Trouble with October 24th PSAT Scoring Curve
The College Board continues to release tests that are too easy and, consequently, have punishing curves. The June SAT was too easy and had a brutal curve, leading to outcries and online petitions. It seems the College Board has released yet another test that ended up being far too easy, with a curve that was equally unforgiving. This time the October 24 PSAT/NMSQT is the culprit. The College Board administered the PSAT to the lion’s share of students across the country on Wednesday, October 10, but roughly 10% of the nearly 1.8 million juniors who took an October PSAT took the October 24th date.
The College Board released the scoring table for the October 10 and October 24th PSATs. Look at how precipitously the October 24 scores fall when students miss only a few questions.
|Raw Score||Scaled Math Oct 10||Scaled Math Oct 24||Difference|
If a student on either date aces the math test, that student will receive a perfect 760; miss one single question, however, and the student on October 10 will have a 750, while the student on October 24 will drop to a 710. Miss a second question and the October 10th tester will have an excellent 740, while the October 24th tester will have dropped to a 670, well out of National Merit range for the majority of students in the country. A third miss will bring the October 10th tester to a 730, while the October 24th tester will be down to a 640. These are jarring score drops on the October 24th test.
The College Board should be crafting tests that do not have this level of extreme variability. The test writers can argue that the tests are fair, and the scaling process, using test equating, keeps every student on a level playing field. But the punishing drops on one test set up an imbalance with real-world consequences. Independent of ability, some students make an occasional careless error. On October 24th, a careless error on math could bring a student to the edge of losing National Merit, and any other error in any other section could close that door permanently. With drops this steep, The College Board is demanding near perfection of National Merit hopefuls, which is a lot to ask.
Comparing the Reading and Writing scores for the two test dates similarly reveal that the October 24th test date was on a much tougher curve all around. For the sake of simplicity, I’ll multiply the scaled scores by 20 to get us to the 760 scale with which we are all familiar.
|Raw Score||Scaled Reading Oct 10||Scaled Reading Oct 24||Difference|
|Raw Score||Scaled Writing Oct 10||Scaled Writing Oct 24||Difference|
It’s evident that the October 24th test was too easy, and the curve subsequently too aggressive. Two students testing on these different days would have very different experiences. Students were not happy with the differential. Check in to any of the various PSAT threads on Reddit, if you are comfortable with profanity, to gauge some of the student responses.
In conjunction with the SAT redesign in 2016, the College Board had shifted away from working with its long-time partner, the Educational Testing Service (ETS), for test creation. It is likely that this shift played a role the College Board’s recent difficulty in producing tests with sufficient numbers of difficult items. The College Board is again working with ETS to create tests that are valid and reliable; however, as we see from this year’s PSAT, the collaboration is not yet producing consistent tests. In time, we would hope and anticipate that these extreme swings will die down, giving students on different testing days a similar testing experience. For National Merit hopefuls who sat for the PSAT on October 24, this is cold comfort. And for the students who will take SAT in the coming months, the College Board’s resolution of this recurrent problem can’t come soon enough.