Could the ACT percentiles be inflated?

Test Preparation
1

<p>Some states like Illinois mandate the ACT, one can assume many will not study for it. If 100,000 or even 200,000 people bomb the ACT like that, can the numbers be inflated?</p>

<p>I find it very hard to believe that only 11% of people who study get a 27 or higher.</p>

2

<p>Yes, the percentiles are being inflated (relative to the SAT) because of that.</p>

3

<p>or do they not take into account the mandated ones?</p>

4

<p>Yes, the percentiles at top level scores are inflated by ~.004 % assuming that 68,000 of the test takers were not college bound (in mandated states, which have about 350,000 total test takers). So the scores are worth about 10 points less than their percentile conversions to the SAT. Note that the definition of college bound is meaningless anyway (people take the SAT all the time without being “college bound”) and 68,000 people is probably a huge over estimation too.</p>

<p>In conjunction, theres a concordance table between ACT and SAT scores that lowers the ACT conversion by about 20 points per score (the extra 10 is just there for no reason) on the SAT because the Collegeboard thought that would be a good idea:
<a href=“http://www.act.org/aap/concordance/estimate.html[/url]”>http://www.act.org/aap/concordance/estimate.html&lt;/a&gt;&lt;/p&gt;

5

<p>

</p>

<p>You’re off by a factor of 1000.</p>

6

<p>

</p>

<p>I said at top scores. The lower the scores, the higher percentile error, but as scores drop on the SAT the percentile difference increases too.</p>

<p>The change always yields around a 8-15 point deduction in terms of SAT score.</p>

<p>I didn’t mean to put the percentile (%) there though.</p>

7

<p>I don’t understand how you’re calculating that figure then. Where’d you get 68,000 from and what did you do with it?</p>

8

<p>I made an arbitrary value of supposed test takers who are “not college bound” and removed them from the total population of ACT test takers. The deduction in score relative to the SAT is small (10 SAT points), regardless of how large the population I remove to about ~200,000 people.</p>

<p>A composite of 34:</p>

<p>1 - (13084/1568835) = .9916</p>

<p>Arbitrary removal of 100,000 people (just for an example):</p>

<p>1- (13084/1468835) = .9911</p>

<p>difference: .00056</p>

<p>This difference yields a comparable score of 2250, rather than 2260 with 100,000 people removed from the total ACT population. </p>

<p>~400,000 people are in mandated states, so an assumption that 25% would not have taken the ACT would result in a 10 points decrease at higher level ACT scores.</p>

<p>This is already reflected in the ACT-SAT concordance table.</p>

9

<p>The percentiles are definitely inflated (in comparison to the SAT), but the practice of forcing everyone in Illinois, Michigan, etc. to take the test is probably not the chief reason for that. I would assume it is more due to the fact that ACT states, as a whole, are simply less high-achieving academically than SAT states. It is a result of different test-taking populations.</p>

10

<p>

</p>

<p>And where is your proof of this?</p>

11

<p>Well you could look at NMS PSAT cut-off scores for each state.
ACT states have lowers numbers.</p>

12

<p>Are you saying that the PSAT is mandatory to take in all ACT states so that the population size matches that of SAT states?</p>

<p>On the actual SAT, Massachusetts has an average score of 1550 and a PSAT of 223 needed for cutoff, while Illinois has as average SAT score of 1780 and a PSAT cutoff of 214.</p>

<p>… Hm, perhaps the size of the population actually matters to understand what the scores in a state mean. Fewer people, fewer qualifying scores, so the score cutoff drops. More people more scores. Score cutoff increases.</p>

13

<p>Wintergreen is absolutely right. 99th percentile on the ACT starts at 32!
99th percentile on the SAT starts at 2200
It’s MUCH easier to score a 32 than it is to score a 2200.</p>

<p>“I would assume it is more due to the fact that ACT states, as a whole, are simply less high-achieving academically than SAT states. It is a result of different test-taking populations.”
Amen</p>

14

<p>^ Actually, 32 is at the 97.1st percentile according to this:</p>

<p><a href=“http://www.act.org/news/data/10/pdf/profile/National2010.pdf?utm_campaign=cccr10&utm_source=profilereports&utm_medium=web[/url]”>http://www.act.org/news/data/10/pdf/profile/National2010.pdf?utm_campaign=cccr10&utm_source=profilereports&utm_medium=web&lt;/a&gt;&lt;/p&gt;

15

<p>While the ACT has a marginal percentile inflation, concordance tables have already taken this in account for a long time so it’s not giving anyone who takes the ACT an advantage anyway.</p>

<p><a href=“http://www.act.org/aap/concordance/estimate.html[/url]”>http://www.act.org/aap/concordance/estimate.html&lt;/a&gt;&lt;/p&gt;

16

<p>^ Unless schools’ internal conversion practices assume that percentiles accurately facilitate concordance.</p>

17

<p>I think the assumptions we’ve made are going off on a tangent of inconsistency.
But I do have to have to agree percentiles are very important in comparing scores. And I just pointed out PSAT cutoff lines as a mere example.
Evidently,
2200 > 32
It takes a lot more work to achieve a 2200 than a 32. : |</p>

18

<p>A 32 is a 97% while a 2200 is a 98% so it makes sense why a 2200 is harder to achieve; a 32 is concordable to a 2120, whereas a 33 is around a 2180 (comparable to a 2200).</p>

<p>

I was under the impression that a college that accepts the ACT would use ACT’s estimated concordance table.</p>

19

<p>I’ve read somewhere that ACT scores are based on percentiles… or is it the other way around?</p>

20

<p>^No. Like SAT scores, ACT scores are not fixed to particular percentiles.</p>