Join 3,494 readers in helping fund MetaFilter (Hide)


When All of the Flashcard Manufacturers Declared Bankruptcy
March 5, 2014 1:35 PM   Subscribe

The College Board announced today that the SAT will be undergoing major changes. The announced changes include the removal of the penalty for incorrect guesses, the essay section becoming optional, and a revision of the vocabulary section.

The changes will go into effect two years from now, in the spring of 2016.
posted by DRoll (71 comments total) 13 users marked this as a favorite

 
Another major change is that every student will receive a coupon for a free ice cream cone after completing the exam.
posted by perhapses at 1:38 PM on March 5 [19 favorites]


All students will also now receive a Participant ribbon at the conclusion of the exam.
posted by Thorzdad at 1:44 PM on March 5 [9 favorites]



What's the damn point then?

I'm of the opinion that today's college degree is the equivalent of 1950's High School Diploma.
posted by Ruthless Bunny at 1:46 PM on March 5 [9 favorites]


Ruthless Bunny: I'm of the opinion that today's college degree is the equivalent of 1950's High School Diploma.

Except it costs tens of thousands of dollars at a minimum! Fun!
posted by Rock Steady at 1:46 PM on March 5 [26 favorites]


What's the damn point then?

I'm pretty sure the test still exists, you know. As does examination of grades and "resume" for applicants.
posted by kagredon at 1:47 PM on March 5


Besides which, overreliance on standardized testing has been the impetus for many of the educational changes that people regard as "dumbing down" when they want to hand-wring over education.
posted by kagredon at 1:48 PM on March 5 [1 favorite]


David Coleman, president of the College Board, criticized his own test, the SAT, and its main rival, the ACT, saying that both “have become disconnected from the work of our high schools.”
Pointless trivia in bubble form and block-checking in place of critical thought? Seems like it's more connected to the work of our high schools than ever before.
posted by Etrigan at 1:48 PM on March 5 [18 favorites]


Seriously though...I had no idea they dinged you for an incorrect answer. I just assumed, y'know, you just didn't get any points for a wrong answer, not that they actually took points away.
posted by Thorzdad at 1:50 PM on March 5 [2 favorites]


“It is time for the College Board to say in a clearer voice that the culture and practice of costly test preparation that has arisen around admissions exams drives the perception of inequality and injustice in our country,” he said in a speech Wednesday in which he announced the changes. “It may not be our fault, but it is our problem.”

I think that this might be the damn point. They're trying to work on a broken system.
posted by Gygesringtone at 1:52 PM on March 5 [18 favorites]


I mean, look at this:

The changes coming to the exam are extensive: The SAT’s rarefied vocabulary words will be replaced by words that are common in college courses, such as “empirical” and “synthesis.”

Nor, he acknowledged, do they inspire much respect from classroom teachers: only 20 percent, he said, see the college-admissions tests as a fair measure of the work their students have done in school.

Sometimes, students will be asked not just to select the right answer, but to justify it by choosing the quote from a text that provides the best supporting evidence for their answer.


Those don't exactly sound like bad changes if the goal is to assess if students are college-ready.
posted by kagredon at 1:53 PM on March 5 [9 favorites]


Seriously though...I had no idea they dinged you for an incorrect answer. I just assumed, y'know, you just didn't get any points for a wrong answer, not that they actually took points away.

They drill that into you pretty hard in the preparation books and classes. The idea is to make it a losing proposal to guess on answers you don't know, so some of the test-takers won't luck into a high score.
posted by Holy Zarquon's Singing Fish at 1:53 PM on March 5


Finally! Now the SAT will actually be a relevant and accurate measure of a person's intelligence and ability to perform well in college.

Oh wait.
posted by Lutoslawski at 1:54 PM on March 5 [2 favorites]


I only got a 900 on my 1980s SAT because I answered (guessed) on every question I couldn't get to. I would love to see the getting dinged for wrong answers thing go away, but then 25% of my positive grade would have been bullshit guesses.
posted by mathowie at 1:54 PM on March 5 [1 favorite]


What's the damn point then?

