I posted this last Friday 1/15/16 on another thread:
"I called NMSC to get some history for TX & was told that the number of NMSF’s don’t change that much (sorry I didn’t ask about other states) but I would venture to say the number awarded for each state doesn’t change that much from year to year. Also she confirmed to me that it doesn’t change that much because it’s based on population (I think graduating seniors is what she meant). The number awarded to TX has been running about 1300.
As far as concordance tables she told me not to use them as a way to figure out cutoffs for this year! Those tables were designed more or less for GC’s & sophomores to prep for junior year. Of course she told me the only thing I can do is wait, LOL! Well I know I’m NOT going to waste time on those tables unless I have a sophomore & I don’t."
@WayTutoring
You seem not to grasp how this works. NM Corp doesn’t pick NMSF by some PSAT national percentile cutoff. It has a more or less fixed number of slots (16,000) and divvies them up among the states, irrespective of a predetermined PSAT national percentile cutoff in that state.
Hello everyone. While there is much debate on cutoffs and percentiles I believe there are some misconceptions I’ve read on here. First of all there I’m fairly certain that there is not a simple fixed 50,000 student sample size to determine the commended cutoff. That is why CB always says around or approximately. .
It’s going to be impossible to know what the cutoffs for NMSF by state will be but without the total number of eligible 10th graders who took the test it is impossible to guess what the commended cutoff will be. Historically it’s been the top rung of the 96th percentile or so on the “Understanding Results Bulletin”. So this year (based on the bulletin) 198-200 would be the best guess. The other critical factor are the typically low scoring states like WY, WV, etc. Some of these states are also very small population-wise so their 1% of the pool amounts to only about 40 or so NMSFs for the entire state. If that 40th student’s score is 198 or 200 then that will be the likely commended SI because CB has never had a commended score cutoff higher than the NMSF cutoff for the lowest state. Commended has been lower and most often equal to the Semifinalist cutoff for the lowest state. I firmly believe they determine semifinal cutoffs for states first. Take the lowest state and see where that score is. Take that score (say 200) and see how many students have that same SI or higher. If it’s 47K, commended will be a point lower to grab at least another 3K or more to reach that 50K. If that lowest state score yields 51K then bingo. You’ve got the lowest state SF and the commended cutoff in one swoop.
I’m with @gusmahler. I would discount any study that forecasts commended at 210. As I said likely 198-200.
@CA1543
Thanks for this summary.
I think Applerouth’s students agree with mine in the sense that the mid-level students had trouble with the higher complexity of the passages and the no-calc math section.
However, the people who get 800 in math level 2 found the math much easier than the math on the old SAT because there weren’t the odd-ball level 5 difficulty questions. Those people did not blink at long divisions or arithmetic of fractions with a pencil.
The people who are very strong readers were not bothered by the harder passages and were helped by the lack of trap answer choices and arcane vocabulary.
So while of course I am not at all certain about this, I wouldn’t be surprised if the strong students do better on this test and the middle students worse. The really low students are going to be helped by the fact that there is no penalty for guessing. And then the whole thing will be fudged by the curve to produce a symmetric distribution. What can’t be fudged so easily is an increase in the number of students who get perfect or almost perfect papers (say up to -5 in total), although CB has tried to do this with the W curve.
I have a big question about the Applerouth link. He was saying how current scores are concording to lower percentiles from prior years (i.e. using the concordance tables). But isn’t CB defining “concordance” to mean a relationship of one score to another given the SAME percentile?:
“The term concordance refers to establishing a relationship between scores on assessments that measure similar (but not identical) constructs. Two scores are considered concorded when the percentage of students achieving each score is the same. For example, if 75% of a group of students achieve a score of X on one test, and 75% of the same group of students achieve a score of Y on a different test, score X would be considered con corded to score Y.”
(Understanding Scores, page 20)
Note what is said: “Two scores are considered concorded when the PERCENTAGE of students achieving each score is the same.” I would take that to mean the percentiles for each score (whether it be current-to-previous or previous-to-current) have to be equivalent. So if someone is concording a current score to previous scores and ends up in a different percentile doesn’t that mean something went wrong somewhere? Either the preliminary concordance tables are off, or the method of concording is somehow wrong, or the previous percentile tables are incorrect - or something?
