He states in response to a question: “Both the percentile scores and concordance tables were developed from studies done prior to the March test. Both are subject to the same concerns about the accuracy of the sample population and how testing was conducted.”
Given his credentials and experience in the testing business, I found this statement credible.
As for a method of converting scores en masse, here’s one that might work. Let’s say I’m Harvard and I have 30,000 old SAT scores from last year. This fall, I will have 20,000 new SAT scores (assume the rest of the applicants took ACT or old SAT). I could randomly sample 20,000 from last year’s old SAT scores. I put the old SAT scores on one histogram and the new SAT scores (with the concordance correction applied) on the other histogram. Compare the curves. If the concordance was good, they should be identical. If the concordance over-penalized the new test, an upward correction of the new scores should get the two graphs in alignment. I then apply that correction to the concordance table to get my corrected concordance.
"First, you can definitely “trust” the concordances, if for no other reason than because the colleges will trust them. At the admission committee table in 2016-2017, the concordances will be considered gospel and a 1530 new SAT will be “worth” the same as a 2230 old SAT or a 34 ACT. (Adam Ingersoll)
“Knell, Both the percentile scores and concordance tables were developed from studies done prior to the March test. Both are subject to the same concerns about the accuracy of the sample population and how testing was conducted. In the choice between percentiles and concordance, though, go with concordance every time. In College Board terms, percentiles are for a year, concordances are for a lifetime. It’s the concordance that links their old flagship test to their new.” (Art Sawyer)
While schools can certainly graph, map and statistically analyze new scores all they want, do keep in mind that they would likely only do so if they had reason to distrust the concordance that College Board provided (how can we possibly know whether they do or not?) and - it’s only for one year’s worth of data (very limited payoff). One can probably make an argument for running an analysis if the majority of other submitted standardized tests are ACT’s and old SAT’s but if 2/3 of all submitted test scores are new SAT’s then not sure there would be a need - how does it change the applicant’s position compared to pretty much everyone else? To use an extreme example - if 100% of submitted scores consist of rSAT’s then they aren’t going to care a whit about how these compare to old SAT’s.
What makes anyone think that any such “adjusting” would be in the UP direction? Why not DOWN? There are anecdotes supporting both directions on this and other threads.
The best way to increase the strength of an application is to submit more than one set of scores to show consistency or just submit your ACT and/or old SAT if either is higher than the concorded rSAT. If you only plan on submitting the rSAT but you don’t like your score then re-take it in June or Oct. In other words, submit your best score unless you need to submit all scores, or submit all those scores that help you. There’s the entire issue of what possibly second guessing the adcoms will do to testing strategy but that’s getting WAY too complicated.
Edit/Update: I’d love to have my D3’s 1500 adjusted up a bit. I’m just not counting on it and neither is she.
Make sure you are NOT looking at the National Representative percentile because it’s not relevant to you, most likely. Your 1460 should have a User Group percentile of 98th or lower. The reason I know this is because my D3 has a 1500 and her User Group percentile is the 98th. CB has not released percentile tables as of yet so all we have to go on is User Group percentiles on the score reports (marginally useful at best) and concordance tables (probably significantly more useful).
If you concord your current SAT to the ACT and previous SAT for reference (use the Score Converter Ap. or the Excel spreadsheets provided by CB) you find that your score converts to about a 32 ACT (at the 98th percentile) and to an old SAT of about 1410/2090 (96% would be below this score). So your score is probably capped at either end by 96+ to 98% using these tables. Unfortunately given that these are estimates it’s difficult to pin your percentile down to something more specific.
Interesting that 1510 must be the cutoff for 99% User Group. D17 took it in March and got a 760 EBRW & 750 Math and her total score is listed as a 99th User Group percentile. Also, her 760 EBRW was 99th User Group, while her 750 Math was only 97th User Group.
@TimEnchanter, that’s actually helpful info so thanks! It will be interesting to see how accurately the User Group percentiles match what is resulting from concordance. So far there seems to be a better alignment compared to PSAT.
Question: Is it ok to take the new SAT twice (I got 1520 the first time)? After I took the old SAT twice (superscore 2260) or does this look bad to colleges?
The new SAT’s inflate the scores about a hundred points. They are significantly easier and the curve is now skewed higher. Everyone is now crowed at the upper scores rather than the middle. Colleges will now report very high scores which are basically fake and inflated.
@crocadile1246 Your 2260 superscore is excellent. Colleges will not superscore across both tests. WHy bother taking the new SAT again - the 1520 equates to a 2210. To score higher than that on the new SAT, you’d have to get a 1550 or higher. Which is difficult. I would focus my time on essays and ECs.
Basically, the colleges don’t trust you not to accidentally/intentionally omit your scores from a particular test date, so they require all scores to make sure that they get the best of everything. This is especially true for superscore schools.
These types of colleges are requiring all scores, but only considering the best scores. Depending on the school and its institutional procedures, the admissions officials who read your application may or may not know how many times you took the SAT.
@ivysource Are these concordance tables going to be taken in full weight by colleges? I got a composite score of 1390, which I originally thought was equal to around a 2070-2080. However, I found that it is worth 100 points lower at around 1970-1980. I am very worried that my SAT scores are going to be considered low for colleges even though this score was seemingly close to a 2100.
@mclaughlin34 those tables were constructed precisely because College Board maintains that the new test is NOT equivalent to the old one. Don’t look at old percentile tables to interpret new scores. You’ll need to look at the concordance table and it’s prudent to assume that colleges and universities will be doing the same (as this is precisely why CB constructed them in the first place).
Indeed, I was chatting with an admission rep from - don’t remember where - and they specifically said that they were planning to just convert the new SAT scores using the concordance table to the old values.
I had a question about how exactly they grade/score the subsections of the test. My question is specific to the math section. I received a 720 for Math and had 6 questions wrong, but my friend got a 770 and had 4 questions wrong. How did he have only 2 more questions right than me but a score 50 points higher than mine? Can someone explain? Thanks.