Well, let's take the three changes.

Making the essay optional just moves the SAT closer to what it was in 2004.

Eliminating the penalty for incorrect answers just changes how they crunch the numbers to arrive at the scores and is being made because they thought test-takers were too reluctant to make even educated guesses.

Changes in the vocabulary that better reflect actual language used in college classrooms instead of, for lack of a better word, trivia words seems a decent enough idea. Except that words like "synthetic" will mean very different things in different classrooms.
posted by ROU_Xenophobe at 1:54 PM on March 5 [5 favorites]


These changes seem reasonable. Don't really understand the snark. The SAT is used as a predictor for college preparedness and it sucks at it. Let's try to make it better.

I took the SATs when the essays were optional (called SAT II Writing) and they were dumb back then. "Write about a time you got an unexpected gift." Sounds like they're still dumb and they're going to try to make them better.
posted by muddgirl at 1:54 PM on March 5 [1 favorite]


Based on the article the changes don't seem too bad. The vocabulary stuff in particular seemed a bit crazy in parts - even well-read kids spent ages studying the SAT words specifically because some were so obscure. I used to teach SAT/LAST/GRE prep and see kids raise their scores a lot after a prep course, which made me feel good but sort of defeated the purpose. For a test that really measures what you are supposed to have learned in HS (plus aptitude), excessive prep classes and test study sort of game the system I think. I took the SAT after doing like 2 practice tests they sent me in the mail - it didn't occur to me to try to get a higher/perfect score by taking a class. Plus my parents could never have afforded it anyway.
posted by freecellwizard at 1:55 PM on March 5 [4 favorites]


drives the perception of inequality and injustice in our country

Oh it's not a perception, sport. I used to work in test prep, and even when a poor school bought a discounted test prep package from us for their students, the students were coming from so much further behind and were so much less prepared to do the work of the test prep class that it didn't matter much. That doesn't mean it was worthless, of course: one of my students went from a score in the 700s to one in the 1100s, which made the difference between getting in somewhere and not getting in anywhere. But this testing system is the shit icing on a manure cake.
posted by 1adam12 at 1:55 PM on March 5 [2 favorites]


My daughter pointed out that the penalty for wrong answers actually reinforced the option of not trying at all.
posted by Old'n'Busted at 1:56 PM on March 5 [2 favorites]


Seriously though...I had no idea they dinged you for an incorrect answer. I just assumed, y'know, you just didn't get any points for a wrong answer, not that they actually took points away.

This makes a huge difference in test-taking strategy - the very nature of the test has changed from "do you know the answer to this question?" to "which of these five (but really more like two or three, once you've ruled out the obviously wrong ones) answers seems most likely?" I'll bet a large number of SAT takers didn't realize this either, or didn't understand its implications, which means that students that can afford the time and money for prep books and courses now have less of an advantage.
posted by theodolite at 1:56 PM on March 5 [1 favorite]


They ding you for a wrong answer so randomly guess when you have no clue doesn't help your score.

If you randomly guess with n options, you should be dinged 1/(n-1) for a failed answer. That way, after n guesses, you should get 1 answer right and your score stays at 0. But if you can eliminate just 1 answer as obviously wrong, then guessing should improve your score. That makes sense and is how multiple choice tests should work damnit.

Grr.
posted by aspo at 1:57 PM on March 5 [7 favorites]


Yeah, if anyone should be mad, it should be those of us who took the SAT in the last decade and will continue to be stared at blankly whenever we mention the writing section or give our score out of 2400.

