Barging in and speaking for @micgeaux, I assume she used the SI table and saw the first 99+ at 214, which we are taking as the 99.5 percentile. This leaves 0.5% (0.005) of the scores and thus 1.6M juniors * 0.005 = 8000.
Sorry for the intrusion
You would think that with all the publicity claiming that CB deliberately published inflated percentiles, CB would have responded already. I think the conspiracy theory is probably just that and page 11 really is it.
If the SI percentiles are to be believed and the total # of eligible test takers hasnât changed that means the commended cut off will be at about 200 and state cut-offs will begin between that and 205.
Yes @ Dave_N, but even if the tables arenât exact, they have to be within some approximate range. For example, say the 99th percentile is really 25,000 students instead of 16-17,000, if you have a few states with cutoffs below 205, you donât have any explaining to do. Regular people will just think âthe test was easierâ; cutoff scores were higher for my state this year. If your lowest state cutoff score is higher than 205, regular people will say, âHuh???â. If this happens, try explaining to GCs and parents how this all works. Because while only a small group is affected by the top 1%, if people can easily determined that the published table is off, then they would question the entire table, right?
Because if there are cutoffs below 205, no one will know if 25,000 students are in the top 1% or > 205.
Testmasters is predicting 210 for commended
@Dave_N - You have hit on the nail with your post #793, with numerical representation of what I believe to be a very likely scenario. We are in a figurative mudhole. You obviously canât borrow from another stateâs unused quota, because that would defy NMSCâs geographical equity rule. I am figuring at the state level, they are really going to struggle to figure out SI level what constitutes a semi-finalist. My gut says this will be more of a problem for middle (aka average) state like Ohio which falls in the middle for a SI cutoff of 215 in 2015, given a national SI range of 202-225, than for a high scoring states like CA, NJ, DC (yes I know itâs not a state) etc.
By the way, I am NOT trying to say that there are significantly more students in the 205 and higher range than 1%. I really think these numbers are on target. If you perform a sample test, and then use the sample test results to extrapolate the results of the actual test, I think you then have to go back and make sure the sample test represented your actual.
If the SI table is correct, then the concordance tables are not. They disagree with one another by up to 4 percentage points, which is huge when weâre trying to discern the top percent or half a percent. The concordance tables are marked âpreliminaryâ and the SI table is not. However, using the concordance tables produce results that yield similar numbers of NMSF candidates when compared to prior years, which is a strong indication that the concordance tables are more accurate.
Take testmasters numbers for instance. They are predicting commended at 210 based on the concordance tables. The SI table shows 210 being in the middle of the 99 percentiles - well above the 97 rank that is commended. The two tables are noticeably inconsistent. Eventually we will find out which is correct.
Pertaining to @Dave_N #793 - The same situation happens in every single state. Thereâs always a cutoff that splits the total number of NMSF allocated (unless luck has it that the number falls exactly on a boundary). The difference is that in states with small populations, the number of slots you would âneed to borrowâ might be in single digits, or perhaps 10s of slots. In large population states it could be hundreds. One of the hazards of compressing the high end and reducing the granularity of the table is that larger numbers of students fall into each slot, increasing the number of slots you âneed to borrowâ.
Since a state canât lend its quota to another, the âneed to borrowâ can only mean either a state under-fills its quota or NMSC makes an exception and allows a larger number of SFâs to go through. Remember we are talking about cutoffs to make SF and not the commended. As you suggest, larger states may have 10s if not 100s of students lurking in the lowest of SI cutoff this year.
Could the states go over their allocations when announcing semi-finalists and then cut them down for finalist?
That is the million $$ question. We know 90% of SF become finalist. Perhap this year with the fat tail, that ratio is much smaller.
Another musing: In post #705 @Plotinus stated âThe reason is that I think the new test is easier for strong students and harder for weak students.â This is likely due to changing from an aptitude test to an achievement test. This could lead to something with more of a barbell-looking distribution, with clusters of scores up high and down low. If this is correct, then perhaps when folks like Jed Applerouth get reports of unusually good performance, that might be due to his connection to high achieving schools. The total number of high scores might not have gone up, but it might be more concentrated in certain types of schools.
I donât know that it is guaranteed that states canât lend their quota. I think that it has been speculated that some mid-size and large states have gone up and down in the past by one point, essentially âtaking turnsâ for which state gets the remainder. That is, taking turns for either having the cutoff value that would put them over or under their population-based allocation. I donât think thereâs been proof, but there has been speculation, and clearly the number canât come out to exactly 16,000.
@DoyleB I agree that the new test is more of an achievement test than was the old test, and hence that students who go to schools that teaches Common Core very well are going to have a big advantage, and students who go to schools that teach Common Core badly or not at all are going to have a big disadvantage. I had students of both kinds do the official practice test, and the difference was very noticeable.
Iâm only familiar with the math portion myself, but if a student has a strong intuitive grasp of basic algebraic material, he will do extremely well on this and find it very easy. If heâs relied on his calculator and a more limited understanding, he will find it challenging. The math more strongly tests concepts behind the math than the old test did.
How many of the math questions were gridded responses? I came across a student recently who didnât answer any of them, yet still came away with a reasonably high percentile ranking, without a stellar math profile overall.
I agree with the analysis in #816. Iâm also going to challenge the assertion that the reading section was easier just because the obscure vocabulary has been eliminated. A college board rep spoke to parents and juniors at my sonâs school last September. She seemed to think that the reading section would be more difficult because it was designed to require âclose reading.â She specifically stated that the âskimming techniquesâ taught in many prep courses were less likely to be successful and we should discourage that approach. I think that a number of very bright kids did not do as well as they usually do in the reading because they didnât read carefully (just like the math kids who went too fast and made sloppy errors). Also, the score reduction per question missed seems very similar between Reading and Math. On the other hand, writing does appear to have been very easy as demonstrated by the much bigger penalty for a missed question.
@higheredrocks
I have students who have done both tests and some have scored higher on the new test, and some have scored lower.
My own subjective experience with the official practice material is that the passages on the new test are a little harder, but the questions are easier with fewer traps. On the old test, I had to be much more careful about traps, and once in a while I would fall into one. On the new test, I donât think I could get any wrong (or at least, it would be much harder to get any wrong).
During the old tests, I am comfortable but cautious. Most of the passages are really uninteresting.
During the new tests, I am hunting for the challenges that are not there. Where are the level 5 questions? I also find the passages more conceptually substantial and interesting.
My conclusion is that readers strong enough to handle the passages are going to ace the new test. Their scores are going to be really, really high. Students who are not great readers are going to have more trouble understanding the passages and will do the same as or worse than on the old test, depending upon how much trouble they have.
Some people claim that the reading on the real test was harder than the reading on the official practice tests. I have not seen any real tests yet.
My feeling is that the percentiles are accurate rather than the concordance table. I would expect accurate cutoffs to be at the high ends of the TestMasters Sliding Scale. Here are some examples:
Massachusetts 214
Maryland 212
New Jersey 213
New York 211
Texas 210
These percentiles correspond with the traditional percentile cutoffs in these respective high scoring states and thus provide the most useful standard for estimating the thresholds come fall. There is no way commended can be 210 nationally when TestMasters estimates 211ish cutoffs in their more rigorous estimation. The statistics continue to point to a lower total.
Many of my friends received 2300+ on old SAT yet struggled mightily on this new format so I wouldnât read too much into the supposed astronomical number of high schorers this year.