<p>This was done before 1941 (every year, the scores would be scaled so that the mean score was 500 with a std deviation of 100). Since then, the mean and std deviation have been allowed to drift. The means were recentered back to roughly 500 starting in April, 1995, but then were allowed to drift again. You cannot fix the mean scores and yet have scores that are comparable one year to the next.</p>
<p>Here are the class of 1981 percentiles (first is verbal, second is math):</p>
<p>The SAT and SAT subject tests deduct a fraction of pt for wrong answer. So only guess if you can eliminate a couple of answers.</p>
<p>ACT and AP tests do not deduct extra for wrong answers, so always guess. AP just ended the guessing penalty a year ago, so that may be what you had in mind.</p>
<p>It is not true that SAT scores don’t go down when you miss one question. (THAT’S good practice for the double negative tricky questions.) On Writing the scores mostly drop at one miss. On Math sometimes you can miss one, sometimes not any. On the June test this year, one miss gets you 790. I know because it happened to my daughter. Other times, one miss is still 800, and 2 is 770.Only on the Reading can you miss 2 or possibly 3 and still get 800.</p>
<p>You’re certain of that? If so, then the CB scores SAT subject tests quite differently, which is odd. For example, based on their “official” study guide for subject tests, you can leave more than 10% of the questions on a Math 2 test blank and still get an 800. On the Physics test, you can leave 11 out of 75 questions blank and still get an 800.</p>
<p>College board allowed calculators for SAT starting in 1994. The kinds of manipulations you need to do for SAT math don’t require calculators as long as you know your multiplication tables, but the test prep people think of all manner of interesting ways for kids who don’t know standard methods of solving problems to use calculators to get answers. Where calculators come in handy is on the ACT, where there is actual time pressure to finish and somewhat harder calculations. Calculators can be programmed to do lots of work for you fast on the ACT’s straightforward math, and some of the tricks can save precious time on ACT.</p>
<p>But I think one of the biggest differences between now and then is the relatively greater weight given to these test scores in admissions today, despite all the controversy about whether tests are worthwhile. You can look up all kinds of research online about it. Naturally kids are more anxious about the tests today. And this kind of anxiety is a self-fueling phenomenon. The more people worry about it, the more everyone else feels they should worry about it. Some of it is ah…out of proportion. Even if you want to go to top ivy league schools, once your scores are ‘good enough’, above 2250 or so, time is better spent on things like crafting the admission essay, making sure that extracurriculars and volunteer activities show leadership and uniqueness of individual, and so forth. Because that is what will make the application, not an extra 30 points on the test.</p>
<p>If it is true that you can’t make any (or very, very few) errors or omissions on the SAT to get perfect scores – and so many kids are getting perfect or nearly perfect scores – then it must truly be a case of the test being substantially easier than before. While I’m not discrediting the whole “test prep” thing (whether that’s studying targeted material that is often tested, learning calculator tricks, or getting into the head of the test makers), it still stands to reason that kids today must be getting easier questions. In a community like CC, it’s easy to forget that not everyone cares as much about scoring a 2300+ on the SAT, nor is every student taking test prep courses. Yet, if it’s not only customary but actually EXPECTED that you score in the mid to upper 700’s in each section, then there is no other explanation. Students 20 years ago who were focused heavily on getting into an Ivy League school put just as much pressure on themselves to score well. I don’t think that the only difference between the students of 1992 and those of today is better preparation.</p>
<p>College board has open archived data for SAT back to '96. They use the most recent test of graduating seniors. Numbers of seniors who have taken the test has gone up from 1.1 m to 1.65 m during that time. Comparing %age with 700+ scores in reading or math (no writing in '96 so nothing to compare)- Reading '96: 4.4%, 2011: 4.6%. Math '96: 5.3%, 2011: 6.7%. I checked years close to '96 and then close to 2011, and results are pretty similar. At the 700+ level, scores have up a tiny bit or stayed the same in reading and have gone up over one percentage point in math. So it’s not radically different over those years. I would guess the math can be explained by a combination of more high level math classes offered these days in high school, and more SAT prep. Right now I can’t find the college board data chart I was looking at recently which listed exactly how many test-takers got each score in 2011, but there were on the order of 300 kids who got 2400 out of 1.6 million, not very many. Keep in mind that the the kids who get high scores on math don’t necessarily get the same high scores in reading. There is some overlap.</p>
<p>I guess that doesn’t address the change after the recentering though. It seems the recentering of '95 did not change math raw to scale scores in the 600-800 range. But the verbal scores were really rocked. 700+ went from 1% to 5% of test-takers and 600-690 went from 12% to 32%. This was not a matter of easier questions, just an increase in scale score for any raw score. So of course universities would expect higher scores to get the same students as before.</p>
