Old SAT scores compared to redesigned SAT scores.

LMM, I’ll be interested to see more results with that level of detail. I think it’s going to vary by college (or at least, tier.) And what draws kids to apply to particular schools, what the admissions pools are made of. Many kids who now score higher on the NS might throw their hats into rings they previously thought were out of range. They still have to pass full holistic muster.

@LadyMeowMeow you’re still working from the faulty assumption that the two groups of SAT takers are equivalent. If the groups aren’t equivalent, then comparing their ranges makes no sense at all. Also, you have no idea how large these groups are relative to each other. Do you really think a 1480 on the old SAT is equivalent to a 1430 on the new one, even though the CB says it is equivalent to 1510 on the new one?

“the CB says it is equivalent to 1510 on the new one?”

I just don’t think that many people think that the College Board has much credibility anymore as it relate to concordance.

“Do you really think a 1480 on the old SAT is equivalent to a 1430 on the new one, even though the CB says it is equivalent to 1510 on the new one?”

Possibly, yes.

I would look at this another way. 1430 New SAT concords to 1380 Old SAT according to CB’s tables. Do we really think that the pool of New-SAT accepted applicants to a selective college scores 100 points lower (on the scale of Old SAT) than the pool of Old-SAT accepted applicants?

Maybe my question isn’t all that different. Just my personal opinion, but it doesn’t pass the smell test - the pools would be too vastly different, the gap too big, to be explained by “the smarter kids were all advised to take the Old SAT fall of junior year and the less-smart ones waited until spring and took the New one.” I am skeptical of the CB’s work product.

I don’t have time to check before I run out of time to edit this post, but perhaps we should check the prior year’s middle 50% for SAT and compare this year’s Old and New to that.

<<and when="" others="" say="" the="" really="" smart="" kids="" took="" old="" sat="" your="" pretty="" much="" saying="" our="" aren’t="" smart,="" which="" makes="" me="" a="" little="" defensive="" i="" read="" it.="">>

Actually, what was being said was that the old SAT (or the ACT) was a much better test than the new one for students who planned to do serious prep. This is an objective fact, since there were like ten previously administered tests available for the old SAT, vs zero such tests for the new one. These actual tests with legit scores were essential for such preparation. Students following this path would choose either the old SAT or the ACT.

Really there would be no reason to take the old SAT, which was only offered through January of junior year (before most students take even their first ACT/SAT) unless you anticipated getting a high score. This makes the old SAT cohort a self-selecting group of high scorers.

In terms of absolute numbers of high scorers, I assume the ACT will have the overwhelming advantage this year as most ppl did not want to deal with the unpredictability of the new SAT. However, in terms of percentage of high scorers, it is reasonable to assume the old SAT cohort was strongest.

Of course, the new SAT would also have high scorers who are just naturally good test takers and didn’t need to study. But to the extent that high scorers across all tests do a lot of prep, the new SAT would have lost takers to the old SAT and the ACT.

My understanding is that colleges are not going to report new SAT mid-50 score ranges on their CDSs until 2018. I believe this is because the ranges this year are too distorted by the mix of available tests.

@keiekei “Do you really think a 1480 on the old SAT is equivalent to a 1430 on the new one, even though the CB says it is equivalent to 1510 on the new one?”

As of now, yes, I think it’s 99% more likely that a 1480 on the old SAT is equivalent to 1430 on the new SAT on than it is to 1510.

Please know that I have sympathy for your argument, since I was a “CB true believer” when they released the “preliminary concordance tables” for the 2015 PSAT. I used all the arguments – they have all the data, they wouldn’t put their credibility on the line if they didn’t know what they were doing, they have expert statisticians – and then watched in grotesque horror as their tables were embarrassingly swamped by empirical data. (I admit that my vitriol was concentrated by the fact that I was rooting for a certain selection index that my DD barely missed.) Believe me, it’s not a crazy idea to think that CB completely blew this round of concordances, too, especially if you think, as I do, that they got it basically right – up until the mark where the CC/ elite college crowd starts to tune in, i.e. 700+ or so. If the concordances at that level were right, then the data from Georgia, Vanderbilt, BC, Williams et al would show HIGHER numbers in the new SAT at the same quartiles as the old SAT. They don’t.

As for “separate pools,” from an admissions standpoint, it starts off as one big pool: who is good enough to take? And the answer, for now, seems to be: to be admitted, students who took the old SAT need to have higher scores than the ones who took the new.

