@SLparent Missouri 218
@walnuthill â NY 218
@Walnuthill this is a great idea!
MN 216
Under 1 Over 0
@suzyQ7 - I totally agree. Thatâs why I said they are trying to get their credibility BACK and raise market share. But in the big scheme of things, at least in my neck of woods, nobody knows or cares about NMSF or even CB as a company. Parents and students might be familiar with the terms âSATâ and âAPâ but if you ask them who is behind that, they will not have an answer.
@thshadow wrote:
"Iâll just point out that the âuserâ and ânationalâ numbers are inarguably pretty darn close at the high end. Look at the report:
https://collegereadiness.collegeboard.org/pdf/2015-psat-nmsqt-understanding-scores.pdf
Page 7 clearly shows both the national and user percentiles for (total) scores. You can see that National 99+ starts at 1430, while User 99+ starts at 1440. National 99 starts at 1370, while User 99 starts at 1390. Itâs clear that going from user to national does not do anything drastic like approximately double your distance from 100.
I agree that the selection index percentiles do not say what the percentiles mean. But even if theyâre âNationalâ, converting the highest scores to âUserâ would likely only make a very small difference (e.g. maybe 99+ starts at 215 instead of 214)."
Youâre not understanding what Iâm trying to say. Let me rephrase and try againâŠ
It doesnât matter if the user and national tables are identical - all that means is theyâve made the assumption that the students who didnât take the test are distributed identically to those who did.
What is important is what each rank means.
For the sake of discussion (because it makes the math easier), letâs say there are 15,000 NMSF
Scenario A: The SI table refers to 1.5 million kids who took the test. To be in the top 15,000, you need to be at or above the 99.0% in the table, which corresponds to a 205.
Scenario B: The SI table refers to 3 million kids across the US. To be in the top 15,000, you need to be at or above the 99.5% in the table, which corresponds to a 214.
The population the table refers to makes a huge difference.
Also, at the top of the SI % column, it states Grade 11 only). So, they must have done something to arrive at that title.
@LivinProof, sounds like you have lost hope. Without hope, we will have Donald Trump or Hillary Clinton as our next president!
Seriously though, if your daughter/son turns out to be 99% instead of 99+% or even 98%, these are some of the highest scores in a country of millions of people and students. Take pride in that.
No one knows what the cutoffs are going to be. Not me. No one. The state summary reports should provide info allowing for refined predictions. But they are just predictions.
Iâm the optimist. Yes, I sound like Iâm defending CB at times. But donât give up hope.
Been following this thread closely and just want to say:
1 - a BIG thanks for all the work of the posters. Itâs been an amazing informal collaborative effort, a kind of investigative data sleuthing to find at solution where thereâs conflicting evidence presented from an authoritative source (CB) along with increasing anecdotal evidence that points to a certain direction (higher cutoff scores). For me, this sleuthing, almost forensic accounting element, has made it so fascinating.
2 - For those of you on the bubble, I am rooting for lower cutoffs.
3 - Of course once we know the commended # weâll have our North Star, know the accuracy of the SI percentiles CB published and be able to triangulate the cutoffs based on those historical percentiles.
4 - Took a look at the 2014 PSAT Understanding Scores and was surprised to see a combination of 2 67s and 1 68 would have made commended cutoff (just). So a tester who got a 68 in CR was 97%, 67 in math was 93%!!! and a 67 in W was 96%. If you average these percentiles they come to a low 96% to make commended.
5- Again I do not know if this confuses the issue more, but even if the current percentiles are inflated by 1% in the top end, as I think they are, perhaps the commended might not be as high as some are saying (i.e Testmasters predicting 210). It looks like historically there has been quite a big percentage spread.
It makes no sense to produce SI percentiles that are based on all 11th graders. It violates precedent as well as basic competency. Has CB ever produced a percentile table that turns out to represent the wrong population group? That would be unbelievable.
@TallyMom2017 wrote: âI seriously doubt that CB would put out ANY numbers that have not been seriously vetted by more than one in-house as well as out-of-house consulting firms/statisticians. Too much hinges on it for themâ.
Well, youâd think. But
(1) More than a week after it was released, the Understanding Scores report still says that a student who got 38 out of 47 right on the reading section of 10/28 test would get a scaled score of 38, a perfect score (page 13).
(2) More than a week after it was released, the Understanding Scores report still says that the maximum # of correct answers on the math section for 10/14 was 47, even though it was apparently 48 (page 13).
