4 ms·

The irony is that universal access to the best prep materials makes the SAT harder.

otoburb 11y ago
The parent refers to the fact raw SAT scores are normalized, thus largely dependent on the relative strength of your cohort's performance[1]. [1] http://en.wikipedia.org/wiki/SAT#Raw_scores.2C_scaled_scores.2C_and_percentiles http://en.wikipedia.org/wiki/SAT#Raw_scores.2C_scaled_scores...
- ikeboy 11y ago
  That's incorrect. https://sat.collegeboard.org/scores/how-sat-is-scored 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 same edition of the test.
  - foobarqux 11y ago
    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 version. That is still consistent with the fact that they do normalize across the entire test taking population.
    - ikeboy 11y ago
      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.
the_cat_kittles 11y ago
id rephrase that to say it's scores become more accurate
ikeboy 11y ago
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 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.
- foobarqux 11y ago
  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. Regardless, because the point of the SAT is ranking, your score is irrelevant. What matters is how you do with respect to your peers. If everyone does better the test has become "harder" because you need to do better to get a higher rank.
  - ikeboy 11y ago
    > 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/content/sat-data-tables http://research.collegeboard.org/programs/sat/data/archived/cb-seniors-2013 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.)
    - foobarqux 11y ago
      > 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?

The parent refers to the fact raw SAT scores are normalized, thus largely dependent on the relative strength of your cohort's performance[1]. [1] http://en.wikipedia.org/wiki/SAT#Raw_scores.2C_scaled_scores.2C_and_percentiles http://en.wikipedia.org/wiki/SAT#Raw_scores.2C_scaled_scores...

ikeboy 11y ago
That's incorrect. https://sat.collegeboard.org/scores/how-sat-is-scored 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 same edition of the test.
- foobarqux 11y ago
  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 version. That is still consistent with the fact that they do normalize across the entire test taking population.
  - ikeboy 11y ago
    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.

the_cat_kittles 11y ago

id rephrase that to say it's scores become more accurate

ikeboy 11y ago

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 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.

foobarqux 11y ago
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. Regardless, because the point of the SAT is ranking, your score is irrelevant. What matters is how you do with respect to your peers. If everyone does better the test has become "harder" because you need to do better to get a higher rank.
- ikeboy 11y ago
  > 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/content/sat-data-tables http://research.collegeboard.org/programs/sat/data/archived/cb-seniors-2013 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.)
  - foobarqux 11y ago
    > 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?