<p>Assuming they have ~2000 undergraduates, and their middle 50% SAT scores at 2090-2350, that means that 500 out of the 2000 incoming freshmen have a SAT score less than 2090! Is that right?</p>
<p>Where did you get these numbers from?</p>
<p>The common app says these are Harvard’s SAT breakdowns:</p>
<p>2,075 |2,225|2,375</p>
<p>I find that unlikely though. >.< The 2075 score anyway.</p>
<p>Athletes, URM’s, donors, legacies. =P</p>
<p>Princeton has you covered if you want stubborn high numbers ;)</p>
<p>Technically, this is what I deduced, and I am wondering if 500, yes 500, out of 2000 freshmen have equal or less than 2090 on the SAT at Harvard, assuming this is the given 50th percentile Harvard has (this is according to many websites).</p>
<p>Anyone want to confirm?</p>
<p>I’d say that’s not a good assumption… Probably less than a hundred do.</p>
<p>If that’s what they share on the Common Data set as their bottom 25 percent numbers, why would you doubt it and what difference does it make anyway?</p>
<p>
</p>
<p>No. If your means of finding the total composite score at the 25th percentile is to sum the sections’ 25th percentile scores, you’re going to underestimate the figure, just as summing the 75th percentile section scores will overestimate the composite 75th percentile score. This is so because knowing that at least one of a matriculant’s section scores is at or below the 25th percentile implies a negative inter-correlation among his or her section scores since they are a factor in an admissions process that generally selects for higher scores.</p>
<p>As far as I know, Harvard does not release this figure.</p>
<p>Using ACT scores is more accurate for the percentiles since they provide a single score. 25% scored at or below 31.</p>
<p>Percent submitting SAT scores
91%
C9
Percent submitting ACT scores
32%
C9
25th Percentile
75th Percentile
C9
SAT Critical Reading
690
800
C9
SAT Math
700
790
SAT Writing
710
800
SAT Essay
C9
ACT Composite
31
34</p>
<p>I’m still confused. If the 25th percentile is a 2090, as many (actually most sites) suggest), then that means 25% of the class (which means lets say 500 out of 2000) have either a 2090 or lower composite score. How does this not make sense? I take AP Stat by the way, so I understand this.</p>
<p>
</p>
<p>If that figure is correct, your interpretation is correct. But those sites’ suggestion of the score does not seem to be founded, since there is no independent way to infer the score and I have seen no reason to believe that Harvard has released it. In any case, those sites’ figure is outdated even if it was once correct.</p>
<p>By the way, Harvard has about 6,600 undergraduates. Since many of its peer schools have similar enrollments, if we were to assume the accuracy of third party sites’ reports of these schools’ score ranges, we would find that they necessarily imply an impossibly large number of high scorers, given the College Board’s data. This suggests that these ranges are wrong (presumably because they err as previously described by summing the section score ranges).</p>
<p>Whatever hairs are split, though, what I sense is your suggestion that a sizable chunk of students at top schools like Harvard have scores well below what many see as almost necessary for admittance is true. This is attributable to a perceived academic competency threshold that the vast majority of applicants surpass and the large role of factors that are not intrinsically quantitative.</p>
<p>Dam, Silverturtle, you never cease to amaze. Where are you at college anyhow (or not)?</p>
<p>
When I have seen such data for other schools (and not many release it), the slot end points end up being about 40 points higher than the sum-of-section-25th’s and about 40 points lower than the sum-of-sections-75th’s. Thus we can estimate H’s “true” SAT range to be about 2130-2310.</p>
<p>
Absolutely correct. Data for cumulative SAT scores is not generally available, but some college data sites present sum-of-sections percentiles as sum-of-total percentiles and aren’t too concerned about the misleading representation they make of it.</p>
<p>
True to a point. A 31 composite ACT converts to about 2060 cumulative SAT equivalent. This suggests a small boost is experienced by the ACT-pool, but probably not due to any preference for that test. More likely it reflects an effort to seek geographic diversity.</p>
<p>A bunch of kids submit scores (it looks like a 20 % overlap based on common data set).</p>
<p>CDS shows 92% submitted SAT and 32% submitted ACT. So it is quite possible some of them qualified based on one or the other although they submitted both.</p>
<p>YES 500 do. The relevance of this statistic to any particular candidate is what is important.</p>
<p>@JackWilshere. Again, that’s not necessarily true. As far as I know (and as mentioned above by a number of other posters), aggregate scores are not reported. Rather, the 25-75%iles are given for each individual section of the test. It’s entirely possible that some students score say 690 on the math portion but high 700s on the other two while other students have an exact opposite breakdown in scores. The 690-700 score maybe the 25%ile in each section of the exam, but that doesn’t necessarily translate to the composite score of 2090 as the 25%ile of the admitted class.</p>
<p>As Descartesz mentioned, the width of the composite score curve is probably narrower than for individual sections.</p>
<p>The fallacy of adding the 25-75th percentiles together has been well explained above. But another issue is that the reported percentiles do not give you enough information to understand the shape of the distribution curves within a section. Let’s take the CR data from Harvard’s 2010-2011 CDS. It reports that 25 percent scored at or below 690 and 25 percent scored at or above 800. For the middle 50% you have no idea how many scored closer to 800 (and there are certainly some, if not many 800 scores in this population) versus those scoring closer to the 690. I do not have the data, but my assumption is that both the mean and median of this “middle 50%” is greater than the average of 690 + 800 because of the skew of the distribution to the right. Similarly, I assume that the distribution of the “lower 25%” at Harvard is probably pretty narrow so that the mean and median of this population is actually not much less than the 690 score.</p>
<p>@cltdad: This is why stats should be required study.</p>
<p>I have not seen real data for H, but saw someone posted some interesting data on the Y forum a while ago. Here are some published numbers regarding the (seemingly) correlation between admissions and SAT scores for Princeton Class of 2012 which should be highly relevant and might give you a pretty good idea about the HYPS admissions:</p>
<p>Percentage of Applicants Admitted, by SAT Range
28.1% 2300-2400
10.6% 2100-2290
6.2% 1900-2090
2.3% 1700-1890
0.6% 1500-1690
0.0% Below 1500
4.4% No CEEB Scores</p>
<p>Middle 50% SAT: CR 690-790, MA 700-790, WR 690-780, CR+MA 1400-1580</p>
<p>As many may also be interesed, here are the numbers regarding the (seemingly) correlation between admissions and GPA for the same group of admitted applicants:</p>
<p>Percentage of Applicants Admitted, by GPA Range
16.4% 4.00
11.0% 3.90-3.99
9.6% 3.80-3.89
6.2% 3.70-3.79
5.1% 3.60-3.69
4.9% 3.50-3.59
4.1% Below 3.50</p>
<p>Other relevant numbers for Princeton Class of 2012:
Total Applicants 21,370, Total Admitted 2,122 (9.9%), Enrolled 1,243
96.8% of the admitted applicants ranked in the top 10% of their class
50/50 Male/Female Ratio
13.7% Legacies
School Type: Public 56.2%, Independent Day 20.8%, Independent Boarding 9.4%, Religious 12.9%, Home Schooled 0.5%, Military 0.2%
Ethnic/Racial Diversity: African American 7.6%, Asian American 16.7%, Hispanic 7.6%, Native American 0.4%, Multiracial 5.6%, International 11.3% (from 45 countries)</p>
<p>Source: Greene, H. and Greene, M. (2009), Making It Into A Top College, 2nd Ed., p284.</p>
<p>I’d be interested if anyone may have any relevant real data regarding college admissions.</p>