Ans is C Will be back for Round 4 @ 8 after gym
you mentioned time series… oh maan!!!
right we can find that serial correlation is possible because the test yields inconclusive results. So we cant reject null of no serial correlation. Does p-value rule out heterskedacity since this is so small? I keep getting confused by this p… The t-stat seems way too high to me, which makes me skeptical.
lookin forward to round 4 later… any how we can def rule out serial correlation here. the DW test is inconclusive right?
The question isn’t asking us to rule out serial correlation, only if there is evidence to suggest it exists, and the answer to that is no.
so the only way we can get evidence of serial correlation is if it falls outside that DW test correct? Or is there another way to collect this evidence?
hey guys – this was a time series problem - and given that the dependent variable and independent variables are the same variable lagged / leading - there will definitely be serial correlation. SWG mentioned it in the top. So DW test is useless and meaningless.
If there is definately serial correlation, why is the answer C, heterskedasticity only?
durbin watson is not an evidence of serial correlation in this case at all. that is what is being tested.
You cannot use Durbin Watson to check for auto correlation; use correlation of errors with its lag divided by 1 /sqrt (number of observations) to see if residual has serial correlation.
I gotcha then… Kind of a weird question, but will not forget now then… Serial Correlation can not be tested in time series by DW (Ive read it a million times and still tried to do it)… Thanks CP
i did the same too… not a problem.
Q4. Qualitative dependent variables should be verified using A. Dummy variable based on the logistic distribution B. A Discriminant model using a linear function for ranked obervations C. Tests for heteroskedasticity, serial correlation and multicollinearity Q5. IND is a dummy variable where IND = 0 means stock from electrical utility industry IND = 1 means stock from biotech industry DPO = divident payout G = growth B = beta * represents - significance at 5% level Intercept----6.75-----3.89* IND----8.00-----4.50* DPO----4.00-----1.86 G----12.35-----2.43* B----(-0.50)-----1.46 Based on these results, it would be most appropriate to conclude that A. Biotech industry returns are statistically significantly larger than electrical utility industry returns B. Electrical utility returns are statistically significantly larger than Biotech industry returns, holding DPO, G and B const C.Biotech industry returns are statistically significantly larger than electrical utility industry returns, holding DPO, G and B const
Q4. Qualitative dependent variables should be verified using A. Dummy variable based on the logistic distribution B. A Discriminant model using a linear function for ranked obervations C. Tests for heteroskedasticity, serial correlation and multicollinearity A…logit, probit distrib … IND is a dummy variable where IND = 0 means stock from electrical utility industry IND = 1 means stock from biotech industry DPO = divident payout G = growth B = beta * represents - significance at 5% level Intercept----6.75-----3.89* IND----8.00-----4.50* DPO----4.00-----1.86 G----12.35-----2.43* B----(-0.50)-----1.46 Based on these results, it would be most appropriate to conclude that A. Biotech industry returns are statistically significantly larger than electrical utility industry returns B. Electrical utility returns are statistically significantly larger than Biotech industry returns, holding DPO, G and B const C.Biotech industry returns are statistically significantly larger than electrical utility industry returns, holding DPO, G and B const A. Biotech industry returns are statistically significantly larger than electrical utility industry returns
Shouldn’t Q5 be C?
^winna’. It’s C for both questions.
c for both?
Yea - I had a death attack too for Q4 to believe it to be C. Q5 was an easy C
verification … read the question dummy cpk…
Q6. Ok some L3 essay practice here. Kill it! So it’s believed in the model misspecification world that using a lagged dependent variable as an independent variable will misspecify the model and render it useless. On the other hand AR§ model goes and keeps adding lagged variables until and unless the autocorrelation of the residuals is not significant. Does that mean AR§ model is inherently misspecified?