I agree that it is an extremely ambiguous question but from my lectures and thinking about the below q's looks like ill treat bo as the known intercept parameter learnt in lectures.

Thanks from OZ.

S.O.S. Mathematics CyberBoard

Need help with Q's on Linear Models

Who is online