(3) More than a week after it was released, the Understanding Scores report still says that the mean SI was 468 (page 11).
(4) Most importantly, the concordance tables, released at the time the PSAT scores were released, and then rereleased as part of the Understanding Scores report, are in conflict with the with percentile tables, by a lot. Neither the concordance tables nor the percentile tables purport to be based on actual scores of the students who took the October 2015 PSAT, so neither is technically false. But either one or the other must be seriously misleading to anyone attempting to guess what the state NMSF cutoff scores will be (an activity that CB would probably say it doesnât condone). Yes, the concordance tables are marked âpreliminaryâ, because data from the March and May SATs will be used to form the final ones, presumably. But these concordance tables may already be being used by schools â for example, to determine minimum scores for entry to AP or honors classes. The percentile tables arenât marked âpreliminaryâ, because they arenât ever going to be updated; they are based on a research sample, and that data isnât going to change. Maybe the actual percentiles from the October 2015 PSAT will be released next year, after the October 2016 PSAT, but whether they are or they arenât, there wonât be an updated version of the percentile tables that we now see.
I think that there is a possibility that the preliminary concordance tables will turn out to be deflationary of 2015 scores, that the actual percentiles for 2015 PSAT scores are somewhere between what is suggested by the concordance tables and what is suggested by the percentile tables. But if theyâre not â if the concordance tables are right on the mark, then CB could simply say that the percentile tables were clearly labeled as based on a research study, and that they are pleased that the research study predicted the actual scores so well (even though they doubled the number of kids who may have thought they were in running for NMSF, theyâre accurate within a percentage point or whatever).
As far as CB doing lots of vetting, letâs donât forget the June 2015 SAT. The effect of that mistake, which could have been avoided if one proofreader had seen the problem and effectively brought it to the attention of someone with power, was incredibly large in comparison with the percentile tablesâ prediction of actual scores being off by a percentage point or so. The screw-up in June 2015 SAT had actual effects on the assessment of every student who took that SAT, unlike simply not updating a research study data with the actual data at this point (for whatever reason).
Iâm becoming weary of this, and I need to get back to work. But before I go, Iâd like to re-visit one point @thshadow made that Iâve addressed before.
I find it very strange that the âuserâ and ânationalâ tables are almost identical at the high end. Essentially that means that when CB adds in all the kids who didnât take the test, the distribution doesnât change much - almost not at all. That seems odd to me. Since the national group is roughly double the size of the test takers, that means when CB makes the national table, they assume that for every kid who scored high enough to become a NMSF, there was another kid who could have scored that high, but didnât bother to take the test. I find that assumption intriguing.
Is that true where you guys live? Around here only a third of the juniors in the state took the test. But from what I can tell, far far more than a third of the kids who could score high enough to become NMSF took the test. For them, unless grandma died or they were really sick, they were in the room. NMSF is prestigious, helps with college admission, and can be a financial windfall. Perhaps things are weird around here - I know some states are ACT centric, and maybe others just ignore the test completely. Thatâs a regional issue, and Iâm referring to a national scale. SoâŠ
Does it make sense to everyone that as just as many of the top, top kids in the US ignore the PSAT as take it?
You probably know choice of scores sent to college. Mean you can specify any scores (both SAT and ACT) on any of your test sitting dates to college for admission or scholarship committee. And college admission officials will mull over the scores you sent.
Reality is at the end, the colleges request final record transcripts to validate your claims (bragging right for certain high scores in SAT or ACT). I saw my brotherâs final record transcripts - it is fully-described every single test he took since freshman year. After that, then after they reviewed final transcripts ⊠then they (college) will sent letter, which inform the award amount of certain scholarships
How does it relate to my next saying?
I believe college people will not look for candidates with sudden fame (one single highest score), but they would consider his/her progress thru years.
Old PSAT scores is 3 digits ( 240), tranfers to Old SAT 4 digits (2400) - just add 0 to old PSAT 3 digits
So the TS (1520) with new PSAT, will closely associate with new SAT (1600). It will help college admission or college scholarship consideration committee to decide.
Itâs kind of easier to view records in term of scores, instead of seeing index of 228 (remind oneself that itâs used to be 240, then relate to present scores)
NMFâs scores are at the very very extreme ends of scores. And most NMFs are looking forward either admission top choice colleges or some hefty scholarship or admission to top programs (like 6 years Medical or 7 years Medical schools without having to apply and interview, Hunt scholarship at SMU, McDermot at UTD and etcâŠ). Only the very few NMFs do not care any anything (snobbish? maybe).
