Hardest colleges to stay in

College Search & Selection
101

<p>
[quote]
The only time superscore makes a difference is if the second sitting gets something like 700 V but 660 M, which is less likely to happen since the whole point of retake is to improve in both sections or at least to maintain one and improve on another.

[/quote]
</p>

<p>Only true for the one sitting colleges.</p>

<p>Not true at all for schools that superscore, and somewhat illogical for some (many) students. If a student applying for a highly selective schools scores a 750 M-670 CR the first time, (which is not uncommon in Calif since the state has so many testers whose Eng is a second lang), which section is s/he gonna focus on for the retake for a superscore school? It would be nonsensical to spend a lot of time trying to get that extra math bubble correct when the CR needs to get above 700 as well. Furthermore, the Writing test cannot be ignored in California since the big two (Cal & UCLA) value it highly; a 700-W is a huge differentiator amoung kids attending the same HS. , </p>

<p>Sam: if you have time, you might want to compare the percentile scores of ACT & SAT applicants to NU.</p>

102

<p>Actually, a lot of students who retook, knowing that their target colleges superscored, nevertheless found that their highest single-sitting score was on the retake. </p>

<p><a href="http://talk.collegeconfidential.com/sat-preparation/373227-what-does-superscoring-do-you.html%5B/url%5D"&gt;http://talk.collegeconfidential.com/sat-preparation/373227-what-does-superscoring-do-you.html&lt;/a> </p>

<p>I think the effect size of superscoring versus not superscoring is not particularly large as it influences comparisons of score percentiles by section (the NACAC-preferred statistic, and the standard statistic for Common Data Set users) between colleges that superscore and colleges that don't. It makes some difference, but not a lot. </p>

<p>Yes, I will remind the friendly College Board representative I met at the NAGC conference in Minnesota in November 2007 once more that I am still interested in the College Board publishing a report of "superscored" SAT scores along the line of the report it now publishes of best single-sitting SAT scores. </p>

<p><a href="http://www.collegeboard.com/prod_downloads/highered/ra/sat/composite_CR_M_W_percentile_ranks.pdf%5B/url%5D"&gt;http://www.collegeboard.com/prod_downloads/highered/ra/sat/composite_CR_M_W_percentile_ranks.pdf&lt;/a> </p>

<p>I don't expect there to be more than about half again as many (150 percent as many) students at a given scoring level by superscoring as compared to single-sitting consideration of scores. Any student who is really worried about this issue can always sit down and take the SAT once more for the win, a straight-up score of 2400.</p>

103

<p>bluebayou,</p>

<p>NU used to have a table showing the applicants/admits/enrollees vs scoring range, just like the one I did for Berkeley in my last post. The last time they did it was for the class of 2009. I remember someone brought it back up a while back so if you know a method of doing the same, you can try. But I am pretty sure the profile looks a lot like the one for Berkeley OOS because that's what I would expect from Northwestern. Otherwise, I would be shocked already (to see that many applicants in the low scoring range).</p>

<p>Also, I believe what I showed you also explains why Michigan's stats is comparable to Berkeley's despite the fact that it has significantly higher admit rate and that they have a lot less students to choose from within its state. The key is 1/3 of its student body are OOS.</p>

104

<p>See last paragraph of post #92; I did mention its different criteria and I was just going to leave it at that. But kyledavid's response made me go deeper and then came post #100. :)</p>

105

<p>I found a site cgaps.com that has rankings for colleges such as top party school, schools with the top professors. I really liked the site and thought it would be useful.</p>

106

<p>
[quote]
Please don't bring in the writing section to try to make your theory look more plausible.

[/quote]
</p>

<p>My bringing in the writing section is not to make my theory more plausible. It's to get through your head that I do not misunderstand superscoring. You keep bringing back the two sections, but what I'm saying is based on the fact that there are three sections. And there are. There have been for a few years. That's just how it works, I'm sorry to say.</p>

<p>
[quote]
note how 12% of applicants have 200-499 and 39% have below 600. it means the group with under 600 is almost twice the size the group with 700-800. also note that 21% of the enrollees scored less than 1200 in their highest sitting.

