Berekely vs Northwestern !?!?

College Search & Selection
241

<p>Barrons- The actual percentage of SAT retakers attending highly competitive colleges may be close to 82% (82% of students applying to three or more colleges will have taken the test more than once - "Retaking the SAT", Clotfelter and Vigdor, 2001 <a href="http://www.pubpol.duke.edu/people/faculty/clotfelter/SAN01-20.pdf%5B/url%5D"&gt;http://www.pubpol.duke.edu/people/faculty/clotfelter/SAN01-20.pdf&lt;/a> )
This is shown on Table 1, page 36 (37/51 on the PDF). Interestingly, the better you do the first time the less chance you have of retaking it. Amongst applicants to 3+ universities, only 17% will retake the SAT with a score of 1500 or more (Table 2, page 37 (38/51 on the PDF).</p>

<p>If you are applying to three or more colleges, income doesn't seem to make that great of a difference. The key separation (and here income may play a difference) of whether one retakes the test is if you are applying to 3 or more schools or only one or two (perhaps those going to local or community colleges).</p>

<p>So on this basis, the true benefit very competetive private colleges have may be closer to 5 points (0.82 x 0.135 x 44). However, the larger the percentage of your student body scoring above 1500 (where only 17% retake the test), the lower your benefit. Most very competitive private colleges have somewher between 25% and 75% of their student body in the 1500 SAT and above range.</p>

242

<p>I took a survey about a year ago. It turned out to be 30-50 points, with the mode at 50 points. However, I realized that students who took it once may have self selected themselves out of the survey. This is hardly scientific, but its better than what we're doing right now. Please answer the following thread. </p>

<p><a href="http://talk.collegeconfidential.com/showthread.php?t=59111%5B/url%5D"&gt;http://talk.collegeconfidential.com/showthread.php?t=59111&lt;/a&gt;&lt;/p>

243

<p>jesus h. christ... is this really what you people do in your spare time? conduct scientific research on SAT scores? At least we know two people who won't be choosing a college based on any "fun" factors. </p>

<p>i mean, im not trying to be judgmental, but i think most of us would agree its a strange hobby.</p>

244

<p>West Sidee: " took a survey about a year ago. It turned out to be 30-50 points, with the mode at 50 points...This is hardly scientific, but its better than what we're doing right now."</p>

<p>"its better than what we're doing right now."
I'm speechless... I've just shown you two research papers on the subject, plus a careful analysis that shows at most a 5 point boost to private college's SAT scores based on best verbal + math and you are going to argue that a survey here on CC is better?</p>

<p>I give up...</p>

245

<p>No "survey" only including CC or a similiarly small sample size has much credibility.</p>

246

<p>TheCity: "i mean, im not trying to be judgmental, but i think most of us would agree its a strange hobby."</p>

<p>Yep, I totally agree. West Sidee made a nonsensical claim, and I just wanted to know if I could actually answer it. It was just a challenge, that's all.</p>

<p>I said, I give up...</p>

247

<p>One thing is that in your analysis of two time test takers, in the upper end of the SAT scores, I believe the volatility is much greater. </p>

<p>Your analysis of three time test takers seem correct. But the two time test takers I will have to disagree on the upper end volatility.</p>

248

<p>Ok, I did find an error (in fairness). My 13.5% figure was based on 45% going up and 30% going down (0.45 x 0.30 = 0.135). But either verbal or math could be the one going up, so that figure should have been doubled.
Also to be fair, should assume 1/3 of the -10/+10 cohort goes down 10 points and 1/3 goes up.</p>

<p>lets plug in just 680-720 scores (highest volatility on down side):</p>

<p>On the verbal, of the decliners (again, all derived from table 5), 2.2% (1/46) will decline 120 points, 8.7% (4/46) will decline 90 points, 26.1% (12/46) 60 points, 45.7% (21/46) 30 points, and 17.4% (8/46) 10 points, for an average decline of 41.6 points. On the math, of the decliners, 7.0% (3/43) will decline 90 points, 23.3% will decline 60 points (10/43), and 46.5% (20/43) will decline 30 points, and 23.3% (10/43) will decline 10 points, for an average decline of 36.7 points.</p>

