| Topic Name: |
This GMAT thing... |
| Message Name: |
All right |
| Date Posted: |
04/18/2001 |
| In Reply To: |
Yabay, I know we are schoolmates (HBS) but I believe that at your time you did not have foundations, so you skipped quant methods. Causal vs. casual correlation is a big theme there. It is everywhere in the literature also. The fact that two variables are correlated may mean just that they respond to similar underlying mathematical structures (eg. similar distributions, with similar parameters affected by some third variable) but still there might not be a cause-effect relation between the two. Let's suppose, A: growth of grass and B: number of umbrellas sold. They are both affected by the amount of rain, right. So, if you run a correlation test, A is correlated to B, lets suppose highly. Now, it is obvious that umbrellas do not make the grass grow or viceversa. They are mathematically correlated, but there is no causation. Examples of this are all around the stock market. Many stocks correlate with each other, but they run completely unrelated business models. The effect is still useful for hedging, as you know (the correlation part), but it is hardly useful to construct a model of any given firm's future earnings, based on another's (the causation part).
I was not "imagining" any causation between GMATs and GPAs by the way. Actually my message could be read as arguing on the contrary, I think you should read it again. I was just arguing against the idea that GMAT is a random test that does not correlate with IQ.
Cheers,
Fellow member, class of 2001. |
| Message: |
You are correct, we did not have Foundations QM. We had a full course on the subject (and not just three weeks), called "Managerial Economics." or ME. I believe it ran for 45-50 sessions first term FY. Now, even that is gone away, since they've folded QM into FY Finance, which will make for better teaching, but much worse depth and scope.
Your original post makes no sense. This post makes a lot more sense, since you seem to know the difference between correlation and causality (not "causal correlation", there is no such thing).
If your point is that the GMAT scores correlate positively with IQ scores, I agree too. However, the correlation is most certainly imperfect, and I'm not willing to make statement as to how correlated they are. As you know, the very valid argument could be made that this correlation is spurious (casual, in your "terminology"), since GMAT scores and IQ scores are also positively correlated with wealth, where you went to high school and even zip code. To try to disentangle all this, and to really ascertain the relationship between GMAT scores and IQ scores requires regression analysis and partial correlation analysis, which wasn't covered in QM, but was in ME. Even worse, regression analysis won't give you a measure of causality, after all, it is still correlations. For causality, you need some serious econometric firepower (instrumental variables, 2-stage least squares), that was definitely not covered in ME either.
By the way, who was your QM prof? I ought to give him/her (if it was Frances) a spanking.
Cheers,
Yabai
|
|
