SAT for 8th Graders

Parents Forum
21

<p>i don't know about it being harmless, like one of the previous posts said, if your kid scores lower then they could feel really bad... it might alleviate further stress come junior/senior year... but then again, stress makes us work harder? haha</p>

<p>in all seriousness, the test itself was pretty worthless when i took it, i know for a fact i didn't care at all and i was dozing off. The new test is 4/5 hours long (depending on the proctor) do you really believe a kid who's 12 will sit through the entire thing and gain anything from it?</p>

22

<p>I was kidding about the ability to tell if an 8th grader is Smart........</p>

<p>They are all pretty smart if you ask me. Most 8th graders that I have seen recently have mastered a second language that I do not understand. (It is called "IM") In addition, many are great communicators and can conserve a lot of time and energy by expressing themselves with monosyllabic grunts, as opposed to actually stringing together words to form sentences when speaking to adults. Many are very gifted design engineers, and they can express their artistic creativity in the way they dress. Many have exceptional people skills, and are often seen traveling in packs. They can multi-task much better than most of us. I have personally witnessed an 8th grader that can talk on a cell phone, watch TV, talk on the computer to ten friends, download music, plan a party, and do homework all at the same time!</p>

<p>I guess even if they are not smart, they can be bold and fresh and wonderful! At least mine is.......</p>

23

<p>to be fair, you can im 10 friends with tabbed windows in trillian etc., download music in queue, watch tv, and do your homework on the computer quite easily :) especially if you have irc and a tv tuner lol</p>

24

<p>My son took the SATs twice for CTY eligibility once in seventh grade (computerized) and once in eighth (paper and pencil). The computerized test was a waste of time, but I thought it was good for him to have a free practice on the "real" test. The scores don't count and it gives them a small dose of reality, so I'm planning on doing it with my daughter (eighth grade) this year. Who knows, maybe he did well on his PSAT last year because he'd had this extra experience?</p>

25

<p>dose of reality? it feels like crap when you get a 720-420 split haha</p>

26

<p>"It is much easier for an eight grader to score 500-550 than it will be for a junior to score 650-700. "</p>

<p>Actually, there is no evidence to support this statement. </p>

<p>As an earlier poster said, the whole purpose of the testing is to spread out the tail of the score distribution from age level appropriate testing into a broader distribution.</p>

<p>Julian Stanley, who pioneered this, DID know what he was doing.</p>

27

<p>When I wrote, "It is much easier for an eight grader to score 500-550 than it will be for a junior to score 650-700." NewMassDad countered with, "Actually, there is no evidence to support this statement."</p>

<p>NMD, while I am sure that I could find SOME evidence to support my claim, let's look at the reverse of my point.</p>

<p>Does this statement rings truer, "It is much easier for for a junior to score 650-700 than for an eight grader to score 500-550." and would you not have posted a similar comment? I do not like to present arguments ab absurdo but I feel that I could find much support in the distribution of the SAT scores to debunk the relative ease of obtaining a 650-700. Unless I am mistaken, only 11% of students obtain a 650 score in verbal and 15% in math. </p>

<p>Further, inasmuch as the CC population hardly represents a sample of the entire population, I am quite certain that very few CC students scored below 500 on early SAT. FWIW, I also believe that there is a huge difference in students scoring 500-550 on an early SAT and the group that most interested Julian Stanley when he started his SET program that identified students who scored above 700 on the SAT Math before the age of 13. At that level, the correlation between math or verbal scores and middle school basic instruction is probably nil, if not negative. On the other hand, I am quite certain that the correlation increases for lower scores, hence the relative ease to score an above average score without being well above average or gifted. </p>

<p>Now, let's go look for some "verifiable" ammunition to shoot back at NMD. :)</p>

28

<p>well lessse- my daughter recieved a 1050 on her 7th grade SAT-
don't remember break out- since her junior year SAT math was 590- I think assuming her math scores 4 years prior- couldn't have been higher than 400. which would give her verbal 650.
interesting -
this kid didn't even qualify for the entry level gifted program in her school district ( by their testing) but qualified to take a supposed comparable college course in middle school ;)</p>

<p>I don't know if this says much about the accuracy of the SAT to predict readiness for college- but I think it does say something about the test that the district uses to qualify students for the gifted program.</p>