Also though most of us are more mad about the immense student debt and lack of jobs though, because no one cares about the SAT after you get into college.
posted by kagredon at 1:57 PM on March 5 [4 favorites]


I think this is a direct reaction to many colleges and universities having given up on the usefulness of the SAT and no longer using it in admissions evaluation.
posted by aught at 1:59 PM on March 5 [2 favorites]


The essay section in particular is a bad joke, from the lazy reliance on the five-paragraph essay to the Taylorized and sometimes outright rigged grading system (Previously on Metafilter).

If you ever want to become especially cynical about standardized testing, teach test prep for a while. Pretty much anything that can be standardized as a measure of aptitude can be reduced to a problem of game theory and from there further reduced to an end run around acquiring general competency in the disciplinary aptitudes supposedly being tested. Decontextualized knowledge unsurprisingly turns out to be no kind of knowledge at all.
posted by kewb at 2:03 PM on March 5 [4 favorites]


That makes sense and is how multiple choice tests should work damnit.

It's clearly how they intended it to work, but it looks like ETS was worried that too many students were reluctant to guess even from two alternatives because they didn't want to risk the penalty.

They can do functionally the same thing on the back end by just changing the transform from raw answers to the scores.
posted by ROU_Xenophobe at 2:03 PM on March 5 [1 favorite]


So much of succeeding in higher education has absolutely nothing to do with academic achievement.

If we really want to consider a student's ability to thrive in college, we should assign college students to random apartments for a month, give them a stipend of $400 and no cleaning supplies, and force them to create a schedule of classes.

At the end of the month, students must hand in a 10-page paper, take two exams, undergo a psychological screening, and have their apartment hosed down with bleach.
posted by evidenceofabsence at 2:03 PM on March 5 [14 favorites]


a problem of game theory and from there further reduced to an end run around acquiring general competency in the disciplines supposedly being tested

Sounds like a firm basis for a career in middle management, then.
posted by RobotVoodooPower at 2:07 PM on March 5 [3 favorites]


I think this is a direct reaction to many colleges and universities having given up on the usefulness of the SAT and no longer using it in admissions evaluation.

If only we could get everyone to stop asking for your scores.
posted by longdaysjourney at 2:09 PM on March 5


I've had two kids take the SAT over the last 4 years. We've yet to find a college that really gives a damn about the new writing section. It was not considered in admissions anywhere. So I'm not surprised they are dropping it.
posted by COD at 2:10 PM on March 5 [2 favorites]


When I was taking standardized tests in grade school in the seventies, one of the five answers you could pick was 'DK,' which stood for 'Don't Know.' So if you were guessin', they'd narrowed it down to four for you. I like those odds!
posted by Sing Or Swim at 2:12 PM on March 5 [1 favorite]


The College Board has also entered into a partnership with Khan Academy offering free online test prep-- Khan will exclusively have the actual SAT questions. It will be interesting to see how this impacts the for-profit test prep industry.
posted by turbomellow at 2:13 PM on March 5 [5 favorites]


Kahn has had that for years, at least for math. It's how both my kids studied for the SAT. Kahn went through a sample test and showed how to do every type of math problem you would encounter on the SAT. Between that and the $15 book you can buy with like 10 old tests in it both of my kids did really well, without us spending more than whatever that one book cost.
posted by COD at 2:37 PM on March 5 [1 favorite]


They can do functionally the same thing on the back end by just changing the transform from raw answers to the scores.

I'm pretty sure that's not possible. If there's no penalty for wild guessing you can't do functionally the same thing. You have to be able to treat a wild guess and a "left blank" as the same thing. Since left blank has a 0% chance of being right, and a wild guess has 1/n, the a wild guess has to be worth -1/nth of a right answer. (Unless you give points for a blank answer, which is the exact same thing as penalizing a wrong answer, just with a different origin.)
posted by aspo at 2:39 PM on March 5


There is good social science behind the removal of the guessing penalty. When you leave people to their own devices about whether to guess or not, they do so at different rates, and in ways that break down along various demographic lines (particularly gender). If you encourage guessing, it is easy enough to correct for that in the scoring ex post and scores are more comparable person-to-person.