[/quote]
</p>

<p>I can't access the site (it isn't working for me), but what you seem to be saying doesn't make sense. 21% scored below a 600 on a section, so therefore 21% scored below a 1200? If this is what you're saying, then I'd tell you again that that means 21% scored below a 600 on one section but not necessarily on the others. They most likely did much better in other sections.</p>

<p>And once again, there are three sections to consider.</p>

<p>
[quote]
this should bring the debate to closure, IMO.

[/quote]
</p>

<p>No, it doesn't. For one, you just "proved" that 21% had below a 600 on one section, not that 21% had below 600 on every section combined. In addition, the whole point here is that superscoring would elevate these numbers.</p>

<p>The data from the College Board supports this.</p>

107

<p>Harvey Mudd?</p>

108

<p>kyledavid,</p>

<p>if you really like 3 sections, you can have 3 sections. i am very open-minded. ;) the problem is it makes comparison difficult as most privates report their average in M+V, not 3 sections. but if the current difference in M+V between berkeley and privates like northwestern, williams, or georgtown is 80-100 points , then the difference in 3 sections will probably balloon to 120+. seriously, it all works roughly in proportion. so it doesn't matter if it's 2 or 3 sections. </p>

<p>
[quote]
For one, you just "proved" that 21% had below a 600 on one section, not that 21% had below 600 on every section combined.

[/quote]

that's the average of M+V of the highest sitting; that means 21% of CA enrollees scored below 1200 (M+V). yes, it could be 450 V and 650V. But it also could be 500 and 500 too. i don't really know how you can say "they most likely did much better in other sections" since scores like 550V/550M qualify also. regardless if the other section is above 600 or not, the total of that highest sitting is below 1200 and this applies to 21% of enrollees from in-state.<br>

[quote]
the whole point here is that superscoring would elevate these numbers.

[/quote]

oh dear, you seriously believe top privates would have 21% of enrollees scoring between 1200? and that the reason it doesn't show up is because they superscore? what can i say? you can believe in whatever you want.

[quote]
The data from the College Board supports this.

[/quote]

i've already explained to you twice it doesn't. note how you are the only person that keeps saying this, not blueboy, not tokenadult, and not powergrid. collegeboard simply says scores improve upon retake and i don't need to see it from collegeboard to know this. now how in the world does it support your hypothesis that superscoring has jacked up the overall average by 60 points for privates? even tokenadult doesn't think superscoreing would have great effect and believe that most students improve in both sections upon retake. </p>

<p><a href="http://statfinder.ucop.edu/statfinder/default.aspx%5B/url%5D"&gt;http://statfinder.ucop.edu/statfinder/default.aspx&lt;/a> try again. it works perfectly fine for me. by the way, if you haven't noticed from my post #100 (statfinder will show you also if it finally works for you), you'll see between the admits and enrollees, there's a noticeable drop in percentage of groups in higher scoring range. this totally contradicts what you said earlier about the admits and enrollees having similar average without any explanation/supporting data. honestly, you've been making statements repeatedly like that without any logic or support and looks like you are caught by real data this time.</p>

109

<p>
[quote]
that's the average of M+V of the highest sitting; that means 21% of CA enrollees scored below 1200 (M+V).

[/quote]
</p>

<p>Where does it say that? What you listed seemed to be single sections. (Unfortunately, I still cannot access the site.)</p>

<p>
[quote]
oh dear, you seriously believe top privates would have 21% of enrollees scoring between 1200?

[/quote]
</p>

<p>No, that's not what I'm saying. You are attempting to claim a huge gap between Berkeley's and top privates' SAT. You use that data as support. My point is that if Berkeley superscored, the data would be different and the gap would not be as big as you try to make it seem.</p>

<p>
[quote]
i've already explained to you twice it doesn't.

[/quote]
</p>

<p>No, you haven't. You've feebly explained away data that hurts your argument by using two sections and saying that the best single sitting is what matters. You haven't explained anything other than your same point, over and over again, and it still makes no sense.</p>

<p>
[quote]
now how in the world does it support your hypothesis that superscoring has jacked up the overall average by 60 points for privates?

