<p>Better than what, hahalolk? A lower score. Well, within the precision of the SATs, yes, in the very limited sense of what the SAT tells us. I’m merely arguing that 1) the precision of the SAT is unclear - but certainly less than many here would argue for. Thus, I would say that a 2400 is not actually better than a 2350 because both are , essentially, the same when you consider precision. 2) Given that prep/retesting/superscoring affects scores greatly, and that not all students prep equally, for reasons both within and without of their control , SAT scores are even more imprecise.</p>
<p>The SAT may be imprecise, but a better score is still a better score no matter which way you cut it. Perhaps there’s the difference isn’t huge in admission, but a better score is always better, because it is simply better. The difference between a 2350 and a 2400 may not be much, but the 2400 is still clearly better, even if the advantage of having the extra 50 points isn’t that huge. </p>
<p>I admit that I can’t quantify how much better higher scores are in admission, but I do know that by their nature, they are better, to at least some degree.</p>
<p>^^I’m not entirely sure I agree with you hahallk. I don’t think it’s that simple. If you take a student who took the SAT cold one time and got a 2350 or a kid who took a prep class and then took the SAT three times ultimately getting a 2400 super-scored it is not the same. I think many pages ago Calmom pointed out “how” colleges could presumably use the super scoring to boost their reportable numbers by selecting the 'best" of each section over a number of sittings, but could just as easily look at the scores differently to find the “stronger” student. So “better” is really a relative statement. 2400 is better than 2350 if everything else is constant (same test prep class, same test taken at the same time for the same number of times) otherwise 2400 may not be “better.”</p>
<p>If it is a 2400 single sitting versus a 2350 single sitting, then it’s certainly better If it was a 2400 superscored, then that might not be the case.</p>
<p>How would 2400 taken in November of senior year by a student sitting the test for the 4th time be ‘better’ than 2350 in March of junior year by a student on their first sitting? Especially if the 2400 student scored significantly less on their first sitting?</p>
<p>Would you really think a 2400 from a student whose parents are college graduates and who attends an expensive, private prep school is “better” than a 2350 from a student at an urban public school whose parents are immigrants who don’t speak English in the home?</p>
<p>Neither the SAT nor any other psychometric test can be used in any context other than the one for which it has been validated. Just because it is a test and has a numerical score doesn’t make it meaningful. </p>
<p>IF we lived in a country that had a national exam for high schoolers that was developed to meet publicly determined curriculum standards – such as A levels in Britain or the New York Regents exam – and that exam was administered in a single sitting with only one score that could be reported – then I might see your point. Those exams are developed to fairly represent the expected curriculum - they are administered in a way that makes it tough to game the system – and they are actual tests of general knowledge in the subjects tested. </p>
<p>But the SAT is a test developed by a private testing company; statistics show that its format favors males over females, and some ethnic groups over others, and that test scores are correlated closely with family income. There is only a very slight correlation between SAT scores and first year college grades – and not even an attempt to draw any correlation beyond the first year grade – and studies also generally have shown that the test has the least predictive value for the same groups that tend to score the worst on them. In any case, the SAT / first year grade correlation is LESS than the high school GPA/ college grade correlation – so to the extent that a college might attempt to use objective criteria, logically any college would want to weight GPA and class rank more heavily than SAT scores. (I am sure that any college admissions officer would agree that a class valedictorian with a 4.0 GPA and a 2200 SAT score is a much better qualified candidate for admission than a student from the same high school with a 3.6 GPA and a 2400 SAT score. )</p>
<p>If it is a 2400 single sitting versus a 2350 single sitting, then it’s certainly better. If it was a 2400 superscored, then that might not be the case.</p>
<p>How can a 2400 single sitting on a third sitting after tutoring possibly be “better” than a 2350 single sitting be “better” than a first sitting score? Yes, the number is higher in the same way that if you get to college and are assigned to live in the 5th floor of your dorm, you will definitely be living on a higher floor than your classmates assigned to live on the 4th floor – but “higher” doesn’t mean “better”.</p>
<p>The most you can say is that the two scores are in a similar range, and both students seem highly capable. </p>