<p>( I didn't really care what she recieved on the test- Ididn't even know about testing middle school kids- till the head of her middle school suggested that her ERB scores qualified her to participate- she did participate in a snail mail writing class in 8th grade- but really since her school had enough challenge for her- and she spent her summers doing non academic work- taking an extra course wasn't necessary- for some kids though, especially if their schools isn't able to give them enough challenge- the classes can be quite interesting- and I know many kids who really enjoyed the summer programs)</p>

29

<p><a href="http://www.collegeboard.com/research/pdf/evaluatating_the_appr_10505.pdf%5B/url%5D"&gt;http://www.collegeboard.com/research/pdf/evaluatating_the_appr_10505.pdf&lt;/a&gt;&lt;/p>

<p><a href="http://www.tip.duke.edu/talent_searches/grade_7/7TSResultsSummary.pdf%5B/url%5D"&gt;http://www.tip.duke.edu/talent_searches/grade_7/7TSResultsSummary.pdf&lt;/a&gt;&lt;/p>

30

<p>My son first took SAT in 6th grade for CTY. It was a nice experience for him, just a "test" with no preparation and no stress. It provided a benchmark for when he took the exam again the following year. Typical improvement for smart kids was 100 to 150 points per year from the base year. He rapidly moved asymptotically toward the limit in math by 9th grade, but still had room to improve in verbal. So by the time he took the exam "for real" in 11th grade, he had virtually maxxed out on SAT I and II. However, doing well early let him know that he was "college entrance" material since he was already well above the median for 12th graders. To be frank, we wasn't thinking about college at age 12. Taking the exam was probably more of a marker for us, his parents, than for him. Time to make sure we were saving enough money.</p>

31

<p>xiggi,</p>

<p>I think I see part of where we differ. </p>

<p>For the person eligible for talent searches, it is, of course, easier for them to get a 500-500 than for the average HS senior to get a 650-700. But that is an apples to oranges comparison. Your statement was "an eight grader", not a talent search participant. Now, if you really meant any ole eight grader, and the rest of your post suggests this, you are mistaken. Keep in mind that the eligible pool is the top 3% of the grade cohort. And even they, on the average, do more poorly than the college bound cohort that takes the SAT. In fact, their probability of scoring in the 500-540 range is only twice that of a college bound HS kid scoring 650-700!</p>

<p>Your other statement, "FWIW, I also believe that there is a huge difference in students scoring 500-550 on an early SAT and the group that most interested Julian Stanley when he started his SET program that identified students who scored above 700 on the SAT Math before the age of 13." is of course true. But so what? In fact, that's the whole reason for using the SAT. The out of level testing enables one to sort out the kids bunched at the high tail of normal testing.</p>

<p>"At that level, the correlation between math or verbal scores and middle school basic instruction is probably nil, if not negative." Is largely true. It is nil, not negative. What the out of level testing appears to measure has nothing to do with past instruction. (Trust me. I discussed this with some of Julian's researchers.) But, the same observation holds for much further down the scale. </p>

<p>BTW, for those that look at the CB research in the link in post 29, be aware that the data presented is combined 7th and 8th grade. The actual score distributions for 7th graders are shifted lower than for 8th graders, by a good amount.</p>

32

<p>
[quote]
As an earlier poster said, the whole purpose of the testing is to spread out the tail of the score distribution from age level appropriate testing into a broader distribution.

[/quote]
</p>

<p>This is a correct statement of what the talent search programs think they are doing when they give SAT or ACT tests to middle-school-age students. But, alas, this is based on a statistical fallacy. In fact, ANY two tests will sort test-takers into a different rank order, and thus you will have a spread of scores (a "bell curve") on the second test if you take a sample from ANY narrow level of scores on the first test: high, middle, or low. </p>

<p>The point above is well explained in an online Power Point presentation by a Ph.D. psychologist, which, alas, is best viewed in Microsoft Internet Explorer with the optional Power Pointer viewer installed. </p>

<p><a href="http://www.iapsych.com/Forrest_files/frame.htm%5B/url%5D"&gt;http://www.iapsych.com/Forrest_files/frame.htm&lt;/a> </p>

<p>The bottom line is that we can be sure that there are some undetected young people who would score very high on the SAT at amazingly young ages who don't come forward to take the talent search tests because they are just a tad too low in their scores on in-grade achievement tests. Some tests are easy for one test-taker, and some for another.</p>