“Either the preliminary concordance tables are off, or the method of concording is somehow wrong, or the previous percentile tables are incorrect - or…”
Or the new percentile table is off.
I hope the new percentile table is accurate. If so, my school is likely to have 4 times as many NMSF this year as we’ve averaged over the last 4 years. If the concordance tables are accurate, we might have one more student than average make it.
I don’t know how to reconcile everything, but any time a student got his score back, he could go to the understanding your scores and see where he fell in the percentages to determine if he had a shot.
My son’s score is 215; it concords to a 213-217. This may make our cutoff based on some of the cutoffs in the past. But, he is in the 99+ percentile, which would always make the cut. So he went from low 99th percentile on concordance to 99+ on the table on page 11.
So, it boggles my mind that he would not make it in TX. Maybe our cutoff will go up relative to the percentiles. But, I keep hoping that of the 6000 kids in CA that make commended, a substantial portion will have a score below their cutoff, but higher than the TX cutoff.
@micgeaux My daughter’s score is 212 (1430, 34 ® 35 (W) 37 (M)) and we are also in TX. So if your son doesn’t make it, for sure she won’t make it. I called NMSC last Friday and was told not to use the concordance tables to predict NMSF’s because those tables were designed to help GC’s and sophomores to help them prep for their junior year testing…whatever that means… I took it as those tables were designed as a guideline and not a “bible”.
Maybe some of you can call NMSC and ask about the concordance tables vs. SI percentiles and share your findings. I spoke to someone in public records dept.
One other thought: it seems like a lot of people think that kids did better with this new test. Maybe we only heard from kids that did better with the new PSAT…perhaps there are kids that typically would do better with the old PSAT (because it’s an aptitude test) don’t do as well on the new version…therefore, net net it’s the same.
@micgeaux Why would CA commended matter to TX SF SI cutoffs? In fact the data could align so that all top 50K commended students are from CA and none from other states, but still have SF from each state per that state’s contribution of graduating seniors to overall student population. So the SI %ile’s you see for 215 at 99+ %tile is at best of all the PSAT 11th graders, but likely among all of the PSAT test takers (freshman, sophomores included). This is where the confusion stems from, you may be comparing your kid against a national pool of test takers as opposed to your own state’s 11th graders. It’s possible TX performs much better and thus SF cutoff exceeds 215, but kind of like you I am keeping my fingers crossed as my daughter’s SI is 215 as well!
My only other thought on this is that some people where unaware on how to study. Maybe some schools WILL do way better this year at the expense of others. I’m trying to determine whether my son should take the March SAT. I don’t want him to sit through ANOTHER standardized test, if he doesn’t need to because his ACT score is fine. (And it seems like everyone he turns around, he is being tested. Benchmarks from the district, tests in his regular classes, etc.)
What I mean by the above is I was talking to another mom, and she said know one that she knows scored above a 1300 because no one knew the test would be so different and how to study for it. College Board sent a representative to our school in September, but school had already started. If you are a kid in AP classes and a fall sport (marching band, tennis, football, etc.), you do not have extra time to study for PSAT. I went to the CB meeting and one of the moms that I know said to me Did you know about this Khan Academy stuff? I wish I had known about it so that my child could have studied over the summer. She is a well-informed mother.
So, while a lot of you might find this hard to believe, it was difficult over the summer trying to figure out how to allocate your time toward studying. I mean, which material.
@Mamelot
The way that ACT/SAT concordance tables were developed was by equating percentiles in a study group, not by equating percentiles in the entire test-taking population.
“Concordant scores are defined as those having the same percentile rank with respect to the group of students used in the study…Concordant ACT and SAT scores may vary significantly across students and colleges. Students included in this study are not necessarily representative of the students at a particular institution.”
Do you think CB used percentile ranks of the entire test-taking populations in 2014 and 2015 to concord scores? If it did, I don’t think we would see the kind of discrepancies that Applerouth turned up. The concorded scores would have to have the same percentile ranks. Ergo, either CB must have used a study group, or the 2015 percentile ranks are (very) inaccurate.
