@Bozusaki - attempting to answer your question in your post #4752:
NMSC must have had a way of equating the old SAT to the old PSAT for situations such as yours. Not sure it was as simple as just looking at the SAT score divided by 10 because that method doesn’t equate the two old tests in terms of percentiles.
Currently there isn’t anything published by College Board that directly allows us to convert the new SAT to a PSAT-equivalent. We do know that when you take the PSAT, that score is supposed to be your SAT score on that day. That’s because the actual SAT is a bit longer and will have “grade appropriate” content that’s supposed to be a bit more challenging than the PSAT. Both tests are scaled similarly (i.e. scores spanning about a 600 point range); however, the SAT distributions for Math and Verbal are shifted +40 from the PSAT. Presumably that means you are expected to increase around 40 points just due to more knowledge/schoolwork/exposure to tougher content as you progress through your college-prep curriculum at school. This idea gets pretty complicated very quickly when you look at their actual benchmarks on the PSAT and SAT reports which are nothing like the midpoint of 460 for the PSAT or 500 for the SAT. But the +40 increase for each section did seem to be their initial goal.
Returning to the issue at hand - below are two ways of tackling the problem. Keep in mind that these absolutely do not represent what you would have gotten on the PSAT had you actually been able to take it. Rather, it reflects what your current SAT says about what your PSAT might have been six months to a year ago when you hadn’t yet covered the same amount of college-prep material in your schoolwork as you had by early March 2016.
Concord by equi-percentiles across the board - you’ll need to print out the old SAT percentile tables as well as the old PSAT tables. Example: New SAT = 1550. This concords to a 2280 (old) which is a high 99th percentile on the historical tables. That percentile, in turn, corresponds to about a 223 old PSAT. We know from the preliminary PSAT concordance tables that the 223 concords to a 1470 new PSAT. Therefore, the PSAT estimate corresponding to a new SAT of 1550 is 1470. That’s about a 218 - 222 SI (new).
OR - MUCH SIMPLER:
Simply subtract 40 from each section of the new SAT (so 80 in total). Using our example from above, if the new SAT of 1550 consisted of 770 (EBRW) and 780 (M), then the resulting estimate for new PSAT is 730 (EBRW) + 740 (M) = 1470. This method has the distinct advantage of easily translating into the NMSQT Selection Index: SI = (2EBRW+M)/10 = (2730+740)/10 = 220.
The two methods are really two ways of doing the same thing. The assumption is that the PSAT and SAT distributions are really identical, only the SAT distributions are shifted +40 to the right. Therefore, when you subtract 40 from an SAT score to arrive at a PSAT score, you are really still at the same percentile. Thus Method 2 just seems to be a shorthand for Method 1 but also doesn’t get into the messy aspect of estimating SI’s from total scores, or trying to equate percentiles of section scores, etc. Therefore, it’s the preferred method, IMHO.
“In some ways, I trust the earlier version, because it was not placed into a forced alignment with the SAT concordance.”
Totally agree with Art on this. The flaw that CB made, I think, is that they equated the old PSAT scores on a percentile basis with old SAT scores. If you look at historical tables of the two you know that’s certainly not the case.
@GMTplus7 I truly believe that’s what those Preliminary Concordance Tables actually were. They must have been actual percentiles from the October tests concorded to the old PSAT using the previous year’s percentiles (which never were published directly). What else could they have been? To do concordance by equating percentiles you actually need two decent sets of percentiles. As the reported percentiles were so off-base, they needed to rely on something else. If not the actuals, then what?
@snicks1234 – I agree on the ACT essay situation - the format was changed a bit recently & I think that threw off kids and the readers. But there has been ridiculously inconsistent grading & rescoring. With practice I think students can do pretty well – and with time for a few ACT administrations, I would think the grading should smooth out a lot.
@snicks1234 The ACT essay is the least of colleges worries this year because of this SAT fiasco. I would stick with ACT and not worry about it. That said, It was not to smart of ACT to roll out this essay changes this year. If they had stayed as is, they would come out of this smelling like roses while the SAT is smelling more and more foul as time goes on.
