<p>It is the pull relative to the number of students who are available to be pulled (i.e., the total number from the state attending private universities).>></p>
<p>And that’s where some of us are disagreeing with you. The students who are available to be pulled are NOT just merely those students who wound up at private unis. There are many California students who would be available to be pulled who don’t wind up going to private unis because they have UCLA & Berkeley right there. That doesn’t mean that they aren’t in the “pool” of people that Harvard is trying to magnetize. The proper measure of the “magnetism” would either be A) against all college freshmen in the other state or B) against all elite-college-material freshmen in the other state. </p>
<p>What you are doing by eliminating the smart Cali kids who wind up going to UCLA and Berkeley from your denominator is that you are making the target population that Harvard could conceivably “magnetize” smaller. Since you’re making your denominator smaller, your index will be bigger.</p>
The disappearance of the UC system would massively increase the number of CA students going to privates. Even assuming that a huge percentage of them went to OOS privates, it would probably at least double the number going to privates. Do you really think Harvard would accept twice as many CA students as it does currently if this happened?
One of the problems here is that students from the Northeast who are not strong enough to go to any of the top privates you are comparing are much more likely to go to another private research university than are students in CA who are not strong enough for these top privates. That is, students from the Northeast with no chance at Harvard or Stanford may very well end up at another private. Similar quality students in CA will most likely end up at a UC. Thus, far more of the comparatively weak Northeast students will count in the denominator in your calculations than will CA students, since more will go to private research universities. Thus, the pull of elite Northeast U’s in CA will seem greater than it really is.</p>
<p>Well put. Collegehelp, if you’re not going to use total college freshmen in a state as your denominator, or you’re not going to use total college freshmen within (say) 300 miles of each college as your denominator, and you want to put in a quality measure, then you have to compare like-quality students in Cal to like-quality student in MA. For example, a metric like GPA > 3.8 or SAT > 2100 or somesuch. “Winds up going to a private uni” is not a measure of like-quality when you’re comparing a state with excellent publics to a state with average publics.</p>
<p>And, conversely, I would surmise that for many families in Mass., attending any top 60 privates (#2 in the OP’s numerology), would be preferable to attending the oft-called “ZooMass” (USNews #98). But, the same is not true in Calif; families would typically prefer Davis & Santa Barbara to say, Tulane, BU, Miami or Rochester, or [fill in the name of a top 60 private]. If one is gonna spend the bucks to attend a private college, the economic rational family would compare the perceived value (and collegiate experience) of that private degree with that of the in-state public at approx. half the cost. (Even UC Riverside is ranked higher than UMass, btw.)</p>
<p>Give the man a break. Perhaps we should wait until he is done with all of the number crunching. Perhaps he doesn’t have any concrete conclusions just yet. Perhaps he is curious to hear from the CC community as to what those conclusion(s) (if any) might be?</p>
<p>I’m not sure what is worse, someone who is trying to contribute by offering some out of the box analysis, or someone who is continually trying to break that guy’s balls every 5 minutes as he is in the midst of providing that analysis.</p>
<p>Yes, it’s not like collegehelp is preparing this for peer review and publication.</p>
<pre><code>He’s putting out some numbers to see what they look like. Maybe they mean something, maybe not because there are too many variables. But is it possible to adjust for certain variables? Is it possible to draw broad if not precise inferences? Can these inferences not be debated until everyone is weary? Why not? Part of the game is to go looking in the maze and either find a way out (a meaningful data point) or get stuck in a dead end.
</code></pre>
<p>And if the argument is that there are fewer students in CA available to be pulled to private colleges out of state because they choose the UCs, shouldn’t this work against the NE unis top 20 and in Stanford’s favor? Fewer students to be pulled out of state should work the same way as the myriad of excellent privates in the NE region offers fewer students for S to attract. So the numbers ought to be less lopsided than they are, shouldn’t they? Either that, or for top students, the UCs are less attractive than one migiht assume, despite the cost savings, and they are going East in droves rather than choosing even a top UC. </p>
<p>And btw, a MI parent made a similar observation in another thread. That for some top students, the excellent public is regarded as a safety and, again, if money is not an issue, they attend only if they don’t get in to any of their other top private choices.</p>
<p>That really doesn’t answer my concern. Collegehelp divides by the total number of students from a given state who go to a private research university. California’s denominator is much lower than Massachusetts’s when you consider how many more people there are in California. A major part of this is because students in California not good enough to get in to top 25 privates go to the UCs while the same type of students in MA go to non-elite privates. When Stanford is compared pair-wise with a NE school, it’s apparent magnetism under collegehelp’s algorithm is lowered by the significant numbers of NE kids who, though not good enough to go to a top private, go to a private U that collegehelp counts for the purposes of his denominator. </p>
<p>To explain. Imagine a very simple university system. CA and MA are the only two states in the country. In CA, there are two schools, Stanford and UCSD. In MA, there are two schools, Harvard and Boston College. There are 10,000 students in CA and 5,000 in MA. Of the 10,000 Californians, 8,500 go to UCSD, 950 to Stanford, 500 go to Harvard and 50 go to BC (the only other school in this simple country). Of the 5,000 Massachusetts folks, 4,000 go to BC, 600 go to Harvard, 390 go to Stanford and 10 go to UCSD. Let us assume, for the sake of simplicity that everyone who is capable of getting in to Stanford or Harvard goes to one of those schools, and everyone who goes to BC or UCSD was unable to get in to Harvard or Stanford.