<p>The results from two tests administered at different times under different conditions cannot be compared the way you suggest. No scientist would premise a research study on such numbers without attempting to factor in the differential. Whenever educational research is done using a pre-test, post-test model, the researchers always have to factor and control for possible practice effects of the post-test. (Typically this is done with a control group). </p>
<p>It would be inane if any college looked at scores the way you suggest. But as I posted above, they don’t – you could contact any ad com at any college and everyone will tell you they look at test scores “in context” – and part of “context” is factoring in differences in testing conditions.</p>
<p>The funny thing is that it’s possible to get as many as 3 wrong and still get a 2400, while you can also get 2 wrong and get 2350. So I’m not entirely sure that a higher score is always better.</p>
<p>Even with same day, same condition testing, the margin of error on the test is 60-80 points for each section, according to the College Board. [Understanding</a> Your SAT Scores<a href=“%22Your%20test%20score%20represents%20a%20snapshot%20in%20time.%20If%20you%20took%20the%20test%20multiple%20times,%20however,%20that%20number%20would%20likely%20change%20—%20increase%20or%20decrease%20—%20on%20each%20test.%20This%20is%20why%20we%20sometimes%20say%20a%20score%20range%20better%20represents%20your%20true%20ability;%20it%20considers%20multiple%20snapshots%20of%20your%20score%20instead%20of%20just%20one.%20Usually,%20%5Bb%5Dyour%20scores%20fall%20in%20a%20range%20of%20roughly%2030%20to%2040%20points%20above%20or%20below%20your%20true%20ability%5B/b%5D.%20Colleges%20know%20this,%20and%20they%20receive%20the%20score%20ranges%20along%20with%20your%20scores%20to%20consider%20that%20single%20snapshot%20in%20context.%22”>/url</a> </p>
<p>So across three tests that leaves a margin of error for the individual of around 180-240 points.</p>
<p>The College Board also provides specific information about the impact of repeat testing on scores; see <a href=“College Board - SAT, AP, College Search and Admission Tools”>College Board - SAT, AP, College Search and Admission Tools](<a href=“The SAT – SAT Suite | College Board”>The SAT – SAT Suite | College Board)</a></p>
<p>(Only 17% of students take the test 3 or more times; 45% test only once; and 38% take the test twice; of repeat testers, 55% improve their scores, with an average score increase of 14 points.)</p>
<p>The College Board itself makes these recommendations to colleges:</p>
<ol>
<li><p>Know enough about tests and test data to ensure that their proper uses and limitations are understood and applied - including basic statistical principles impacting accuracy and reliability of the score. </p></li>
<li><p>Use SAT scores in conjunction with other indicators, such as the secondary school record, interviews, personal statements, writing samples, portfolios, recommendations, etc., in evaluating the applicant’s admissibility at a particular institution.</p></li>
<li><p>View admissions test scores as contemporary and approximate indicators rather than as fixed and exact measures of a student’s preparation for college-level work.</p></li>
<li><p>Evaluate test results and other information about applicants in the context of their particular background and experience, as well as in the context of the programs they intend to pursue.</p></li>
<li><p>Ensure that small differences in test scores are not the basis for rejecting an otherwise qualified applicant.**</p></li>
<li><p>Guard against using minimum test scores unless used in conjunction with other information such as secondary school performance and unless properly validated.</p></li>
</ol>
<p>** Quote from College Board:
</p>
<p>So the bottom line is that the College Board – which designs the test – says the same thing I do about the value of the test. </p>
<p>I can understand why high school kids stressed out about getting into college would like to believe that an individual score retest has more significance than it does, but the college ad coms are professionals who have a very good understanding of the limitations of the tests – and the College Board puts out materials and conducts workshops and training seminars to make sure that the above points (and more) get across.</p>
<p>^^^^
calmom, for some reason I think that kids that worship the CollegeBoard for developing a test and giving them a goo score, will find a way to ignore itsr analysis when it goes contrary to what they want to believe.</p>
<p>And rather than kids stressing about getting into school, a lot of this stuff is probably posted by kid’s for whom a high SAT score is the major accomplishment of their lives.</p>
<p>Thanks, calmom. I keep failing to remember to dig up the SED data. Now, I think the 14 points that CB gives as the average for improvement is a slightly misleading figure because I think (but certainly have no figures for) that there is a huge variance in score change for retakes. But your point is quite clear.</p>
<p>^^^
First, this is only for the verbal section. Assuming the increases are reasonably close for each section, (I unerstand this is just an assumption) thats 42 over all sections. </p>
<p>Also, with “superscoring” the effect is enhanced. This is an average, averagesome students actually do worse on retake, contributing to the average. However, the only scores considered for admissions woul be improvements.</p>
