Discrepancy between different practice test scores

Post Reply
schwiss
Posts: 108
Joined: Fri Aug 14, 2009 3:49 am

Discrepancy between different practice test scores

Post by schwiss » Tue Sep 28, 2010 3:39 pm

Hey.

Last sunday I did the 0177 practice test, and got an incredibly bad score of 620 (41%). Today I did the 8677 test and got 710 (69%), a difference of 28% and a difference between a loser and okay-er. What's curious is that the test conditions were very much the same and both were timed. The only difference is that during my 2nd test I took a *** break in the middle with the clock stopped (but did not look at the next question before break, so there was no help). I also did no studying at all between them, apart from going over the answers of the first test.

Now my actual point was not to describe my shitting habits during tests but ask if the percentile of the second test (calculated 1981-1984) is anymore accurate or not. Eg, was my better score because the quality of test-doers was worse in 80s or was it really that helpful to go through the wrong answers.

I mean if it was, and of course because as a physicists we always assume linearity, I could expect to get something like 97 percentile in my next practice exam and something scary in the actual exam.


Summary: Have you noticed a trend that older tests give better scores, or not?

User avatar
HappyQuark
Posts: 762
Joined: Thu Apr 16, 2009 2:08 am

Re: Discrepancy between different practice test scores

Post by HappyQuark » Tue Sep 28, 2010 4:55 pm

schwiss wrote:Hey.

Last sunday I did the 0177 practice test, and got an incredibly bad score of 620 (41%). Today I did the 8677 test and got 710 (69%), a difference of 28% and a difference between a loser and okay-er. What's curious is that the test conditions were very much the same and both were timed. The only difference is that during my 2nd test I took a *** break in the middle with the clock stopped (but did not look at the next question before break, so there was no help). I also did no studying at all between them, apart from going over the answers of the first test.

Now my actual point was not to describe my shitting habits during tests but ask if the percentile of the second test (calculated 1981-1984) is anymore accurate or not. Eg, was my better score because the quality of test-doers was worse in 80s or was it really that helpful to go through the wrong answers.

I mean if it was, and of course because as a physicists we always assume linearity, I could expect to get something like 97 percentile in my next practice exam and something scary in the actual exam.


Summary: Have you noticed a trend that older tests give better scores, or not?
I can't comment on the scoring/scaling of the tests but I can say after having taken all 4 exams myself that I noticed a very significant difference in terms of content between exams 8677 and 0177. It seems to me that ETS did a better job of keeping test takers from using dimensional analysis and limits on the 0177 test. For example, I recall a number of problems on the 0177 test in which the possible solutions only differed by some constant coefficient whereas it was extremely common on the previous tests to have all possible solutions be distinct from the others by some important variable involved in the problem. This could all be in my head, but based on the amount of time that passed between the 86 and 01 tests, it really would surprise me if the tests weren't significantly different.

schwiss
Posts: 108
Joined: Fri Aug 14, 2009 3:49 am

Re: Discrepancy between different practice test scores

Post by schwiss » Tue Sep 28, 2010 5:40 pm

HappyQuark wrote:
schwiss wrote:Hey.

Last sunday I did the 0177 practice test, and got an incredibly bad score of 620 (41%). Today I did the 8677 test and got 710 (69%), a difference of 28% and a difference between a loser and okay-er. What's curious is that the test conditions were very much the same and both were timed. The only difference is that during my 2nd test I took a *** break in the middle with the clock stopped (but did not look at the next question before break, so there was no help). I also did no studying at all between them, apart from going over the answers of the first test.

Now my actual point was not to describe my shitting habits during tests but ask if the percentile of the second test (calculated 1981-1984) is anymore accurate or not. Eg, was my better score because the quality of test-doers was worse in 80s or was it really that helpful to go through the wrong answers.

I mean if it was, and of course because as a physicists we always assume linearity, I could expect to get something like 97 percentile in my next practice exam and something scary in the actual exam.


Summary: Have you noticed a trend that older tests give better scores, or not?
I can't comment on the scoring/scaling of the tests but I can say after having taken all 4 exams myself that I noticed a very significant difference in terms of content between exams 8677 and 0177. It seems to me that ETS did a better job of keeping test takers from using dimensional analysis and limits on the 0177 test. For example, I recall a number of problems on the 0177 test in which the possible solutions only differed by some constant coefficient whereas it was extremely common on the previous tests to have all possible solutions be distinct from the others by some important variable involved in the problem. This could all be in my head, but based on the amount of time that passed between the 86 and 01 tests, it really would surprise me if the tests weren't significantly different.
Now that I think about it, this seems very reasonable. I'm much better at using dirty tricks and reasoning than remembering formulas. Too bad the trend seems to go the wrong way...

