@OHToCollege - I don’t follow? 50% of what? Approximately 25% of students have TS >= 1400. Are you asking about how to get an SI from 1400? Some (who do better on reading) will have more than 209. Some (who do better on math - which might be more common at a school named “math and science”) will have less than a 209…
TS associates with certain SI
Score Min Max %tile
1520 228 228 99.86
1510 226 227 99.83
1500 224 226 99.81
1490 222 225 99.77
1480 220 224 99.73
1470 218 223 99.69
1460 216 222 99.64
1450 214 221 99.60
1440 212 220 99.54
1430 210 208
1420 208 216
1410 206 214
1400 204 212
at 1400 TS, Min SI 204, Max SI 212 (or is it 216?), mid point appx 208-209. If you set SI cutoff it at 210, you can have a score lower than 1400 and still be at an SI of 210. A school like Il Math & Science likely has many students between 50th to 75th percentile pushing an SI of 210. So my conjecture, if you had to pick an SI at a TS of 1400, I would lean on the higher end of SI.
Not sure if I can add any further analysis, but maybe using the max SI for a 1400 would be a good baseline assuming that IL Math and Science is following a traditional trajectory as in year’s past of ~25-30% of students making NMSF. So based on that, we could parse out that 212 would be the lowest? Unless, DS’s class just happens to not be following the traditional trajectory.
A 1400 could be obtained by 760V/640M or by 640V/760M. So wouldn’t SI range for a 1400 be 204-216?
@Speedy2019 - unless I’m misunderstanding, his graphs aren’t that interesting. They’re just scores vs percentiles (sometimes concorded). The main point he’s making is that if you concord the score first, and look up the new score in the 2014 table, you get a much lower percentile. Which we all know already.
The one interesting comment he made based on his data is:
He’s in Georgia, right?
Their cutoff last year was 218, which was in the middle of the 99’s (before the 99+) on the previous chart. So his implication that anyone in the 99th percentile was a “national merit hopeful” is a little bit misplaced. He said that 95 juniors scored >= 205. Based on percentiles, the middle of the 99th percentile on the current SI chart is around 209. IOW, the percentile table predicts that ~209 would be the Georgia cutoff. He states that if the cutoff for Georgia was 205, they’d have 95 NMSF. When he uses his concordance table though, he gets 37. He didn’t give the details, but my guess is that that corresponds to a 214 on the new test (because new 214 concords approximately to old 214, and old 214 is the first 99th percentile). Again, he’s probably using the first 99th percentile (which is how he seems to define “national merit hopefuls”). But the fact that he’s presenting is (and I’m putting it in bold because it’s important):
Applerouth’s data (presumably from Cobb county?) predicts a 214 SI cutoff for Georgia
(This assumes that his statement that 37 “is much more typical” means that 37 is about how many NMSF these schools typically have.)
His data does not predict massive inflation, i.e. that the cutoffs will remain the same (or even go up). His implied 214 is still above what is predicted by the percentiles, and a little bit above my “midpoint” prediction of 212. But it’s less than what is implied by the concordance tables by themselves.
Min SI for 1400 TS is 640 R/W + 760 M, SI of 204. Max SI at 760 R/W + 640 M is 216. Thus a range of possible SI at 204 to 216. Testmasters may be valid at their prediction of 215 for IL.
In response to “how do you convert 1400”. That’s not really what we’re trying to find. What we’re trying to figure out is, if you sorted by SI, what would be the 25th percentile? Obviously some 1390’s will have SIs > 209 - but some (many in fact) 1410’s will have SIs < 209.
The most rational way to guess the 25th percentile of SI, if you know the 25th percentile of TS, is to guess a “normal” SI for that TS, IMO. Which is why I said 209.
I would be extremely surprised if the 25th percentile of SI for this school was not in the range of say 208 - 210.
@thshadow. I was thinking the chart was actual SI scores for 2015 for the high schools which sent him .csv files. If so, that would number a few thousand, I would estimate. But you are saying the 2015 chart isn’t for actual 2015 psat test takers?
Here is my input from Florida:
Anecdotal:
Mu D’s school (large public) had 6 NMF last year (~1%). This year the school ran a three day PSAT Boot Camp for about 30 students who had high PSAT scores in their sophomore year. The school and the district had follow-up sessions through the fall; about 12 stayed with the program throughout, some went to private tutoring. The expectation was an increase in NMF for the class of 2017.
After the test, several of the historically top performers stated there was math on the PSAT that they had not seen before. One of the areas my D was not familiar with was Conics, she has since covered that topic in Pre-Calc.
Florida made a disastrous attempt a launching a Florida Statewide Assessment (FSA) end-of-course exam for Algebra II last year. Teachers were attempting teach to new standards that were in flux until shortly before the exam was administered. There was a huge emphasis on use of graphing calculators only to find out a few weeks before the exam that the FSA only allows scientific calculators, so a mass rush to re-teach ensued. My D only got 74% correct on the end-of-course exam, yet she was in the 98 %tile for the state. She got a 630 in math on the PSAT which is 93%.