<p>Also, I don’t believe, this takes into account prep. Where I worked for a couple years -
[About</a> Ivy West](<a href=“ivywest.com”>ivywest.com)
I regularly saw kids increase their scores by over 100 if they just put in a little work. THey can also do this just by studying themselves. Many kids just go in and retake with no prep, so that also scews the data down.</p>
<p>My son’s data sheet for his SATs also says that there is a slightly higher chance that his scores would go down on a retake. I believe that once you hit around 750 or so , that’s the case. And if there are a number of kids in the 750+ range who keep trying for 800, that might drag the average increase down a little bit. Probably, the effect is minor because the absolute number of kids doing this must be small.</p>
<p>
Well, the same document I got that 14 point average from, said that there is a 100 point variance, either way, for about 1% of retakers.</p>
<p>An average is an average - to get a 14 point average it must mean that for a lot of retakers there is little or no variance --which actually tends to validate the original score. That’s how I saw it – my daughter took SAT’s back under the old system in the spring of her sophomore year, got a score we saw as disappointed … retook in the spring of her junior year under the new system, with roughly the same score on the CR/Math parts, significantly higher on writing – and even with the changes in the test, I took the 2nd round as an indication that a substantial change with #3 would be unlikely – and discouraged further testing.</p>
<p>If it is a 2400 single sitting versus a 2350 single sitting, then it’s certainly better. If it was a 2400 superscored, then that might not be the case.</p>
<p>This is the third time I have posted this, because I don’t think you understand what I’m trying to say. After three sittings, a 2400 may not be better than a 2350. I literally said if they are both single sitting, then the 2400 is better. I agree that a 2400 after taking it three times isn’t nearly as impressive as after one try.</p>
<p>Bovertine – re your post #330 – I agree completely. </p>
<p>I just get irked by the idea that these kids who are posting either have managed to attain really strong SAT scores – or are engaging in wishful thinking hoping they will get such scores – and yet seem so deficient in analytical skills when it comes to understanding numbers and their significance. Because I grew up in the stone age, in order to study science at college, we had to demonstrate mastery of the slide rule – and when we were taught how to use the slide rule, we also learned not to expect results that were more accurate than the fuzzy numbers that went in or the rough-result tool we were using to calculate the answer. A whole lot of time spent on the concept of significant figures. </p>
<p>I wonder if a generation raised on calculators and computers is still being taught these basic concepts in high school. </p>
<p>(Also, back in the stone age when I was in school, SAT scores weren’t rounded as they are now… it was very possible to take the test and get a score back of, say, 638 – and 638 would have been considered a very respectable score. But I doubt that anyone would seriously have thought that earning a 638 was significantly different than earning a 630 – we talked about scores in much broader terms those days. As in, “Wow, I got over 600 on both sections!” - where “over 600” meant everything up to “almost 700”. And I remember being very delighted and amazed 4 years later when I got “over 700” on my LSATs, which practically guaranteed admission into any law school I wanted.)</p>
<p>
</p>
<p>I’ve got to admit that I don’t entirely understand what you mean. If you mean a 2400 single sitting score is HIGHER than a 250 single sitting score, then I wouldn’t imagine you even need to post it. If you mean you could wave a magic wand and add 50 points to your score would that be better? Sure. Significantly better? Nobody can say for sure. I sort of doubt it. Worth the effort to study and take the test again? Likely not.</p>
<p>If you mean it necessarily even indicates a better command of the material, I suggest you read calmom’s post 329.</p>
<p>
</p>
<p>Hahalolk – when you get to college, please sign up for a course in statistics – this is especially important if you plan to major in sciences or social sciences.</p>
<p>I appreciate that we are now talking about comparing single, onetime sitting scores. That clarifies the discussion somewhat. </p>
<p>However, a 2400 is “higher” than 2350 but not “better” because of the margin of error for the test. According to the College Board’s own published data a 2400 is <em>the same as</em> a 2310. (In the same way that if the Gallup tracking poll has a 3% margin of error shows candidate A has 46% support this week, and next week reports that candidate A has 48% support – the polls are indistinguishable, no matter how much the cable news pundits might want to make of the 2% “rise” in numbers.) </p>