@Mamelot So you think the prelim concordance tables were based on actual scores of the October test concorded to percentiles from previous tests, but then they threw in a red herring on the SI percentile chart by using the research study and this is why the SI percentile chart released in January using the research study does not jive to the concordance released in January?
@suzyQ7 For us it kind of equates. He took the old SAT and got 2230,then new one 1540,so results were similar if tables are correct. It is messy generally though and I wonder if CB even knows how many parents are as well informed as these on here. I would think if they did there would be some kind of response to the questioning of the concordances not being available.
Parents are one thing, but counselors are another. As far as I know, they have the same info we do and must be just as confused/annoyed as we are. At least the ones who have time to think about PSAT and NMS - which may be very few of them.
But the SI percentiles and concordance on the original released info do not equate for anyone, isn’t that right?
@SuzyQ in #4766. Yes - that is what I’ve been concluding of late. It’s pretty obvious that they didn’t use the “Research Study” percentiles in their Preliminary Concordance Table, or else that table would have concorded a previous 201 or 202 to a current 201 or 202, rather than 208 or 209 (which turned out to be where commended is).
You need two sets of percentiles for the concordance tables to be generated: a “current” set from the new test, and a “previous” set from the old. After all, the method of concordance is to equate percentiles between the old test and the new. What are the possibilities for a “current” set? Well, actual percentiles of course, or perhaps something representative like percentiles generated from an underlying research study. But we know that the research study generated a PSAT distribution that was markedly different from how it actually turned out. Is that because CB’s focus was alignment between PSAT and SAT? Or because CB bungled the research study? Or both? Not clear (yet).
The “red herring” was simply CB following its standard operating procedure which is to hide current year actual percentiles for a year or so. In the past they have lagged by one year in their reporting, and so we’ll probably see this year’s actual percentiles presented in next year’s Understanding Scores report. This year’s report had to show something else so it naturally showed the results from the study group even thought they were way off base and generated from a distribution that was (for whatever reason) markedly different from the PSAT’s actual distribution.
@mamelot This is a good theory to explain the madness and inconsistency that we see in these numbers. Totally makes sense. Thanks!
Edited to say that College Compass Art used the initial concordance (which we now think was based on actual PSAT percentiles vs old PSAT) and actual test results he had to come up with his estimates, so his numbers seem to be the most accurate we’ve got.
Has anyone checked the online score report lately. It seems to me the user percentiles may have changed. I know my son had one that was a 98 and it is now a 99. Not sure if that means anything. The downloadable report now says “Fall 2015 percentiles were set based on a research study. Fall 2015 benchmarks were preliminary benchmarks. Benchmarks have been finalized for Spring 2016 assessments and forward, and user percentiles will now be based on administration data.” What is administration data? Does that mean actual percentages?
@snicks1234 The scoring for the ACT essay was very odd. I contemplated having my son’s score reviewed because I don’t believe the domain scores correlate to the writing score. I decided against it but I don’t know if they were giving higher scores based on re-reviewing essays as much as I suspect they may have been re-calibrating domain scores to the writing score. Since the report shows what the domain scores are immediately below the writing score, I decided it wasn’t worth it and any college that is looking at his ACT report will hopefully look at the domain score. I believe where ACT went wrong was in converting the domain scores to the overall writing score because, in my opinion, my son’s domain scores justified a higher writing score than the ACT gave him. LOL. Oh well…
@knowledgeless “administration data” refers to actual administrations of the test. So future percentiles will be based on data from actual tests, most likely lagged by a year (or two).
@paveyourpath yes,and I found the whole essay scoring procedure for SAT to be very subjective also.I would suggest that college admissions might just read the essay and come away with a very different take than the reader who scored the essay. Either way,I believe his application essay will prove to be a stronger indicator to college admissions anyway. Hopefully.lol!
The percentiles did not change. same fake percentiles. >:)