Top score at her school was 1490, top 5 appear to be in the 1450 – 1490 range.
Her school got an A+ rating this year which means scores on the state’s standardized tests for last year compared well to the rest of the state.
So, a top performing Florida public (by State of Florida standards) but no appearance of stellar PSAT scores.
Theory:
Florida cut-off will not have a big increase this year due to lower than usual math scores, and probably a decrease in English scores as well. Private schools that do not have the FSA burden may increase, but typical top performing publics will not. Therefore I think Testmasters has overstated the Florida cut-off at 215.
For converting “25th percentile TS is 1400, what is 25th best SI?” I just ran some simulations. I picked some somewhat arbitrary means for math and verbal, with a standard deviation, such that about 10% of the time the 25th best TS was exactly 1400.
If I pick a math mean of 61, and a verbal mean of 57 (with a sd of 25), I get results like:
For mathmean 61, verbmean 57, average SI is 209.47
For mathmean 61, verbmean 57, average SI is 210.56
For mathmean 61, verbmean 57, average SI is 209.73
If I make the average spread bigger, I get results like:
For mathmean 62, verbmean 56, average SI is 209.15
For mathmean 62, verbmean 56, average SI is 209.34
For mathmean 62, verbmean 56, average SI is 209.83
If I make the average math equal to the average verbal (which seems unlikely), I get slightly higher SIs:
For mathmean 59, verbmean 59, average SI is 211.94
For mathmean 59, verbmean 59, average SI is 211.60
For mathmean 59, verbmean 59, average SI is 210.66
@Speedy2019 - it seems like he has access to 323 juniors, because he said “95 of the juniors in our sample (29.4%!) …”. If you’re imagining that he computed the percentiles among his 323 juniors and graphed that, it would look freakishly jagged and non-normal. And if he did that, he would have explained it.
I believe that what he did was graph the percentiles that was reported to his juniors - which is kindof stupid, because we (now) know what percentile is given for any arbitrary score.
His green line (concorded 2015 scores) is somewhat jagged, because a given SI will concord quite differently depending on the subscore breakdown - and I think he used the breakdowns that his actual students had. So that’s how he “used” his data…
I don’t think any of his graphs show in any meaningful way how many of his students got say a 215, vs a 214, or a 213, etc… Which is I thought what you were saying it might reveal…
Ok, so for the IL school, a 210 is higher than the sliding (-12), but much lower than testmasters. I’m assuming is higher than going by percentiles, but I am not familiar with IL historical scores.
So, this would evidence would fall somewhere in between the concordance and the percentile tables. Right?
@dallaspiano I’m just trying to summarize this piece of evidence. But, I’m still trying to be a true believer, dream weaver…
@thshadow With due respect, I beg to differ from your conclusion. Yes, IMSA data is not sufficient or more confusing as they given only total score instead of index score. They do admit about 240/year. For 2015 graduating class they had only 48 NMSF’s so it might be possible they fall at 20% this time which means SI might be higher than 211.
For example at my daughter’s school they have 33 students with 215 or higher. As per your midpoint 211 IL cutoff prediction it will be somewhere 40+ students getting NMSF from my daughter’s school which never happened before. Maximum they had is 23. Look at my posing #2079 which has NMSF for my daughter’s school since 2011.
Since IMSA didn’t reveal the SI averages, we need to look for some other School from Illinois to come to a more close approximation.
From the actual data received so far, it appears NJ and VA are the two state which still have high SI score reported that is more than what is being predicted here. This two states data reported so far are mostly higher than testmasters prediction. TX seems to be in line with testmasters, and we did not hear too many high score from CA lately at all, surprising a bit considering #NMSF awarded. Anyway CA is not much lower than testmasters 219, can we assume that , may be 217 lower end. (considering we do not see CA reports being reported, means the scores in CA not too high?).
First time here.
Testmaster said today it “obtained a new data set which we will use to generate an updated cutoff estimation,” and would “post a second update with more data, and that further improves our belief that the concordance table estimates will be fairly accurate.”
@thshadow, thank you for the analysis and comments.
@WGSK88 You can look up the total number of graduates per class here to calculate the % of NMSF per year. https://en.wikipedia.org/wiki/Illinois_Mathematics_and_Science_Academy
For instance, the class of 2015 had 48 NMSF out of a graduating class of 195, which gives a 24.6%. The class of 2016 had 63 NMSF with a class size of 206 which gives 30.5%.
I hope more people are able to post SI score distributions from their schools to help narrow it down.
@suzyQ7, post #630 - nothing as they don’t talk about, and nothing from GC
@NathanBN…did you see where they mentioned Texas cutoff going up to 219? If so, I wonder why?