<p>I used all available numbers that I have so far including % of the test takers in 75-80 and 70-74 range, mean score, Ns of the test takers at both state and national level, etc. from 2005 to 2012. Then, I tried to identify the best fitting equation to predict the cut-off score. I used a statistics program and checked all possible combinations of the available numbers and found that surprisingly simple equation predicted cut-off scores between 2005 and 2012 with 93% reliability. The % of test takers in 75-80 and 70-74 categories and the state mean were the significant predictors. By using the equation from the stat program and the new released state summary information (i.e., % of test takers in 75-80 and 70-74 and the state mean), I could estimate the cut-off of WI for the class of 2015 that was 209. Although the equation that I got from the analysis of the data from 2005 to 2012 did not predict the cut-off scores 100%, still 93% prediction is pretty nice. So, based on my analysis, my conclusion was 209. </p>
<p>I learned that the N of the semi-finalists in each state is decided based on the N of graduating students in each state and national level. I think I can make a better prediction, which would be close to 100%, if I have information of the N of graduating students, not the N of PSAT test takers. While I was looking at the data, I realized that the ratio of the N of students (state/national) would be another factor of the cut-off score because it is the critical factor deciding the N of semi-finalists in each state. One of my concerns is that the N of test taker in WI was almost not changed from last year, but the N of national level was increased little bit. If it reflects the pattern in the N of graduating students, it might imply the N of semi-finalists in WI would decrease. Then, the cut-off could increase than prediction. </p>
<p>Overall, although my analysis was a very simple level (I don’t have much time to do more articulate on at this point) and there could be another factor influencing the N of the state’s semi-finalists and the cut-off score, I’m concluding that the score will be 209 this year. Maybe I want to believe it and stop here and move on to other things that I’ve been set aside for a while. No excuse for procrastination now! </p>
<p>If you have a graphing calculator, you can throw all sorts of data in your lists, and play around with regression formulas. It’s kind of fun. I got some pretty good data with a simple linear equation for PA using just 75-80 totals. (That one predicted a one point drop for PA.) It’s all in fun. There are too many unknowns to be certain no matter how accurate the historical data may be.</p>
<p>Boy, my method of guessing (and I am guessing) the cut-off score for CA is much cruder than pom2015’s. Let me describe it and then if someone thinks there is a better way to refine it, I’m all ears. All I do is look at the number of test-takers that scored in the top category (75-80) in each subject, and I compare that to prior years to predict a directional trend (cut-off score up if the total number is meaningfully greater; down for the opposite). While I know not all NMSFs from CA will have scored 75-80 in all three categories (the minimum score would have to be 225 minimum then, right?), the cut-off score is generally pretty high and there are many test-takers that score in the top tier for each subject. So bringing in the 70-74 scores seems like it would not add much for CA prediction (but I’ve not tested that assumption).</p>
<p>Even as I typed this I realized their probably some things to look at that might improve the prediction. For example, CA almost always has way more test-takers get a 75-80 on Math than the other two subjects. For example, this year CA had 5,585 juniors that scored 75-80 in Math. That compares to 2,180 juniors that scored 75-80 in Critical Reading. Math was WAY UP (almost 50%) and Critical Reading was somewhat down (17%). So it would be interesting to see if results in one of the subjects (or two but not all three) are more predictive that the aggregate of all three subjects 75-80 results. I might look at that to avoid doing some work or chores.</p>
<p>I’ll end by asking for feedback re other States in terms of math scores. I know folks have said that the math scores went up in their States, but do they compare with CA, where this year the number of Math high scorers is up from 3,800 last year to 5,585 for this year? </p>
<p>Pom, you convinced me. That is SO much more sophisticated than my feeble attempt. I hope you are right. My S is a few points above all-time high, but I’m interested, because, well I can’t help myself is all. And some of S’s friends are sitting around 209, so I’ll hope for them too.</p>
<p>In WI 5.2% of test-takers scored 70+ in math this year, the highest ever. Before this the most was 4.2%. Last year it was 3.4%. I suppose that pales in comparison to performance of kids in CA and MA and etc., but it’s pretty good for us.</p>
<p>One thing to keep in mind is that sometimes states pay for all junior/sophomore students to take the PSAT. If a state started the policy this year, that could explain the increase in test-takers, rather than the increase being in correlation to the number of graduating seniors.</p>
<p>As for Texas, it looks a lot like 2010 numbers (219 cutoff) when I look at some of the data. And it looks a lot like 2012 numbers (also 219 cutoff) when I focus on other data. But the math scores really throw me. In 2013, there were 2400 juniors scoring in the 75-80 range and 3580 scoring in the 70-74 range. For 2012, those numbers are 1808 and 2684. The top math category doubled from 2011! </p>
<p>@pom2015 and @PAMom21, ya’ll lost me at the time stamp.</p>
<p>My analysis does not sound as sophisticated as Celesteroberts’. I plotted (in excel) the cut off score vs. the