[/quote]
</p>

<p>And I explained this before.</p>

<p>From the data, it is apparent that a retake can boost the score in a section. I'm assuming a few things: a) most of those admitted to Berkeley are going to have taken the SAT more than once, and b) they are going to have a higher average increase in each section than the national average. With that in mind, it's easy to see that a person's SAT were to be higher if superscored (for those who retook and did not improve in all three sections). This would also help to explain why Berkeley's range is so wide (as I've explained multiple times, so if you want elaboration on that, go back and re-read).</p>

<p>
[quote]
honestly, you've been making statements repeatedly like that without any logic or support and looks like you are caught by real data this time.

[/quote]
</p>

<p>I could say the exact same thing of you. (You try to write off the College Board data, even though it makes sense.)</p>

<p>
[quote]
without any explanation/supporting data.

[/quote]
</p>

<p>I mentioned the difference between the 2070, 2040, and the 1990.</p>

<p>I think we can end this discussion here. It's clear that there is no definite resolution to this matter with the current data. Until we have more information, it's pointless to continue.</p>

110

<p>
[quote]
Where does it say that? What you listed seemed to be single sections. (Unfortunately, I still cannot access the site.)

[/quote]

It says on the website. It says it's best viewed in IE 6 or higher and Firefox 1 or higher; your browser may be outdated.

[quote]
No, that's not what I'm saying. You are attempting to claim a huge gap between Berkeley's and top privates' SAT. You use that data as support. My point is that if Berkeley superscored, the data would be different and the gap would not be as big as you try to make it seem.

[/quote]

Didn't you say they came out to be about the same based on your analysis a while back? Finally you acknowledged there's a gap. :)</p>

<p>
[quote]

No, you haven't. You haven't explained anything other than your same point, over and over again, and it still makes no sense.

[/quote]

Care to clarify what "makes no sense" in post #109?<br>

[quote]

From the data, it is apparent that a retake can boost the score in a section.

[/quote]

Never said otherwise and that's pretty obvious.

[quote]
I'm assuming a few things: a) most of those admitted to Berkeley are going to have taken the SAT more than once, and b) they are going to have a higher average increase in each section than the national average. With that in mind, it's easy to see that a person's SAT were to be higher if superscored (for those who retook and did not improve in all three sections). This would also help to explain why Berkeley's range is so wide (as I've explained multiple times, so if you want elaboration on that, go back and re-read).

[/quote]

Your assumptions got very little to do with collegeboard data. Those come from your own thinking and words. Even tokenadult doesn't think superscore has great effect on <em>overall</em> SAT average. I don't think so either. The collegeboard data alone reveal nothing about which assumption is closer to the truth. Now go back to your above statements; it's easy to see they have little substance:

[quote]
It's easy to see that a person's SAT were to be higher if superscored (for those who retook and did not improve in all three sections).

[/quote]

You are simplying stating the obvious--what superscore does. We all know that. What you have not done is to show that "those who retook and did not improve in all three sections" contribute sigifnicant fraction of the enrollees. As the collegeboard data shows, about half of the test takers had one sitting and another 40% or so retook once. I don't think it's so far-fetch to say most people who retook improved in all sections. Also, you also need to show one more thing: among those who retook and did not improve in all sections, most of them have a drastic drop in one of those sections. Why? Let's say we have 3 people: A took one sitting, B improved in all sections, and C improved in one but drop in another. How much bump from superscore on C is needed to have a bump of 30 on overall average? 90! Try to play with some combination of first sitting and 2nd sitting scores and you will see you have to come up with atypical result to get 90-pt bump from superscore (Given a first sitting scores, I actually find it impossible sometime to come up with the second sitting scores that would produce such result). Not to mention it's likely less than 1 out of 3 applicants actually fall into that category. This is why I don't believe superscore would have great impact on "overall" impact (despite the fact it can on certain individuals). This is also why ACT has been corresponding to SAT fairly well in Michigan and Northwestern data.</p>

<p>
[quote]
This would also help to explain why Berkeley's range is so wide

[/quote]

I disagree. Let's take a look at the following data from collegeboard:
Michigan (public): CR 580-690 M 630-730 (no wide range)
Cornell (private): CR 630-770 M 660-730 (wide range in CR)
Pepperdine (private): CR 550-670 M 560-680 (wide range in both)
Tufts (private): CR 670-750 M 670-740 (narrow range)</p>

