@LadyMeowMeow, the interesting note from your info is TX at 951. Total agree with you, Texas has traditionally had a higher state cutoff compared to the average. Last year was 220. While Texas has a low mean, there are areas that really excel.
Really hoping testmasters got actual test scores from Houston ISD. There will be a good mixture in HISD with many bad schools but also some high achieving students. Houston has a sizeable Asian population (maybe 6% or so), so you will get high achieving scores from that population.
@speedy2019 I would be so thankful and appreciative of any estimates for Indiana. If you look on everyoneâs data collectingâŠI am the only one from Indiana posting psat scores. Some think it might be a mid state that increases dramaticallyâŠhowever, I cannot imagine it going down to 206. Any help would be appreciated.
@kikidee9, NOT trying to falsely raise your hopes for a 206 for Indiana. However, my recent post is the Only explanation posted to this site that can make both the SI% table and Concordance tables work. If other people can make both work, please post your explanation.
As for Indiana, 213 out of 240 last year, 206 out of 228 this year. Sounds plausible on its face.
May I suggest you analyze the 2015 and 2014 SI % tables and make your own guess as to a cutoff for Indiana (assume 2015 SI % table is correct). You can take that as 1 possibility, then get the low end and high end ranges for the 213 concorded number. You will have 3 numbers that will give you a good estimate for your state. Beyond that, everyone is just guessing.
Now, if testmasters updates their Texas prediction with a significantly lower number compared to their previous 217 estimate, that can help guide you in determining which of those 3 numbers are more likely.
@Speedy2019 Good morning. I would love to see the cutoff goes down so my kid would have a higher chance. But we are here to find out the truth. There have been so many cases to show that if the SI percentage table were correct, a lot of schools would have several fold NMSFs. Possible, but highly unlikely. Just like you, I noticed Testmastersâ sudden change of mouth. As I stated earlier, unless they got a ton of data that totally changed the picture on Wednesday night, and they analyzed it overnight and reached a totally new conclusion, I can not convince myself there is a âchange of heart.â
@NathanBN, luckily I donât need to convince you. You can make your own decision.
The comment âwe are here to find out the truthâ, implies Iâm not.
It is very logical to take testmasters comment that 205 represents the 99th percentile as seen in a very large pool of data and conclude the SI % table is correct. BECAUSE THAT IS WHAT THE SI% TABLE SAYS IS 99TH PERCENT! Assuming the SI% table is correct or incorrect, people, you, me and everyone else can make their own predictions.
If you donât believe the SI % table is correct, that is your right. I donât care.
If you have predictions, post them so we can âfind out the truthâ. If you have explanations, post them so we can âfind out the truthâ.
@Speedy2019 I really have no intention to imply you are not here to find out the truth. But I would not take ONE comment from testmasters as proof of anything, because for any statement from them, you can find an opposite one across the "border.â By âborderâ I mean sometime yesterday when they took a 180 degree U-turn.
What makes me wonder is why the turn. If I have to make predictions, I think the Commended should be around 208. The cutoffs are very close to Teammastersâ updated (concorded) estimates (as they said on Wednesday to be âcloseâ) with higher states 1~2 points higher and lower states 1~2 points lower. My predictions are no better or worse than othersâ. We are here to figure out together because we are concerned parents.
But youâre going in the other direction, from 2014 â 2015, and using the tables on page 27.
In particular, for example, your writing table shows that 80 down through 77 all concord to a 38. Whereas the table in the other direction says 38 goes to (only) 78. So your algorithm could try many different 2014 scores for a 2015 38, while mine would only try 1. So itâs logical youâd get a bigger range than me.
Not that this technique is really scientific or anything. Why would it make sense to convert your SI to some extreme equivalent version, and then concord that?
@NathanBN - hereâs what I think. Clearly they (just) got some big chunk of data, from 1 (or more) schools. They then have to figure out how to morph that into a nationally representative sample. This is actually quite tricky. If itâs one (big) school, the easiest / most obvious thing would be to guess that this school would have the same number of NMSF as last year (call that n), and then predict a cutoff of the nth highest score.
