@itsgettingreal17 When Testmasters wrote a blog post creating a new estimate for TX, they used a set of about 8500 TX scores that they swore were valid, useful, and typical, at least for TX. They didn’t publish the data set, but they did post a graph of the data. The graph was detailed enough that a couple of us counted the data points and reverse-engineered the data. My percentiles are based on two things: 1) 3% of scores nationally fell between 209-228, and 2) I assumed that those 3% were distributed in the same way the scores between 209-228 were distributed in the TX data set. That second assumption may be wrong. The TX data may not be representative of the national data, but it is the only real data we have so far.
“Are we really sure that the shape of the curve would be the same nationally as for TX? I believe Testmasters conducted PSAT prep for the district, no? Wouldn’t the TX scores be slightly higher than a typical district elsewhere?”
This is a good question and all we have to go by in terms of confirmation would be anecdotal data (which have either supported or at worst, failed to disprove) and the preliminary concordance.
So take a look at TX. Last year’s cut-off was 220 and using the preliminary concordance tables that number corresponds to a current total score of 1460 which in turn is going to correspond to some SI range. We can easily compute the midpoint: its going to be 219 (just assume both sections are 730, drop the “0”, multiply the verbal by 2 and add the math: 73*2 + 73 = 219).
And 219 is, of course, exactly the Testmaster’s cut-off conclusion from their Houston Public School data set.
I have no idea whether this is a lucky occurrence but given that there are 8,500 data points, obviously that data set will capture some underlying truths about this current PSAT (as have the concordance tables, presumably).
Actually, we know that the scores in the Testmasters data set are lower than the national average. Just under 2% of the scores in the sample are 209 or above. That doesn’t mean, though, that the distribution of the scores doesn’t mirror the national distribution. I am assuming it is at least close.
@Mamelot and @candjsdad, thanks for the explanations. It makes a bit more sense now. My D has a 219 in CT where the cutoff is typically 220 each year. I can make a case either way for her to make or lose NMSF by a point. It’s going to be a long wait until September!
@kikidee9 Finally found my old post. It was in a separate thread: PSAT Discussion Thread 2015 p. 98 #1470. People from that thread switched over to this one. Please excuse the column headings. I didn’t know how to space things, so the three column headings are squished together. They are
Old PSAT score in each section
Old Selective Index
New 2015 Selective Index
(Now I know that it is called Selection Index, not Selective Index. Sorry)
My examples for simplicity used the same score in each of the three sections of the old PSAT.
My examples showed (converting from old score to new score):
201 converting to 206
210 converting to 215
216 converting to 216
225 converting to 222
@candjsdad There must have been a little misunderstanding because I never questioned the 209 commended score. The fact that 209 is approximately the 97th percentile lends further proof that the percentiles that were provided based on the separate test group don’t match the actual test takers’ percentiles. In the chart provided by CB based on the test group, the percentile for 209 is listed as 99 (on p. 11).
https://collegereadiness.collegeboard.org/pdf/2015-psat-nmsqt-understanding-scores.pdf
@BunnyBlue what concerns me now is that the “no change” SI number isn’t 216 anymore. It might be more like 219. That would indicate that everything up-river last year might come down but not by as much as people were hoping. And that there are a whole lot of cut-offs from last year which will need to come up. Not just everything below 209, or even 216, but everything below 219.
Edit/Update: I really hope I’m wrong.
thanks @BunnyBlue I appreciate you finding it. I hope we are all being overly cautious.
@Mamelot Yes, my method of choosing the same score for each section just provides a general picture of how things are, not the precision that you and many others have been attempting to provide. Mainly I just wanted to emphasize that the division between the percentile faction and concordance table faction is over, and there is no going back to trying to draw conclusions from the percentile table.
Sorry I forgot to include the link in my post #4104:
Definitely got commended, but hoping that Ohio doesn’t make some crazy jump…thinking 220 will cut it, though.
@smilingalong 215
@Endorcer88 OH and MN are usually very close in cut-offs. My D3 has a 220 as well. Fingers crossed for both!
I am not sure HSD data even tracks TX very well, leave alone national. We already know HSD data does not correspond to 209 SI at 97th percentile as required by commended cutoff.
Just thinking aloud here, does the spread of commended to highest cutoff for NJ in 2014 tell us anything? I am assuming pushing up of commended level from 202 in 2014 to 209 in 2015 already accounts for “the test was easy” students. From there on, I am hoping the distribution of SI’s in 2015 may track 2014 SI’s. If so, in 2014, NJ’s cutoff was 225 and it’s 23 SI’s away from 202 SI at commended. And the spread of max 240 SI to commended 209 is 38 SI’s. This year, the spread of commended to max is 19 SI’s (228 - 209). The resulting ratio of spreads 19/38 or 0.5.
So now, if 2014 cutoff for NJ was 225, this year we can assume 209 + ((225-202) * 0.5) = 220.5, rounding to a preferably odd number so cutoff at 221. For OH, this would be 209 + ((215-202) * 0.5) = 215.5, rounding up to nearest 216. But we don’t like even cutoffs too much, so perhaps a 217, but god willing, a 215!
I like your formula!
@OHtoCollege it would work fine if there were no bumps in the curve. Last year’s distribution was normal (I believe - we really don’t know but it has been so in previous years). I’m betting that over time there will be a definite relationship between the average old curve and the average new one. This year might be hard to establish that relationship.
Still, the results are pretty good. They certainly agree with the Testmasters results for the most part! I’m just thinking those happen to be point or two low for some of the mid -to-high states. Again, I hope I’m wrong.
I’m going to go ahead and say that Kentucky almost certainly won’t be under 215 after looking at my school’s data again. Under 215 would cause my school to have way more semifinalists than commended students, which I don’t think has happened before. I’ll be back at the end of May to see if the new tables align with my prediction.
So do you guys think Texas can possibly be 215? That’s what I got and I’m worried I won’t qualify.
Sorry @neoking, I’d be highly surprised if 215 made the cut for Texas. I think Testmasters’ guess at 219 is likely.
@destined4harvard can you please post that data again, now that we know commended of 209?
Is 220 safe for IL do you think?