33

<p>tokenadult,</p>

<p>Do you actually understand what your link is saying? I should quote from the final slide:</p>

<p>"For most children with cognitive disabilities (those with below average IQ scores), it is NOT possible to predict individual levels of expected achievement with the degree of accuracy that would be required to deny a child the right to the high standards/expectations "&lt;/p>

<p>I doubt that any competent psychometrician would argue with that. Nor do I expect many educators to argue that test performance correlates very well with achievement, any more than ability in anything correlates very well with output - you have a lot of slackers and a lot of hard workers.</p>

<p>But the talent search discussion is not about that. And they are NOT based on a statistical fallacy, Forest Gump analogies or not.</p>

34

<p>"(Trust me. I discussed this with some of Julian's researchers.)"</p>

<p>Oh, I do trust you. I remember from a past thread on a similar subject that you discussed this issue with the SET researchers before dediding on the benefits of accelerated programs for your D. And that, in spite of this, I would still decide to argue the subject must represent some kind of death wish. :)</p>

<p>Well, since I am not afraid of being shot down in flames, allow me to wonder about some finer points: </p>

<p>*At that level, the correlation between math or verbal scores and middle school basic instruction is probably nil, if not negative." Is largely true. It is nil, not negative. *</p>

<p>Here I assumed that the were would not be much correlation between the "instruction" and the math and verbal scores. Isn't it correct that there are cases of negative correlation between math and verbal scores? </p>

<p>*"What the out of level testing appears to measure has nothing to do with past instruction. But, the same observation holds for much further down the scale." *</p>

<p>Here again, I assume that basic instruction should account for a fraction of the "primary points" being from 200 to 450 in average, and from 200 to 550 for selective high schools.</p>

<p>Lastly, regarding the pools. Are you positive that that the statistics quoted by SV2 only compare an "eligible pool is the top 3% of the grade cohort." with the entire cohort. I understand that a comparison of 200,000 students taking the early tests loses value when the standard is established for an overall population of 2,000,000. However, for some reason, I do not see how the statistics for the early testers represent the TOP 3% of the available population. In other words, I believe that the samples may have different sizes and different attributes, but that it may NOT represent such a high selective number (97 percentile). </p>

<p>If the early testers really represented the top 3% of the student cohort for that grade, the comparisons between the early scores and the senior scores would HAVE confirmed your point by a huge margin. If 15% of seniors earn a score oof 650 but only 20% of the sample of early testers score a 500, the comparison really becomes 15% versus 3% of 20% or .6% when controlling the population. </p>

<p>I realize that the TIP program is atypical. However, I have good reasons to believe that not all the students who took a SAT in 7th or 8th grade had been "qualified" or "preselected". I know that at the middle school I attended, everyone had to sit for the Duke TIP. </p>

<p>Am I missing something?</p>

35

<p>I am completely aware of what the Power Point presentation I recommended is saying, and I wish I could point to an online site that makes the same STATISTICAL point with such visual clarity in the context of discussing gifted education. I am still looking for such a site. But the point is general, and it is easily found in any GOOD book about statistics. (Indeed, a rather similar point appears in one of the late Professor Stanley's books about educational testing, written when talent search testing had hardly begun.) I refer online to sources that are easy to look up online. (You've already had time to look up the reference, and I appreciate your comments about it). I will stand by my original statement: there are young people who are fully capable of scoring well into the middle of the pack, or even into the high end of the pack, on the SAT or the ACT at middle-school age who do not obtain scores in the top few percent on in-grade standardized achievement tests. Different item content favors different test-takers. WHICHEVER test you give first, then selecting out the highest scorers on that test to take a second test, you will find a range of scores on the second test. </p>

<p>To be sure, one particular individual who scores high on all tests he takes (I have known a few individuals like that over the years) has a very strong Bayesian inference case to make that he will score high on the next test, and on the test after that, and so on. But it's not a sure thing: not all of the "math-high" kids in the Study of Exceptional Talent score particularly well on the AMC tests, for example. Not all young people with high SAT scores at a young age grow up to be college graduates with high GRE scores. And so on and so on. </p>

<p>I have a "gut reaction" about some of the various tests that middle-school-age students take that I will have to test empirically this school year. Xiggi may be on to something, at least as it pertains to standard scores on tests given generally to older students.</p>

36

<p>
[quote]
WHICHEVER test you give first, then selecting out the highest scorers on that test to take a second test, you will find a range of scores on the second test.

