I don’t get discrimination among states, why a 202 scoring kid from Utah gets to become a NMSF but a 220 scoring kid from CA is snubbed?
“We can do similar analysis when the 2015 reports are available. I’m hoping they break down the 3 subjects by 2 point intervals, 37-38, 35-36, etc.”
@Speedy2019 I’m really hoping that as well. I suspect, however, that we will see the two converted section scores - EWR and Math - but no test scores (the ones on the 2-38 scale). Obviously it’s very easy to work with the section scores to arrive at SI information - the real challenge will be those intervals. Guessing 10 points will be a bit too “informative” and they will go with something a bit wider - though if they go too wide then it’ll look funny (being that the top is so “top-heavy” this year). Hoping for 20 point intervals or smaller.
I would like to thank some posters, before I start my comments
To @WorryHurry411: I verfify your number (100) of PSAT perfect scores. It’s correct and it came from some school reports (2012, 2013), but not by CB official report
To @addgap (post # 126) with “220 – 0.31%” and “Remember 220 or above is 0.31% (=.0031x 1.7m = about 5200) still a lot nationally”
To @Speedy2019 (post#1361) about explaining curves, cautions not to discount SI tables, about CB data with “Meaning their research establishes the percentiles and then the curves are set to meet the target”. Also Speedy2019’s analysis to discount the data inflated by CB to gain back popularities or whatever - conspiration theory. A beautiful analysis
We are talking about NMF cut off (the very very extreme ends of score ranges, SI ranges, %tile ranges – all relate to each other.)
I assume that due to the redisgned PSAT, the new test favors to the top end test takers since we used to have 100 perfect scorers, now we have 500+ (use past SAT data tables)
How the data tied to @addgap’s analysis. According previous SAT, at 0.31% or 99.69 %tile. Usually at this point we normally have only 2800 test takers. That’s why @addgap said “5200 still a lot nationally”
From previous SAT tables, at the lowest 99+%tile we usually have 7500+ test taker.
Again I assume new PSAT data behaves like old SAT data, then at lowest end of 99+%tile (both SI and TS tables) we would have 7500+. So where is the rest 7500 (of total 15000 NMF)? It should dip into top 3rd of 99%tile ranges in both TS and SI table (2015) – lower than SI 213, lower than TS 1440. I believe that where cut off happens
@crazym0m – does your school typically have many NMSF? 224 is a remarkable score!
Thanks for putting this data together @Speedy2019 - this shows Connecticut to be closest to the 99.5+% state percentile cutoff among PSAT program entrants, while most state hover close to the average of 98.9 %ile or better for the top 1.1% of statewide entrants. Interestingly, CA is just average as far as making the cut for SF. Unless participation rates change significantly, this will continue to follow patterns from past cutoffs.
My prediction for OH SI of 215 or possibly 216.
@SLparent - add that to your list. My kid in OH 215 SI, 1440 TS. Her school may send 1-2 SF (though not every year) and she and one other student appear to be top 2, with the other kid being an SI of 216 and I believe TS of 1450.
@Speedy2019 You cannot just look at the number of test takers and awards to get percentages. The percentage is not based on test takers. it is based on graduating seniors. So in the states where you are stating 2%, it simply means fewer 11th graders took the test. But the likelihood is that the top performing students are fairly stable. (Some states might provide testing for all 11th graders, whereas some states only those who see themselves as likely qualifiers might take the test. Look at the Cobb County data. Huge variants within just those schools.)
@WorryHurry411 NMSC’s purpose is to recognize top students by each state. NMSF are typically the top 1+% of students in each state.
It is no different than other scholarships with designated parameters (county, minority status, financial need, GPA/test scores, major, etc)
@Speedy2019 post 1396. Agreed. I think Applerouth and other posters as well as concordance charts show there was about 5-8% inflation using the national percentile # along a good section of the middle of the curve. The national percentile # is rubbish. I think this thread, however, is mostly interested in the 99% part of the curve (the 1%ers) and here it seems the inflation using the reported user and SI percentiles is around 1-2%. This is based on concordance and anecdotal evidence. However, the reports of the extremely high median in FLA as well as some of the GA schools seem to be tipping the scale more toward "highly probable’ perhaps even conclusive rather than anecdotal that we are going to a high commended cutoff (207 - 210) and very slight drop from the high scoring states. I’m hoping I’m wrong but this was an especially cruel test. 1 or 2 wrong answers on the reading is 2-4 SI index points. Huge.
@Mom2aphysicsgeek - states with low participation rates, also have low cutoff for SI compared to states with high participation rates. In fact, they hover close to the commended SI cutoff.
