are colleges lying about being need-blind?

Parents Forum
41

<p>Marylandmom - the stats are well researched and easily borne out. They do not predict achievement on an individual basis: you can not say how well child x will perform based on x's parental income. But they are incontrovertable on a group, statistical basis. Plot out all SAT scores on a graph related to parental income, and you will get a straight, linear progression; as family income rises, so do the scores.</p>

<p>So yes - using SAT's as a screening devise is a way to control for income & need. The higher the bar for admission in terms of SAT score, the higher the average family income of admittees. There is a much smaller percentage of students needing financial aid coming from the the group of SAT scorers in the 1500-1600 range than, for example, in the 1200-1300 range. </p>

<p>The ONLY way around this would be to reduce reliance on SAT's or to deviate from need-blind to need-aware to create a recalculated SAT. In other words, to use a formula that adds in SAT points based on economic status. (Which is kind of how the UC admission works, though the extra points are not added directly to SAT - but that in turn is why UCLA has so many more kids qualifying for Pell grants than Harvard).</p>

<hr>

<p>As to the purported 5% of students with family incomes in the $40-$100K range, the reason for that figure is not necessarily admit rate, but the realities of financial aid. Families with incomes under $40K are more likely to get close to a full ride with financial aid, and they will qualify for and need aid at both public and private colleges; the private colleges are likely to be more generouse with grant aid, so those kids, when admitted to top colleges, also have a financial incentive to attend: the private colleges end up costing less than or equal to the public schools. </p>

<p>In the $40-$100K range, the situation changes: as the income rises, there is less likelihood that the financial aid will reduce the cost of a private education to the equivalent of an in-state public. However, in that income range, families do not have a lot of discretionary income, and are less likely to have significant savings to draw on for college. Any kid who can get into Columbia can also get into their state university - the state university is also likely to offer merit aid, thus increasing the differential in costs. So the choice becomes between a private college the parents cannot reasonably afford, and the public institution that they can easily afford. It's a no-brainer for most: kid goes to state U. After all, if the kid's so smart, the parents know s/he will do just fine at Berkeley or UT Austin or wherever, and it makes more sense to conserve resources for grad school.</p>

<p>Above $100K we start to reach the range of families who can afford the private colleges without impacting significantly on their lifestyle -- and, in most cases, families who also have been able to put away significant savings to fund college. So the financial differential is not important - but at the same time prestige is far more important, because the upper class and wealthy are surrounded by peers who place high value on prestigious, private education. So the percentage of admitted students in upper ranges who actually attend is going to rise dramatically, especially since the wealthy student is not comparing costs with the state u any more, but now comparing the cost between Columbia and whatever private college or LAC is the safety. </p>

<p>So what I am saying is that the rate of attendance does not necessarily reflect rate of admittance -- rather, it reflects the combination of differential admit rates and differential yield rates among groups.</p>

42

<p>i was considering barnard but i dont want only columbia to be the reason as to why i pick barnard, thanks for the suggestions i will def check out smith and wellesley</p>

43

<p>
[quote]
all other things being equal, there being TEN TIMES as many top 5%ers as there are middle income students attending (at Williams and Yale, closer to 12x)

[/quote]
</p>

<p>Sadly this is what you would expect from schools that draw their students from the top end of the SAT distribution. It is also why places like Princeton and Harvard, with generous financial aid packages for those who qualify, have so few Pell grant recipients. Income and SAT are highly corellated. There are far, far more students in the high income group with Williams/Yale SAT scores than there are in the lower income groups. So many that places like W and Y could easily fill their classes with these kids with essentially no sacrifice in these quantitative measures. </p>

<p>The heavy tilt towards high income at these schools is a result of the academic requirements for admission, not an end run around need blind. They really are need blind, but very few lower income people meet the academic requirements. Those who do face a level playing field for admission. Again, see "Equity and Excellence..." Of course, the family may not agree that the "need" as calculated by the school equals what it would take to make attending one of these places possible.</p>

44

<p>calmom,</p>

<p>You were writing while I was, and making the same point, only more carefully. I would add that reducing the reliance on SAT would not help much unless one completely changed the definitions of academic qualifications. As long as admissions looks at academic preparation the high income kids have a huge advantage. They almost always have gone to excellent schools- either public in wealthy regions, or top privates. They have taken more advanced courses, they are far more likely to take AP courses, they have more and better exam prep options, they have more music/art/dance lessons, more and more interesting foreign travel etc. In short everything that looks good on a college application is more likely to be there for a student with a family income of 200k than 50k. Not just SAT I.</p>