Under collegehelp’s algorithm, Harvard’s magnetism in CA would be 500/1100(the Harvard class size, since CA and MA are the only states in this country)/1500(the number of Californians in private school) *100,000 (collegehelp uses 1,000,000, but with our numbers this is unnecessary). This produces a magnetism of 30.3.
Stanford’s magnetism in MA is 390/1340(Stanford’s class size)/4,990(the number of Massachusettsans in private schools) * 100,000. This is 5.8, far less than Harvard’s magnetism.
Yet, consider that of the 990 MA residents good enough to get in to Harvard or Stanford, 390, or 39%, went to Stanford, while of the 1450 Californians good enough to get into one of those two schools only 500, or 34% go to Harvard. Once you correct for Stanford’s class size, the numbers are basically the same (dividing 39% by 1340 students produces 0.029, while 34% over 1100 students is 0.031). Yet, collegehelp’s algorithm would tell you that Harvard was far more magnetic than Stanford, when actually, they were about equal.</p>
<p>To apply this scenario to the real world, in the Northeast, of the many students unable to get into top 25 private research universities, a great deal will go to other private research universities, while in California, most will go to UCs. This creates significantly larger denominators for Northeast states, making CA schools, in head to head comparisons, seem quite weak, much weaker than their actual ability to attract students.</p>
<p>your discussion makes sense, but it misses the NE opportunity cost factor. UMass is lowly ranked; Cal ain’t too shabby at 20, and half the cost of privates. Even 'Cuse looks better than UMass for many. Thus, Calif kids will flock to HYPM bcos the perceived value is higher than Cal or UCLA, but will NOT flock to 'Cuse over UC Santa Barbara. OTOH, a New York resident would take 'Cuse over SUNY-Bing.</p>
<p>collegehelp is cranking out data, but needs to go back to AP Stat 101, IMO. Its not the Calif kids that have public options, but the NE kids that have fewer. c-help’s data is all private schools, so it will be skewed.</p>
<p>There was nothing simple about that example my friend. I never realized that a country with two states and four schools could get so complicated, I nearly broke out Excel… that said, it’ll be interesting to see what TourGuide has to say about BC being, at worst, the 4th best school in the nation.</p>
<p>Sorry, it was as simple as I could make it. To be more accurate, there should also have been a UMass school and perhaps a weaker CA private, but the example I used got across the basic point. Hopefully it wasn’t too confusing.</p>
<p>take a look-see at the OP’s first post: the last denominator is private school attendance. In real numbers, let’s assume that Mass residents had a highly-ranked public to attend, and that 1,000 more Mass residents attended that public than do so today. If so, then the “animal magnetism” of S & H would be roughly equal. It’s the OP’s denominator that is driving the stats…</p>
<p>This is a great discussion. I hope to reply to everybody’s posts as soon as I finish up some other work that has deadlines looming. Meanwhile, I’ll try to get all the information out on the table for discussion. Maybe I can tally the wins and losses and get a ranking. Maybe I can look at who has “beaten” who and get a sort of pecking order hierarchy, like sports teams. I’ll try to do a more formal analysis of the results once I have a chance to study them.</p>
<p>I am open to suggestions.</p>
<p>I have really enjoyed everybody’s stimulating, high-level discussion. There are lot’s of really smart people on CC.</p>
<p>take a look-see at the OP’s first post: the last denominator is private school attendance. In real numbers, let’s assume that Mass residents had a highly-ranked public to attend, and that 1,000 more Mass residents attended that public than do so today. If so, then the “animal magnetism” of S & H would be roughly equal. It’s the OP’s denominator that is driving the stats…>></p>
<p>Which is the crux of the problem … because the magnetism of a private uni is, in collegehelp’s algorithm, positively affected if the school is in a state with relatively poor quality state schools and negatively affected if the school is in a state with high quality state schools. It’s not Stanford’s “fault” that the UC system is so good, and it’s not Harvard’s doing that makes the UMass system so “bad” (relatively speaking, as my intent is not to dump on UMass). </p>
<p>The ways to correct this are threefold, but ALL involve changing the denominator away from “students who attend private unis.” They are as follows:<br>
all college bound students in the state;
college bound students who live within X miles of each of the target colleges (regardless of whether that X crosses state boundaries or not); or
elite-college-material students in that state (regardless of whether they go public or private in the end).</p>
<p>Another way of putting this is that attending a private uni is in no way correlated with “top 25 private uni worthiness.” So in MA, collegehelp has kept in the B student who goes to Syracuse but in CA, he has deleted the A student who goes to UCLA or Berkeley.</p>
<p>Here’s another way that may help you think about it, collegehelp.</p>
<p>Let’s assume that Harvard’s target is the top 5% of all CA students and that Stanford’s target is the top 5% of all MA students (and, of course, that they also target the top 5% in all other states, including their own).</p>
<p>If you could somehow identify all these students, this would be your “perfect” denominator, right? What % of each of these top kids are willing to cross country to go to the other school? </p>
<p>The algo makes a serious error in trying to serve as a proxy for these students, however.</p>
<p>Because you have made your denominator “those who go to private unis,”
what you did in California is that you cut out a certain number of these 5% students who are indeed H / S material, but wind up at UCLA, Berkeley, etc. You arbitrarily withdrew them out of the 5% target that Harvard is trying to magnetize. You made your CA denominator smaller, thus making the Harvard index bigger.</p>
<p>AND, because you have made your denominator “those who go to private unis,” what you did in Massachusetts is that you expanded your denominator BEYOND these 5% kids who are H / S material and included kids who are going to second-tier privates. You made your MA denominator larger, thus making the Stanford index smaller.</p>