I believe CB and NMSQT start selling ideas of TS, even they says SI will be used for selection of NM
Before concluding from the relatively generous math and reading curves that there wonât be clumping of scores at the top or that no statesâ cutoffs might go up, we need to consider the effect that having a less generous curve in those areas might have on the balance of boy and girl NMSFs. I am not saying that any state will have a 228 cutoff, but once you get into lower cutoffs, who knows?
Suppose NJ ends up with a 224 cutoff (not predicting that, just using it as a simple example). There would be various combinations of errors and scaled scores that could get you that score, but on the 10/14 test, you couldnât miss more than 2 in writing and still get a a 224, and you couldnât miss more than 6 in math and still get a 224. What if the curves of the three tests created approximately 50/50 girl and boy NMSF in NJ? What if the curves approximately balanced the number of girl and boy NMSF in most states and the total number of girl and boy NMSF?
There are definitely some top kids who do not take the PSAT. I think some kids just do not understand the importance of it and our guidance office does not promote it much other than putting it on the daily announcements. That being said, I still do think the majority of the kids taking it in our school are the above-average or top students. Most of the students in the bottom half of the class do not take it so I would expect user percentile to be much higher than national percentile, at least for our school.
@dallaspiano Schools do not care about progress on test scores over the yrs. They only care about your highest score, either from a single sitting or super-scored. And kids who âdo not careâ about NMF are most likely planning on attending schools who do not offer scholarships to NMF or are aware of similar level scholarships w/o NMF status.
@Moma2physicsgeek, not mean to anything disrespect at all. Have you ever worked with any top choice colleges under department of College admission or Scholarship Committee? If you do and if you did, I take your words at face values and you make me very happy since I can brag more and more with them (college admission office).
FYI not to brag (perfect scores in Math both ACT and SAT, very high on Reading SAT)
@dallaspiano No,I do not work for admissions, but my 11th grader is my 5th 11th grader. I have been through the scholarship hunt and college app process multiple times. So, yes, I do have a slight understanding about college admissions.
Then you probably know and understand the words âholistic reviewâ at UT Austin and Rice (I am in TX)
@Mom2aphysicsgeek #1594. I think you are right about this. Most colleges want the highest scores their applicants provide, whether they come from a single sitting or super-scored. This is b/c it affects their US WNR ranking. In other words, they have a vested interested in disregarding the lower scores or averaging all the scores. Perhaps the top tier donât have to do this b/c they are getting mostly SAT scores in the 97%-99% anyway.
Another note: I think some posters have tried to reverse engineer percentile probabilities based on # of wrong answers between the 2014 and 2015 tests. Offhand I seem to recall it was still possible to make most cutoffs with 2 99% and 1 98% score on the 2014 test â I am not sure how many errors this came out to. I know the tests are different (2015 no guessing penalty and less choices to make mistakes) but it had more questions and was longer, if Iâm not mistaken. It seemed you could get more wrong on the 2014 test and still come out ok. 1 mistake could be very expensive on the 2015. Was there any meaningful extrapolation done from comparing error counts between the tests or was it decided it was a meaningless exercise based on the inherent difference between the tests?
Thanks to everyone for all of their analysis. It is a complex and perplexing issue and one of great importance to many!
The Most Convincing Evidence about whether the NMSF cutoff will be closer to SI% charts or Concordance charts, IMO, is the actual data we have from Walton high school in GA and the Oklahoma high school that @CAtoOK mentioned. Both report very high scores from their juniors this year. The current SI% tables make if look like they are going to have 2 to 3 times as many NMSF as they historically average. This seems Highly unlikely that these two schools double or triple in NMSF numbers. Thus, I think the cutoff numbers will be higher than the SI% charts predict and more like the revised predictions of testmaster and prepscholar which factor in the concordance tables.
@DoyleB does a great job summarizing the OK school case in #1563. @Thshadow does a great job discussing the Walton case on the thread âstatistical (monte-carlo) analysis of Cobb county.â (Although, I feel certain the argument is summarized elsewhere. If anyone knows where the best summary of that is, please post it.)
Bottom line: We can speculate about the SI and Concordance charts and go back and forth about reasons why CB would make each of them. But letâs talk about the data from the Cobb county and OK school: Is there ANY reason to believe that these schools could double or triple their # of NMSF?
I would love to hear someone come up with a reasonable argument for this.