Just to repost for the folks who don’t go on the main PSAT discussion page, here’s the link to the updated Prepscholar predictions. What do you guys think?
http://blog.prepscholar.com/national-merit-scholarship-cutoff-2015-2016
The reason we can’t figure out what’s going on is that CB doesn’t WANT us to figure out what’s going on. I also suspect that CB does not know what is going on.
@garyasho2: “What do you guys think?”
I think they’re still using the numbers from two years ago instead of last year’s cutoffs. Aside from that, I think those are pretty close to testmasters and based on concordance.
@PAMom21
Student reports about the test I am currently analyzing (Oct. 14) are that R was hard and M was easy. It may be that the other version (Oct. 28) had easier R and harder M. This would also be indicated by the scoring curves, which are slightly harsher for Oct. 14 M and Oct. 28 R, and slightly gentler for Oct. 14 R and Oct. 28 M. I don’t currently have access to the Oct. 28 test.
I have finished going through all the Oct. 14 math. I agree that the math was very, very easy for students who have solid math skills and who had covered the relevant material well in school, with a tutor, or via self-instruction. I was actually blushing as I read some questions. Of course, careless mistakes are always possible, and sometimes even more likely when the problem is easy, so it is possible that careless errors could keep the scores down. This is a bad way to pick NMSF students.
The writing was extremely harsh at the top. Ix = 2 selection index points, 2x=4, etc. Stupid mistakes there were very expensive.
Maybe this is a stupid question, but since the tests are scored “harshly or gently,” why is it assumed that as the SIs were created that the outcome distribution was not the actual scoring goal?
@suzyQ7 The PSAT W curve has often been brutal, even in the past. For example, on the 2014 Saturday test, -2 W was 71. Students in high cutoff states need to be really careful in everything, but really, really careful in W.
@Mom2aphysicsgeek they can set that scale to be anything they want (-1 is a 37, -2 a 36, etc. or -1 a 38, -2 a 37, -3 a 35 etc.). So yeah, they are going to “norm” it as close to the reference group as they can get (because it’s the reference group). It’s possible, however, that the actual performance was so strong that this was a hard fit - although I do think that that would just result in a less forgiving scale. But maybe it was still a hard fit nonetheless. Not sure how the upper tail is affected by all this.
If the math was really that “easy” the scale would be harsher. I think College Board knew exactly what they were doing when they set the scales, and there is not some huge number of kids that have perfect scores. The scales are set to get the distribution they want, that is the whole point of the scales, and explains why they are different for the two tests.
I’ll throw in my two cents. I agree with @ tammy21. They had the data for awhile. If they set the curves before the test in anticipation of a certain level of performance and got a different level of performance, wouldn’t they adjust the curves?
That’s what I keep thinking too – hopefully those state summaries will be forthcoming and shed some light.
Plus, the state summaries are based on actual test results, aren’t they? X% of state residents scored in this range, etc.
Edit/Addendum: they report actual numbers too, of course.
Just fyi- the Today Show (NBC) is about to have a story about a move by colleges to make the SAT/ACT optional … –
@Plotinus I agree. How students perform on these tests is interesting. I thought the math was quite easy as well. Yesterday I was going over the results of one student. He missed 3 on reading, none on writing, and 10 on math (9 of them “hard”). I was surprised by that because he’s an excellent math student. I printed out the 10 questions he missed and handed them to him. Just a few minutes later he gave them back, with correct answers on each. He couldn’t believe he missed them on the test.
Time pressure? Test anxiety? Careless reading errors? Who knows.
@Mom2aphysicsgeek You are right. It is possible to lower the number of high scores by making a curve that is very brutal at the top. The Oct. 14 2015 curve is definitely harsher at the top in R and more gentle in M, but not really harsher overall compared to the old curve. By the time we get down to -4 per section (-12 in all), the new curve is getting friendlier overall than the old curve. The W curve is pretty much the same. This might have been done to get the right number of students at the different scaled score levels.
