Q5
"Hi, regarding
part b) should the calculations in the solution stops at 9.1198 which I thought it's the "intercept term" asked in the question?
part c) will any credit be rewarded for stating "include the interaction term, and then assess p-value and std error ..."
part e) on 'full credibility of data,' would it be reasonable to argue in the direction of GLMM that there may be fewer electric vans, which could result in the predictor being not significant if there is a large number of cars or trucks?"
thank you in advance!
Comments
b) This is a new question for us and it probably should be worded differently. The current wording means it would be acceptable to stop at ln(u).
c) The intent here was for candidates to present their understanding of perspective plots. The current wording does mean an answer along your lines would receive full credit.
e) This question is focusing on the limitations of a GLM rather than asking for an alternate approach such as a GLMM. We know a GLM treats the data as fully credible so we're looking for a reason why this actually isn't likely to be a valid assumption. Here we have to know that EVs were relatively new in the timeframe of the GLM dataset and this means the EV category is unlikely to contain a sufficient volume of data to be considered fully credible.