<p>I have been following this thread for a few weeks and am very impressed with all the number crunching. I also get a warm feeling that so many parents are pulling for not only their children but the children on other posters as well. My daughter took a review course and spent many hours studying on her own in hopes of becoming a NMSF. The week before the actual test she took at practice test at her review course and scored a 228. She did not let that go to her head but keep studying. Her actual score is 218 and we live in Texas so she is right on the border. She told me she doesn’t want to think about it anymore until September so I am not talking about it with her. I, on the other hand, am drawn to analysis so I want to thank all of you for sharing yours! By the way, in addition to crunching the numbers for NMSF I am trying my hand at winning the billion dollars in the NCAA bracket competition. Are any of you working on that as well?</p>
<p>SO SORRY! I had it backwards…You are all correct. No * is a GOOD thing!</p>
<p>Ha! Didn’t mean to start asterisk panic, but after reading Numbersfun’s post, I was convinced my “almost sure thing” son would have done something stupid, LOL! He’s been known to miss problems on the right side of the page, or maybe on the back of a test before, <sigh> . I’m glad I double checked, and can scratch that worry off my list.</sigh></p>
<p>Lessismore, welcome! Wouldn’t that 228 be nice right about now? Fingers crossed with you that the 218 will be enough. If not, there’s always the billion dollar option, LOL! Who needs scholarships with that much money?</p>
<p>I called the National Merit Corporation to see if I could get the annual report for priors years and was told that in the past organizations have used their data for purposes other than recognizing the merit students (educational funding, etc.) so they don’t release the numbers after the old reports are replaced by the new reports. </p>
<p>The number of test-takers does matter, particular if there is a big blip, a one year fluctuation up or down.If that happens, really need to look more at numbers of students in the various scoring categories, instead of percentages. I know some here are doing that already as their basic method. I prefer to look at %ages if the test-taker numbers are stable. Not sure why. I guess they are easier to quickly eyeball for a basic idea of things.</p>
<p>I was looking at Kansas data a bit after someone posted here from Kansas. They have, in general, a similar profile over time to my state, WI, with a very slow but steady erosion in # of test-takers. Their population is rising, albeit at a rate about half that of US as a whole, and under 198 is rising similarly, so it isn’t a loss in potential pool. It must be like WI, the test becoming less popular. But the Kansas decrease is less than a 100 students a year on average.</p>
<p>Anyway, for some reason Kansas had a huge decrease in test-takers this year, about 8%. That tends to depress cut scores in general. A blip up tends to raise them. If you analyze the KS data with #students, not %ages, it’s looking like a 2 point drop this year. Could even be 3, but that’s a stretch. As PAMom noted, not exact science here. But it does really look like a drop. Have to be a bit careful using #students as you go back further into time comparing with years when population of students/test-takers was smaller. Or in the case of WI/KS, more test-takers further back. I guess a real data cruncher would exactly be correcting for that. But you can get a pretty good idea by eyeballing.</p>
<p>Barfly, I was thinking of that what you said about non-citizens not being program entrants. </p>
<p>Is anyone willing to look at South Dakota’s numbers? I am a mom without a background in statistics. My son has a 213. Based upon a simple comparison of % fluctuations in the 75-80 score group over the years, I am guessing that the cut off will remain the same or tick up a point. Thanks!</p>
<p>Elk1122, took a quick look at Alabama (why work when there is State data to look at, right?). Comparing the 70-80 high score precentages for this year and last, looks like a drop from 211 is possible/likely. Adding in 65-69 percentages the years are roughly equal indicating possible no change in cut-off. So, with all prior caveats, I say drop likely, no change possible, and increase not likely. Good luck with 212 but hopefully won’t need luck! </p>
<p>I think I’ll look at Ohio next, because they say as Ohio goes, so goes the Nation, right?</p>
<p>This is a busy thread, just saying. I guess colleges have the PSAT scores from college board now. The mail is coming from Oklahoma and Bama. What I really want are offers to waive the application fee. </p>
<p>My KS post is supposed to say ‘under 18’, not 198, too late to edit,darn.</p>
