Lot of past graduates from my S’s school who ended up @ Ivy Leagues were not NMSFs or NMFs. Knowing that lot of top schools don’t value PSAT for admissions, they don’t pay attention to SAT/ACT early in junior year. Most these kids ended up having 2300+ SAT or 34+ ACT scores. Even my S didn’t prepare for PSAT but relied on his ACT prep.
@speedy2019 It’s a new morning and hope springs anew. I am extremely proud of dd’s accomplishment and view it like we are holding a lottery ticket in the pocket…we didn’t not win yet. Now with DoyleB going back to the card table we don’t have to rationalize away all his logical math based reasonings; we can just live in our fantasy world of 217 cutoffs.
Seriously, even if her 99+ designation is erroneous it has kick started a new more positive outlook towards her upcoming ACT. She got a 31 on a practice in fall and she now views that as “must have been an off day”. She now thinks she should get a 35 or 36. She has no plans to take the SAT.
@LivinProof, my daughter got a 33 on her latest practice ACT test. She thinks that 34 is so close. Taking the real ACT on Feb. 6th for the first time. No plan to take SAT unless the psat state summary reports show the SI% table is correct or nearly correct. So hoping she can wrap up the standardized testing this school year. Then summer and senior year can be all about the college apps and interviews.
Take care.
@likestowrite – I think so far the Okla data and the points DoyleB is making in post 1563 seem to line up with the Test Masters cut off predictions - OK goes up this year to about 212: http://collegeadmissions.testmasters.com/update-psat-scores-cut-national-merit-2016/
But perhaps the cut off could even be a little higher - hopefully the elusive state summary reports will shed a bit more light on it. There are some thoughts about the Cobb County & Walton & Wheeler data in previous posts as well – but it may turn out the GA’s cut off is not going down from 218 to 216 as predicted by Test Masters - we need much more data to really be able to predict this but you can see that Test Masters (and now Prep Scholar) and others are estimating states that had cut offs of 215 last year stay the same, states that were below 215 are predicted to have cut offs that go up and above 215 they start to go down a bit, with SI’s topping out at 220 in DC. But this is all estimates based on limited actual data and/or a research study group & the inflextion point (SI of 215 - where no change is predicted) as well as amount of decrease in the top cut offs may well be off once actual info is published. I do not think schools like those in Cobb or OK are at all likely to suddenly have a “huge” number of additional NMSFs than in the recent past - many more students there might end up though close to (but below) the cut off point. The overall state cap remains close to the same as it was – it could be that schools that had produced one or a few NMSF’s get none bc of the clusters of very high performers who prep at certain schools - another possibility. Our school does not take the PSAT very seriously - students are pretty high performing but this test is not one that many prep for. Many similar schools might end up with very few if any NMSFs - will be interesting to see if there is more “concentration” than in the past in certain schools in some states. Thanks to all for all the number crunching, super-sleuthing & theory generating - it is very interesting to consider so many aspects of this one test and its relevance to so many other things - I am still wondering why my son’s report says that among the AP courses recommended for him as future engineer is Music Theory - seems CB is trying to “recruit” more SAT takers & more AP students through the PSAT reporting.
Here is a link to an interesting article from Compass Education Group. It summarizes most of what this group has stated regarding the scoring report, percentile inflation etc.
@DoyleB - If I understand you correctly, I don’t think I agree with what you said. In particular, in scenario B, 99.5% would get you into the top 15,000 of the 3 million kids - but half of them didn’t take the test and hence aren’t eligible for NMSF. You need to be in the top 1% of the user sample in order to make NMSF.
Let me state something slightly differently. First of all, assume that the “user” percentile is the actual percentile in kids that took the test. (I think we agree that we don’t have that exact number, but rather CB’s estimate of it. I don’t think you’re currently arguing that CB’s estimate is way off - if so, that’s a different argument.) Obviously given the (true) user percentile, we could estimate NMSF cutoffs pretty accurately.
Now consider 2 scenarios
- CB estimates that kids that don’t take the test would do exactly as well as kids that do. Then user percentile equals national percentile.
- CB estimates that kids that don’t take the test would all do bad / worse than the top scoring kids that did take the test. So now the national percentile would be half the distance from 100. e.g. 96 user would be 98 national, 98 user would be 99 national, 99 user would be 99.5 national, etc.
If we have the user percentile, we’re done. Modulo various state adjustments, the NMSF cutoff is around 99.0%.
If we have the national percentile, we just need to convert it to the user percentile. In scenario 1, being in the 99% for national means you’re also 99% for user, which is enough for NMSF. In scenario 2, you need to be in 99.5% national to get 99 user, to make the NMSF cutoff.
I think our data shows that we’re definitely closer to (1) than to (2). So even if the SI percentile tables are national, the conversion to user at the high end isn’t that harsh.
Restating what you said: The population the table refers to makes a huge difference only if the percentiles shown vary significantly from the true user percentile.
@thshadow, my head is spinning out of control :-* :-*
Can you explain in plain words if possible? So top 15000 counts both 10th and 11th? Top 15000 just for 11th only
I know you know what you post, but please help
Compare http://www.cobbk12.org/news/2016/PSAT2015.pdf with http://www.ajc.com/news/news/local-education/cobb-students-outperform-state-national-scores-on-/np8K5/ The Walker / Wheeler numbers are not the same.
Thanks @Shelt29 ! That article:
http://www.■■■■■■■■■■■■■■■/wp-content/uploads/2016/01/Problems-with-the-New-PSAT-Compass-1-26-2016.pdf
was very informative and well written / organized.