Here is an exercise. I counted the number of raw points lost, and then I converted the corresponding R, W, and M 2015 scores to 2014 format just by doubling them (not using the concordance). This has been done by others in different ways, but I did it like this to make it look more like a 2014 curve. Here is what we get from the score curve for the Oct.14 2015 test:
PSAT Oct. 14 2015
R M W SI
-0 76 76 76 228
-1 76 75 74 225
-2 74 75 72 221
-3 74 74 70 218
-4 72 74 68 214
-5 72 73 66 210
PSAT Wed. 2014
-0 80 80 80 240
-1 80 78 75 233
-2 78 73 71 222
-3 76 71 69 216
-4 74 70 67 211
-5 72 68 65 205
@DoyleB, @Plotinus, I crunch a lot of test problems (PSAT, SAT, ACT) for my tutoring business, and I make my most mistakes on the easier tests. It is a fact that part of your brain turns off with repeated math well below your level. In the past, I have always made more mistakes on an ACT practice test than an SAT one…all stupid and careless. I’m making those same “ACT” type mistakes now on the new SAT. Maybe it really is more like the ACT. Granted, I’m only averaging 1-2 per test, not 10, but young men in particular seem more likely to clock out in these situations.
The student comments I was was referring to were linked somewhere here on CC, but I didn’t save the link. They spoke of impossible decimal math and not enough time to compute it. Having witnessed a student trying to calculate the mean for a question that asked which would change the most…mean, median, or range, I know it’s possible for students to do unnecessary math. I have other examples of students complicating the math considerably, because they are missing the obvious shortcut. Also, I’m guessing 9 out of 10 students would use “substitution” for 2 equations, 2 unknowns, when “elimination” is a faster strategy on almost every SAT (PSAT?) problem.
I agree with @FrequentTraveler post 891. I also went through this two years ago with a daughter who ended up NMF. The uncertainty can eat you alive if you let it. My oldest is on a NMF full ride so it really helped and greatly influenced her final college choice. A lot of those from two years ago didn’t end up at schools which give generous NMF scholarships, so while it is a very nice honor, it isn’t make it or break it for many. I just hope the CB gets it right when they finally release the commended number and the, eventually, the state numbers and that they would err on the generous side for those states without clear breaks for their appointed numbers.
If you have a student younger than a Junior, make sure they review their old math before their Junior PSAT. My oldest didn’t need to and I didn’t realize how much it would’ve helped my youngest. She found out AFTER the PSAT as she was finally studying for the SAT that she had forgotten a lot of the math since she is in AP Calc now. Simple review would have helped a ton to refresh, help eliminate silly errors and speed up problem solving. Too bad we can’t go back. Oh well, she will be so much better prepared for her SATs and ACTs since she is actually preparing for them.
I am proud of her regardless of her status. She improved by 30 points from last year to this year.
@DoyleB fatigue could have been a factor - this format is more like the ACT (though less time pressured) and apparently there were more calculations on the non-calculator portion than the practice test? That’s what I heard, at any rate. And it was at the end, rather than scattered throughout the test like before. The science section of the ACT was always the potential nemesis for my kids - just due to fatigue. They really had to work on their stamina to make sure they didn’t tank on that one last section!
And then this probably doesn’t impact your student but something else that I learned - at least at my daughter’s school. Apparently a LOT of kids - as well as the proctor, if you can believe it! - were completely unprepared for the length of the new test. Seems a whole bunch of people were mentally checked out when the counselors visited classrooms to discuss the new test and hand out practice materials. We don’t have any feedback yet from the GC’s on results but I’m thinking that this basic lack of preparation HAD to impact scores at her school by quite a lot, esp. in math which was at the end of the test, and pretty much about the time when all of the complaints about the “length of this test” started to surface. Kids really were expecting a shorter test which means they were talking to older siblings or friends about it and not really looking at the new practice test or at Kahn, etc. Even a kid who is really strong in math may do poorly if he/she isn’t prepared for the length and/or getting all the math questions at the end.
@PAMom21 @Mamelot That’s interesting. His frustration was evident. Some of the mistakes were simple carelessness (I forgot the units! I had to invert the answer!) On others he was incredulous (The answer is A. What did I put? B. Why did I pick that?) He feels that he should have been in the mid 220s.
“They spoke of impossible decimal math and not enough time to compute it.”
@PAMom21 Did they take the Oct. 14 or Oct 28 test? The hardest non-calc decimal calculation I could find on the Oct. 14 test is in question 12, the second from the last of the multiple choice non-calc questions. You have to do
300=(.06)(30000-x), solve for x. People who are calculator-dependent might call this “impossible”. Even CB rated this question “hard”. I don’t think the US is going to be globally competitive if this is a “hard” math question for a junior in high school headed to college. Division of decimals is 7th grade math, at best.