# Estimate of the New SAT I Percentiles

<p>This is a takeoff of an idea from another thread. If you want to try and estimate the percentiles for the new SAT, post in a format like below:</p>

<p>COMPOSITE:
99th Percentile - 2270
95th Percentile - 2110
90th Percentile - 1980
80th Percentile - 1830
70th Percentile - 1730
60th Percentile - 1630
50th Percentile - 1550
40th Percentile - 1460
30th Percentile - 1380
20th Percentile - 1280
10th Percentile - 1140</p>

<p>MATH:
99th Percentile - 760
95th Percentile - 700
90th Percentile - 670
80th Percentile - 620
70th Percentile - 580
60th Percentile - 550
50th Percentile - 520
40th Percentile - 490
30th Percentile - 460
20th Percentile - 420
10th Percentile - 370</p>

99th Percentile - 750
95th Percentile - 700
90th Percentile - 660
80th Percentile - 610
70th Percentile - 580
60th Percentile - 540
50th Percentile - 510
40th Percentile - 480
30th Percentile - 450
20th Percentile - 420
10th Percentile - 370</p>

<p>WRITING:
99th Percentile - 760
95th Percentile - 710
90th Percentile - 670
80th Percentile - 620
70th Percentile - 580
60th Percentile - 550
50th Percentile - 520
40th Percentile - 490
30th Percentile - 460
20th Percentile - 420
10th Percentile - 370</p>

<p>I think the errors that are possible are that the scores for composite and math might be too low, and scores for critical reading and writing too high.</p>

<p>I hope you were right about everything except Math...</p>

<p>No, the composite scores are way too high. Just because 99th percentile is 760 in all three sections doesn't mean 2280 is the mininum 99th percentile. Not all 760 Math people score 760 in all of the other sections. This is what makes 99th percentile in all three sections unique. A 99th percentile on the New SAT is like a 2170-2180. How do I know? 1470-1480 was 99th percentile on the old SAT and 2170-2180 is about equivalent to that. Its a longer test, you know, mor fatigue, more sections, lower scores.</p>

<p>yea this is thoroughly flawed.</p>

<p>Composites are primarily based off of this chart: <a href="http://www.collegeboard.com/prod_downloads/about/news_info/cbsenior/yr2003/pdf/table_3b.pdf%5B/url%5D"&gt;http://www.collegeboard.com/prod_downloads/about/news_info/cbsenior/yr2003/pdf/table_3b.pdf&lt;/a> and it's looking like the writing distribution will probably be centered higher than the math and cr distributions, so that's how I come up with the composite estimate. CB did a study on the effect of length versus performance and found no statistically significant result that would support the claim that performance was diminished by a longer test. Most of the percentile estimates are similar to the released data for the old test, slightly modified.</p>

<p>I concede my 99th estimate is a high, but it should be no lower than 2230.</p>

<p>

</p>

<p>If you find it flawed, provide your own estimate. I never said my estimate was conclusive.</p>

<p>And I forgot one more percentile estimate:</p>

<p>Composite 99+ percentile: 2320
Math 99+ percentile: Maybe there will be one for once at 800
CR and Writing 99+ percentile: 800</p>

<p>Writing scale is off. Writing is so much harder now...</p>

<p>Here's my guess for the composites:</p>

<p>99th 2180
95th 2050
90th 1950
80th 1800
70th 1700
60th 1640
50th 1530</p>

<p>

</p>

<p>From the data I've collected, it seems that even on CC, the writing distribution is higher in general by 10-20 points over the math and critical reading distributions.</p>

<p>ALot of us with perfect old sat scores struggled alot on the writing section. 740 now is like 780-800 before</p>

<p>Yup. Us awesome 1600ers with terrible writing scores. Oh too common.</p>

<p>While top 800 scores may decrease, 700+ scores should be of greater frequency in this section compared to the frequency of those in math and critical reading.</p>

<p>where did you pull that from?</p>

<p>i think 2110 is higher than 95th .. more like 97th</p>

<p>

</p>

<p>I should have made myself clear on that statement. That was based on data collected from CC and compared to math and critical reading data collected from CC. I understand it's a small sample and hardly representative of the population, a paired difference of means helps to filter out the fact it is not representative of the population.</p>

<p>agree with Icedragonad, seems like those would be the percentiles</p>

<p>What data did you use? I collected about 50-55 results from kids who took the June SAT & it was consistent w/ the data I had from the March SAT: Math was 15 points higher than reading. Reading was 10 points higher than writing.</p>

<p>And also, 1470 on the old SAT was 99th percentile. With writing being the toughest section, 99th percentile will be below 2200 (but not much).</p>

<p>Here's my guess...</p>

<p>Composite:
99th = 2190
95th = 2060
90th = 1950
85th = 1870
80th = 1800
75th = 1750
70th = 1700</p>

<p>Math (same as old SAT):
99th = 770
95th = 710
90th = 670
85th = 650
80th = 620
75th = 600
70th = 580</p>

<p>Reading (same as old SAT Verbal):
99th = 760
95th = 700
90th = 660
85th = 630
80th = 610
75th = 590
70th = 570</p>

<p>Writing:
99th = 750
95th = 690
90th = 650
85th = 620
80th = 600
75th = 580
70th = 560</p>

<p>Seems pretty accurate</p>

<p>I agree with Stem. Heca's is too harsh.</p>