Can anyone who has done PGRE lately confirm it's more like 0177 than 8677?

User avatar
HappyQuark
Posts: 762
Joined: Thu Apr 16, 2009 2:08 am

Re: Discrepancy between different practice test scores

Post by HappyQuark » Tue Sep 28, 2010 5:59 pm

schwiss wrote:
HappyQuark wrote:
schwiss wrote:Hey.

Last sunday I did the 0177 practice test, and got an incredibly bad score of 620 (41%). Today I did the 8677 test and got 710 (69%), a difference of 28% and a difference between a loser and okay-er. What's curious is that the test conditions were very much the same and both were timed. The only difference is that during my 2nd test I took a *** break in the middle with the clock stopped (but did not look at the next question before break, so there was no help). I also did no studying at all between them, apart from going over the answers of the first test.

Now my actual point was not to describe my shitting habits during tests but ask if the percentile of the second test (calculated 1981-1984) is anymore accurate or not. Eg, was my better score because the quality of test-doers was worse in 80s or was it really that helpful to go through the wrong answers.

I mean if it was, and of course because as a physicists we always assume linearity, I could expect to get something like 97 percentile in my next practice exam and something scary in the actual exam.


Summary: Have you noticed a trend that older tests give better scores, or not?
I can't comment on the scoring/scaling of the tests but I can say after having taken all 4 exams myself that I noticed a very significant difference in terms of content between exams 8677 and 0177. It seems to me that ETS did a better job of keeping test takers from using dimensional analysis and limits on the 0177 test. For example, I recall a number of problems on the 0177 test in which the possible solutions only differed by some constant coefficient whereas it was extremely common on the previous tests to have all possible solutions be distinct from the others by some important variable involved in the problem. This could all be in my head, but based on the amount of time that passed between the 86 and 01 tests, it really would surprise me if the tests weren't significantly different.
Now that I think about it, this seems very reasonable. I'm much better at using dirty tricks and reasoning than remembering formulas. Too bad the trend seems to go the wrong way...

Can anyone who has done PGRE lately confirm it's more like 0177 than 8677?
Here is another thread on that topic: http://www.physicsgre.com/viewtopic.php?f=1&t=984

schwiss
Posts: 108
Joined: Fri Aug 14, 2009 3:49 am

Re: Discrepancy between different practice test scores

Post by schwiss » Tue Sep 28, 2010 6:02 pm

Thanks.

And damn.

User avatar
HappyQuark
Posts: 762
Joined: Thu Apr 16, 2009 2:08 am

Re: Discrepancy between different practice test scores

Post by HappyQuark » Tue Sep 28, 2010 6:36 pm

schwiss wrote:Thanks.

And damn.
Agreed. If it's any consolation, one thing I did notice about exam 0177 is that most of the problems can be solved with some sort of trick or clever reasoning, provided you see the trick. The problem with the 0177 exam is that they've done a much better job of hiding those tricks from the test taker and it's rough, if not highly stress inducing, deciding whether you want to spend that precious minute and a half you get per problem on solving it mathematically or searching for a clever hint.

geshi
Posts: 200
Joined: Tue Dec 01, 2009 12:01 am

Re: Discrepancy between different practice test scores

Post by geshi » Wed Sep 29, 2010 12:06 am

I can't remember which exam it was, but there was one exam where I score something stupid like 70th percentile in practice. The other exams, not nearly as well (maybe like 40-50th?). It was definitely one of the older exams (one of the oldest 2). I threw out all my old study material, so I can't look it up.

User avatar
grae313
Posts: 2296
Joined: Tue May 29, 2007 8:46 pm

Re: Discrepancy between different practice test scores

Post by grae313 » Wed Sep 29, 2010 4:04 pm

My friend took the PGRE last October and he said that almost every question could be answered with limits, units, or some other clever insight/trick. So not all of the new tests are like 0177. Being more familiar with what timing/pace you'd need and the way the questions are asked could account for some of that difference, but probably not all.

schwiss
Posts: 108
Joined: Fri Aug 14, 2009 3:49 am

Re: Discrepancy between different practice test scores

Post by schwiss » Fri Oct 08, 2010 7:01 am

Yeah buddy, 80 percentile on the last practice exam (sadly, it was 9277). Still, lightweight baby!

CarlBrannen
Posts: 381
Joined: Mon May 24, 2010 11:34 pm

Re: Discrepancy between different practice test scores

Post by CarlBrannen » Fri Oct 08, 2010 3:03 pm

Funny, I did best on the 0177. I had driven to the university where I will take it for real tomorrow, and took the test in the same building at the same time. I took a bathroom break but it was on the clock.

