@engineer, do the SI percentile tables somewhere say “sample” instead of “actual”? The concordance tables are in the document and marked “preliminary”. For the SI tables, Is there reason to believe they aren’t “actual”?
@engineur The Understanding the PSAT report seems to always be based on a sample, so I don’t expect that there will be any updated version of this report.
@engineur Do they update this every year? Then, why would they bother publishing the first version. I’m really liking the published data so I don’t want it to change much. 
@Speedy2019 oh my bad, the total score percentiles seem to be the only things marked nationally representative, not the SI percentiles! I just don’t want to get my hopes up too high at this point since this chart says I’m in the 99+% but all predicted cutoffs are higher than my score.
@billchu2, what is the definition of “sigma”?
Edit/Update: if it’s just the standard deviation, then never mind.
So, then for my state, Illinois, based on 2014, Cutoff = 215. 50th percentile = 141, 215 - 141 = 74 distance from median.
“normalizing” 74 / 31.0 = 2.39 stddevs from median.
For 2015, median + 2.39 stddevs = 146 + 2.39 * 2.54 = 207.
But most “predictions” say the number will remain 215.
I don’t really understand stats that well to know what logical error I’m making.
So, NMS says there were still about 1.5 M juniors taking the test in the latest report. There seems to be some score compression, stddev = 25.4 vs 31.0 over the same range (48 to 228 vs 60 to 240). Are the scores not really normally distributed?
All these predictions makes me doubt my son’s 211 in AL for NMSF. Any thoughts?
Two years back you needed a 224 to hit the 99th+ percentile. Yet if you take an old score of 214 and convert it from page 24, you now get a score of 99th+ on the conversion. What am I missing?
Put in another way, if I take a non-commended old score of 200 and convert it (page 24), I get 1370. This is now roughly a 99th level score. Hmmm…
The Calcualtion is for the 99+ Percentile
1520, 228, 99.95th Percentile
1500, 225, 99.90th Percentile
1490, 224, 99.85th Percentile
1480, 222, 99.80th Percentile
1470, 220, 99.70th Percentile
1460, 219, 99.65th Percentile
1450, 218, 99.60th Percentile
1440, 218, 99.55th Percentile
1430, 211, 99.50th Percentile
1420, 210, 99.45th Percentile
I feel much better about the boy’s 1430/216 than I did yesterday but I’m still not going to believe anything until September.
At least,you feel much better that count. Not till 9/2016 but to be sure 100% at 3/2017 when you receive NMF letter
We are hopeful that husband’s company merit scholarship will increase his chances at the NMF. But that’s too far in the future to stress about yet. I just needed to know his numbers looked solid for Florida to stop obsessing For now.
At this stage, no test prep center, no one can give u solid answers … it’s all about guessing. I am a junior (same age with your son). Based on the data behavior (not logic), my opinion your boy’s 1460 very solid for FL. He’s in 99.65%tile
@mamelot, “the reality is that those sophomores taking the PSAT are a more select group of kids who are choosing to take the exam”. I don’t know about this year, but actually in recent years that has not been true. In fact, more sophomores than juniors have been taking the PSAT. College board data for 2014: http://research.collegeboard.org/programs/psat/data/cb-jr 1,595,486 juniors took the PSAT/NMSQT. 1,812,143 sophomores took the PSAT/NMSQT. So, if anything, the juniors have been a more select group. And of course anyone trying to win NM would take it as a junior but not necessarily as a sophomore.
In 2014 http://www.bernardsboe.com/UserFiles/Servers/Server_3096886/File/Jill%20Shadis/Ridge%20Counseling/Standardized%20Testing/Understanding%202014%20PSAT-NMSQT%20Scores.pdf juniors needed 76 CR 79 M 79 W to score 99+ whereas sophomores only needed 72 CR 76 M 71 W.
For comparison, this year juniors need 73 ERW 75 M and sophomores also need 73 ERW 75 M. Those are the “user” percentiles which might be affected by the test taking group. But the college board claims that the “national” percentiles are representative of students as a whole, and on these percentiles, this year, sophomores need 72 ERW 74 M to score 99+ whereas juniors need only 70 ERW 74 M to score 99+. The entire point of the National percentile is that it’s supposed to represent all students and be free of bias due to who is taking the test. And by this measure the college board is asking us to believe that the verbal skills of the top .5% of students are going down from sopnomore to junior year. I haven’t looked through the rest of the percentile charts.
1480 scores may produce SI of 220 or 222 depending on the math score vs R-W score. so 99+ calculations may not be correct.
OK. I imported the 2015 percentile data from College Board and modeled it as a function of SI scores from 82 to 213 . I then compared it against the cumulative percentile plot of a normal distribution with different standard deviations as a parameter. The standard deviation curve of sigma = 28 best fits the data. Previously I had only used the 50% and 99% points to model (that is why I got sigma = 25.4 before). I wish I could export the plots, but alas…
Do we know how many eligible juniors took the PSAT this year so that we can calculate the percentile for the 16,000 cutoff?
@srk2017, every student only gets one SI - whether 220 or 222. For NM qualification, ignore the total score and use SI. In the case you mention, both 220 and 222 are 99+. The minimum for 99+ is 214.
At my son’s school the GCs were encouraging juniors to take ACT and think hard if it would make sense for them to take PSAT at all. Maybe it’s not typical, but I would think that the number of juniors taking PSAT this year would be lower than last year and they would be more heavily skewed towards the top. I guess it didn’t pan out this way.