[/quote]
This is absolutely true, and applies as well if you select out the lowest scorers to take a second test. And in fact, on average what will happen to the group of high (or low) scorers is that their second test scores will "regress to the mean."</p>

<p>This principle applies not just to test scorers, of course. It applies equally well to the batting averages of baseball players.</p>

37

<p>". I will stand by my original statement: there are young people who are fully capable of scoring well into the middle of the pack, or even into the high end of the pack, on the SAT or the ACT at middle-school age who do not obtain scores in the top few percent on in-grade standardized achievement tests."</p>

<p>Of course. WE have no argument on that statement. And a version of the same issue is why some NM semifinalists do not make finalist. It is also why some kids that qualify to take talent search exams do very poorly on the SAT.</p>

<p>Because some aspects of testing are statistical - subject to random variation, it is arguable statistically that a kid could score way below average in one sitting of a test and way above average on another sitting. Possible, just not very likely. </p>

<p>An important statistic for any standardized test is test reliability - how close the scores will be for someone repeating the test. Knowing this calculation, one could even calculate the probability of such a score rise. BTW, this is not Bayesian.</p>

<p>Xiggi,</p>

<p>I have no doubt some schools test more broadly. Most do not. I suspect, but do not know for certain, that the searches have an agreement or understanding with the CB regarding this. I wish I could find stats regarding how a random distribution of 7th graders would do on the SAT. I don't. But other data from the CB, if I recall correctly, shows about a 50 point rise in the median score per year for a random sample of kids. Working backward, you can imagine the average would be pretty low. </p>

<p>It is also interesting to note on the CB study linked above how different the score distribution is for talent search kids compared to HS kids. I'm not enough of a statistician to know what to conclude. </p>

<p>Finally, Xiggi, the math performance of the high scorers is rather amazing when you consider most of them have never been taught any of the skills necessary to solve the problems in a formal way. In fact, Stanley's work found most of them could not tell you how they knew the answer, only that they DID know it! I realize it may surprise some folks here to hear that most of the high math scorers had not received accelerated math instruction, but it is true.</p>

38

<p>FWIW, College Board data on 7th and 8th graders from 1999-2000 and 2000-2001. Previous data was from 1997-1998 and 1998-1999.</p>

<p><a href="http://www.collegeboard.com/prod_downloads/about/news_info/cbsenior/yr2002/pdf/four.pdf%5B/url%5D"&gt;http://www.collegeboard.com/prod_downloads/about/news_info/cbsenior/yr2002/pdf/four.pdf&lt;/a&gt;&lt;/p>

39

<p>NMD, I am not sure that finding the exact numbers is really worth our mutual time and energy. At this stage, I'll settle for your explanation and ... an unmistakable gut feeling to be correct. If, at a future date, I find data that supports your position -or mine- I'll make sure to let you know. </p>

<p>Without going through too much research, I believe that the number of total students who supply SAT scores to the TIP should be available somewhere. The reports quoted by SV2 indicate "The Results Summary reflects the 72,906 scores received by the Duke University Talent Identification Program Duke (TIP) which represent approximately 90% of all the 2004–2005 Talent Search applicants." I read in another report -about starting a TIP in Spain that the number in the 90s was closer to 60,000 a year**. It seems that the data from TCB shows a number well above 130,000 per annum. </p>

<p>Number tested SAT SAT
Grade in 1991–2000 & 2000–01 ==== FWIW, a very rare error by TCB in a report
7th graders 202,551
8th graders 65,382
Both grades combined 267,933</p>

<p>We'll figure this one soon or later.</p>

<p>** An interesting element of that Spanish report was that there was a great gender difference in students scoring above 500. Surprisingly the GIRLS did much better than the boys.</p>

40

<p>There are several regional TIPs (talent identification programs) around the country. Duke's report refers only to those who participate in Duke's TIP. They present data on 90% of their (Duke's) participants, because they had not yet received all the scores at the time they put together that report.</p>

<p>In addition to Duke's, the following talent searches also allow younger students who score highly on a grade level standardized test to take and submit an SAT score for consideration for their programs: CTD - Northwestern, CTY - Johns Hopkins, & Rocky Mountain Talent Search - University of Denver. The College Board has SAT scores for all the 7th and 8th graders (and younger students) who take the test, regardless of regional considerations.</p>