This is basic classical test theory: a latent ability score predicts each question's success, your test instrument is a larger-n sample of ability. A simple count of number right is inflated relative to the true number right, but is still a valid measure of the latent ability. You're not getting away from randomness whatever you do (the test is a sample) so it's good to avoid introducing more sources of bias.
posted by shadow vector at 2:47 PM on March 5 [17 favorites]


B.
posted by Fizz at 2:51 PM on March 5 [1 favorite]


the perception of inequality and injustice in our country

I chuckled at this complete denial of reality. They won't solve their problems until they are honest about what is actually going on.
posted by srboisvert at 2:52 PM on March 5 [2 favorites]


The idea is to make it a losing proposal to guess on answers you don't know

Unless that bit has changed since I took it in the late 1980s, the idea was to make it exactly a break-even proposition, statistically, to guess on a question you had no idea on as to leave the question blank.

It was pretty well understood when I took it that if you could eliminate even one of the choices, guessing among the remaining choices was a net positive.
posted by DevilsAdvocate at 2:53 PM on March 5


My daughter pointed out that the penalty for wrong answers actually reinforced the option of not trying at all.

Well, yeah — that was the whole point. It was to keep you from guessing. Either you know it, or you admit you don't know it and leave the question blank; you don't guess. (Unless you could eliminate two of the answers, then you guess!)

I think the reasoning was that discouraging guessing made the test more "reliable", because someone can't achieve a better score by blind luck. Although there are probably other ways you could fix that, without the wrong-answer penalty.

Off the top of my head, you could have no wrong-answer penalty, but (assuming you get 5 choices for each question, and each question has only one correct answer) you just curve the grades so that getting 20% of the questions right is a "zero" in terms of the final score.

Although at least when I was in school, many colleges seemed to ignore the actual SAT "score" and focus on the percentile. (The exception seemed to be big state schools who had some sort of minimum score requirement to keep people who were going to put pencils up their nose out.) There was a lot of discussion among students as to whether you could somehow improve your score by mass-poisoning the cafeteria meat in a few large highschools the year before on test day, since percentile scores were based off of the last year's test group.

It seems like maybe there's more focus on the actual score today though, from the people I know who have highschool-aged kids.
posted by Kadin2048 at 2:56 PM on March 5 [1 favorite]


The best predictor of future university performance has to be previous university performance. Everyone should be required to take the same first year at any community college, with a core set of courses that could be compared to everyone else's. Baccalaureate-granting universities could then look at what you made of that whole first year in terms of core courses, extra courses, professor evaluations, class projects, and extracurricular activities.
posted by pracowity at 2:58 PM on March 5 [2 favorites]


I don't get why they can't add a disclaimer to the top of the test:
Incorrect responses will reduce your score more than skipped responses. If you can eliminate two incorrect choices, it may be to your advantage to take an educated guess.
I'd imagine that most college-bound high-schoolers could parse the above sentence.
posted by schmod at 3:22 PM on March 5 [2 favorites]


That's a really good idea, pracowity. I think a lot of universities are in the business of overenrolling without regard to their students' academic well-being, but if higher learning really were more about educating young minds and less about making a buck, that would be an excellent way to go about preparing and assessing applicants.
posted by Ice Cream Socialist at 3:23 PM on March 5


aught: "I think this is a direct reaction to many colleges and universities having given up on the usefulness of the SAT and no longer using it in admissions evaluation."

That was why they added the essay section in the first place. (Where "added" means "folding the SAT II Writing into the SAT".)

I want to say (though I could be completely misremembering) that the UC system threatened to drop the standardised testing requirement, which would have been a massive blow to the SAT racket. (But I also vaguely remember the UCs requiring the writing test, so who knows.)
posted by hoyland at 4:00 PM on March 5


Yeah, we're really rolling in bucks over here.
posted by aaronetc at 4:00 PM on March 5 [1 favorite]


Pracowity, it seems to me like you're describing the entire K-12 system. There is no reason a "13th grade" at community college would indicate anything different than what a high-schooler's senior year already shows.