average percent in the 75-80 category (add the 3 and divide by 3) for each year. The line fit (R-squared) was about 80% and predicted an Ohio cutoff slightly above 213. Since only integers allowed, I am predicting (hoping) for 2014 vs. last years 2015. What worries is me is the assumed decline in Ohio’s number of high school graduates and the affect it might have on Ohio’s allocation of slots. My son needs a 214 cutoff, so hopefully my estimate is not wishful thinking.</p>
<p>I realize now that my method is fundamentally the same as described by PAMom21. Graphing vs. the total % in the 75-80 or the average of the three will net the same prediction. </p>
<p>For NJ Here is my analysis
National NJ CR75-80 % CR 70.74 % M 75-80 % M 70-74 % W 75-80 % W 70-74 %
2012 1551059 68679 934 1.4 2006 2.9 1384 2 2101 3.1 1038 1.5 1503 2.2
2013 1579720 78745 732 0.9 1271 1.6 1949 2.5 2255 2.9 753 1 1851 2.4</p>
<p>Based on The above comparative stats of NJ, Looks like there is a significant drop(2%) in 70-80% for CR, an increase of .3% for Math and Drop in .3% for W. So the SF cutoff will most probably go down a notch.</p>
<p>My highest “r” value was 0.88, and that came from only using data from the past 5 years, which of course has its own flaws. The best and easiest correlation for PA though seems to be distance from the 99th+ cut, which we won’t have in time :(( .</p>
<p>My “worst case” estimates…which assume terrible things happening…only have PA going up by one point. Most of my fair playing around seems to point to a one point drop. I will be really surprised if PA goes up, even with the increased math numbers.</p>
<p>I’m REALLY excited about the potential for getting the commendation cut in April. Whether that goes up, down, or stays the same will be telling I think.</p>
<p>I also was close to an R value of 90% if I just used the last 5 years. I took data all the way back to 2002. The R square got worse, but the prediction was about the same. Adding in the 70-74 group reduced the R square, so I dropped that. I also tried number of students in the 75-80. No correlation. I also noticed that Ohio has a fairly consistent gap between cutoff and commended (12), so getting the commended number in April will be interesting.</p>
<p>CO was all over the place but the actual numbers - whether you look at 75-80 scorers, 70-80 scorers, percentages in each category, actual numbers in each category, mean scores overall or mean scores for each subtest, look more like a highER cutoff year rather than a lowEST cutoff year. I am looking for a one point drop from 215 cutoff to 214 cutoff for class of 2015.
As you have all noted: M scores were higher, CR scores were lower. Writing scores were also lower. This holds true for the high end and the mean.
Test taker numbers were up slightly from last year but still significantly lower than the highs of '08.
Good luck to all! </p>
<p>I don’t have the statistics know-how of many of you, but I would love to hear your opinions on Texas. My son is sitting on a 219, so it will be a looooooon wait.</p>
<p>But here is my most recent method of analysis. Not scientific, but I think it is logical. </p>
<p>I looked at last year’s data and tried to figure out whether the actual number of test-takers who scored above my son in each category is higher or lower than it would have been last year. My goal, of course, is trying to figure out whether my son is going to be in the 1250 or so NMSF’s from Texas. My son got a 74 in Writing. Anyone who scored in the 75-80 range this year scored above him, and this year there are fewer of those than last year. His scores in Math and Critical Reading were in the middle of a range, so I had to guesstimate. But I think more people scored above him in Math this year than would have last year, but not many more since he had a 78. For CR, I think fewer people scored above him than last year. </p>
<p>So, using this technique, I think his 219 looks pretty good. </p>
<p>Any opinions? Am I just searching for ANY method that makes a 219 look good in Texas? Yep, probably. :-S </p>
<p>I found some data online that showed a graph of the increase in Texas graduates. It has been steadily rising, but the rate of increase seemed to have slowed down for the last year of data, which I think may have been class of 2012. I also found data showing a big decrease in number of dropouts in Texas. So the increase in grads is probably due to a combination of factors, such as increase in population and decrease in percentage of dropouts. But maybe all states have an increase in grads as population increases. So I don’t know that Texas has an increase in relative to the whole county.</p>
<p>Also, I don’t know if NMSC changes the distribution of NMSFs yearly, or occasionally, or randomly. Does anyone know whether the number of NMSFs per state is actually based on the most recent figures, like the prior year? </p>
<p>Barfly, I took a look at the Texas data for you, and while my methods are certainly not full-proof, I think there is a really good chance that Texas will remain at 219. We are in a similar situation, as we have the same 217 that was the cutoff in PA last year. Would you be willing to look at the PA data, and weigh in with an honest “guesstimate?”</p>
<p>Barfly - the only thing I would do for your son’s mid-range scores guesstimate is to look at the Score Conversion Tables and see how many scores are actually in the range. For example, for 2013 math scores in the 75-80 range, there are only 3 scores, 76, 78 and 80. I would arbitrarily allocate 1/3 of the scorers in that range to each score. I would do that for each year for which you have a score conversion table and try to see if those numbers of scorers are more closely aligned to the Texas cutoff number.
