if anyone’s done with quant and is doing schweser 2008 materials- pg 256 right at the end of quant, question # 2 (it’s a long vignette so would be painful to post it all)… if anyone really understands and can walk me through those 2 hypothesis tests and results, that would be great. i’m a bit confused. second question and this i hope will be very basic and easy for JDV or someone to answer- so looking at formulas and remembering old L1 stuff, the std error of an estimate = sq root of SSE / n-2 (apologies if parenthesis not in right spots, but hopefully we’re all on same page. ok, so now i’m chugging along doing a question at the end of the chapter that’s forecasting sales based on a mult regression model, b0 + b1(blah) + b2 (blah) + b3(blah) + E. and there were 180 months of data used (n = 180). so here I had to calculate SEE and the answer used sq root of SSE / n - k - 1 instead of the earlier n-2 basic denominator. so my basic question- how do I know when to use n - 2 or n - k - 1. i’m sure this is something pretty easy that i’m just not thinking about b/c i’m not all that great at stats, but if someone could illuminate when and why those are different, that’d be excellent. i finished the schweser quant stuff and the CFA quant stuff- haven’t done any qbank yet, plan to do a few problems soon. I prob will go to FI or derivatives from here since we have book 5 schweser now. I would think that after that book, it’s FSA for me around late Dec or new years. any help on these 2 questions above would be much appreciated. thanks!

I believe n-2 was used when you are doing a simple regression (i.e. a single (k=1) independent variable), so that n - 2 = n - k - 1 = n - 1 - 1.

what a minute. how do u even have book 2? schweser says it wont be ready till mid november

it was book 1, my bad… but thank you for the n - k -1 part. makes sense now. if anyone has book 1 and has read til the end- they’re the very end review quant problems where I have the question.

As clarified above, the n-2 is for simple regression where we use Intercept (B 0) and slope (B 1). That means we are estimating only two co-efficients (intercept and slope). That is why n-2 (2 stands for number of parameters we are estimating). In multiple regression, we estimate intercept PLUS beta co-efficients (b1, b2, b3 … bk). In such a scenario we use n minus k (which stands for beta co-efficients) minus 1 (stands for intercept as there would be only one intercept irrespective of whether we use simple or multiple regression). Hope I understand correctly and this helps you. In case I am wrong, please correct me.

I have a very limited understanding of Quant at this point, but it has to do with the number of parameters you are etimating. I remeber Joey answered this question for me in Level 1, didn’t make sense to me back then but the CFAI text will clear it up for you. check pg236 of volume 1

FRM2cfa Wrote: ------------------------------------------------------- > As clarified above, the n-2 is for simple > regression where we use Intercept (B 0) and slope > (B 1). That means we are estimating only two > co-efficients (intercept and slope). That is why > n-2 (2 stands for number of parameters we are > estimating). > > In multiple regression, we estimate intercept PLUS > beta co-efficients (b1, b2, b3 … bk). In such a > scenario we use n minus k (which stands for beta > co-efficients) minus 1 (stands for intercept as > there would be only one intercept irrespective of > whether we use simple or multiple regression). > > Hope I understand correctly and this helps you. > In case I am wrong, please correct me. This is correct…