The writing section was somewhat foolish. There can't be a "perfect" essay, and so an 800 on that section doesn't make a lot of sense. There's too much subjectivity involved.

Removing the penalty for incorrect guesses is something I take a little more issue with. The fact that too many kids were not making educated guesses is just another indicator of the students' lack of ability to come to conclusions based on the relatively simple mathematics involved. There are similar warning to the one schmod mentioned above. I never took an SAT prep class, but I was well aware of that particular rule and its implications.
posted by papayaninja at 4:02 PM on March 5


I'm pretty sure that's not possible. If there's no penalty for wild guessing you can't do functionally the same thing. You have to be able to treat a wild guess and a "left blank" as the same thing.

In a rough and informal way, you can accomplish the same goal by encouraging everyone to answer every question, even if it's a guess, and then requiring a higher number of correct answers to get scaled to a 500. Doing it for real probably means something that falls out of a big IRT model of the sort that shadow vector mentioned.
posted by ROU_Xenophobe at 4:05 PM on March 5


I don't get why they can't add a disclaimer to the top of the test:

They did. ETS seems to have data suggesting that it wasn't enough, and there was some sort of detectable bias in who was guessing how much.

Removing the penalty for incorrect guesses is something I take a little more issue with. The fact that too many kids were not making educated guesses is just another indicator of the students' lack of ability to come to conclusions based on the relatively simple mathematics involved.

But the verbal SAT isn't testing that, and even the quantitative section is intended to assess general mathematical skill rather than skill at dealing with probability.

I think the reasoning was that discouraging guessing made the test more "reliable", because someone can't achieve a better score by blind luck.

You could absolutely achieve a higher score by blind luck. It's just unlikely.
posted by ROU_Xenophobe at 4:14 PM on March 5 [1 favorite]


The SAT is used as a predictor for college preparedness and it sucks at it. Let's try to make it better.

Alternatively, because it's a bad predictor it should be abandoned. Don't try to spruce up the condemned building with a little paint, burn it down and try to collect the insurance money.
posted by justsomebodythatyouusedtoknow at 4:20 PM on March 5 [1 favorite]


The thing I've always been curious about is the fact that the expected value for guessing is 0. Clearly you should save your pencil lead rather than guess on large numbers of questions. But what does guessing on a small number of questions look like?

I suppose I can just make R do this for me and resolve the things I wondered about in high school. Back in a minute...
posted by hoyland at 4:22 PM on March 5


The thing that would screw me, if I ever take the GRE again is, you do it one question at a time, on a computer now (right?) and can't move around, answering things you know cold, assessing how much time remains, and then doing a second and even a third pass. This is how I learned to take tests back in the rotary phone days. So, I'd need to do a lot of practice tests, I think, to get acclimated.
posted by thelonius at 4:42 PM on March 5 [2 favorites]


>>>They can do functionally the same thing on the back end by just changing the transform from raw answers to the scores.

>>I'm pretty sure that's not possible. If there's no penalty for wild guessing you can't do functionally the same thing.

>There is good social science behind the removal of the guessing penalty.


A bit like graduated income taxes, the math behind the various scoring schemes for these standardized tests, and how the scores are adjusted and renormalized and so on, is pretty much dead easy and simple enough to explain to anyone who's successfully completed, say, Algebra 1.

But. Like the graduated income tax, relatively few really understand it clearly1 and even among those who do, nearly everyone will respond to it on an emotional level in some way. Just for example, A PENALTY for guessing naturally makes one cautious about guessing whereas NO PENALTY for guessing makes one prone to guess each and every time because why not. But I'll wager that our thinking about it is far more influenced by that emotionally charged word "penalty" than the actual math behind how the scores are calculated.