One thing it brought up which I hadn’t seen mentioned before. They have changed the definition of what “percentile” means!!! 99% used to mean you did better than 99% of the population. In this PSAT, they changed it (to be more consistent with ACT) to mean that you did better than or equal to 99% of the population. AFAICT that means that for any percentile-based comparison to a previous year, you need to shift the score by 10 off the bat to account for the percentile definition change.
Yes, that article was interesting. I think the change in the way percentile is defined helps explain why counselors are reporting more kids scoring at the 99% this year. It isn’t necessarily that the percentile charts are wrong (although they could still be) it is that they mean something different by it this year.
@Shelt29 I have such a headache!
@likestowrite - I think your summary is relatively accurate.
There are (lower) cutoffs based on percentiles (which based on what I just said about percentile definitions, these need to be bumped up by a point), vs (higher) cutoffs based on concordance tables.
The argument for percentiles is mainly that those tables are not marked preliminary, while the concordance tables are.
The argument for concordance tables is mainly based on the anecdotes being reported.
An argument against both of them is that they’re based off of a research study - not the actual test results.
An argument against the anecdotes are:
- Some of the percentile “inflation” is based on GCs incorrectly looking at the national percentile instead of the user (which doesn’t seem to be a big error, but it does have an effect)
- Anecdotes are inherently unreliable to extrapolate from. Conceivably the schools that are issuing press releases are the ones that did the best. (For example, Cobb county never released a press release like this before AFAICT. Were they motivated to release the report because they did so well?)
One final point - I think CB wants the (absolute) scores to about equate to old scores. (I think I read that in one of their publications.) They want a 1400 PSAT to predict a 1400 SAT score. Which might imply that SI scores are directly comparable to past SI scores, and hence cutoffs would stay about the same as in past years (and would agree with concordance tables).
Yes, this whole report and especially the percentile changes from past PSAT’s have been very frustrating to try and decipher. I have been following this thread for a few days as I wanted to see where my twin D’s fit.
My biggest issue is that we have a 214 and a 212 for MN (same TS’s though) and it looks like we probably won’t make the cut, but nothing substantial to verify anything. The only SAT date they can likely take the test is the March one because of their activity schedule. I also hate to put them through another test. They’ve already taken ACT’s 33’s and their school will automatically test them for the ACT again in the spring.
I think I needed reminding that anecdotes Are inherently unreliable. So, there exist anecdotal stories that make the cutoffs look like they will be higher. But, as a category, anecdotal evidence is unreliable. It does not mean they are incorrect–just can’t assume they are written in stone.
I just read the CompassPrep article @Shelt29 linked. Wow! I wonder how understanding the definition change of percentile will help us in our analysis.
@Shelt29 just so you know, they should not have to take that MN “school-mandated” ACT. The schools by law have to administer the test but I don’t think that the students are mandated to take it - they can opt for SAT or ACT off-site, as your children have done. My D2 did not take ACT last year with the rest of her class and she meets the grad standard just fine with her SAT. For the class of 2016 at least the grad standard is satisfied by any major standardized test regardless of where it’s administered. You might want to make sure that nothing has changed now that the transition year has passed so check with the counselor to make sure or check online (it’s a pretty quick google search if I recall).
Another thing to consider: If their 33’s are fine, why must they sit it again and risk doing worse? that can’t help in the admission process (in case any of their schools of interest ask them to send all their scores). They might wish to be focusing on other things for their applications at this point.
It’s a convenience to be offered the test and school districts often tout the “free” aspect (conveniently skipping over the part about the lack of essay which is actually required for a lot of schools!). Knowing you most likely have a choice in that is something to consider.
@Shelt29, great article. Thanks for posting.
The definition change, as others posted, was the big surprise to me.
Definition A: The percentage of students scoring below you.
Definition B: The percentage of students scoring at or below your score.
My first thought when I read about that change was that there are twice as many students in the 99% range now using Definition B. Is that the correct way to read it?
Then a 99+ with Definition B really equals just 99 with Definition A?
Am I thinking about that correctly?
@thshadow wrote
Which might imply that SI scores are directly comparable to past SI scores, and hence cutoffs would stay about the same as in past years (and would agree with concordance tables).
I now lean toward believing this, but there is one perplexing hurdle to completely believing…if this was their intent, then why put out an SI% table that in no way at all supports it?
@Speedy2019 #1614 – I think you have it right. I think I mentioned before that instead of 16k sitting in the 99% slot you have possibly 35k, perhaps 40k kids in the SI 207-218 range who reports all show them to be in the SI TS and US 99% slot. This is not inconsistent with the definition change but makes predicting NM very hard.
Thanks for the info. We are considering opting out of the school ACT. The colleges they are looking at are local to the Twin Cities (U of MN and Macalester) and their 33’s are good enough to get into those.
So MN used to do BETTER than Low-to-Mid 99% for it’s cut-off.
Now, if MN stays in the same 213 - 215 range it will do AT-OR-ABOVE low 99+%.
I don’t see a problem.
Comparing 2013 SI percentiles to page 11:
200: Better than mid-96th percentile in 2013; at or above low 97th in 2015.
205: Better than top of the 97th in 2013; at or above low 99th in 2015.
210: Better than mid-98th in 2013; at or above mid-99th in 2015.
215: Better than low 99th in 2013; at or above low 99+ in 2015.
220: Better than mid-high 99th in 2013; at or above mid 99+ in 2015
If I poke around the lower scores I find that the page 11 percentiles are actually a bit more “harsh”! Compare:
195: Better than mid 95th in 2013; at or above low 95th in 2015
190: Better than mid 93rd in 2013; at or above low 93rd in 2015
Am I missing something or are the page 11 percentile tables a bit more reasonable now?