Also, Texas had 1348 NMSF last year vs. 1237 the prior year. The number of NMSQT entrants was lower and the total number of NMSF was higher so TX actually received a higher % of the total NMSF awards compared to the prior year. We’ve discussed before the problem of which cutoff score for each state produces about 16K NMSF, but from economic data TX is one of the strongest states in the country which implies population growth (and graduation growth) above other states growth (or decline). I wish I had NMSQT annual reports for a few years before 2012 as that would show the long term change in TX NMSF numbers. NMSF must use old graduation data because they announce NMSF in September,before they know how many kids will actually graduate in May/June. I would guess that NMSC uses 2 or 3 year old graduation data as it takes government entities forever to get out any kind of information and NMSC basically begins the cutoff process when it sends out the notices to schools in April about high scorers. They have to have a very good idea of the commended cutoff and state cutoffs before they send out those April notices otherwise they would be raising the hopes of a lot of kids unnecessarily. When we get those notices we will begin to truly understand the direction of state cutoff scores.</p>
<p>As for my opinion, I think your son is good at 219. I will toast him tonight at dinner.</p>
<p>Barfly, I took a quick look at Texas data for last two years and I think Texas has a good chance of going DOWN. Certainly better than California’s chances (although I now am feeling bullish on no change or 1-point drop for CA). On the 2012 test, 0.6 percent of Juniors scored in 75-80 on CR section (and a total of 2.2 percent scored in the 70-80 range), 0.9 percent scored 75-80 on the Math section (total of 2.2 percent scored between 70-80) and 0.7 percent scored 75-80 on the Writing section (2.2. percent scored between 70-80).
For the 2013 test, only 0.5 percent scored 75-80 (and only 1.4 percent scored between 70-80) on CR, 1.1 percent scored 75-80 on Math section (2.7 percent scored between 70-80) and 0.5 percent scored between 75-80 on Writing (1.9 percent scored between 70-80). Of course, you knew all that.
My reading of those percentages points to a high chance of at least a slight drop. The percentage of high scorers in Math is up, but not by as much as in some other State and not enough to even offset the drop in the other two sections. Ditto when adding in the 70-74 scores. So, while one always wants to be cautious when predicting because it creates hope/expectations for those near the cut-off, it would be unexpected for the cut-off to go up from 219, which was already a big jump from prior years, when the overall percentages in both the 75-80 and 70-80 groupings are down.<br>
Now guessing how many points the cut-off might drop to would require looking at additional data (and still it is a guess, of course). There was enough of a drop in the 70-80 overall percentages though to think that at least a one-point drop will occur. Sure hope so for all the 218ers in Texas!</p>
<p>@numbersfun, earlier today I had looked at just that thing! Great minds think alike! I did that not just for the midrange scores, but for all the scores, but only for 2010 and 2012, which have similar numbers in Texas. But I have the conversion charts for all recent years and will take your advice and look at those as well. My son’s 78 in Math is interesting because for Wednesday test takers, the highest possible scores are 80 and 78. For Saturday test-takers, the highest scores are 80 and 75. That makes the 78 score look even better as only the perfect scores beat it, and Saturday test-takers couldn’t even tie that score. </p>
<p>For his 74 in Writing, the highest Wednesday scores were 80, 78, 74, 71 and 70. For Saturday, the highest scores were 80, 77, 72, 71 and 70. So while he is at the top of the 70-74 percentile range, the closest Saturday test-takers in the 70-74 range scored a 72. </p>
<p>Not sure how to interpret this, but it seems like good news. Although, I don’t know if many Texans take the Saturday test - I’ve never heard of any. How would all of you statistics gurus interpret this for your states?</p>
<p>I did not know there were 1348 NMSFs for Texas last year. That is quite an increase from the 1237 of the prior year! If anyone has historical info on number of NMSFs in any states for particular years, that would be interesting to see.</p>
<p>I do believe the 1600 NMSFs and 1500 NMFs stays relatively consistent from year to year, regardless of number of graduating seniors and regardless of number of NMSQT entrants. If NMSC were to increase those numbers, they would surely get many complaints from the colleges and companies that promise to give scholarships to all NMFs. If NMSC were to decrease those numbers, they would surely get many complaints from all the colleges who boast in their literature about how many NMFs they enroll each year. </p>