We are in shock, D21 had a perfect score! She is a great student, but always makes some mistakes by overthinking certain questions. Not this time, SI 228. We are in Texas but she was able to see her score on her Khan Academy portal. We verified to just make sure 
@Totoros Congratulations!!! That’s awesome!
Wow! Congrats to everyone. We are in Calif so will have to wait until tomorrow. She said the math section was tough so my hopes aren’t too high. She Only scored a 1380 on that paper practice test so waaay too low. But at least she will be able to use it for practice for her Spring SAT.
@EGHopeful try earlyscores psat. We are in So Cal and we got DS21 scores first thing this morning on that site. He got a 226! We are crossing our fingers that is good enough as it meets all historical score cutoffs for CA but I guess we will have to wait until the Fall? Ugh!
For those of you with kids who will be NMF, are you planning on looking at the big merit schools? I often wonder what percent of kids take advantage of those. Of the 30 NMFs at our school in 2019, not one went to a big NMF merit school.
@homerdog absolutely! I just booked a ticket to Miami for spring break this morning based on DS21 scoring a 226. We are visiting U Miami which is private (and my alma mater) but is likely to be comparable merit-wise with the public schools based on what I’ve read (Benaquisto match plus Singer/Foote scholarship), and then we are heading up to UCF and UF. We are not planning on looking at UT-D but we might look at Texas Tech (but not on this upcoming spring break trip).
And 30 NMFs at your school last year? That is incredible! Was that an unusual spike or do you guys put something in your district water?
@sherimba03 that didn’t work it just show Pending. So I guess I have to wait until tomorrow.
2019 - https://collegereadiness.collegeboard.org/pdf/psat-nmsqt-understanding-scores.pdf
A score of 43 of 48 in Writing last year yielded 38 and 36 on the two test dates. This year it yielded 37 and 36
A score of 46 of 48 in Reading last year yielded 37 and 36. This year it yielded 37 and 36.
There’s no way to lose 8 points with the same raw score, assuming the same (not mentioned) math score. It either went down 2 or up 1.
But this is actually irrelevant. Cutoffs are based on a percentile of scores. If the test was impossible and everyone got at least 20 wrong, then the students with 20 wrong received 760s, 228, and be the high scorers and NM qualifiers.
It would not be the case that no one in the country got above a 1050 and there were zero qualifiers just because those were last year’s conversions of a 28 raw score.
The numeric cutoff may be lower, but roughly the same percentage of students qualify.
Putting aside that it’s irrelevant, I looked at the two score conversion tables this year. Comparing the same raw score to converted score, the difference are:
-2: 7%
-1: 19%
-0.5: 10%
0: 34%
+0.5 : 4%
+1: 24%
+1.5: 1%
That’s a weighted average difference of less than 0.3 index points of 228.
36% lower, 30% higher, 34% the same. Not sure who is calling this “bad/brutal/forgiving”.
And again, this is all assuming you get the exact same number of problems wrong, regardless of how hard the test is and how others do.
Art at Compass Prep has said it would be statistically impossible for a cutoff to be higher than 224. This year’s cutoff for CA is 222. Even if the cutoff could ever be 227, it would be unheard of to have a 5 point jump.
Just in the interest of accuracy, there were 3 test dates last year and this year, as well.
And just as a follow up to my post, above, pointing out that there were 3 tests both last and this year:
For the October 13, 2018 psat, I’m not sure if there was an official release for the curve, but on reddit, users compiled a curve that indeed suggested -2 in reading was a 37 and -5 in writing was a 35, for a total combined score of 720. (Also, there are 44 questions in writing, and 47 total questions in reading. So, -5 in writing is 39, not 43; and -2 in reading is 45 not 46.)
In comparison, this year, missing 5 on the writing yielded scores of either 33 or 32 on one of the two published curves, and missing 2 on the reading yielded either a 35 or 36, respectively, a drop from 720 to 680 if we are looking at the Oct 13, 2018 test.
Assuming a score of 38 on the math for October 13, the selection index for 37r, 35w and 38m would be 220.
This year, the same raw scores would have yielded: 35r, 33w and 38m (106 X 2)=SI of 212 for Oct 16 sitting; 36r, 32w and 38m (106 X 2) = SI of 212 for Oct 30.
So, yes, given the same raw score, it was definitely possible for an individual to lose 8 SI points from last year to this year.
Sorry if this was convoluted. I just wanted to make sure the analysis of the change from last year to this year was accurate. This was written in a hurry while I’m making dinner so I apologize if I have made an error. I think you get the idea.
