Maybe ya’ll have figured this out already and I’m late to the party. But if the lowest cutoff score, as predicted by testmasters is 210, and 205 is the lowest 99th percentile report on the understanding your scores, that wouldn’t even be 16,000 kids. The cutoff would end before the 99th percentile ends.
So then we are saying that the actual test results (SIs) don’t concord, at all, to the published test results.
Testmasters should probably take a look at those SI percentiles and update their predictions accordingly. To me, a cut-off range of 205 - 225 seems about right, with most of the highest ones being closer to 220 than 225. All else equal the concordance tables seem to suggest a slight downward adjustment at the highest end (225 becomes 222, 223 becomes 220, etc.) - but that of course could be offset by an upward trend trend in scores that was happening anyway. Commended will be around 200. This is all based on 1.6 million NM-eligible test takers. If the pool is higher than that then we should see an adjustment upward at least at the lower end.
So I predict nothing higher than 225, and nothing lower than 200. This prediction should be quite accurate. 
I would hazard a guess that CB doesn’t provide SI percentile tables based on the current year’s students BECAUSE they provide too much information. You would know the commended number within a point, and could accurately estimate each state’s cutoff, likely within a point as well. That’s something NMSC would prefer to put out in their press releases later on instead.
Haven’t they always published the Selection Indices in the Understanding your Scores?
I feel like I am totally missing something here. IF you perform a sample test and then give an actual test, and then realize that the actual results do NOT concord with your sample, WHY would you calculate every student’s scores on the sample test - total, subject, and SIs? Then issue score reports and publish scores on this data that doesn’t concord to the sample. And maybe they did this prior to giving the actual test, but they also had to calculate concordance tables to prior years.
So, when the actual results come in and nothing concords to the sample, why would you still issue the report? I’m sure it takes a while to process all this data, but you would rather run the risk of printing data that corresponds to a sample that doesn’t concord to the actual, then just scrapping the sample and calculating the actual???
As @PAMom21 and @Plotinus pointed out, in the Understandng Scores document they’ve published previous year data, not current year data (for the last few years they’ve done that, anyway). Indeed, when I compare my state’s cutoffs to the year later SI chart instead of the SI chart published for the cutoff year, the cutoff value is in the same relative position every time.
Yes, I agree. But there must be a fairly high degree of correlation from year to year.
I suspect that the more the years pass, the greater is the correlation from year to year. Wondering whether next year’s SI percentiles will be based just on this years, or on an amalgam of this year and test data.
Hi Y’all. New to this CC forum. It’s fascinating to read through all this NMSC cutoff discussions now that it’s pay-up time for a 2017 college bound kid!
My question is wrt high-expected cut-offs each state’s SI’s. As the tail is expected to be FAT, what happens if there are far more candidates at the lowest SI (let’s just say 215) than are positions available? For OH the number of semi-finalist was 618 in 2015, out of 16227 semi-finalist nationwide. So let’s say the SI cut-off is determined to be 215 and clearly everyone with >= 216 is in, but there are far too many students at 215 SI than slots available for semi-finalist selection. Would CB then invite all of 215 SI students to participate forward in the competition or take in fewer than 618 semi-finalist for those who scored >=216? I anticipate this will become an issue for many states as the tail is expected to be fatter than it’s been traditionally. My hope is that the appx 16000 SF is just that, an approximation.
@OHToCollege, what actually happens is kind of the opposite. If there are 618 NMSF allotted to OH, they will pick the cut-score such that 618 or fewer meet the criteria.
What we don’t know is if we have hit the “nuclear option”, where more candidates for a state achieve a perfect 228 selection index. Based upon the SI percentile tables, this seems exceedingly unlikely, but now there is some doubt about whether the SI percentile table distributed to date is representative or accurate.
@OHToCollege – I have that question too - really wonder how NM deals with the situation bc that could happen. Seems on this thread in the past there might have been a precedent for treating kids with the same SI differently but seems unfair. Once the State summaries are out maybe we can get a better picture of those who scored at the top ranges of scores for the subsections.
