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

Khan Academy SAT Prep

khanacademy.org

Post reply on HN
Print view View original

21–29 of 29 posts

Re: Khan Academy SAT Prep

#21

Post by ikeboy »

foobarqux wrote: post #18
ikeboy wrote: post #14

Earlier quoted context omitted.

That's incorrect. https://sat.collegeboard.org/scores/how-sat-is-scored >In our statistical analysis, equating adjusts for slight differences in difficulty between test editions and ensures that a student's score of, say, 450 on one edition of a test reflects the same ability as a score of 450 on another edition of the test. Equating also ensures that a student's score does not depend on how well others did on the sa…

Intuitively it must be the case that your score decreases if the SAT taking population gets better at taking it because the entire point of the SAT is to rank students. I don't quite understand what the discussion in the link is trying to say but I take it to mean that they normalize between different versions of the test in the same year in a way that doesn't depend on the ability of the cohort that took a specific…

No, it's not the case.

The point may be to rank students, but they don't give lower scores if everyone does better. It's possible for everyone to get perfect, or for no one.

>I don't quite understand what the discussion in the link is trying to say but I take it to mean that they normalize between different versions of the test in the same year in a way that doesn't depend on the ability of the cohort that took a specific version. That is still consistent with the fact that they do normalize across the entire test taking population.

That's not what's happening.

There are different versions of the test. ETS (the org running it) wants a score to mean the same thing no matter which version or year you took it in (well, besides for when they make major changes, like 1995.)

So every test includes one section that doesn't count. You don't know which one it is. Let's say everyone who took Math non-counter-A gets an average of 540. The same people get an average of 520 on Math counts-B. So we know that 520 on B is just as good as 540 in whatever year A was given. So we scale, and anyone who got a 520 on on B is given 540 instead. This is extended for other scores, using math.

(Actually, this is probably done before converting away from raw scores.)

If everyone does poorly, that will make the average go down for the actual test, and also make the average go down for the non-counting section. Those factor out completely, meaning others' performances have zero effect on your score.

Hope that helps.

Re: Khan Academy SAT Prep

#22

Post by ikeboy »

foobarqux wrote: post #20
ikeboy wrote: post #15

Earlier quoted context omitted.

No, it just increases average scores (assuming people actually use them and they're minimally effective)). https://sat.collegeboard.org/scores/how-sat-is-scored >Equating ensures that the different forms of the test or the level of ability of the students with whom you are tested do not affect your score.

I understand that to mean that they do normalization between different versions of the test in a given year, not that they do not do normalization across the SAT population in that year. It would surprise me if they didn't do population-wide normalization, otherwise too many scores would be at or near the upper limit, making the score uninformative, and also you could not compare the rank of students between years. R…

> It would surprise me if they didn't do population-wide normalization, otherwise too many scores would be at or near the upper limit, making the score uninformative, and also you could not compare the rank of students between years.

Be surprised. http://research.collegeboard.org/content/sat-data-tables http://research.collegeboard.org/programs/sat/data/archived/...

You can't compare percentile, which is why they're released anew each year. You can compare scores.

>Regardless, because the point of the SAT is ranking, your score is irrelevant.

That's a better point, perhaps. Your race also matters (minorities can get into Ivies with much lower scores than other people). Besides, if people get better, colleges might have more room.

(For people with perfect scores, it doesn't really matter.)

Re: Khan Academy SAT Prep

#23

Post by silencio »

film42 wrote: post #13

This is great! I'm not sure if it was discussed on HN or not, but Khan Academy recently released an MCAT course with "more than 900 videos and 2000 practice questions." This is incredible, because last summer my wife's online MCAT course cost us around $2,500. It was a good course, but Khan Academy's course seems extensive and thorough. I hope it gets more publicity among med-school applicants! Link: https://www.khan…

I'm getting ready to take the MCAT and that's pretty widely known already on premed forums. It helps that it's a partnership between AAMC and KA and was one of the first resources out there for the new edition of the MCAT. I see a lot of folks still taking courses and using other resources too, but yes - definitely somewhat of a moneysaver :)

Re: Khan Academy SAT Prep

#24

Post by pkaye »

CodeSheikh wrote: post #16

SAT and all these similar "standardized" tests are a big joke. This outdated testing methodology needs to be evolved. College admissions should be somewhat similar to job interviews. Students walk in, meet admission committee panel (composed of a member of registrar office and professors/lecturers), talk about expectations, solve a few problems, talk about their essays, present design work, piece of literature etc. S…

What you propose is not scalable. How would you execute this for overseas admissions?