Yes, my NMSF dd20 has applied to a number of big merit schools this year, and my dd21 definitely intends to apply to big merit schools next year. Where they go will depend on where they get the best aid, and where they feel is the best fit. The likelihood is that they may both end up at either Alabama or one of the Florida schools. If one of the other schools they applied to for competitive merit comes through with a comparable package, that will have an impact on their decisions but we know those are long shots for everyone. Not counting on any of those long shots.
I’ve read this post several times and I’m still not clear on what numbers you are using to get the percentages you list and how you come to the conclusion about a “weighted average difference of less than 0.3 index points of 228.” Can you clarify what your calculations were here?
FWIW, as we’re discussing scores/cutoffs . . . Clearly, the scale is harsh as missing just one costs a good bit of points. Missing one on math cost 20 points (4 SI points). Missing 3 on English cost 40 points. (8 SI points).
My dd missed 4 questions in total. 2 in reading, 1 in writing, 1 in math.
Reading 45/47 = 35 (98%)
Writing 43/44 = 37 (99%)
Math 47/48 = 37 (99%)
English Score: 720 (99%). Math Score 740 (99%). Total = 1460 (99%).
All that said, it really shouldn’t matter much since they take the top 1% (or so) of students, no matter how tight the curve. What stinks is that having SUCH a tight curve for NM cut off makes is so pretty much test taking accuracy is absolutely critical. One minute of distraction or a couple “stupid” errors can be absolutely critical. I do think it’d be much wiser to have a tougher test to spread out the top end a bit, for more accurate results in segmenting these top students.
But, it is what it is. I’m very thankful my kid had a good day and didn’t make many (well, not more than 4, lol) silly errors.
Son got one wrong in reading lost 20 points and 2 wrong in Math lost 30 points, that’s awful…
I’m not sure that anyone should be hoping for SI cut offs for NMF to go down. It just doesn’t happen that often. I get the thought process around top 1% and how lots of kids probably made a few mistakes so top 1% of scores might be lower but the curve was bad last year too and I don’t think many states had SIs go down. Doesn’t look like Art at Compass Prep is predicting any SIs getting lower this year.
Art has some interesting comments in his new article. (Can I quote even though I can’t link it?)
I’m glad he graphed the scales, as I was tempted to do the same, albeit too lazy.
I’m also happy that he called out College Board on their fake percentiles. It’s worth noting that the same is true for the SAT.
Yes, for those of us desperately trying to read the tea leaves on our own, the guru has spoken. Art Sawyer updated his blog last night and gave state cut-off estimates for 2021.
His estimated cut-offs remain the same for most states. He doesn’t think that the Oct. 16 harsh curve will result in lower state selection indexes in most cases, so there must have been enough close-to-perfect scorers to keep the estimates the same for most states. (Of course, he has a disclaimer that these cuts are only estimates etc. etc.)
The curve is indeed brutal for those in high scoring states. Kids who took the Oct. 16 test could only miss a total of two questions on the entire test if even a single mistake was in either reading or writing to get to a selection index of 222 . Wow.
Even speaking as someone whose kid is a great test-taker rather than a genius and thus likely squeaked in at the expense of kids who knew the test content better but made silly mistakes, I still don’t think a curve like this is ‘fair’.
Random chance played too big a part in the scores this year. As I said earlier, kids are human, not computers, and some small room for human error should be built into any high-stakes test like this. One single mis-bubble will mean the difference between a full ride or not for lots of kids. And while that’s true every year, it just seems unfair that this year’s curve means there’s a higher chance than usual that the error was due to chance rather than truly not knowing the answer.
Count me in as another parent pouring over Art’s post this a.m.
The curve is indeed brutal for those in high scoring states. Kids who took the Oct. 16 test could only miss a total of two questions on the entire test if even a single mistake was in either reading or writing to get to a selection index of 222 . Wow.
Even speaking as someone whose kid is a great test-taker rather than a genius and thus likely squeaked in at the expense of kids who knew the test content better but made more silly mistakes, I still don’t think a curve like this is ‘fair’.
Random chance played too big a part in the scores this year. As I said earlier, kids are human, not computers, and some small room for human error should be built into any high-stakes test like this. One single mis-bubble will mean the difference between a full ride or not for lots of kids. And while that is true every year, it’s unfair that this year’s curve means there’s a higher chance than usual that the error was due to chance rather than truly not knowing the answer.