<p>The other problem is that, much as most us may hate to admit it, the SAT and these other measures of academic preparation really do predict college performance. The lower SAT scores of lower income students, on average, result in lower college GPA's, on average. If a college were to throw out SAT and the other academic measures, it could increase the proportion of lower income students, but only at the risk of having a higher proportion of them running into academic trouble. For any individual the SAT is a poor predictor, but across a class, it is depressingly good.</p>

45

<p>Afan, you said "The heavy tilt towards high income at these schools is a result of the academic requirements for admission, not an end run around need blind. They really are need blind, but very few lower income people meet the academic requirements. Those who do face a level playing field for admission."</p>

<p>Your statement is a conclusion. It is one that quite a few of us here would not accept. No one has told Harvard that they must only take the top SAT scorers. Given what research, even of the CB itself, says about the SAT, one could easily make a case that Harvard, even more than U. Calif., should junk the SAT and use other better selection criteria, including ones that don't so heavily discriminate against lower SES kids. (and your comments in post 44 make me suspect you have not read the ETS's own research. If you want, I can send you links.) That they don't do so is telling. It tells me that they love to give lip service to SES diversity, but fear it. They fear it because if they embraced it, the school might lose its allure for the elites who pay full freight and donate heavily.</p>

<p>YES, It IS about money.</p>

46

<p>Afan, actually SAT's do NOT predict college performance - they predict only first year grades, and then it is only a weak correlation, far weaker than high school GPA. The predictive value of SAT's also dissipates for subgroups who tend to score less well on the test -- so basically while very high scorers are likely to do well in college, the converse is not necessarily true. </p>

<p>When it comes down to it the SAT is really highly effective but lazy way to weed out candidates. It pretty much guarantees that all the students admitted will be very bright, but it also weeds out many equally bright students or students with higher potential. In terms of potential, the most telling features would also be the hardest to measure on an objective scale, because it would come down a lot to personal characteristics, with qualities like personal initiative and work ethic probably being paramount. But then, the more the process is subjectified, the less likely it is to be need-blind or free from bias -- because you couldn't really get a sense of those personal qualities without knowing something about the student's background. </p>

<p>I mean, the kid who manages to sustain a high A average and a heavy, college prep courseload while at the same time holding down a 20-hour-a-week job and being responsible for after-school care of younger siblings very likely is a much better college prospect than the kid who attends a prestigious private school and whose parents have hired private tutors to help bolster grades and test scores. One kid is basically fulfilling expectations, the other kid is suceeding in spite of limitations, performing well beyond expectations - and the SAT score may reveal little more than the hours of prep time involved. But the minute you read the biographical facts, you know which kid needs financial aid.</p>

47

<p>Calmom, </p>

<p>Good points. We both forgot to add the coachability factor. The high SES kid is also the one for whom the Kaplan SAT prep course is chump change. These kids go for private SAT tutors. Guess what that tutoring does to the scores? Guess who gains?</p>

48

<p>Unfortunately, and I wish it were not true, SAT does predict college grades, and not just in the first year. There is lots of data on this, For recent reviews, look at Equity and Excellence, Bowen and Bok "The Shape of the River" and "The Black White Test Score Gap"</p>

<p>ETS, and others who have no stake in the issue, come to the same conclusion. The other measures of accomplishments that also predict grades- like AP scores are also income-corellated. Although it is appealing to assume that those who maintain high grades and get high scores from lower income backgrounds would do even better in college than their more priveleged peers with equal grades and score, there is no evidence that this is true. </p>

<p>Yes, as I said, the colleges could deemphasize grades and test scores (not just SAT, coaching works for SAT 2, AP, IB, etc), but the evidence is that this would result in a less academically sucessful student body. Places like Harvard are unlikely to bypass the most reliable predictors of academic sucess available (far from perfect, but better than subjective assessments of "drive") and risk a drop in the academic qualifications of its students. If anything, the range of acheivement considered acceptable is getting ever more narrow and crowded at the top. Those who meet these standards in spite of their backgrounds get admitted need blind, but not many lower income students can leap these hurdles. It is not a matter of intelligence, it is the lifetime of opportunity provided by weathier parents up to the point of college application. </p>

<p>It is very hard to qualify for Harvard even with every advantage. Without the top schools, parents with graduate degrees, etc it is even more difficult. This is the major policy point raised in Equity and Excellence. The authors concede the magnitude of the high income advantage and argue for more aggressive admission of lower income, first generation college students than "need blind" produces. They do not argue that this would occur without a price, and certainly not that it would result in a more talented student body- by the standard academic criteria, but that it is good social policy. I agree, but until someone implements their suggestion, the academic standards imposed by the most selective colleges ensure that the student body will be drawn disproportionately from the top of the income distribution.</p>