<p>If two students are being compared – and if one has a 2350 and the other has 2250 – then according to the data specifically published by College Board to explain the numbers – the two students are the same. There is no “significant” difference between those two scores. (If large groups of students are compared – such as the average scores of all students from Nevada compared with the average scores of all students from Maryland… then the range for significance is considerably narrowed. This enables college ad coms to draw valuable information from reported average scores for <em>groups</em> that they could not draw from individuals).</p>
<p>Keep in mind that the word “significant” is a term of art that means something a little different when used to discuss mathematics or statistics, than in ordinary speech. It may be emotionally very “significant” for you if you go in on a retake and increase your overall score by 60 points – you can rightly go out and celebrate your accomplishment – but those 60 points simply are not mathematically significant because of the known margin of error for the test. On any given day, two students who are exactly the same in terms of ability might score 40 points apart on each section of the test – simply because the test itself is not deemed to be all that accurate. Again… my numbers come from the College Board itself. (As a matter of science, its quite possible that its worse than that – what we are missing from the equation is peer review or an outside audit or analysis of the same numbers – that is, College Board has a vested interest in providing best case scenario numbers. )</p>
<p>You seem like a very bright and inquisitive kid. I honestly don’t know how much is taught in high school these days about statistics and understanding concepts like standard error – and as you are only now entering your senior high school I wouldn’t fault you for not yet knowing something you have not yet studied. </p>
<p>But the issue is that the SAT is now and has always been a “blunt tool” – the scores on each subtest have always meant, “X, give or take 30 or so points” – so that a 630 CR score becomes the functional equivalent of a 600-660 score. And again, the college ad coms know this.</p>
<p>It is true that there are sometimes very bright lines drawn when it comes to certain benefits. My son scored exactly at the cut off for National Merit in our state. My son did NOT score significantly better than all the students who scored 1 point below him – there is no way that a single digit up or down on the PSAT can be imbued with that level of accuracy. But because there is a scholarship contest with a particular score cut off - as a practical matter (if not as a mathematical one), my son’s score became very significant. So it would certainly be worthwhile for you or any other student to retake a standardized test in the hope of a marginal increase, if you knew that a particular benefit was tied to achieving a particular score, and that you had a reasonably good chance to achieve that. </p>
<p>It is also true that while there probably is no “significant” difference between scores that are 30 points apart on a given SAT subtest – most schools probably do have a rough cut off, below which they won’t consider students. If you look at the College Board document I linked to, you will see that there is some discussion about why this would be appropriate – not to weigh the abilities of a particular student, but more as a way of the college setting its own minimum standards, based on the cumulative data of the thousands of scores that are submitted. So if you had scores that are low enough to be around that bottom line cut off – I certainly see the value of a retake. Based on the data gleaned from the CDS for various schools, I think that bottom line for the Ivies is roughly around 600, maybe very high 500’s. That doesn’t negate the fact that most admitted students score a lot higher than that. As I and others have pointed out, the students with low end scores on one test probably have higher counterbalancing scores on other tests. I’m just saying that if a student had two high end scores in CR and writing but a math score of 540 – that particular student would probably be well advised to retake and bring up the math score. But if the disparity was more along the way of 750+ on CR and writing, and a 660 in math, there would probably be no particular benefit of a retake – unless, of course, the student was a prospective engineering major.</p>
<p>hahalolk - I think several of us (calmom, boventine and myself at least) do quite well understand what you are saying. We just don’t agree with it. A 2400 is higher than a 2350 - well, yes, we can all do that much math. We all think that such a difference is not significant. Measurement is limited by calibration - and although you can have SAT 1 totaled scored that vary by as little as ten points, that doesn’t mean that those ten points matter at all.
A physical world example - human body temperature. When your temperature is taken in the hospital, the number will be recorded as xxx.yy. Does that mean that a difference in body temperature of .01 degrees matters in any reasonable way? No.</p>
<p>Okay, calmom I think we’re finally on the same page. I agree with you now. I said earlier that 50 point difference isn’t that big, and you are saying the same, so I think we have reached a consensus.</p>
<p>PS: I do understand standard margin or error, it’s just that we were discussing different things.</p>