We don’t know that yet, though, just based on a numbers report. At my school, they’re looking for higher New SAT scores.

Last year’s 700 needs to be higher on the new test, to be seen as functionally equivalent. Roughly and depending.

Try not to look at this as the mid range is lower, so it equates absolutely. Many factors could be in play. And that was Early.

@lookingforward “At my school, they’re looking for higher New SAT scores.”

On what basis are they doing that?

<<the pools="" would="" be="" too="" vastly="" different,="" the="" gap="" big,="" to="" explained="" by="">>

The pools can change radically in response to perceived advantages and disadvantages, especially at the extremes. For example, the frequency of a 36 score on the ACT increased from 1 in 12,000 in 2001 to 1 in 935 last year, as the test gained in perceived legitimacy and market share among the most competitive applicants. What changed was not the difficulty of the test, but the composition of the applicant pool.

As I have explained, circumstances last year created a clear set of incentives for a self-selecting group of competitive students to take the old SAT while it was still around. That doesn’t mean every smart kid took it. It means the cohort that took the test was self-selecting and had a different composition. This can be true and it can still be the case that the children of the wonderful parents on this thread are brilliant and did great on the new SAT.

@keiekei “As I have explained, circumstances last year created a clear set of incentives for a self-selecting group of competitive students to take the old SAT while it was still around. That doesn’t mean every smart kid took it. It means the cohort that took the test was self-selecting and had a different composition.”

At this point I’m ready to bow out and agree, as politely as possible, to disagree. I think the crux of the issue is that I’m not as convinced as you that the incentives to take the old SAT outweighed the incentives to take the new one, at least not enough to form significantly different pools. Anyway, thank you for your informed perspectives on all these matters. Soon we’ll all know more and the matter will be moot for this year’s students.

Has anybody seen any statistics showing the % breakdown of tests submitted, Old SAT vs New SAT vs ACT, at various schools? Highly competitive schools? That would be interesting.

@LadyMeowMeow It’s too early to know what they actually found in RD. And Early is a self selected group of applicants. But based on that, and knowing high schools, their rigor, standards, patterns of what kids had been submitting (grades/rigor and scores) in recent years, they saw the NS scores trend higher. In some cases, kids are submitting both old and new scores. None of this is enough to declare anything yet.

It’s more complicated than the assumption a traditional test and a modified one somehow equate. Even GA didn’t tell (afaik) how many in each pool. It’s possible more of the better qualified kids were ready to take the old test, earlier. Not gaming, but moving through the steps. They had their scores in jr year or earlier and saw no need to take the new test.

Just because GA’s early admit NS mid range is lower tells us zip.We have a lot yet to learn and this whole convo could take place later, when more is published.

Not trying to make it more complicated, but I will try. Students stronger in Math might see it that half their score will be higher, so take the New SAT in which Math counts for 1/2 not 1/3 of the score on the old SAT. Maybe some student “self selected” this option of taking the new test. Yes, the “self selecting” characterization irks me, big league. Does anybody have any information on how many kids took both the tests?

@LadyMeowMeow <<please know="" that="" i="" have="" sympathy="" for="" your="" argument,="" since="" was="" a="" “cb=”" true="" believer"="" when="" they="" released="" the="" “preliminary=”" concordance="" tables"="" 2015="" psat="">>

I don’t know what a CB true believer is, but I don’t think I’m one, considering I told DD to skip the new SAT back in 2014 because I suspected there could be issues with the new test. That is just a recognition that creating a large-scale standardized test is really hard. Why be a guinea pig if you don’t have to?

In any case, the argument I’m making doesn’t actually depend on the published concordances being “correct;” the only question is whether the adcoms will follow them.

<>

I’m going to guess this is why they attached the qualifier preliminary to the preliminary concordance tables.

<<it’s not="" a="" crazy="" idea="" to="" think="" that="" cb="" completely="" blew="" this="" round="" of="" concordances,="" too,="" especially="" if="" you="" think,="" as="" i="" do,="" they="" got="" it="" basically="" right="" --="" up="" until="" the="" mark="" where="" cc="" elite="" college="" crowd="" starts="" tune="" in,="" i.e.="" 700+="" or="" so.="">></it’s>

There is an empirical assertion in there that may or may not be true.* But just maybe you are making this assertion for subjective reasons, given your stated bitterness toward CB.