<p>Let's use the precise up/down percentages for 680-720: v 46% down, m 48% up (0.221), m 43% down, v 46% up (0.198)</p>

<p>That's 0.221 x 41.6 = 9.2 points for verbal decline
0.198 x 36.7 = 7.3 points for math decline</p>

<p>So average gain is 16.5 for 680-720 scorers who retook test.
71% of 1300-1490 retook test who applied to three or more schools (Table 2 from "Retaking the SAT", Clotfelter and Vigdor, 2001).
0.71 x 16.5 = 11.7 points gained</p>

<p>I think you will agree 680-720 is maximal volitility (remember only 17% who scored 1500 or above retook the test - gains for private colleges for these students would only be 0.17 x 16.5 = 2.8, Could use slightly higher number than 16.5 but doesn't change figure much).</p>

<p>Since at least 25% of private competitive college are made up of 1500+ students (I know, some may have reached 1500 on the retake but some elite schools have 50% or more at the 1500 level, so let's factor in just 25%.</p>

<p>0.75 x 11.7 = 8.8
0.25 x 2.8 = 0.7</p>

<p>Average points gained by competitive private colleges (by using best verbal + best math score instead of best sitting): ~9.5 points (or somewhere pretty close). Also remember, that this assumes randomness on up/down scores, when probably people who score higher on one (verbal or math) will probably score higher on the other, so the actual points would be lower. So 9.5 is probably maximal.</p>

<p>Not near your original 50-60 points but higher than I actually thought.</p>

<p>I posted because I did find an error and for West Sidee, I thought it was fair to correct my post.</p>

249

<p>oh god, this is going too far lol!</p>

250

<p>
[quote]
Let's use the precise up/down percentages for 680-720: v 46% down, m 48% up (0.221), m 43% down, v 46% up (0.198)</p>

<p>That's 0.221 x 41.6 = 9.2 points for verbal decline
0.198 x 36.7 = 7.3 points for math decline

[/quote]
</p>

<p>Actually, when you take into account that .46(v down)* .43 (m down) = .198 and .46 (v up) * .48 (m up)= .221, then you have to figure that .198+.221 = .419 to figure out the percentage that doesn't benefit from second time test taking (because they both went down or both went up) So 58% of students see a rise in their SAT score when they take it a second time. Because all you have to do is subtract the .419 from 1 to find the population of those that increased in one, and decreased in the other. </p>

<p>So if 58% gained from an average of lets say 39.2 (avg of math + verbal), then the average gain is 22.7 for these students. </p>

<p>Now in my estimate, I would say that 60% of students at Berkeley took the SAT twice. 25% took it 3 times, and 15% took it just once. </p>

<p>(remember that Berkeley has a lot of Asian Americans, and Asian parents are notorious for making their sons and daughters retake a 1550 SAT score)</p>

<p>So</p>

<p>.60 * 22.7=13.62
.25 * 44 = 11
.15 * 0 = 0 </p>

<p>That would be an appoximately 25 point difference for Berkeley. </p>

<p><a href="http://osr.berkeley.edu/Public/STUDENT.DATA/PUBLICATIONS/UG/ugf04.html%5B/url%5D"&gt;http://osr.berkeley.edu/Public/STUDENT.DATA/PUBLICATIONS/UG/ugf04.html&lt;/a&gt;&lt;/p>

<p>The 25th -75th here is 1200-1450 for Berkeley in 2004, the most recent data available. </p>

<p>When I used 1240~1250 - 1490~1500 It was a 40-50 point difference. Not a 50-60 as you stated. </p>

<p>So although it seems about 25 point disadvantage for Berkeley using best one sitting method, I would be much more comfortable looking at the actual student records.</p>

<p>BTW, thanks for the mental excercise. I needed a balance, I've done way too much qualitative over quantitative recently. Its because the world is getting too political, but anways... latez! ;)</p>

