<p>-2 = 740.</p>
<p>:(</p>
<p>Looks like Tenplenesis’ prediction was correct.</p>
<p>800
770
740
720 = -3 (my son - would have liked just one less missed)</p>
<p>Got 5 wrong and got 680 no omit.</p>
<p>1 wrong- 770 .harsh curve!!!</p>
<p>The curve has changed markedly over the past 3 years…one could miss 2-3 questions 3 years ago and receive a perfect 800. Now if one misses 2 questions, one will receive a 750 to 760. Three points down (1 omit and 2 misses) yields 730-740.</p>
<p>It’s a whole different ball game for you students now!</p>
<p>
Not true. The most one could ever miss and get an 800, at least on QAS months, in SAT I Math was a single question, never two or three. The curve is a bit harsher now, but is definitely not markedly different.</p>
<p>^The most generous curve I’ve ever seen is -1=800 on math, i.e. you can miss one question but certainly not 2 or 3 questions.</p>
<p>Darn, ACTTester was faster.</p>
<p>Remember that a harsh curve means the questions were overall easier than average – all scores are standardized so that a 770 one session will equal a 770 in another session.</p>
<p>Of course, this still works against you if you knew the material inside out yet made a couple of sloppy mistakes.</p>
<p>It seems to me that the curves have gotten harsher without the tests getting that much easier… I though the Jan SAT math section was the same difficulty as the bluebook 3rd test that shows a -1 800 curve.</p>
<p>
</p>
<p>That may actually be possible over different years. The top test takers have been getting better at taking the test these last 5 years – the number of perfect scores per million has doubled or more. Colleges are interested in how a given student compares to other students in his class, not necessarily how he compares to his old brother who took the test 4-5 years ago. So it’s possible that a handful of questions have gotten harder or that the very top of the scale has gotten tougher to achieve without a perfect score, as a way of offsetting improved test-prep skills by each new year’s group of juniors.</p>
<p>My son took the ACT last June and got a problem with imaginary roots, which pretty much needed to be solved using synthetic division. Who teaches synthetic division in high school outside of a math competition? No such problem was in any test prep book I saw. I suspect it was tossed in just to take the edge off of getting too many test-takers with a perfect score in math.</p>
<p>Synthetic division is pretty commonly taught, but not until (typically) later in algebra II or in pre-calc. But that is what the ACT is about: it is mostly an achievement test, and less a reasoning test than the SAT.</p>
<p>Here are some statistics of question difficulty using QAS tests. Note the consistency of overall test difficulty (last column).</p>
<p>
test L1 L2 L3 L4 L5 ave</p>
<hr>
<p>200503 10 10 17 13 4 2.83
200510 9 13 16 11 5 2.81
200601 11 10 14 12 7 2.89
200605 9 10 22 9 4 2.80
200610 11 10 20 8 5 2.74
200701 9 14 17 9 5 2.76
200705 10 14 15 12 3 2.70
200710 10 13 19 9 3 2.67
200801 8 15 16 10 5 2.80
200805 11 9 22 8 4 2.72
200810 8 10 22 11 3 2.83
200901 11 8 20 10 5 2.81
200905 8 13 18 12 3 2.80
200910 11 11 17 10 5 2.76
201001 9 15 14 13 3 2.74
201005 9 12 18 11 4 2.80</p>
<p>ave: 9.6 11.7 17.9 10.5 4.3 2.78
</p>
<p>
</p>
<p>But how does one determine “question difficulty” when the test-taking group is ever-changing and growing ever more sophisticated? It may well take a harder question given to a group of one million juniors in 2010 to yield the same right/wrong stats than an easier questions given to one million juniors in 2005.</p>
<p>Each question is tested not long before use (perhaps up to a year before but certainly not 5 years) to determine the difficulty level on the basis of the percentage of test takers getting the question correct.</p>
<p>In SAT math, the evidence that juniors in their entirety are “more sophisticated” is lacking: wouldn’t this imply a rise in average SAT math scores?</p>
<p>
</p>
<p>The average doesn’t necessarily have to change, just the very upper tail and the number of test takers in the very upper tail is too small to shift the average more than a small fraction of a point.</p>
<p>There IS evidence that the very upper tail is being breeched more frequently. Five years ago, a 36 ACT (i.e., 35.5-36.0) was achieved by only 1 in 5500 test takers; last year, it was achieved by 1 in 2500 test takers. I think the SAT numbers have risen sharply also. I would imagine that near-perfect scores rose, too, but I didn’t investigate it (my younger son got a 36.0, so I was curious as to its rarity).</p>
<p>This implies a small “bump” in the upper tail of the distribution curve. Given all the test prep books, test prep seminars, SAT/ACT tutors and the like, plus multiple retakes, it shouldn’t come as a surprise that a portion of 1% of the test takers now know these tests and their tricks inside-out.</p>
<p>One way to dampen this “bump” at the upper tail is to simply increase the difficulty of one or two problems in each section. A few test-takers will be hit with unfamiliar material and others will simply run out of time working through the last problems, or have no time left over to double-check for sloppy mistakes.</p>
<p>Ah, what you really meant to say was: the top echelon of test takers are increasingly more sophisticated. Yes, I’d agree with that. The percent of SAT test takers scoring 700 or more in math has in fact risen since the new SAT started, from around 6% to about 7%.</p>
<p>But the “bump” in the distribution at the top isn’t a problem (and is a side-effect of limiting scores to an upper limit, namely, 800) unless you feel that the goal of the SAT is to provide information distinguishing the very top students (i.e., the weak 800s from the strong 800s), which it is not meant to do.</p>
<p>If they wanted to distinguish among the top students better, they would instead put more hard questions on and keep the curve going down by no more than 10 at the top end rather than penalizing 30 points for careless mistakes.</p>
<p>
</p>
<p>There is a certain marketing magic to the rare and elusive 2400. If present trends continue unchecked, there could be a couple thousand 2400s every year by the end of the decade, eroding its exclusivity and newsworthiness. Thus, an effort needs to be made, tweak by tweak, to subdivide the current low 800s from the high 800s.</p>
<p>Consider the SAT II scores, where an 800 is obtained by 10% of the test-takers in some fields (physics, math II). This erodes the test’s ability to make finer distinctions.</p>
<p>
</p>
<p>Unfortunately, that would lower the median score sharply, which would invalidate its longer-term norm value, which is primarily based on consistency near the average – at the far tail, results are always less valid due to a small population (standard IQ tests also suffer from this problem). The goal is to only make those questions more difficult that most average students would get wrong anyway, but that would now trick some of the previous 700+ test-takers.</p>
<p>
I don’t see why the College Board would or should be interested in such an effort; AP tests as well as the math subject tests already provide at least some information to distinguish between these types of students. With 1.5+ million seniors taking the SAT, it is hard to see how having fewer 2400s would be a useful goal for the CB from a marketing point of view.</p>
<p>Im glad finally people are catching on. </p>
<p>The SAT in 1996 tried to make the upper end tail more indistinguishable. So “weak” 800s and “strong” 800s were pretty much the same. for instance a 1520 on the SAT in 1994 was 99.991 percentile or 1 in 20,000 whereas in 1995 a 1520 on the SAT I would be 1 in 1000. </p>
<p>Im glad to see the SAT try to distinguish them again. It is crucial to our nation’s education and future to try to distinguish upper tail students so that they can build bridges, make discoveries, research and development, public policy makers, nobel prize winners so that our country can succeed in innovation and create jobs for everyone. </p>
<p>I am glad to see the trend reversing, however, I am not sure a big penalty for a few questions wrong is the answer. I think the answer is try to put in harder questions.</p>