Adding a variable in Quant

Does anyone remember this question? What were they getting at? What was the answer??

Adding an independent variable as a function of a dependent variable is a model misspecification error. Your regression results will be biased and inconsistent.

Is that what they were asking? What variable were they adding? I don’t remember how I answered this one. So frustrating.

Eh. Its one question. And just cuz you don’t remember how you answered it doesn’t mean you answered it wrong. I like AF because I already know how I did. I don’t know where you all get your memory from.

it was about whether you should take oil level or the difference in oil prices. Level of oil is not covariance stationary so if you take level then statistical results will be invalid, I picked C

ok, I’m talking about another question then - I thought there was one where you had to add a variable. In the oil level question I got the same answer you did. 1 95% Confidence Interval - upper bound was either -1.03 or -0.48 2 Misspecification - effect of making change of model 3 T-statistic (2.1 or 0.9) 4 Dummy Variables - use 2 5 Correlation coefficient -0.416 6 Effect of adding new variable??? I’m talking about question number 6 in this list. I think that you’re talking about #2 right now.

you had to find the value in next period, don’t remember the exact value, could have been B

No. #6 was the plug n chug.

“I like AF because I already know how I did. I don’t know where you all get your memory from.” wait, you like the forum because you already know how you did?

**Correction*** I have an idea of how I did.

Guys, i do not recall seeing any question on adding a variable… I had a weird plug question with JUne sales for change in oil sales given and asked us to use employment from May data in the above regression etc… Could they have done something funky like this?

this is the same question we are talking about, inflation in June

financial_novice Wrote: ------------------------------------------------------- > Guys, i do not recall seeing any question on > adding a variable… I had a weird plug question > with JUne sales for change in oil sales given and > asked us to use employment from May data in the > above regression etc… > Could they have done something funky like this? I had the change in oil prices as well, nothing on adding a variable. Maybe others had slightly different exams… after all we are the only two peeps who saw “cash on hand” on the eps after share buyback problem.

RStewart, it seems i will see you again next year either on L2 or L3… :slight_smile: