breaking the SAT

if you don't like the scoreboard? rig the game.

one of the key embarrassments that woke and DEI seek to avoid are the obvious, demonstrable, and durable differences in performance between groups. this movement keeps getting rugpulled by things like “standards” and “test scores.” claims about how racist, sexist, classist, and every other kind of “ist” of which one may conceive the SAT is increasingly look facile and contrived. despite their endless attempts to break these assessment tools, the fact is that most of them are actually quite good and highly fit for purpose so long as your purpose is “accurate measurement.” alas, this seems less and less to be the case and underlying intent appears more focused on muddying waters than clarifying them and when choosing “objective truth” over “my truth” this gang has long since revealed its preference:

where objectivity is the enemy of narrative, those with sufficient commitment to the story they are telling and the ideology into which they were inculcated will, when faced with evidence they do not like, seek to eliminate objective assessment.

don’t like the scoreboard? rig the game.

the SAT has been debased and slanted to prevent such outcomes for ages, but it still remained a basically sound albeit probably too easy and too limited a test. but this next step is the literal end. it removes all pretense of objectivity and becomes an entirely rigged and tuneable structure to prevent information from emerging. what was once a gold standard is about to become pure pyrite.

the general gist is this:

the "digital SAT" will now be broken into sections, 2 for each subject.

if you do poorly on the first part of a section, the second part will "adapt" and present you with easier questions in the second half.

if you do well, it will show you harder questions.

this is a common practice in tuneable tests to cover a broader ranger of ability, but those tests weight scoring more highly for harder questions and score less for easy ones.

what will break the SAT is that the number of easy vs difficult questions will have no impact on your score.

this is nonsense. of course it will skew your score. the literal point of harder questions is that it is more likely you will get them wrong. you are not taking the same test as others will, so your score will no longer be comparable. this is obviously done by design and a deliberate departure with malice aforethought from most sorts of sophisticated adaptive testing weight more difficult questions higher and less challenging questions less in order to cover a wider range of ability while still providing continuous and comparable scores. that sort of structure makes functional sense.

this is designed to do the opposite.

there is fundamentally nothing wrong with adaptive testing that adds harder or easier questions based on performance. it’s actually a great way to extend the range of tests. a set of IQ test questions that can both reliably tell 69 from 70 and 129 from 130 is basically impossible. but an adaptive test can. but, doing so requires one very important difference to this new SAT model: it requires scoring to be weighted by question difficulty. solving a differential equation needs to count more than knowing 4+6.

it’s removing that piece of the adaptive model that breaks it and turns if from a enhancer to a degrader of assessment.

these changes will have a clear effect: it will pull all the middle scores in toward the median and make the test non-comparable across students as your exam might have had many hard questions and mine all easy ones. it will compress the point range dramatically.

we all start the bike race on flat ground.

those who ride quickly have to finish uphill.

those who ride slowly get to finish downhill.

but we mark everyone’s time the same way.

seem like a good measure of ability?

yeah…

the interesting (and likely unexpected irony) here is that this is going to really favor the top students and severely penalize the worst. for some small number of students, all SAT questions are easy. for others, all are difficult. this is going to segment and cluster them with far greater clarity and you’ll get a very different shape than the traditional bell curve.

the middle will compress due to the leveling effect of the adaptive questions, but the far tails will pop and become far more evident because the 20-80 percentile range will be pulled away from them.

essentially, the whole middle will see scores and standard deviations compress, but the ends are not going to be affected much.

for the very smartest, it’s going to make their scores all the more noteworthy. it will be much harder to hit a 1500 or 1600. meanwhile, getting a 600 despite all easy questions will be indelible stigma.

(sidenote: it’s going to be hilarious the manner in which this inadvertently highlights asian outperformance at top percentiles)

those benefiting most will be the 20th to 40th percentiles.

those hurt most will be the 60th to 80th.

a score of 700 becomes 900, but a score of 1300 becomes 1100.

this “adaptive” variance in difficulty is fully tunable. still don’t like the results? slant the field further. it’s a full on fudge factor that will be basically invisible from outside and could easily be applied selectively if one desired to do so.

push this hard enough and the gap between 70th and 30th percentiles could become miniscule and basically a 600 could become a 900 and and 1400 an 1100. this is the injection of pure, customizable results output and it has the potential to make the whole center of the SAT an unreadable mess.

they will call this "fair,” but it's ridiculous.

the purpose of this test is to measure ability.

this test can no longer do so in an objective, comparable fashion.

the purpose of this change is to essentially remove all resolution from the middle of the SAT's bell curve and make 20th to 80th percentiles blur together in one indistinct morass devoid of discernment in which group variation can be masked.

this matters greatly because we are are dumbing college down tragically, mostly because so few folks arriving there are ready to be there.

pretending this is college work instead of middle-school serves no one and adulterating the test that is supposed to weed out those unable to write at a 7th grade level will just make it worse.

and this just keeps spilling forward until you wind up with social justice MD’s who cannot do the work and Diversity Engineering Incompetence that makes airplanes fall from the sky.

you cannot validate credentials by nerfing the tests to get them and rigging the tests for objective reality will not change that reality one whit. it will just put people into positions where they are over their heads, and past such peter principles the world comes unglued pretty darn quick.

so let’s keep a simple truth in mind: