<p>
[quote]
Give me a kid who gets 2200 without any prep over a 2400 3rd time anyday.
[/quote]
</p>
<p>Hi, 2250 on a practice test the first time, 2220 on actual exam after a prep class that probably gave me advice that would have been helpful for someone in the 1900 range.</p>
<p>Then I took a bunch of SAT Subject, AP, and final exams, did a bit of the blue SAT study book myself (sans parent intervention, my mom says I can go to community college), and overshot my goal of 2300 on a second taking.</p>
<p>
[quote]
gellino writes: Not only were scores recented in 1995, which effectively adds 70 points to verbal scores, but the percentiles themselves have changed in this age of no child left behind, self-esteem boosting. I get the impression after adding 70 points that 1430+ isn't as high as top 1% and 1350+ isn't as high as top 5% with new score compression, which makes it harder to stand out which has caused added competitiveness.
[/quote]
That's a pretty cynical interpretation, and one that reflects a misunderstanding of what scores and percentiles mean.</p>
<p>The SAT score before recentering tells how you did relative to a group of kids who took the test in the early 1940's. Back then college was for kids who tended to be affluent, a group not reflective of society as a whole. The SAT scores are based on standard deviation so 1/2 of that 1940's group scored above 500 and half below on each test, 2/3rd's scored between 400 and 600 which is 1 standard deviation from the mean, and so on. If you go to any stats manual and you know your score on the old SAT (or use the adjustment tables to find it from the new scale) you can find out where you fall relative to those kids by looking at how many standard deviations your score is above/below the mean of 500.</p>
<p>The percentile table for each administration of the test tells how you have scored relative to current test takers, not the original population. The current generation of test takers does not score the same as those in the 1940's. Before recentering only 7% got a 600 or above (you'd expect 1/6th or 16% to do so) and a whopping 42% scored 400 or less (again in the 40's only 16% did). See this info at <a href="http://tinyurl.com/yjayvq%5B/url%5D">http://tinyurl.com/yjayvq</a> </p>
<p>What recentering did was adjust the SAT score so that it regained it's old statistical properties of a mean of 500 and a standard deviation of 100 for the 1990's population used to recenter it. What it did not do, contrary to gellino's claim, is change the percentiles in any way. If you were in the top 5% before recentering you were in the top 5% after recentering. The score used to indicate that percentile may have changed, but so what? If 1350 (V+M) isn't top 5% on the new scale that's because it's not enough standard deviations from the mean of 1000 to put it there. But the simple fact remains that if you are a top 5% student you would have gotten a 1350 on the old scale and will get whatever top 5% corresponds to on the new scale.</p>
<p>Before recentering there were too many standard deviations available (for the SAT score) above the 50% percentile of current test takers and not enough below it. On the verbal scale, for example, 50% of the population was scoring 420 or less before recentering. The original intent of the SAT was to lump everyone who was 3 standard deviations above the mean (top .1%) together with the score 800, but because the population mean was not 500 anymore the test could make even finer distinctions than that. That's why back in the day there were only a handful of kids in the entire country who got a 1600 and they got invited to the White House. And with an average of 420 there were not enough score values below 420 to the floor of 200 to spread out the low scorers. Hence the motivation for recentering. 800's are more common today because they reflect the original intent, top .1%. If you don't like that, tough.</p>