Durbin-Watson help me.

A questions gives critical Durbin Watson stats as 1.59 and 1.73 The regression durbin watson stat is 1.9566 The explanation sais We fail to reject the null of no positive serial correlation (great i agree with this). But then the selected correct answer is: B-significant serial correlation is present… Can someone confirm that the correct answer should be, serial correlation is not present?

Schweser page 199, you have a nice Decision rule summary what if you have a dl of 0.56 and a du of 2.21 this “mental image” no longer works, your decision ranges would be messed up…?

1.9566 is greater than the upper bound of 1.73, so we reject & say that serial correlation exists, specifically negative serial correl.

Andrew, you use Schweser, look at the decision rule on page 199 above d-upper is Do Not Reject talk to me via email if you can, my email got hacked today email me to adzt@samedomainasmyhackedemail.com

AndrewUNH Wrote: ------------------------------------------------------- > 1.9566 is greater than the upper bound of 1.73, so > we reject & say that serial correlation exists, > specifically negative serial correl. Unfortunately, you are wrong.

gulfcfa Wrote: ------------------------------------------------------- > Schweser page 199, you have a nice Decision rule > summary > > what if you have a dl of 0.56 and a du of 2.21 > > this “mental image” no longer works, your decision > ranges would be messed up…? Don’t go with the mental image. Draw a line and do the calculation of (4-Dl and 4-Du). EDIT: The reason why this is a Fail to Reject is because the test value falls between the FTR H0 range of 1.73 to 2.41

^Boss, so does that meen am I right?

Iginla2010 Wrote: ------------------------------------------------------- > gulfcfa Wrote: > -------------------------------------------------- > ----- > > Schweser page 199, you have a nice Decision > rule > > summary > > > > what if you have a dl of 0.56 and a du of 2.21 > > > > this “mental image” no longer works, your > decision > > ranges would be messed up…? > > > Don’t go with the mental image. > > Draw a line and do the calculation of (4-Dl and > 4-Du). thats what i was going for, but if you have a dl of 0.56 and du of 2.21 that line gona look messed up …?

Why messed up? Do 4-0.56 = 3.44 and 4-2.21 = 1.79 Then look at what your test value is and plot it on that line. If it falls between 1.79 and 3.44 (which it does in this case) you conclude: Test not conclusive.

It seems as if you have stated something incorrect. This should be do not reject the null which means statistically there is no positive serial correlation.

well i say meed up cause like in the typical line you see it goes in this order 0, dl, du, 4-du, 4-dl with 0.56 and 2.21 you gona have dl=0.56 du=2.21 4-du=1.79 4-dl=3.44 on the line you would have 0, dl, 4-du, du,4-dl do the 4-du, and du got switched around and i dono how to interput the ranges

Mike_alts Wrote: ------------------------------------------------------- > It seems as if you have stated something > incorrect. This should be do not reject the null > which means statistically there is no positive > serial correlation. Did you at least spend time to read ALL the posts? The answer to Gulf’s first question is Fail to Reject (read my post 4 posts above) The answer to his second question (where the limits are 0.56 and 2.21) is INCONCLUSIVE. Get it?

thanks guys…I was misinterpreting it…don’t they often give us the DW stat and then give you a DW table and say is it significant or not, therefore eliminating this exercise above?

Iginla2010 Schweser has these rules DW

“thanks guys…I was misinterpreting it…don’t they often give us the DW stat and then give you a DW table and say is it significant or not, therefore eliminating this exercise above?” lol, tell you boss oh i dono how to do this, they usally give us if it is significant or not

------ | X | O | X |--------- + dl du 4-du 4-dl - | inconclusive | No CR | inconclusive |

shit, format is messed up

Never mind Iginla2010 I think I get it the ranges wont overlap its just that the do not reject range will not exist in this case it will always be inconclusive or reject…agree ? (in the case of 0.56 and 2.21 )

Also while awaiting the answer from Iginla2010, the rest of you dont bother with this thread you will only get a val of du above 2.21 when k=5 and samples less than 20 Something I doubt you will see on CFAI exam, so the range on page 199 will work just fine.

gulfcfa Wrote: ------------------------------------------------------- > with values of with 0.56 and 2.21 > > you draw the line and you get over lapping ranges Hmmmm, I agree. I just re-arranged the values on the line. I think you are referring to a case where K = 5 and n = 15, right?