If they have similarly detailed data from a previous year, though, they can do more complicated things. Maybe last year the cutoff was 1.2 sdevs above the average (for that school). They could do a similar calculation on the new data.
Maybe the data they had from a previous year was more limited, like the top 50 scores from the school. You could still make guesses, but it would be quite complicated, and I would imagine different techniques would yield different results.
Regardless, there are lots of little details in all of these techniques. I think they have either been changing their mind / vacillating between a detail, or maybe they think something they did is possibly wrong, or maybe theyâre questioning if their data is complete⊠Or it could be as simple as they got (multiple) complete schools, and they predict significantly different cutoffs (because their performances changed significantly since last year).
The major amount of backtracking theyâre doing (weâll release a new table tonight! no, tomorrow night! no, forget a whole table, weâre just going to update 1 state, and not until this weekend) implies to me that theyâre conflicted in the one state they have data for. They probably get a high number one way, and a low number a different way. So they donât even know what to say for (presumably TX), much less how to extrapolate that to other states.
@thshadow I really appreciate your effort to reconcile the two vastly different versions given by testmasters. From a large set of data, it is virtually impossible to reach two mutually exclusive conclusions (like 217 to 219 at the same time with 205 being 99%).
Letâs look at the data from Walton (TX), with the average of top 100 SIs being 216.6. If the cutoff is 212 or 214, how many SF would that generate?
I am concerned because I need low cutoff. I hope you are right.
I might argue that a lot of these discussions is more about our own personalities than anything else. Presumably most of us have kids somewhere in the (giant) bubble. (I do.) Some of us want to predict low cutoffs, presumably because weâll be happy in the short term (though possibly disappointed when the real cutoff is released). Some of us want to predict high cutoffs - because by setting low expectations, you can only be pleasantly surprisedâŠ
Yeah, yeah, we all want to find the âtruthâ - but there isnât really any truth to be found as of now. Itâs all just a lot of guessing by people who have never been faced with this exact problem before (yes, that includes the âexpertsâ like testmasters).
WRT your question about Walton (which is from the Cobb press release, so itâs GA, not TX). Yes, that data implied high cutoffs. I think my simulations showed (guessed) that even keeping the same 218 cutoff that GA had last year would increase the # of NMSF from Walton et al.
But keep in mind that Cobb county issued a friggin press release! How weird is that?? I couldnât find any evidence of them issuing a press release about PSAT scores ever before. So you know they were super-stoked by their results. Now maybe they were wrong to be excited because all scores are inflated. Or maybe for whatever reasons (e.g. they prepped their students specifically for the new psat in classes) their kids did do way better than theyâve done before.
Also note that there are other anecdotes that go in the opposite direction (though somewhat rare). e.g. @mphill1tx posted:
If the TX cutoff stayed at 220, they would have less than 5, instead of â10 - 12â. From his anecdote, I guessed a 213 or 214 TX cutoff would yield 10-12 from his school.
Regardless, (A), Iâm not typing in another set of concordance tables :-), and (B) what I did gives somewhat of a middle ground between percentile tables and concordance tables, while still allowing my daughter to make the cut :-), so Iâm sticking with what I have⊠<:-P
@thshadow âBut keep in mind that Cobb county issued a friggin press release! How weird is that??â
My sentiments exactly. Iâm ready to be corrected by people closer to the scene, but to me it looks like they issued a press release for the first time this year, and did so very quickly, based on data received January 5, i.e seemingly before the students received their results. In fact their press release was out long before many students nationwide ever received their scores.
Iâm ready to be corrected again, but it seems like this was also the first year that Walton prepped for the PSAT with Applerouth.
I asked Testmaster about 205 being 99% percentile and commended still around that SI. This at some point they noted, been bit confusing/contradictory. Also what happens to CA cuttofff now that TX is being projected at 219. (Last year CA was 3 score higher than TX):
Michael says:
February 5, 2016 at 1:55 pm
Hi, it is correct that 50,000 students will be Commended, which usually corresponds to the 97th percentile. The 50,000 students is what drives the commended cutoff, not the percentile itself (the percentile is the dependent variable). Itâs entirely possible that California will have a cutoff higher than a 219, or it might even be that Texas is finally catching up to the Golden State!