  • I think the way we will know about the validity of the concordances is to check the CDS score ranges for the HS class of 2018, when distortion caused by the old SAT is minimal and universities finally start reporting new SAT score ranges (they won't this year, on the CDS). But again, that is a separate issue from whether it is reasonable to believe, right now, that the adcoms are using the concordances.

<<if the="" concordances="" at="" that="" level="" were="" right,="" then="" data="" from="" georgia,="" vanderbilt,="" bc,="" williams="" et="" al="" would="" show="" higher="" numbers="" in="" new="" sat="" same="" quartiles="" as="" old="" sat.="" they="" don’t.="">>

You are now making another assertion, namely, that the adcoms at the schools you mention, despite lacking any expertise in statistics or access to the data used by CB to calculate the concordance tables, have decided, on the fly, to establish new ranges (based on what??) that are completely out of whack with the ranges provided by the CB. In the old days, they used to say that nobody could get fired for buying IBM. In this admissions cycle, I’d say the default assumption for most adcoms is that nobody will be fired for using the concordances provided by the CB itself. I would not say that the default assumption is for non-experts to magically create ad hoc score ranges.

If you follow Occam’s Razor here (that the adcoms will use the CB concordances), the score range differences can be explained by other factors, some of which I enumerated before.

<<as for="" “separate=”" pools,"="" from="" an="" admissions="" standpoint,="" it="" starts="" off="" as="" one="" big="" pool:="" who="" is="" good="" enough="" to="" take?="">>

I wasn’t talking about it from an admissions standpoint; I was talking about it from the test-takers’ standpoint: what might motivate some testers to choose one test over the other, and could this result in appreciable differences in score distributions among the various tests? While adcoms will concord these groups (using the concordance tables, as mentioned), back when these cohorts were planning to take their tests, there were specific reasons some students took the old SAT in particular. This was a self-selecting group of stronger than average students. If this group is stronger than the new SAT group percentagewise, then it is no surprise that their scores are higher.

That doesn’t mean that everybody who took the old test is a genius, and everybody who took the new test is a dope. But there could be a higher—much higher—percentage of high scorers in the old-test cohort, and I have outlined why. In simplest terms, unless you were anticipating/desiring a high score, there was no reason to take the old SAT. Just imagine what it does to the score distributions if the bottom half of scorers nationwide skip the test. And then at the top end, imagine what happens if, say, 15% of people in the top 0.5% across all tests (old SAT, new SAT, ACT) select the old SAT, which cohort might amount to less than 3% of all test takers in the HS class of 2017. Of course, you can use WAG numbers other than 15 and 3 here, but I would argue that the first is significantly greater than the second.

<<and the="" answer,="" for="" now,="" seems="" to="" be:="" be="" admitted,="" students="" who="" took="" old="" sat="" need="" have="" higher="" scores="" than="" ones="" new.="">>

In college admissions, some groups always need to have higher scores. For example, unhooked kids. Precisely the groups who are likely to have favored the old SAT.

“This was a self-selecting group of stronger than average students” Whats the proof? If your stronger in Math than English, one strategy is to take the New SAT. If you don’t need to study that much, you may want to go up against (take New SAT ) kids who don’t have time to prep much and or don’t have expensive tutors. These kids who don’t need much prep to do well aren’t stronger? I think your focused on one group that may be stronger, but there are a variety of reasons kids took the different test.

One reason to take the old SAT is kids who were recruited for athletics, and wanted early scores to give to the coach to give them a tip in college admissions to a school they fall below the average to get in all on their own. Also kids who are better at English and Writing, and want those scores to count more may take the old SAT.

I was trying to concord Old CR with New EBRW and I can’t figure it out without the Old Writing being added to CR or knowing the New reading and writing subscores that I’m trying to concord to Old. It’s much more straightforward to compare math scores alone or even to compare total SAT scores between 1600 and 2400 scales than it is to try to compare Old CR + math to New EBRW + math. But, CB has the page 4 table concording Old and New total scores both on a 1600 scale - feels like something is missing.

On motivations to take Old vs New: New SAT math is half the total rather than one-third; this is an absolutely critical reason that I would have advised my student to take the New SAT instead of the Old one if my student had been a junior last year. My student seems likely to score higher on math than on EBRW. Conversely, for students who tend to do better on reading and writing than on math, it would have made sense to take the Old SAT rather than the New one.

I agree w/ @keiekei that the Common Data Sets will not be informative until the Sets for the 2017-2018 admissions season are published and that really stinks for the current junior class as they try to craft their college lists. I am still trying to find out when these are typically published - will it be early summer or late summer/early fall?