<p>That was exactly my point… if the test itself isn’t getting easier, then the scoring scale has to be more lenient. I had made the assumption that the entire curve was shifted upward, but I guess that was incorrect… only that the curve’s mean is shifted upward but the entire curve was also pinched.</p>
<p>I don’t know what parts of all of this are true. We’re all speculating based on past experience, anecdotal evidence, and whatever statistics the CB feels like giving to us. The bottom line is this: based on what I have seen and heard to be true, there seems to be a lot more people getting scores in the individual sections that would dwarf what was the norm 20 years ago. In addition, it seems that what is considered a “good” score nowadays is substantially higher than 20 years ago. I’m going to go out on a limb and say that kids today aren’t “smarter” than they were 20 or 30 years ago. Perhaps it’s partially preparation or maybe it’s due to proficiency is test taking. But I would guess it’s probably mostly due to a more lenient scoring curve. I’m not sure what is gained by the CB by having an easier curve. Maybe it’s like the sizing of clothing… if a designer makes a dress that is a size 8 roomy enough to fit a size 10, it’s good for our psyche as a consumer.</p>
<p>We’re probably spending way too much time worrying about what kind of mathematical equation helps us compare '92 apples to '12 oranges.</p>
<p>There is a perspective issue here, I think. We were raised at a time when a perfect score on SAT seemed like winning the lottery. That is a historical artifact of the times in which we grew up, when CB had neglected to adjust scoring as needed for many years. SAT was originally designed to have a normal score distribution with mean 1000 and standard dev. 200, range 400-1600. During late 60’s and 70’s the scores dropped, a lot, particularly verbal, which had a drop in mean from 500 to 425. Test-taking pop. changed- many more kids taking test, and even the brighter kids were growing up with lower skills.Little fixes were tried, but nothing major until the score re-centering. Now the score curve is roughly normal again, mean 1500, SD 300. For this distribution 0.19% of tests should be above 2.9 SD- that is 2370. In 2011 about 1300 of 1.65 million graduating seniors had their last SAT test score at 2370 or above, 0.08%, less than half of normal distribution. So the numbers aren’t unreasonably high at all. The test makers HAVE also made some questions a bit easier over the years. You can read the research. We are all getting dumber. But the point of the test is not to measure something fixed, but rather to fit the test-taking population into a normal curve, spread enough that the users (college admissions officials) can differentiate between students. Whether or not you accept the premise that testing does reveal something useful, which is a topic for another thread, it does fit the scores into such a curve. The admissions people know what the test is like and how the scoring works. If they think they want kids more into the tail because kids at the mean are dumber and test is too easy, they can set their cutoffs higher. My daughter just took SAT, missed 6 questions out of 170, which earned her a score of 2280, which seems just right to me. On the ACT she missed 3 questions out of 215 and got a 35, again, seems right. These tests are pretty grueling, much longer than when we were kids, for one thing. With all the proctor’s yakking about rules and getting people settled, ACT was 5 hours! The kids earn their scores.</p>
<p>I think the stats I quoted before about the effects of re-centering on V scores are not accurate. I should have verified source before posting. Those #s are much too large, have to go back and check that. The #s I just put up above in last post are correct, checked them from independent sources.</p>
<p>I’m not aware of the format of the score report nowadays. Do they tell you specifically what you got right and wrong?</p>
<p>I think my perception of the scores is tainted by the endless supply of posts on CC about complaining about “only” getting a 2350. I have to question how many of those scores are legit if we’re talking about only fractions of a percent that even achieve those types of scores.</p>
<p>On the report, questions are grouped into general topic and difficulty categories and you can see incorrect/correct/left blank for each of these groups, but you don’t see results for individual questions.</p>
<p>^That is the table I was used to get #s in my last post. Now I figured out what was wrong with the re-centering stats from my earlier post a couple of days ago. The person who used them was misusing #s from a normal distribution table, I think. Didn’t expect that. It was from a research paper about re-centering, but not one of CB’s. Anyway, 700+ should go from 1% before re-centering to 2.5% after. Both tails together, including 300-, would be 5%. And 600-690 should be going to 13.5%. The 32% would be 600-800 combined with 200-400. These are standard #s from normal distribution. I should have caught that when I first saw it right away.</p>
<p>My daughter is thrilled with her scores, especially the ACT, where she struggled with science prep and managed to get it down to only 3 errors. Maybe she would have liked to break 2300 on SAT. But complain?? She is ecstatic to only need to take tests once, wouldn’t dream of retaking, pretty thankful I insisted on all the prep. Everyone she knows takes it at least 2, and often 3 or even 4 times, and she didn’t believe she could raise her score enough by practicing at home to avoid that. Now she says there is no way you’d get her back in there to do that again. Tests on consecutive Saturdays and 2-3 weeks prior prepping at home. It was a really draining experience.</p>