251

<p>Hmm, I hate to keep this going, but you missed the point. If both scores (verbal and math) went up or both scores went down, then there is no advantage in points gained by the private school (calculating in best verbal or best math) and Berkeley (calculating in best sitting). The 58% that went up (actually ~ half that since you forgot the 58% includes students where both scores went down) do not benefit the Privates over Berkeley and thus do not give the privates any bonus to their reported SAT over that of Berkeley (the students higher scores (2nd sitting) are counted fully for both.</p>

<p>Ex:</p>

<p>1st sitting: 600v 600m</p>

<p>2nd sitting 650v 650m
(both Berkeley and Private report 1300)</p>

<p>2nd sitting 650v 550m (this was like the 36.7 m decline above)
(Berkeley reports 1200, Private 1250)</p>

<p>2nd sitting 550v 650m (this was like the 41.6 v decline above)
(Berkeley reports 1200, Priv 1250)</p>

<p>2nd sitting 550v 550m
(both Berkeley and Private report 1200)</p>

252

<p>
[quote]
Now in my estimate, I would say that 60% of students at Berkeley took the SAT twice. 25% took it 3 times, and 15% took it just once.

[/quote]
</p>

<p>I hope you never become a statistician. Your estimates are no more credible than a dog's. </p>

<p>I can estimate that 50% of the citizens of India like red blowpops. Am I correct? Probably not. </p>

<p>My point is that you cannot just go making up data that fits your argument. This is a lost cause because neither one of you know how much the different recording of SAT scores affects the overall scores. Don't waste your time on achieving the impossible. (Unless you want to survey EVERY single Berkeley student).</p>

253

<p>uc_benz: "This is a lost cause because neither one of you know how much the different recording of SAT scores affects the overall scores."</p>

<p>I disagree, statistically speaking I essentially took the worse case scenario. Yes of course the exact number is impossible to come by, but the final number will not be too far off the 9.5 points I gave (a point or two, and likely downward).</p>

254

<p>no no. The .221 represents when both go up. and the .198 represent when both go down. We are essentially using the same numbers, but I just rather do a ( 1 - (both go up+both go down)) to find out the statistically significant number of .58. See my point? There are various scenarios which your number doesn't count for, such as the M stays same, and V goes up in second test, which mine DOES count. </p>

<p>Does this make sense now? It already addressed your second to last point.</p>

255

<p>I don't see your point, and when M stays the same and V goes up, it shouldn't count...</p>

<p>1st test
600 v 600 m</p>

<p>2nd test
650 v 600 m
Both Berkeley and Privates report 1250
(no advantage to SATs reported by privates)</p>

256

<p>^ Actually, you're right. My bad. I was getting the same V and same M chance of .7% confused with something else...</p>

<p>Well, ok. Then lets estimate that .58 - (.08+.09)= .41 </p>

<p>.41 * 39.2 = 16.1 </p>

<p>Now lets redo my calculations.</p>

<p>.60 * 16.1 = 9.66
.25 * 44 = 11
.15 * 0 = 0</p>

<p>So that still is a 21 point SAT difference. Either way, given Cal's incoming class of 3500-4000 every year, having a 75th percentile at 1470+ is pretty good. I suppose that the 80-85th percentile is at 1500+, so that does still mean that Berkeley has potentially 600-800 students with a 1500+ SAT score. Given that they are mostly concentrated in difficult majors like Engineering, Math, Haas, Econ, PEIS, CS, Statistics, etc... and these department majors are all ranked in the top 5 if not #1 in their fields, I believe that the competition and education at Berkeley can serve someone as well as Harvard, but with a Berkeley feel to it.</p>

257

<p>People, this is ridiculous. Anybody in Academe and the corporate world will tell you that Cal and Northwestern are awesome schools. We can argue this point until the second coming, but it we will not come to a conclusion. There is no way to prove that Northwestern is better than Cal or that Cal is better than Northwestern. It boils down to preference and fit. I love both those schools and when I had to chose, I would have probably picked Cal over Northwestern because it fits my personality better. But I would never say that Cal is a better university than Northwestern. Nor can anybody say that Northwestern is better than Cal...because it isn't. They are both top 15 universities...point finale!</p>

