-Book 1, Page 257, Self Test:Quant, #5 I was hoping a fellow Schweser user would be able to clarify a question for me. The answer they have listed in the back of the book is B which I understand. I was wondering why it couldnt also be C (Positive serial correlation). I might be misunderstanding the application of the Durbin-Watson Test but here goes my thought process. The sample size is large because n=180 so DW=2(1-r). r for this would be 211.6/357.2=.592385 making the DW stat .81523. The DW Critical Values would be around DL=1.61 and DU=1.74 (Tables only go up to 100 in the back of the book when DOF is 180 and k=3). Due to the DW stat being less than the lower boundry, this is evidence of position serial (auto) correlation which is what I originally thought. The more I look at it the more I feel I’m missing something obvious, so any help/insight would be greatly appreciated
as far as I see it, if the test-stat is less than D(lower) critical val, then it falls into the left-hand region of the 0-4 scale. (ie 0 to 1.61), so it’s positive serial correlation. I don’t think there is anything more to it. Also I notice that generally the questions have a large sample size - so we can avoid the nasty DW test-stat calc and just use DC = 2(1-r) instead - thankfully!!
sorry I meant “DW=…”
this is troubling me now… R-sq is moderate: 0.59, but F-stat is very high: 85. (there is a typo in the answer - should be Q3 instead of “Q9” I presume) But none of the slope coefficients have strong test-stats - therefore not individually significant as specified --> therefore model is mis-specified - due to some correlation between coefficients. Hence B. so far so good. also - looks like you calculated DW-stat using R-sq instead of r. R-sq = RSS/SST = 0.59. So r = root R-sq = 0.768. Therefore DW-stat = 0.463. However, the questions provide extracts of the tables needed + sometimes they include red herring tables as well. Since neither the DW table nor the DW critical vals are provided, and since B is a strong answer with the info given and the tables provided, stick with B. Not totally satisfactory, but it works…
Thanks for the clarification, it helped a lot and makes more sense. Like you said I guess we assume that since they have not given us the DW Table or the critical values, we don’t worry about it.
Sorry to be a little late in asking this… but did u get question 2 from this self test correct? I have a doubt in the first part i.e. Hypothesis 1 … the calculation for tstat is done as 11.2/2 = 5.6 … shouldnt it be 11.2/3.5(the std error of DOL?) … I havent understood where the 2 came from… clarification would be greatly appreciated.