Lastly, our state flagship put current freshman middle 50% scores on its admissions website separately for Old total, math + CR, and New total, math + EBRW, with the New score range being precisely concorded from the Old using the page 4 table where both are on a 1600 scale. The middle 50% scores were put up on the state U’s website early last fall, prior to early admissions. Apparently, this university was expecting to use the CB’s concordance tables. (I haven’t seen any recent admissions data for this school.)

@keiekei “I would not say that the default assumption is for non-experts to magically create ad hoc score ranges.”

I don’t disagree with you on this point, but I think that one big unknown that we just don’t (and won’t) know about UVA’s, William’s and UGa’s admissions criteria is how big an impact SAT scores had on admissions decisions for their EA cohorts. If it had a large % impact then maybe they are ignoring the concordance; however, if SAT scores are a relatively insignificant factor then the lower new SAT scores for the admitted students is a red herring - as we all know, correlation does not equal causality!

@Akqj10 <<“This was a self-selecting group of stronger than average students” Whats the proof?>>

I would suggest you read my earlier posts on this thread. Not going to repeat myself.

<<if your="" stronger="" in="" math="" than="" english,="" one="" strategy="" is="" to="" take="" the="" new="" sat.="">>

I think it really depends on what level of school and score ranges you’re shooting for. At many top schools, Writing is discounted or completely ignored. UChicago is very up front about this. And for example, Williams doesn’t give a crap about the old SAT Writing—you can google “Williams Academic Rating” to see a rubric of the nine AR levels into which applicants are categorized (spoiler: you need to be an AR1 or AR2 to get in unhooked). The AR used by Williams, and the similar rubric used by Amherst, do not use the SAT Writing. Also, for example, the Academic Index used by the Ivies for athletic recruiting equates the verbal with the math (they added Writing and CR together, and divided by 2).

Secondly, at the very competitive level of student and institution implicitly under discussion here, the SAT Math is not nearly as important as the verbal. Look at the 2015 percentile distributions by ethnicity for the unhooked—800 is only 95th percentile for Asians, whereas 800 CR is 98th percentile. Trying to hide a weak verbal score behind an 800 Math isn’t going to fool anybody.

Also (and this is something I suspect much of the white parent-age readership on CC does not fully appreciate), even if you’re smart and can easily get an 800 on the Math, getting the same on the verbal is no cakewalk if you live in a non-English speaking household, or one where the parents are not college-educated native English speakers. This is a huge handicap for many Asian students, one that in most cases can only be overcome with a great amount of preparation. Again, this points to the strategic import of knowing well in advance which test you’re going to take junior year.

<<if you="" don’t="" need="" to="" study="" that="" much,="" may="" want="" go="" up="" against="" (take="" new="" sat="" )="" kids="" who="" have="" time="" prep="" much="" and="" or="" expensive="" tutors.="" these="" do="" well="" aren’t="" stronger?="">>

I actually discussed this before. Obviously obvious, as I pointed out. I never claimed that every group of stronger kids took the old SAT. But the old SAT group was probably MUCH smaller in terms of total number of takers, so having one or two sets of strong kids strongly prefer the old test can have a huge impact on its score distributions.

<<one reason="" to="" take="" the="" old="" sat="" is="" kids="" who="" were="" recruited="" for="" athletics,="" and="" wanted="" early="" scores="" give="" coach="" them="" a="" tip="" in="" college="" admissions="">>

This is a really small group, in terms of the total test-taking population. It would be utterly swamped by the much larger number of below-average takers (i.e., half of the total test-taking population), nearly all of whom would have taken either the new SAT or the ACT last spring.

Even if math now counts for half the total, it’s still its own number. Each score can matter more than the total, depending on the school and the potential major. That stem kid better score well on the SAT Math. He/she might indeed get a little leeway on the verbal. Same with the ACT, where composite is nice, but the subscores matter.

 I agree that some kids who took the old SAT were great students who had prepped for many years and had a good chance to score high. But to make a 100 point difference seems statistically inconceivable. 

 I think the true concordance is somewhere in the middle. I don't think and old SAT is 50 points better at the 1400 range ( concordance table ), and I don't think a new SAT is 50 points better than the Old ( Early decision results from select colleges. I'm feeling right now the gap is about 10 points. In any case, can we agree that since we are discussing it this much and can't figure it out, and since it can have a pretty big impact on large merit scholarships, not to mention admission results, that college board really screwed this up.