<p>
[quote]
Sakky, don’t be pedantic. I’ve known more about modeling than you can possible imagine. Not sure? Do you want to compare our CV?
[/quote]
</p>
<p>Absolutely. PM or email me and we can do an exchange, if you insist on getting into a contest of pure academic machismo. </p>
<p>Besides, that message has to do with the people who continue to misconstrue the use of models. Not necessarily you, but others here. They demand perfection in modeling, when the fact is, no model has ever been perfect. The question is not whether a model is perfect, but whether it gets you closer to the truth, relative to whatever else is available today. I believe that Hoxby is at least as useful as USNews, and arguably more useful. For example, Hoxby is one of the few ranking systems that I know of that actually compares RU's and LAC's in a unified scale. </p>
<p>
[quote]
Indeed, there may be a refined math model that is more directly applicable to this. A more robust model 1) that can account for corrupted, imperfect data set, and 2) that can account for un-modeled phenomenon. This robust math model may yield a much better result even in the presence of imperfect data set. Alas, the simple model (in the paper) is not it!!!
[/quote]
</p>
<p>Nobody is saying that this is a 'perfect' model. Of course you can always refine the model. But that's not the point. The point is, I am not aware of a better paper out there that deals with this subject. If you are aware of one, please present it. Furthermore, I have no reason to believe that the Hoxby article is any less useful than any other ranking system out there, i.e. USNews, Gourman, or what-have-you. Maybe the Hoxby model is too simplistic and doesn't capture every nuance. But * neither do any of the other rankings out there *. What sort of modeling did USNews do?</p>
<p>What I am saying is that it's all relative. The Hoxby model is not perfect. No model is. If perfect were what was required, nobody would ever be able to do anything in academia. The real question is not whether the model is perfect, but whether it's better than anything else we have available, at least as a first approximation. Newtonian classical mechanics as a model in physics is not perfect, and never was. But for hundreds of years, it was used as the best available model in physics until QM and relativity were discovered, and to this day, Newtonian mechanics is * still * used as a useful first approximation. I am sure that some day, we will find that QM and relativity are "incorrect" and some new model will be discovered. But until that day, we should continue to rely on the current models.</p>
<p>I put the ball in your court. What model out there is better than the Hoxby model? </p>
<p>
[quote]
I believe the sample is not representative of the underlying population, which includes non-applicants. All the people who like BYU applied there, and they would and did prefer it . Their preferences were revealed. The underlying population include a far higher proportion of people who would not apply to BYU, and would not attend it if they were admitted vs. anyplace else. The preference of these non-applicants was not adequately revealed, in my opinion, judging from the results. Because the sample of applicants does not perfectly reflect the behavior of non-applicants. Whatever these equations say, this nuance was not captured appropriately in the results, which imply a preference for BYU in the underlying population of college applicants that is miles ahead where, IMO, BYU is really preferred by the mass population of applicants at large.
[/quote]
</p>
<p>First off, the study never says that the model is representative of the population at large, if by that you mean every single student in the country. The study explicitly stated that their data was from a group of top high school students. I think that actually makes the model better. After all, I don't want to sound harsh, but I am not particularly interested in the preferences of some of the irresponsible, unmotivated kids that I went to high school with. I remember one of them loudly proclaiming that he just wanted to find the college that will allow him to smoke as much weed he can, and so he was seriously looking at going to college in Europe, especially in the Netherlands. Do you think that his "preference" is a preference that others should care about? I'm sure that some of the Playboy Magazine Top Party Schools are highly preferred by many students, but do you think that that's a meaningful preference to model? The truth of the matter is that plenty of American kids are not particularly motivated to learn. They go to school to have fun on their parent's dime. </p>
<p>Secondly, I still don't see why it is so unrepresentative that BYU should be preferred highly. I don't see why it's hard to believe that many of the most motivated students in the country happen to be Mormons. Mormons are very hard-working people. Certainly a heck of a lot harder working than many of the kids back in my high school. </p>
<p>But more importantly, like I said above, if you don't Hoxby, what else do you got? USNews? Gourman? Like I said, the measure of a model is not perfection, but whether it's better than anything else available. If you have no good alternative, you have to go with what you got until such time as somebody creates an alternative. I'm sure that every single thing we know about science today will be proved wrong at some time in the future. But so what? That doesn't mean that we just give up and claim to know nothing at all. Our models are never perfect truth, but they get us * closer * to the truth.</p>