<p>Michigan doesn't superscore yet it's range is narrower than Cornell in CR and Pepperdine. This shows you the range has very little to do with superscoring. It has everything to do with the range of the applicant pool. The Ivies that are more selective than Cornell show narrower range. On the other hand, many 2nd-tier privates have wider range. The reason why Michigan doesn't have as wide of a range as Berkeley is partly because they take 1/3 from OOS whereas Berkeley takes only 9% from OOS. As statsfinder shows <a href="http://statfinder.ucop.edu/default.aspx%5B/url%5D"&gt;http://statfinder.ucop.edu/default.aspx&lt;/a>, Berkeley OOS pool has a narrower range and if Berkeley takes a larger piece from that group, it's range would look more like its OOS pool, i.e. narrower. Makes sense?</p>

111

<p>Let's simply end this discussion here. Neither I nor you gained any more insight by the past few posts. I daresay others haven't, either. We simply don't have the data. Even for things like this:</p>

<p>
[quote]
I don't think it's so far-fetch to say most people who retook improved in all sections.

[/quote]
</p>

<p>We don't have the data to make such an assumption. So we will have to be resigned to that unless incontrovertible data comes along.</p>

<p>(By the way, though, I never said that the wide ranges were due only to the superscoring; I simply said that contributed to it.)</p>

112

<p>Actually KyleDavid - most people do increase in all sections. And quite frankly, nearly everything you have said has been extensively countered with data. I think it is quite clear that the superscoring advantage is minimal.</p>

<p>I'm just frustrated that they don't do the same for the ACT.</p>

113

<p>
[quote]
Actually...most people do increase in all sections. And quite frankly, nearly everything you have said has been extensively countered with data.

[/quote]
</p>

<p>What data? Where? Please post it so we'll all be enlightened. </p>

<p>Statistically, increasing in every section is not likely for "most people," so I do think it is "far fetched." Every CB SAT report includes a summary that says something to the effect: upon a retake, at your score level xx% of students increased by yy points, and zz% decreased by aa points. Thus, CB's own public data indicate that some scores go up and some go down on a retake. But, where is the data to show that those increasing scores are the same students, increasing on all three tests?</p>

<p>fwiw: my fact-free assumptions is that superscoring helps, but not by much at the highly selective school since the SAT scores are skewed left.</p>

114

<p>^i agree that my "not so far-fetched" assumption isn't yet proven and i agree with what you said in the rest of your post.</p>

<p>the key is not only thinking about whether it's likely most people improve in all sections but also whether it's likely that those who don't improve in all sections get flip-flopping results. superscoring has great impact only when your retake gives you an opposite result (big improvement in one but big drop in another). given the caliber of students at highly selective schools and the test taking skill they have, i'd be surprised if many of them get that kind of result. you raised a good point about the limited impact due to left-skewing scores. </p>

<p>that's why i think the impact of superscoring among those that benefit from it is rather limited on average and when you average that impact across the whole student body, including those that take it once and those that improve in both sections, the impact is further diluted significantly. for example, if superscore bumps up the average score of those that are affected by 20 points, once you average that 20 points across all others that aren't affected, the net bump would be a single digit.</p>

115

<p>Why do schools superscore, even if it doesn't statistically help. What is the motivation for doing it? Have we seen any statements from universities why they superscore SATs?</p>

116

<p>Note to all…
Sorry to be behind the curve but what do you mean when you use the term “superscore SATs”</p>

117

<p>^ Taking the best score for each individual SAT section across multiple test attempts. Some schools only count SAT scores in a "single sitting".</p>

118

<p>thank you ......</p>

119

<p>Schools do this so that they appear to be 'working in the best interest of the students'. Also, I don't see it as that different from 'the best single sitting', you are just giving the benefit of the doubt. If you only accepted version 1.0, then yeah, superscoring would seem odd to me, but that's not the case.</p>

120

<p>Colleges just set a rule to know how to deal with multiple sets of scores from the same applicant. They can set any rule they like, but most privately operated colleges have settled on the rule of considering each section of the SAT I separately if an applicant took that test more than once.</p>