<p>celesteroberts - This is where all the cool people hang out…or obsess.</p>
<p>The post by @celesteroberts reminds me of a question I have: Does anyone know what exactly colleges can get from College Board? I have always heard they can just get all the kids in a certain range, or something like that. It seems like the mail this time is more specific (such as OU’s “Congratulations on your incredible score on the PSAT” flyer) than when my other 2 kids (classes of 2010 and 2012) were at this stage of the NMSF procedure. </p>
<p>Thanks to all for sharing your thinking. I’ve been playing with the numbers as well, so I thought I should share what I found. I did a regression analysis of Illinois data using the past cutoff scores as the dependent variable, and the past numbers of NMSF’s and the total number of scorers in the 65-80 categories as independent variables. Excel has a data analysis add-on that’s pretty simple to use. </p>
<p>It seems to me that the variability of the number of NMSF’s in a given state from year to year needs to be accounted for. It was tough to find the past numbers of NMSF’s for Illinois - I had to do some googling to find them.</p>
<p>I used the last 5 years data, and R-square is a very strong 99% for Illinois. Using 70-80 or 75-80 did not produce as strong a correlation. It predicts this year’s Illinois cutoff to stay at 216. I had to project this year’s number of NMSF’s (I used this doc <a href=“http://www.wiche.edu/info/publications/knocking-8th/knocking-8th.pdf”>http://www.wiche.edu/info/publications/knocking-8th/knocking-8th.pdf</a>), but even when I varied it pretty widely, it still projects to 216. </p>
<p>As someone has already suggested, the best range of data to use (65-80, 70-80, etc.) probably depends on the typical cutoff score - the states that are above 220 would probably be most accurate using the 70-80 data.</p>
<p>Wow, interesting. I’m a bit bummed because my Excel spread sheet doesn’t have room on a single page to include the 65-70 data, LOL! Plus…I’d have to go dig it all up. Hard to imagine getting to 99% though. Do you remember what you had for 70-80? I had my best numbers with just 75-80, but that was only 88%.</p>
<p>Hofmannmd, I looked at SD’s numbers, and they are up slightly at the highest levels (75-80), but down if you consider the next categories (60-80 and 65-80 categories; this might make sense given SD’s cutoff is lower). I wouldn’t be surprised if the cutoff even dropped a point from last year’s 206. Either way, your son should be absolutely safe at 213.</p>
<p>PA, I believe 70-80 was 79%, and 75-80 was 68%. I agree that the 99% seems too good to be true. One of the P-values was too high (.2), so it could be the stars aligning with few data points. I’m not enough of a stats person to know what to make of it. We are also looking at different state’s data.</p>
<p>Barfly, Start here.There is a lot to look at.</p>
<p><a href=“College Board Search”>College Board Search;
<p>Well, I just threw in the data for the last 12 years, and the correlation was horrible. I plotted PA’s cut against total high scorers in the 65-80 range, and came back with only 37.8%. Maybe I’ll try using just the last 5, which gave me the best results for the other ranges. For what it’s worth, our cut is typically in the 214-216 range.</p>
<p>Are there any other threads that are this intense about making the PSAT especially if you have a student on the “bubble?” It is hard to imagine, but I have seen some references to the other threads.</p>
<p>Okay, I repeated the PA regressions for all three groupings for just 5 years, and got the following:</p>
<p>Full 65-80 totals, r = 39.7%, which is terrible, so I didn’t even predict this year’s cut
Full 70-80 totals, r = 85.8%, predicts a PA cut of 216
Full 75-80 totals, r = 88.5%, also predicts a PA cut of 216 </p>
<p>Even at 88.5% though, it predicts incorrectly for 3 of the 5 years, though it’s always within a point.</p>
<p>OK…I can’t hold back, I have to weigh in again:) So…like PAMom, when I did the analysis for Ohio I only used the top ranking and got around 80% R2. Adding more rankings just made things worse. Full disclosure, I threw out 2004 because it was such an outlier for Ohio. If I use the equation to predict the remaining years, it does it incorrectly 5 times of 10 times, but…the method I use was to always round up the prediction to the next highest integer on the logic that there is no such thing as a decimal cutoff. The years that were wrong were predicting high by one point, so conservative. My equation predicts slightly over 213. So by my method I predict 214 for Ohio, right where I need it to be in my wishful thinking:) Remember, figures never lie, but liars figure!:)</p>
<p>Correction to my post above:</p>
<p>total scores for 75-80, r2 = 79%
total scores for 70-80, r2 = 63%</p>