<p>If you think that is depressing, consider this. Even after controlling for high school AND college grades, parental income significantly predicts income for the offspring. So consider two kids with identical grades and test scores from high school, identical grades in college. The one from the wealthier family is likely to earn more money after graduation. Not just for the initial job, but over a career. There is lots of data on this as well. If you like, I can provide the references.</p>

49

<p>Afan,</p>

<p>You miss the point, and perhaps did not read the research carefully enough. The question is not whether the SAT predicts grades. It does. The questions are (i) how does it compare to other predictors and (ii) is it an unbiased predictor? It does poorly on both counts. No need to rehash the research here. Suffice to say that even SAT proponents have been forced to reluctantly concede that it predicts poorly, and not as well as HS GPA.</p>

<p>Regarding your statement "hard to qualify for Harvard with every advantage", you obviously forgot the legacy advantage. With a 40+% acceptance rate, I'd not call it "hard".</p>

<p>Income of parents predicts income of offspring? Shocking. Gee, I never thought having money would make it easier to make more...and, I don't see the relevance.</p>

50

<p>Think of it this way: kids with financial aid resources due to low income of their parents take on debt through loans before they even graduate and make it out into the working world. That debt is increasing.</p>

51

<p>I guess we do need to rehash the research here. </p>

<p>Look at Shape of the River. In table D.3.6 they report predictors of college class rank for 32,000 students who entered a set of 28 selective colleges in 1989. This group included 4 Ivies, other top privates like Duke and Stanford, as well as LAC's like Williams, Wellesley, Bryn Mawr... the authors controlled for race, gender, parents income, selectivity of the college, high school class rank, choice of major, and other factors. After controlling for all of these, SAT was a significant predictor of class rank cumulatively through college, not just for freshman grades. Family socioeconomic status also was significant after controlling for all these factors. All these effects were significant at p<0.05.</p>

<p>Then look at Table 13.2, pages 468-469 in Black White Test Score Gap. This reports results for 10,558 students at 11 universities and LAC's. The authors included the controls discussed above, as well as parents education, and whether the student was an athlete. The conclusions: SAT predicts college cumulative GPA, and students of parents with college educations had higher GPA's. The SAT predictions were significant at p<0.01.The parental education significance ranged from p<0.01, for mothers college degree to p<0.1 for fathers' advanced degree. </p>

<p>For one of these colleges the authors also had the complete admissions file including the admissions officer's "academic rating" and "personal rating" for each matriculant. They found that the full academic rating did predict collge GPA better than SAT alone, but only raised the R squared by 0.09. They found that the personal rating did not predict college grades at all.</p>

<p>Yes, legacies have better admisisons chances than others, and this effect persists when controlling for SAT score. However, it is much smaller after controlling for SAT score. Legacies at these schools come from highly educated backgrounds (most graduates of the top schools get some sort of advanced degree), and they are from higher SES families. The legacies who apply have higher SAT scores. In short, they show the benefits of growing up in an affluent familty. They do benefit somewhat from their legacy status as well, but most of the effect on admissions is the accomplishments they show on their applications - see Shape of the River, p28-29.</p>

<p>These data are particularly helpful because they are derived from the sorts of schools with which CC posters are obsessed. However, the same findings are made by people who look at a wider range of colleges. The literature on this is extensive, and Bowen and Bok, and Jencks and Phillips provide reviews. If you don't like the results I quoted above, look at the data as a whole. The conclusions will not change.</p>

<p>Similarly, the income benefits of growing up in a high SES family persist for a liftetime. If this is used as a proxy for career sucess- and it is difficult to find other quantitative measures of career sucess- then the high SES background is a predictor of sucess in later life, after controlling for high school and college academic performance. </p>

<p>Look, there is no sense in attacking me for pointing this out. I am not saying it is a good thing, that it is fair, or that it is desirable. But it is a fact that on every measure that has been shown to predict college GPA and class rank growing up in an affluent family offers advantages. If colleges select students by these measures, then they will select a pool of largely well-off applicants. The point of Equity and Excellence was that, this being the case, need-blind alone cannot level the playing field. Those from lower income, lower educated families have been fighting uphill all the way to high school senior, simply judging them on an equal basis with the affluent does not compensate for this.</p>

