Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

New SAT, New Problems

theatlantic.com

Post reply on HN
Print view View original

31–40 of 45 posts

Re: New SAT, New Problems

#31

Post by IndianAstronaut »

yummyfajitas wrote: post #16
BlackJack wrote: post #3

Here's the College Board's page on the redesigned SAT: https://www.collegeboard.org/delivering-opportunity/sat/rede... Here's a 200 page PDF on the new format with lots of sample questions: https://www.collegeboard.org/pdf/sat/delivering-opportunity/... The sentence completions are gone; now we have words in context. Essay is twice as long. No independent writing section. Math is more algebra and prob/stats focused,…

More coachable is probably the goal. A big problem for the college board is that insofar as the SAT is difficult, math driven, and g loaded, Asian people tend to score too high. This creates political pressure for colleges to stop using the SAT in admissions, an existential threat to the college board.

They have always tweaked exams to meet with new(old for the rest of us but new for college admissions departments) demands. A few years ago they changed the GRE to include more statistics and probability.

Re: New SAT, New Problems

#32

Post by comex »

nilkn wrote: post #29
mynameishere wrote: post #15

Earlier quoted context omitted.

The answers are so ungrammatical and mystifying that it almost seems like the designers were trying to kill the test on purpose. The "correct" answer is this: The predicted height increase in centimeters for one centimeter increase in the first metacarpal bone ...as far as I can tell, the answer written for sentient humans would be something like: As the length of the metacarpal bone increases, so does height. But my…

I think their answer is perfectly sensible and perfectly grammatical. It's also clearly correct according to even the most basic notion of slope as rise over run. I honestly don't see the issue with this question. Not only does your answer not fit grammatically, but it's also less precise.

When I skimmed the list of answers, they also looked wrong to me, but this is because I saw the answers (which start at the top of the page) before the question (which is at the bottom). Because the answers started with capital letters, I initially read them expecting a full sentence, mentally replacing "increase" with "increases":

"The predicted height increases in centimeters for one centimeter increase in the first metacarpal bone." (True, although the height also increases in any other unit for any amount of increase in the bone.)

"The predicted first metacarpal bone increases in centimeters for every centimeter increase in height." (Ditto, but this looks more correct because it has it increasing for every/any centimeter increase in height, rather than specifically one.)

Judging by the parent comment, I think they may have done the same, but failed to notice the correct interpretation? In any case, while as a native speaker, I only saw it that way for a few seconds, I can certainly imagine that someone speaking English as a second language could be thrown off.

The answers also do have odd wording - they refer to the bone itself "increasing", rather than its length, and the first answer's "for one centimeter increase" could be interpreted as either "for (one centimeter) increase", which lacks an article, or "for one (centimeter increase)", which is not ambiguous but not proper terminology either.

I think the design could be improved - first by avoiding the weird inverted question/answer position if at all possible, second by avoiding capital letters, third by using easier to parse wording:

"the predicted increase in height, in centimeters, for every one centimeter increase in the length of the first metacarpal bone"

Re: New SAT, New Problems

#33

Post by bsder »

BlackJack wrote: post #3

Here's the College Board's page on the redesigned SAT: https://www.collegeboard.org/delivering-opportunity/sat/rede... Here's a 200 page PDF on the new format with lots of sample questions: https://www.collegeboard.org/pdf/sat/delivering-opportunity/... The sentence completions are gone; now we have words in context. Essay is twice as long. No independent writing section. Math is more algebra and prob/stats focused,…

> The main issue is that you kind of have to dumb yourself down to answer the questions - if you start actually thinking critically, you'll run into trouble.

This has always been true on the SAT--especially the math section.

In the math section, it is almost always faster to eliminate answers using backward reasoning than getting the correct answer using forward calculation. For the sample question, answers C&D relate to intercept--answers A&B relate to slope. The question asks about slope so you immediately discard C&D. Now, even a guess is a plus for you.

By the way, that sample question would get removed after statistical norming. It's sufficiently confusing that it won't pass the standard--poor students always miss--mediocre students sometimes hit--good students always get it right--that is required for the SAT to work statistically. The confusion is that choice of slope is arbitrary--there is no particular reason for why which length is on which axis and which is cause and effect.