258

<p>How's That Whole 50-60 Point Thing Coming Along?</p>

<p>What's It Down To Now?</p>

259

<p>meltingsnow: "How's That Whole 50-60 Point Thing Coming Along?
What's It Down To Now?"</p>

<p>Glad you asked :)</p>

<p>It's about to be 4.75 points.</p>

<p>I know, I know, I totally agree with Alexandre. This is more meant for the Math/Stat geeks following along.</p>

<p>One thing I am doing is subtracting the full negative score when one score goes down and the other up. This is fine when the up score is greater than the down. For instance:</p>

<p>1st test
600v 600m</p>

<p>2nd test
580v 680m
Berkeley reports 1260, Privates 1280: Privates gain 20 points</p>

<p>But if the decline score is is greater than the gain score:
1st test
600v 600m</p>

<p>2nd test
520v 620m
Berkeley reports 1200, Privates 1220: Privates again gain 20 points
not the 80 points of the declining value as I previously allocated.</p>

<p>The point gain of the privates is actually the value of the positive score if the positive score up is less than the negative score down. If the positive score up is greater than the negative score down than the point gain by the privates is the value of the negative score. In other words, the point gain of the privates is the value of the positive score up to the decline in the negative score.</p>

<p>So for a 100 point decline in the verbal on the second test, the point gain by the privates would be any positive point gain on the math up to 100. 80 decline verbal, privates gain anything up to 80 rise in math</p>

<p>ex:</p>

<p>1st test
600v 600m</p>

<p>2nd test
500v 720m
B=1220 P=1320, Private gain=100</p>

<p>2nd test
500v 700m
B=1200 P=1300, Private gain=100</p>

<p>2nd test
500v 680m
B=1200 P=1280, Private gain=80</p>

<p>2nd test
550v 720m
B=1270 P=1320, Private gain=50</p>

<p>2nd test
550v 650m
B=1200 P=1250, Private gain=50</p>

<p>2nd test
550v 630m
B=1200 P=1230, Private gain=30</p>

<p>2nd test
590v 720m
B=1310 P=1320, Private gain=10</p>

<p>2nd test
590v 650m
B=1240 P=1250, Private gain=10</p>

<p>2nd test
590v 610m
B=1200 P=1210, Private gain=10</p>

<p>So as the decline in verbal gets less, the privates lose any gain from the math over the the value of the verbal decline</p>

<p>100 v decline is limited to math gains up to 100, so the range of private college gain is the average for math increase on the second test between 0 and 100, any math increase above 100 is lost.</p>

<p>50 v decline is limited to math gains up to 50, so the range of private college gain is the average for math increase on the second test between 0 and 50, any math increase above 50 is lost.</p>

<p>10 v decline is limited to math gains up to 10, so the range of private college gain is the average for math increase on the second test between 0 and 10, any math increase above 10 is lost.</p>

<p>You can see that the range of point gain by the privates is the average of the math score rise on the second test below the verbal decline score. This works out to half of the verbal decline.</p>

<p>verbal decline 100, math decline 100 - 0, average private gain 50
verbal decline 50, math decline 50 - 0, average private gain 25
verbal decline 20, math decline 20 - 0, average private gain 10
etc.</p>

<p>So the previous 9.5 score should be half that or 4.75 points.</p>

260

<p>Slight correction. I'm beginning to realize this is one of the toughest problems I've tackled, I just want to get through with it.</p>

<p>While just averaging the range of the verbal losses works for the upper limit say 100 point loss (private college point gain is average of 0-100 range), you can see that at the lower reaches say 20 verbal loss, virtually all math gain scores will be 20 or above (private college point gain very close to 20). At the midway point of the verbal loss range, say 50, half the math gains will be below 50 (so one can average the 0-50 range), but the other half will be above and count as 50 (50% of the private college point gain would be 50, the other 50% a 0-50 range average)</p>

<p>Therefore, mathematically speaking, half the calculated score should count as the decline range average (9.5) and the other half as the average of the maximum range score (4.75)</p>

<p>Therefore, (.5 x 9.5) + (.5 x 4.75) = 7.125</p>

<p>~7 points is gained by private colleges.</p>

<p>sheesh... I'm changing my major to math... or maybe not</p>