<p>So a college can genuinely, really, truly admit students without regard for financial need. However, if that college uses factors that predict academic and career sucess in admissions decisions, then the college will offer admission to a high income group. The college could reduce the weight given to these factors and enroll a more diverse student body. However, since these factors do predict academic and career sucess, a college that did this would have a less sucessful student body as a result. This may be good social policy, it may be the moral thing to do, perhaps selective colleges should try to ameliorate these class effects on sucess. However, the colleges that take this approach will enroll students who will get lower grades in college, be less likely to get advanced degrees, and have lower career incomes.</p>

52

<p>To get back to the OP's question. Yes, they really are need blind. You should not let the high incomes of the student body convince you that the colleges explicitly select for high income students. If you have the grades, scores, and EC's to be competitive in that pool, then your need for aid will not reduce your chances of admission. Your odds are as good as any other similarly qualified non-legacy (I assume you are not a legacy, since you did not mention this), non-athlete, and non- URM. </p>

<p>If you are an athlete recruited by the coach, then the admissions process is entirely different and your admissions chances skyrocket. If you are a URM who is competitive in this pool, then your admissions chances go up substantially, but not as much as if you are an athlete</p>

<p>If you want to go to Columbia, apply to Columbia.</p>

53

<p>For a more recent review of the University of California data, see Zwick, et al. "California and the SAT: A Reanalysis of University of California Admissions Data"</p>

<p><a href="http://repositories.cdlib.org/cshe/CSHE-8-04/%5B/url%5D"&gt;http://repositories.cdlib.org/cshe/CSHE-8-04/&lt;/a&gt;&lt;/p>

54

<p>I'm reminded of an old New Yorker cartoon, in which one captain-of-industry type remarks to another "Money is life's report card."</p>

55

<p>Afan:</p>

<p>"Look at Shape of the River. In table D.3.6 they report predictors of college class rank for 32,000 students who entered a set of 28 selective colleges in 1989.... the authors controlled for race, gender, parents income, selectivity of the college, high school class rank, choice of major, and other factors. After controlling for all of these, SAT was a significant predictor of class rank cumulatively through college, not just for freshman grades....see Zwick, et al. "California and the SAT:..."</p>

<p>First of all, the table in Shape of the River did NOT control for any variables. It was an ordinary least squares regression with a predictor variable of % class rank. I note that its parameter estimate for SAT was 5.929, and for top 10% HS class, 10.819. </p>

<p>Zwick et al, page 5, state "Using SAT I scores alone (without HSGPA) to predict UCGPA produces an R-2 of only 0.084 (Model 2). On page 8, they state "First, the results show that, given the set of predictors considered herer, high school grades are the single most effective predictor of UCGPA (i.e., its average standardized regression coefficient is highest), followed by the SAT II Writing Test...The remaining test scores (SAT I math and Verbal, SAT II Math) contribute little, given the predictors included in Models 9 and 10."</p>

<p>Look, I have not said the SAT does not predict college success. What I have said, based on the research of various sources, including the College Board itself, U. California, and Bowen and Bok, is that the SAT is a poor predictor compared to other factors, most importantly HS GPA. And, given how highly corrrelated the SAT is with family income/SES, more so than HS GPA, one MUST wonder why elites continue to use it so extensively. And they do, as Avery's book Early Admissions Game showed, among others.</p>

<p>So, your statement "So a college can genuinely, really, truly admit students without regard for financial need." is entirely accurate. They don't regard need as a factor because they use proxy variables that accomplish the same end.</p>

56

<p>That is what "control for" means. In the OLS, with the other variables in the model, SAT was a significant predictor.</p>

<p>SAT is also a predictor (significant) in Zwick. Other combinations of predictors are better than SAT alone, but SAT remains significant, again, the way you analyze this is to include other predictors in the model, and see whether SAT is significant. </p>

<p>To quote a little more extensively from Zwick</p>

<p>"While these results corroborate the findings of UC and the SAT that SAT II scores are superior to SAT I scores as predictors of UCGPA, the differences in the proportion of explained variance are tiny. Also, as discussed below, the predictive power of the SAT II is largely attributable to the SAT II Writing test. Using SAT I scores alone (without HSGPA) to predict UCGPA produces an R2 of only .084 (Model 2). The R2 values for the remaining models range between .110 and .126."</p>

<p>So the proportion of variance predicted by SAT is substantial compared to the total predictability in the models.</p>

<p>Further, Zwick did not confirm, and Geiser and Studley subsequently modified, the assertion that SAT II is less conditioned by SES than SAT I. </p>

<p>Finally, any other measure of academic accomplishment also will favor the wealthier students. So you could eliminate the SAT I completely, and rely on HSGPA (which corellates with family income), SAT II (which corellates with family income), AP course and test participation (which corellates with family income), and you would get the same result: more admissions offers to higher income students.</p>