@Dave_N - Interesting! If that is true (fixed allotted statewide SF’s), then I expect there will be a lot of disappointed candidates and possibly unfilled slots because the one higher SI doesn’t quite get you to the magic number. I would hope that NMSC chooses to err on the side of more SF’s to go forward as opposed to less. That will allow other academic performance indicators come in to play to determine overall eligibility of finalists.
To keep it fair, CB may need to develop fractional SI’s!
<<what we="" don’t="" know="" is="" if="" have="" hit="" the="" “nuclear=”" option",="" where="" more="" candidates="" for="" a="" state="" achieve="" perfect="" 228="" selection="" index.="">>
My guess is that if there is anything like this even possible we are going to hear about it here. And it’s been 10 days so far w/o a serious indication that a state topping out at 228 will be the case. For sure not everyone - not even the majority - have a clue as to what their SI is because they haven’t accessed their online score report yet - and they won’t know till the paper reports are distributed. But those aren’t going to be kids at risk of scoring 228.
I suppose that one POSSIBLE explanation is that entire schools of kids without access are topping out at 228 - but that’s taking the conspiracy theories a bit too far :)]
A slightly more plausible is a case where we have a high population / high cutoff state. What happens if the test was too easy (for high achievers) and there is clustering at the top?
Hypothetical state, PSATObsessionVille:
“budgeted” NMSF slots: 500
228: 10 490
227: 20 470
226: 70 400
225: 150 250
224: 300 ==> Now what? Do we go over budget and “borrow” from another state?
225: 500
If the presented SI percentile table is accurate, there might be enough 99+ score slots available where this shouldn’t be happening, but still too early to tell.
This helps illustrate why the DC cutoff is so high historically, I think. The proportional number of slots available to DC is small and there are many excellent private schools there.
I doubt very seriously that there are going to be any issues with staying inside the 16,000 and state ratios. It could very well be that is what caused the delay in releasing the scores to students. (CB knows the numbers that NM needs to deal with.)
I don’t think there were too many perfect scores. CB could have prevented this by not handing perfect section scores to kids who got one wrong. I assume they would have done so if they were seeing too many perfect scores. However they are awfully close to the situation where they cannot fix a problem of too many perfect scores.
Y’all are overestimating the test-taking population and the ease of the test. Even on CC there are very few astronomical scores.
@MatzoBall I agree.
If you look at the first understanding your scores after the redesigned test, the cutoffs for each state (based on where they fall in the percentile rankings) has pretty much tracked for the next 10 years.
My prediction is that the page 11 is IT. If you then want to make assumptions about states having extremely high cutoffs, then off to the races.
But, between 214-228, you have 8,000 students. Some state cutoffs have to be below 205, or otherwise you wouldn’t have the the top 1% of test takers, let alone by state.
The average parent will not understand this complicated concordance. Try explaining how the cutoff for the top 1% is less than the top 1% as published.
Problem is there is a lot of anecdotal stories out there about unusually high scores. That, combined with the concordance tables, makes it seem like the published percentiles are off as well. The question is whether CB will publish new percentiles (or better yet concorded percentiles) along with the final concordance tables in May. It looks like for now CB is in a deep hole because a 209 SI is being reported as middle 99% score on user and SI published index. But on concorded 2014 it is high 96/97% and might not have made commended cut off in 2014 and before. In other words, CB is telling someone with a 209SI you are in the 1-16k range of scorers for now as well as the 45-55k range of scorers. When dealing with 1.5 mill population this might be a rounding error but for SF contenders it’s critical. My sense is a 209 SI equates to the 45-55k band not the 1-16k and I do not know how CB will explain this.
I suspect that NMSC uses the international and boarding school selection group as the “plug”. There is no data to determine how these areas get allocated finalists. Unlike states, they likely do not use total number of graduating seniors (they can’t count every senior in a country). My guess is they use some type of proportional allocation for number of NMSF and just set the cutoff (arbitrarily) to whatever the highest is (NJ/DC).
@micgeaux "But, between 214-228, you have 8,000 students. "
How did you figure that out?