<p>No, Predator, I'm pretty sure you are wrong. This is dependent on the particular test you take. Some sittings you can get one wrong and still get an 800, and some you can't. I will stand by my original statement on that.</p>
<p>You can see that this is likley to be true by comparing the number of students who got above a 1580 on the "old" test (pre-1995) and the "new, recentered" test. </p>
<p>According to Wikiperdia:</p>
<p>"The "old" SAT (prior to 1995) had an incredibly high ceiling. In any given year, only seven of the million test-takers scored above 1580. If one makes the reasonable assumption that all of the very brightest people in that U.S. age group, which numbers 3 million, took the test, then a score above 1580 has a rarity of about one in 400 thousand, equivalent to the 99.9997 percentile."</p>
<p>In fact, the year that I took the test (1988), there were only three people in the whole United States that received 1600 (no questions wrong). (I knew one of these, who was my classmate at Harvard, by the way. He's a math professor at U. Wisconsin now.)</p>
<p>Further:</p>
<p>"In 2005, the test was changed again, in response to various criticisms. Because of issues concerning ambiguous questions, especially analogies, certain types of questions were eliminated (the analogies disappeared altogether). The test was made marginally harder, as a corrective to the rising number of perfect scores. A new writing section was added, in part to increase the chances of closing the opening gap between the highest and midrange scores. Other factors included the desire to test the writing ability of each student in a personal manner; hence the essay. The New SAT (officially the SAT Reasoning Test) was first offered on March 12, 2005, after the last administration of the "old" SAT, the 1994 revision, in January of that year."</p>
<p>Finally:</p>
<p>"The March 12th 2005 test was reported to have 107 perfect "2400" scores, above College Board estimates but at a far lesser rate relative to the old proportion of perfect scores on the 1600-point test. (meaning between 1995-2005, when the number of perfect scores was much larger--Greg)"</p>
<p>Now, pre-1995, there were about 3 people per YEAR in the entire US that received perfect scores--- meaning NO questions wrong. That was summed over about ten sittings for the test over the entire year, meaning that on each test sitting, on average, about .3 people--- a fraction of ONE person--- who got a perfect 1600. Compare this with the SINGLE test sitting of March 12, 2005--- where 107 people got "perfect" 2400. On that SINGLE sitting. Which means, each year, around 1000 people are now getting "perfect" 2400s.</p>
<p>What we can surmise from this is one of two things (assuming that the average intelligence level of the testtakers has not improved dramatically):</p>
<p>(1) Either the actual test questions have gotten much much easier than they were in 1995--- and therefore more people would get every single question correct. However, this cannot be true, since when they revamped the test in 2005, "The test was made marginally harder, as a corrective to the rising number of perfect scores," as the Wikipedia article states.</p>
<p>Or:</p>
<p>(2) It is now possible to get a "perfect" 2400 but still get several questions wrong.</p>
<p>I grant you that even under this scenario, it may be so that you must get all math questions correct to get an 800 on the math section--- but I think this unlikely, since the number of perfect scores (ie 2400) is SO MUCH GREATER--- about 300-1000 TIMES greater--- in 2005 than it was in 1995 that I do not think it possible that this could have possibly been reached by allowing you to get one wrong on the reading and writing sections alone.</p>
<p>I have not been able to find solid data to back up my assertion, but the numbers today versus the numbers in 1995 are SO much different (3 a year "perfect" versus 1000 a year "perfect",) that I highly highly doubt this could have been achieved without allowing--- at least on some sittings--- for people to get one or two wrong on the math section and still get an 800. </p>
<p>-Greg</p>