57

<p>Look, however, you slice it, among the major private universities, the following schools had more than 50% of the student body receiving no need-based aid, meaning that they come from families in the top 5% of the population, with a minimum income of around $150k, the median being much higher (though quite variable):</p>

<p>Georgetown, Notre Dame, Yale, Vandberbilt, Tufts, UPenn, Johns Hopkins, Brown, Princeton, Duke, Boston College, Emory, Washington U., Dartmouth, Brandeis USC, Harvard, Northwestern, Stanford, Brandeis, UChicago, Columbia, Cornell, Rice, and Tulane. (Significantly absent from the list are Caltech and MIT.) </p>

<p>Could do the same for the major LACs. What this means is that 95% of the population has less representation at these universities than the top 5% (and it ain't because they aren't applying). In each case (with the very notable exception of USC), those with incomes below $40k (Pell Grant recipients) represent roughly 10% of the population (with a low at Harvard at 6.8%, and likely a high at Cornell, in their "public" component.) Doesn't mean that there aren't plenty of low-income students who could do the work - the 39% of high-achieving Pell Grant recipients at Berkeley (where there is very little grade inflation) pretty much proves otherwise. But, for whatever reason, they aren't recruited, didn't apply, didn't get in, or didn't get the support they needed to attend.</p>

<p>Which leaves less than 40% of the school populations at these schools from the "broad middle" - $40k-$150 - with the vast majority (at least at Harvard and Williams, and I doubt it is much different elsewhere) coming from the upper part of it.</p>

<p>"So, your statement "So a college can genuinely, really, truly admit students without regard for financial need." is entirely accurate. They don't regard need as a factor because they use proxy variables that accomplish the same end."</p>

<p>I think that is precisely correct. As to the complaints from Winnetka that only one or two a year get into each of the Ivies, well, in my town, a state capital with a well-educated, but much poorer workforce and "good schools", in recent memory the only students who have gone to HYPS have been athletes. No vals/no sals/no 1600 SAT scorers (to be fair, a couple of years ago, we had two turn down H for Brigham Young.)</p>

<p>But, from the private school's perspective, it would be hard to think of good reasons why they should do it any other way - and it is hard (and expensive!) to change, as the experiences of Amherst and Smith so well indicate.</p>

58

<p>afan,</p>

<p>you need to learn the difference between statistical significance and practical significance.</p>

<p>I agree that the SAT scores reached statistical significance in the studies we discussed, at a 0.05 confidence level. So what? As a practical matter, the same models showed much higher r2 values for other factors, including, depending on the study, being in the top 10% of HS class or HSGPA.</p>

<p>Now for the biggest so what! None of the models accounted for much of the variance in the sample population. It is completely accurate to say that most of what was trying to be predicted, whether it be FYGPA, class rank, satisfaction (yes, River looked at that) or whatever, was not explained by any of the regression models, so must be due to unidentified factors. </p>

<p>This leads us to the following discussion points: </p>

<ul>
<li><p>colleges use a criterion (SAT) that is lousy for picking a class (it does not well predict any higher ed success factor)</p></li>
<li><p>the same criterion IS highly correlated with SES.</p></li>
<li><p>therefore that criterion, the SAT, is a good proxy for SES. QED.</p></li>
</ul>

<p>Now, if perhaps, this problem were accidental, we might see colleges somehow compensate. This would lead to a higher probability of admission for lower SAT/GPA sets of low SES kids compared to higher SES kids. BUT, the data says NO. With comparable records, recent studies have shown no "tilt" toward lower SES kids.</p>

<p>So, in summary, elite colleges use criteria, including the SAT, that have a high SES correlation. Knowing this ( I must assume that if WE know this, then adcoms do!), colleges ignore the issue, providing no compensation.</p>

59

<p>And that's the whole point - what incentive would they have for doing otherwise? What possible incentive is there for taking a kid who is not poor (i.e. family income below $40k), whose SAT scores can easily be matched by those with higher incomes requiring less ($100-$150k) or no (($150k plus) financial aid? The answer is obvious - there isn't any - unless s/he bring something else to the table (football, crosscountry, tuba, skin color). But, all things being equal again (forget equalizing SAT scores), there are many incentives to take wealthier kids (ranging from relations with the private school GC, to the need for lacrosse players). </p>

<p>And so adcoms do what is necessary to carry out the institutional mission. If you were in their position would you (or could you) do differently?</p>

60

<p>wow if colleges really think this way then enough with the 'education is equal and important for everyone' crap! i will just apply and try my luck, thanks to all the parents who replied</p>