@DoyleB - yes, or the current percentile table is off, that must be included as a possibility.
Here are the issues as I see them:
previous percentile tables: I can’t recall what Applerouth was using for his previous percentile tables - he may well have been using ones from the previous “Understanding Scores” reports but I just don’t recall whether he disclosed that or what exactly he disclosed, nor how he did his calculations (point estimate or range?). So his analysis may be flawed although I think a couple of folks on these threads replicated some of his results which supports what he is saying.
The concordance tables themselves are the result of analysis, computation and estimation as well as being preliminary - so there is room for error there. They will be finalized in May and it will be interesting to see whether everything hangs together a bit better at that time.
Current SI table (page 11) - there would be no excuse for messing these up because the actual data is available. The problem is that CB didn’t disclose exactly what percentiles are based on so no one has a basis for saying that they are good or bad. We simply don’t know, but it doesn’t look good to remove an explanatory bullet point.
When I concord my D3’s 220 I end up with a range of 219 - 226 with a most likely estimate of 222-223. That corresponds to just below the 99.5 percentile (2014 SI table) compared to the 2015 table (page 11) of well in the 99.5. So they are off. But not that much. When I concord the MN cut-off from last year, 214, I end up with a range from 213 - 218 with a most likely estimate of 215-216. Last year’s SI was in the low 99th percentile, this year in the low 99+. Again, off a bit but not by much. This is the only concordance analysis I have done but both conclusions seem very reasonable to me and suggest that the concordance tables, while not exact, seem to be decent tools for the task at hand. I didn’t get the large swings that Applerouth claims to have found.
I’d caution against using the current percentile table to predict a cut-off at this time because no one knows yet how many eligible 11th graders sat the test. An additional one million took the test this year but we don’t know how much of that increase, if any, can be attributable to the NMSQT pool. Therefore, there could be anywhere from 8,000 to 13,000 kids at that 99.5 percentile!
@Plotinus I guess the next question would be how exactly did CB derive the concordance tables? They have an entire history of previous tests so presumably they can draw samples from that data. What, then, are they using on the “current” end to establish the relationship? The actual results? The “research study” results? My guess is “research study” has to suffice until they get the results from the first SAT - they then will have three sets of data: one hypothetical (research study), and two based on actual results (PSAT, March SAT). Wouldn’t surprise me to find those concordance tables being fined tuned just a bit.
Mme, I agree and respect your opinions . I believe data (SAT and PSAT) released CB for the 4 years are very REAL and Data for last 4 years behave in a very unique way
If one can access SAT data from 2012 to 2105, you can always see lowest 99+%tile is very close to 99.54%. Plus the student ranked at 16000th is usually at high of bottom 3rd of 99%tile (just divide range 99%tile into 3 parts - one nest to lowest 99+%tile). - based on Total Scores
How do I transfer the above SAT data to predict PSAT cut off.,
I look at PSAT data from 2012 to 2014, I believe it behaves same way the SAT data, translate that — mean the last 16000th (considered for NMSF) will be at the low of Mid 3rd in 99% range (next to the lowest 99+%tile). Translate again, the cut off SI for NMSF is at the low of Mid 3rd - look at past cut off
Conclusion
1- If you are in states with higher cut off. Very good chance for NMF, if you are at mid of top 3rd of 99%tile range, and up
2- If you are in states with lower cut off. Very good chance for NMF, if you are in high bottom 3rd of 99%tile range, and up
I am a junior. If I am wrong, please ignore. I can say most posters in this CC thread are very concerned parrents, but not test takers that much. Discussion of statistic math makes dizzy
@Mamelot, I think the increase was 500K. From 3.5M to 4.0M PSAT test takers. Some of the increase will be 8th graders who didn’t test in previous years.
This is going to sound confusing but I want to hear your guys’ thoughts. My residency is in Texas but I attend a Nebraska high school, and I took the PSAT at an international school in Chile. Which cutoff scores do I fall under? I put in my Nebraska high school code on the test form along with my Texas home address. I’m not sure which I should be looking at but I did score in the 99th percentile.