Result is that removing the penalty for guessing is bound to increase the amount of guessing a lot, and that is bound to increase scores by a noticeable amount--because if you even vaguely understand a question and the available choices you're going to be able to guess the correct answer a fair bit better than random chance.

So--why not? And double why-not when you figure that different personalities and sub-groups are going to have a different emotional response and risk assessment of the "penalty" which most likely ends up influencing their score. We're not trying to measure emotional response and risk assessment of the scoring system here.

1As amply illustrated by comments above.
posted by flug at 4:48 PM on March 5


I'm so glad they removed the "essay" question, which was apparently graded in ways that were worse than whimsical. As an English teacher, I prefer that the supporting evidence be true and not made up out of thin air, and I prefer actual content to hot air and vague stabs at structure.
posted by Peach at 4:58 PM on March 5


Pretty much anything that can be standardized as a measure of aptitude can be reduced to a problem of game theory

I think the idea is that the ability to figure this out probably correlates with ability to succeed in college and perhaps intelligence. The system works!
posted by save alive nothing that breatheth at 5:11 PM on March 5


I want to say (though I could be completely misremembering) that the UC system threatened to drop the standardised testing requirement, which would have been a massive blow to the SAT racket. (But I also vaguely remember the UCs requiring the writing test, so who knows.)

Not sure about the SATs, but the UCs dropped the SAT II (subject test SAT) requirement in 2006 or 2007, IIRC. Which was a good move: it was essentially an extra standardized test that was duplicative of the AP exams most UC-bound students were already taking, and cost money a lot of students' families didn't have. I know I took my SAT IIs in the subjects I had just taken my AP exams in, to kill multiple birds with one stone. Though getting a certain score on the math SAT IIs might still count towards testing out of a required course at the UC.
posted by yasaman at 5:35 PM on March 5 [1 favorite]


The thing I've always been curious about is the fact that the expected value for guessing is 0. Clearly you should save your pencil lead rather than guess on large numbers of questions. But what does guessing on a small number of questions look like?

Well, since it is a small number of questions the maximum possible effect on your score would be small. And since the whole scheme is designed to have no net effect on the score, the most probably results will all be clustered right around zero. So "small and most likely close to zero" are the two redundant concepts that you really need to understand.

But just for laughs let's analyze, the possible results of guessing on 1 question. So it has five choices. You get +1 for a correct answer and -.25 for an incorrect answer. This gives you: The interesting thing here is that the upside is larger, though less probable, than the downside.

Now how about guessing on two questions (again, same setup--5 choices for each, etc). Here there are more possibilities: Again the potential upside is quite a bit larger, but far less probable, than the potential downside.

I won't bore everyone by taking you through the various cases--3, 4, 5 questions and so on--but the pattern is clear enough: The upside is always larger but less probable.

One conclusion: If you're a gambling sort, you're always going to guess. Why not? The downside is likely but relatively small. The upside is large and though it may be improbable--but why not give it a whirl? Getting it would be like winning the lottery . . . and even if you miss, you're only out a few points at most.

However: I haven't factored in the time spent guessing on many questions rather than spending time answering one or two more accurately. Almost certainly, spending time getting an answer or two actually correct is a better use of your time.
posted by flug at 5:40 PM on March 5


Not sure about the SATs, but the UCs dropped the SAT II (subject test SAT) requirement in 2006 or 2007, IIRC.

Yeah, that was after the new SAT with the writing rolled in came in. I think I needed the SAT II writing but no others for UCs. (I applied 2003-2004, which was right as the SAT changed. I took the old one (so 2002-2003?), but I want to say people the year behind me took the new SAT.)
posted by hoyland at 5:41 PM on March 5


"I'm of the opinion that today's college degree is the equivalent of 1950's High School Diploma."

My college degree taught me how to read articles, and how to not misplace apostrophes.
posted by klangklangston at 5:52 PM on March 5


The idea of the essay questions always seemed farcical to me, so I'm glad to see them go. The change to less esoteric vocabulary seems good also.