Thank you to all of you knowledgeable posters, you are very helpful to rest to us and to students at CC.
@Mom2aphysicsgeek, I agree the number awarded to a state is based on graduating seniors.
However, the percentages as stated are correct.
Percentage of SF in a given state, is number of SF in that state divided by test takers in that state
Percentage of SF from the nation, is number of SF in that state divided by national test takers total.
I’m in no way saying HOW the SF numbers are divided between the states, only how many SF are recognized as a percentage of the respective state and the nation.
@OHToCollege That may be, but the number of awards per state is not based on participation rate.
@thinmints Thanks - his school usually has a couple NMSF’s and 15-20 Commended kids
@speedy2019 Thanks for clarifying. I see what you are saying.
I called the GC at my D’s school. Only got a little info as they are still going through the data. It’s a mid-sized HS in NJ that is typically ranked in the top public schools in the state. We usually produce around 20-25 commended and 2-3 SF candidates.
She wasn’t able to give me any aggregate data but I asked about my D’s score 1500/224. She sorted the results for the school and said the score was “really good” and “near the top” but that’s about it. Didn’t offer any congratulations for having a NMSF candidate LOL.
So having the conversation didn’t make me any wiser or feel any more comfortable about her being on the cusp. I don’t feel like 224 is a certainty by any means. Hopefully the state reports come out soon. I think they will be instructive.
Thanks so much @Speey2019!!
@Tgirlfriend I think someone had said around 1.7 million in a previous post. As far as # of NMSF’s awarded to TX, it looks like it’s going down. I was giving rough numbers by NMSC when called. Per Speedy2019’s post, TX had 1353 for 2014. I was given 1325 for 2015 and I think roughly 1300 for 2016…so TX’s pro rata share of 16,000 is going down.
@Pickmen, you could be right, I just haven’t been able to convince myself of the high numbers yet.
When I hear “principals and GCs say too many 99%”, I bet most of them are talking about the National % which is inflated because that part of the study assumes all juniors in the country took the test. Applerouth’s chart of this inflation was for National %. In the end, it is the SI % numbers that matter It would have been great if Applerouth charted out SI %. But he did not.
Remember, if FLs mean is higher than the nation, some other state is lower than the national mean. Can’t escape that fact.
I seem to be in the minority in my view. But there HAS to be a reason the curves for Writing was tough and Reading and Math were more lenient. If the scores were super high, the curve for Reading/Math would have been much more harsh.
CB attempts to make each test (psat, sat, etc) equal between different test dates. So a 2100 SAT in 2007 is equal to a 2100 SAT in 2012. They do this with the scoring curve. It doesn’t matter if a test is “easy” or “hard”, Curves adjust for that.
Now, since 2015 is the first PSAT, CB needs to establish a baseline for their equating process. The 2016 PSAT can be equated to the 2015 PSAT. But for the 2015 PSAT, what to equate to? I’m guessing they equate to the research sample of 90,000 students. Why even do the research study? The have to equate the 2015 PSAT study to something - be it some “historical data”, the research study, some desire to close the gap in cutoffs between the states (as someone posted previously). They had to decide on curves to make the scores match something.
If they are using the research study for equating purposes, then it is possible the SI and User data could be correct.
I do think there were 1.7 mill. of 11th grade testers in 2015 and I understand the number of SF slots per state is based on population in the 11th grade in each state, not on test-takers. NY went down recently from 1012 SF slots to under 1000 - population changes are the reason but can also be that the SF slots and cut off are done to the closet amount that works - not an absolute number of students - that was in some NM statement someone quoted above in this thread.
@dallaspiano 1422. Nice post and as said before trying to corollate the percentile ranges for the SAT with the given user percentiles does make some sense, at least it seems to be the way CB operates (it curves the tests to get a certain amount of testers into certain 99 or 99%+ slots.) but part of the problem is you can’t work from TS percentiles with the new SI index. I think that’s where also the problem with concordance is arising. For example, on the old test you had 3 scores that were your projected SAT scores (just add a zero) and when added together gave you your SI. This allowed, I think, you to match or compare SI percentiles and TS w/out any issues. But on this new test a 700 or 2 35s on the CR/W is supposed to be a 700 on the SAT (although this test is scored out of 760) but on the old PSAT this score concords to about a 670-680 out of 800. But maybe I got this wrong and would happily stand to be corrected.
@CA1543, I think it is based on graduating seniors instead of 11th grade population per state. From my understanding.
If the slots allocated to Texas is going down, then another state is probably going up. Texas school population is probably increasing but the graduation rate may not be keeping up - too many kids not graduating? Just guessing.