Re: Khan Academy SAT Prep

#25

Post by IndianAstronaut »

tokenadult wrote: post #12

I'm glad to see that the official URL for the Khan Academy partnership with the College Board is now live. https://www.khanacademy.org/sat I have been looking forward to being able to tell all my friends that they don't have to worry about test prep for the SAT anymore, as that is now free and online through Khan Academy. For people who are still inclined to worry about standardized tests, I will mention that this pr…

When I used to teach SAT classes, my number one recommendation was to start reading Smithsonian magazine. That magazine has a lot of the types of articles you find in the SAT.

Re: Khan Academy SAT Prep

#26

Post by foobarqux »

ikeboy wrote: post #22
foobarqux wrote: post #20

Earlier quoted context omitted.

I understand that to mean that they do normalization between different versions of the test in a given year, not that they do not do normalization across the SAT population in that year. It would surprise me if they didn't do population-wide normalization, otherwise too many scores would be at or near the upper limit, making the score uninformative, and also you could not compare the rank of students between years. R…

> It would surprise me if they didn't do population-wide normalization, otherwise too many scores would be at or near the upper limit, making the score uninformative, and also you could not compare the rank of students between years. Be surprised. http://research.collegeboard.org/content/sat-data-tables http://research.collegeboard.org/programs/sat/data/archived/... You can't compare percentile, which is why they're…

> You can't compare percentile, which is why they're released anew each year. You can compare scores.

What does "compare" mean? Does it mean that two people with the same score but in 1996 and 2015 respectively would be expected to achieve that score in a 2005 test?

Re: Khan Academy SAT Prep

#27

Post by ikeboy »

foobarqux wrote: post #26
ikeboy wrote: post #22

Earlier quoted context omitted.

> It would surprise me if they didn't do population-wide normalization, otherwise too many scores would be at or near the upper limit, making the score uninformative, and also you could not compare the rank of students between years. Be surprised. http://research.collegeboard.org/content/sat-data-tables http://research.collegeboard.org/programs/sat/data/archived/... You can't compare percentile, which is why they're…

> You can't compare percentile, which is why they're released anew each year. You can compare scores. What does "compare" mean? Does it mean that two people with the same score but in 1996 and 2015 respectively would be expected to achieve that score in a 2005 test?

It means that someone in the Xth percentile on one test/year may have done either better or worse than the same percentile a different year.

Your second question is yes.

Re: Khan Academy SAT Prep

#28

Post by dang »

jeresig wrote: post #10
dang wrote: post #8

Earlier quoted context omitted.

That's its home page, but I suppose the submitter chose the other URL because it gets you into the details straightaway, which an HN audience tends to appreciate. Both links are good.

Not exactly. It's linking to the old SAT practice material that's for the old (current) SAT exam. The new material, for the upcoming 2016 exam, which was just launched today, is available at: https://www.khanacademy.org/sat If you want to "get into the details" you'll need to be logged in and complete the onboarding, through this URL: https://www.khanacademy.org/mission/sat I mean, linking to the current URL is fine,…

Ok thanks, that's dispositive. Url changed from https://www.khanacademy.org/test-prep/sat.

(Sorry khirasaki; I get what you meant now.)

Re: Khan Academy SAT Prep

#29

Post by mirithekiwi »

tokenadult wrote: post #19
CodeSheikh wrote: post #16

SAT and all these similar "standardized" tests are a big joke. This outdated testing methodology needs to be evolved. College admissions should be somewhat similar to job interviews. Students walk in, meet admission committee panel (composed of a member of registrar office and professors/lecturers), talk about expectations, solve a few problems, talk about their essays, present design work, piece of literature etc. S…

You are advocating an unstructured interview process for college admission. That would be hugely expensive, as the first reply to your comment mentioned, and it would also be less likely to find strong students for colleges. There is a lot of research on this. The current mixed admission system that includes standardized tests actually works better for colleges, at less expense, than a system of in-person interviews…

I would say you are probably correct, although the college admissions process isn't exactly cheap as it is.

It normally costs ~$50 to take the SAT once, plus ~$12 per college you want to apply to. Then there are the SAT subject tests, which cost less ($30?) but it's not uncommon for a student looking to get into a selective school to take >3 subject tests. Then, maybe the student needs to retake the SAT, which is pretty standard, so add another ~$50 dollars and now we're looking at several hundred dollars in exams. This isn't counting the $50-100 application fee per school. Having a professor interview applicants would still be impossibly expensive of course. Do you happen to know how well the alumni interviews work? I never had one, but if I understand correctly, they are a part of Ivy League tier applications where a graduate of the school deems if the applicant would be a good fit. I've heard there's not much weight placed on them, however.

I personally think the SATs are better indicators of socio-economic standing than scholastic aptitude, but agree that there's probably nothing better. I kind of wish that the application process was similar to that for graduate school, where there's more weight placed on the relevant subject test and letters of recommendation.

Post reply on HN