If the impact of removing the guessing penalty is to lesson some kind of bias (such as the value of high-dollar coaching that gives richer students higher scores because they better understand the strategy), then that's positive also. I'll be curious how that one works out in the real world, though.
posted by Dip Flash at 6:02 PM on March 5


I won't bore everyone by taking you through the various cases--3, 4, 5 questions and so on--but the pattern is clear enough: The upside is always larger but less probable.

The probability of a nonnegative outcome depends on the number of questions, though. For example, if you guess on four questions, there's a 41% chance of getting all four wrong, but that's the only outcome where you lose points.
posted by hoyland at 6:14 PM on March 5


Point being there are numbers of questions where the probability of a nonnegative outcome is greater than half.
posted by hoyland at 6:16 PM on March 5


The idea is to make it a losing proposal to guess on answers you don't know

Keep in mind that the guessing 'penalty' isn't really a penalty at all but is cleverly designed so that the net effect of random guessing is zero.

It's not really that you lose points by large scale random guessing--just that you're not going to gain points, either.

So a person who guesses on questions they didn't have time to complete will get some plus points for the correct guesses and some minus points for the incorrect guesses. Almost certainly the sum of the plus and minus points for guessing will total up to zero or very close to it.

In short, you don't gain anything by random guessing but you don't lose anything, either. For large-scale guessers, the final result will be the same (or very, very close to it) whether they random guess or just leave them all completely blank.
posted by flug at 6:43 PM on March 5


The probability of a nonnegative outcome depends on the number of questions, though. For example, if you guess on four questions, there's a 41% chance of getting all four wrong, but that's the only outcome where you lose points.

Point being there are numbers of questions where the probability of a nonnegative outcome is greater than half.


You're quite right. Ain't probability wunnerful!

Because that seems like a surefire winner--and why doesn't everyone make a point to randomly guess on just exactly four questions, because that seems to be a case where the odds are stacked in your favor?

Here is what the breakdown looks like for four questions: So you could say, w00t!!!1!!! 59% chance of getting 0.25 points or more! That looks like a surefire winner!!!11!!

But on the flip side, you could say: Equal chances of getting -1 points or 0.25 points. That's obviously a sucker's bet--equal chances of a small gain or a MUCH larger loss.

The only thing that evens up those odds is that you also have a (smaller) chance of getting larger gains as well: 1.5, 2.75, or 4 points.

Any way you dice it, the overall probabilities add up to zero. (Multiply 0.2% X 4 + 2.6%* 2.75 etc and you'll see the grand total is exactly zero.)

The system is deliberately set up so that you can't really game it.

The only thing really notable here is that you do have a slim chance of racking up 2.75 or 4 points, whereas the potential loss is limited to one point. So if you're the gambling sort, you might want the roll the dice on the very slim possibility of a 'big' win when your loss is stopped at a reasonable level. This is true whether you're guessing on 2, 4, 10 or 50 questions . . . larger potential upside as the number of questions grow and smaller chance of hitting it. Which should make this whole scheme more and more appealing to the gambler in all of us as the number of guesses grows, right? Hit the jackpot and GUESS your way into Harvard. Best lottery in town!
posted by flug at 7:29 PM on March 5


Whenever anybody mentions the SATs, all I can think of is this Andy Bernard quote from The Office:

"I’ve always been the guy who can rally other people to rebel. In high school, I organized a walk out over standardized testing. Got over 500 students to just skip the SATs. At the last second I chickened out, took it anyway got a 1220. Always regretted it… I feel lachrymose."
posted by discopolo at 9:01 PM on March 5 [3 favorites]


Does the penalty or lack thereof make much of a difference when the test is standardized, anyway? I mean, apart from cross-year dick-measuring and the bare-minimum cutoffs that were mentioned upthread, but if 99th percentile winds up being at 780 instead of 760 (which could happen anyway from normal fluctuations in difficulty year-to-year), wouldn't that be functionally the same? Or am I missing something?
posted by kagredon at 9:04 PM on March 5


I'm not from the US so I don't have any experience with the system but everyone here seems to agree that the SAT is a poor predictor of college performance.