This was actually one of the unique things about the SAT for me. While "sample" questions always had lots of confusing answers, the SAT questions were absolutely crystal clear. Any SAT section with more than one confusing question was obviously the experimental section.

Re: New SAT, New Problems

#34

Post by bsder »

keypusher wrote: post #14

Can someone explain why both answers A and B are not valid on the metacarpal problem? If the line indicates a linear relationship between these two values, and the choice of axis is arbitrary, it seems to me you could just as well predict one from the other as vice versa.

Both answers are valid. This question would not survive the statistical norming process that the SAT uses to make sure that questions are "well-behaved" with respect to their answers.

By convention, X is supposed to be the independent variable and Y is supposed to be the dependent variable. Since it is not clear which variable really should be "dependent", the question would get thrown out.

Re: New SAT, New Problems

#35

Post by bsder »

copascetic wrote: post #27
keypusher wrote: post #24

Earlier quoted context omitted.

This then seems like less of a test of math ability and more of a test of whether or not you have effectively absorbed the jargon and conventions around certain math concepts. Maybe that's what they want, but it doesn't really seem like problem solving at all.

See my other response, but I don't think you're giving the question enough credit. The other responses don't mention the fact that the line would likely be different if we swapped the variables. The question tests both whether you know what a line of best fit represents, and also whether you know that the correlation in one direction is not necessarily equal to the correlation in the other direction. This is somethin…

Except that there is no reason to choose the correlation in one direction versus the other in the given problem.

It's a bad question.

Re: New SAT, New Problems

#38

Post by jlarocco »

keypusher wrote: post #24
Chinjut wrote: post #19

Earlier quoted context omitted.

Sure, you could just as well predict one from the other. But the question asks "What is the following is the best interpretation of the _slope_ of the line of best fit in the context of this problem?", and you are expected to interpret "slope" in the usual jargon sense (as the the ratio of vertical change to horizontal change, rather than vice versa). (By "vertical" and "horizontal" here, I mean with respect to the g…

This then seems like less of a test of math ability and more of a test of whether or not you have effectively absorbed the jargon and conventions around certain math concepts. Maybe that's what they want, but it doesn't really seem like problem solving at all.

To be fair, knowing the jargon and conventions is almost just as important as knowing how to solve problems.

If nothing else, in college, most of the problems will be posed in terms of standard jargon and conventions, so it helps to know it.

Re: New SAT, New Problems

#39

Post by harryh »

camelite wrote: post #28
harryh wrote: post #22

Earlier quoted context omitted.

It's actually become less about wrote manipulation of symbols and more about a deeper understanding of actual math concepts. The latter is, indeed, harder.

One thing rote learning is good for is how to spell in languages whose pronunciation and spelling are inconsistent, like English. (sorry)

Hahahah. Yes. You got me there!

Re: New SAT, New Problems

#40

Post by graeme »

yummyfajitas wrote: post #16
BlackJack wrote: post #3

Here's the College Board's page on the redesigned SAT: https://www.collegeboard.org/delivering-opportunity/sat/rede... Here's a 200 page PDF on the new format with lots of sample questions: https://www.collegeboard.org/pdf/sat/delivering-opportunity/... The sentence completions are gone; now we have words in context. Essay is twice as long. No independent writing section. Math is more algebra and prob/stats focused,…

More coachable is probably the goal. A big problem for the college board is that insofar as the SAT is difficult, math driven, and g loaded, Asian people tend to score too high. This creates political pressure for colleges to stop using the SAT in admissions, an existential threat to the college board.

>More coachable is probably the goal. A big problem for the college board is that insofar as the SAT is difficult, math driven, and g loaded, Asian people tend to score too high.

How would this help? I've taught SAT courses, and Asian students are overrepresented in those prep courses. Those cultures tend to have a habit of using external prep.

Note for posterity: I'm not saying Asian overrepresentation is a bad thing. Just stating a fact about courses I taught. If the college board's motive is to reduce Asian scores, I don't think they've picked a good method.

Post reply on HN