Is there any actual public data on this? How well do upper-division grades correlate with SAT scores? How does the essay part compare with the multiple-choice part in this regard? Does the College Board measure this in any detailed way, in order to decide which specific questions are most predictive, or to optimize question structure?

I think one fundamental issue, aside from correcting for structural biases, is that being able to prepare well for a test is a necessary but not sufficient skill for university success. At my institution we correct for this by using exam scores as a pre-filter and then basing admissions decisions on rather rigorous in-person interviews. Of course this is expensive, and introduces all sorts of further possible biases, which we try to correct for with mixed success. It is always remarkable to me how well some students can do on a standardized test, and yet be unable to analyze a situation slightly outside their sphere of preparation.

The ability to approach new ideas independently is hard to test on paper, and even harder to teach in a reliable, scalable way. I'm not exactly sure how to improve the way we teach secondary students that type of skill (though I have some ideas), but I am sure that obsessive test preparation interferes with it.
posted by Pre-Taped Call In Show at 5:13 AM on March 6


but everyone here seems to agree that the SAT is a poor predictor of college performance.

If the SAT could somehow factor in the prospective student's susceptibility to binge drinking, it might have a better handle on predicting college performance. (Speaking as someone who crushed the SAT back in the day but had a rough first couple years in college.)
posted by aught at 5:36 AM on March 6 [1 favorite]


Is there any actual public data on this?

There is a huge amount of research on this -- googling "SAT predictive validity" will show you the tip of the iceberg, for example.
posted by Dip Flash at 6:08 AM on March 6 [1 favorite]


Any way you dice it, the overall probabilities add up to zero.

Yes, this is the observation I started with (though I was certainly not clear what I thinking beyond that). I suppose what I'm asking is whether that's the question to ask in terms of impact on your score (rather than your point total). If you break the outcomes of guessing into "noise", "positive" and "negative" in terms of impact on your score, is the expected value "positive" in some cases? Is that the more relevant expected value than how many points you should expect to pick up by guessing?
posted by hoyland at 6:19 AM on March 6


First the Common Core and now the SAT. This Coleman. WTF is he up to. Who the hell is really paying him.
posted by oflinkey at 6:23 AM on March 6


Dip Flash: "googling "SAT predictive validity" "

Yeah, but predictive of what?

These studies imply that we can accurately measure student performance at the University level. A lot of university administrators would probably (anonymously) disagree with you, while quietly adding that they have no clue if they're adding any value to their students (particularly among the "top performers").

It's a big problem.

Oh, and don't even get me started on the GRE, which doesn't seem to be a good predictor of anything -- it's not even good at predicting future GRE scores. At the very least, students who take the SAT multiple times tend to score very similarly across multiple attempts. The GRE fails at even the most basic tests of statistical validity.
posted by schmod at 6:36 AM on March 6


Huh? I'm not saying it is predictive of college success, but rather that that question is heavily researched and is a huge policy issue. Part of why colleges are increasingly making it optional is exactly that lack of strong predictive power.
posted by Dip Flash at 7:13 AM on March 6 [1 favorite]


Oh, and don't even get me started on the GRE, which doesn't seem to be a good predictor of anything

It's a good predictor of your ability to spend $200.

I work for a chain of colleges. No SAT or ACT required. No essay. Just $50 (optional) and a 2.0 in high school.
posted by GrapeApiary at 7:13 AM on March 6


Thanks Dip Flash, that provided several hours of entertainment.
posted by Pre-Taped Call In Show at 4:17 PM on March 6


« Older If the Moon Were Only 1 Pixel...  |  Stephen Glass was a well-known... Newer »


This thread has been archived and is closed to new comments