I would really like to ace this test because I would hope to get a fellowship for my first year. I'm considering eliminating the bathroom time. Adult diapers.

User avatar
HappyQuark
Posts: 762
Joined: Thu Apr 16, 2009 2:08 am

Re: Discrepancy between different practice test scores

Post by HappyQuark » Fri Oct 08, 2010 9:22 pm

CarlBrannen wrote:Funny, I did best on the 0177. I had driven to the university where I will take it for real tomorrow, and took the test in the same building at the same time. I took a bathroom break but it was on the clock.

I would really like to ace this test because I would hope to get a fellowship for my first year. I'm considering eliminating the bathroom time. Adult diapers.
Trust me that is a bad idea. I thought I was being clever when I tried to use the diapers but you won't believe how distracting it is squishing around in your own waste for half the test.

CarlBrannen
Posts: 381
Joined: Mon May 24, 2010 11:34 pm

Re: Discrepancy between different practice test scores

Post by CarlBrannen » Sat Oct 09, 2010 10:52 pm

There was a correlation among my practice test scores that I didn't notice until I took the October test just now. I definitely got better scaled scores on the tests that had easier questions. This is even more true if I take into account how much study and trivia I'd memorized at the time of taking the tests. I took the 8677 just to see what kind of questions were on it, then the 0177 after a little studying. When I took the 9677 after quite a bit of studying I was 50 points below the 0177 score. It was depressing.

Here's the raw scores required for reports:

Code: Select all

      500 600 700 800 900 990
8677   18  32  45  59  72  84 = 98%
9277   16  28  40  53  65  76 = 97%
9677   07  19  32  44  56  67 = 97%
0177   16  30  44  58  73  85 = 98%
      500 600 700 800 900 990
It might be useful to see how your scores on the four tests correlate with difficulty. My problem on the more difficult tests was time management.

I'm curious about the 990 area, so I've also added the percentage reported for the 990 score. Interestingly, it seems that the tougher tests, i.e. 9277 and 9677, get maxed out by 3% instead of 2%. I think that there's certain number of people who can answer almost all the questions no matter how ridiculously hard they are. I would guess that in terms of score distribution, the test I took will be similar to the 0177.
HappyQuark wrote:Trust me that is a bad idea.
Hmmm. It's probably worth it to pay for the most absorbent brand. Maybe this would run into issues with: "Disruptive behavior in any form will not be tolerated; the test administrator has sole discretion in determining what constitutes disruptive behavior."

Anatoliy
Posts: 3
Joined: Tue Oct 21, 2008 4:03 am

Re: Discrepancy between different practice test scores

Post by Anatoliy » Sun Oct 10, 2010 7:32 am

CarlBrannen wrote:

Code: Select all

      500 600 700 800 900 990
8677   18  32  45  59  72  84 = 98%
9277   16  28  40  53  65  76 = 97%
9677   07  19  32  44  56  67 = 97%
0177   16  30  44  58  73  85 = 98%
      500 600 700 800 900 990
Thanks for this correlation table. It is interesting how it looks like for April 10, November 09 and October 09 tests. Specially columns "900" and "990".

User avatar
quizivex
Posts: 1031
Joined: Tue Jan 09, 2007 6:13 am

Re: Discrepancy between different practice test scores

Post by quizivex » Sat Oct 15, 2011 1:07 am

CarlBrannen wrote:There was a correlation among my practice test scores that I didn't notice until I took the October test just now. I definitely got better scaled scores on the tests that had easier questions.
Interesting, my experience was the opposite. 0177 was a disaster for me. I thought there were lots of really hard questions (the last ~10 or so were especially miserable. The ones I remember off the top of my head were the questions on the power dissipated by an expanding sphere of charge and the angular spread of a refracted beam). Even the "easy" problems on that test seemed to have tricks that would make me mess up factors of two (and of course those wrong answers would always be one of the choices).

On the other hand, I found it a breeze to get more than the 67 needed for 990 on the 9677.

I think in theory, there are two ways people get questions wrong... a) don't know how to do the problem right, and b) stupid mistake (factor of two, minus sign, giving radius instead of diameter etc).

For me, I was much more likely to do type b) mistakes than type a). The chance of making a stupid mistake is probably independent of difficulty, so for me the number of questions I'd get wrong on a PGRE is fixed (independent of difficulty) to first approximation. Therefore, I was much more likely to do well on a "hard" test with a generous curve than an easy test where you needed 85 fora 990.

norma19
Posts: 1
Joined: Mon Oct 24, 2011 5:31 am

Re: Discrepancy between different practice test scores

Post by norma19 » Mon Oct 24, 2011 5:33 am

And how big of a difference was your score from the FIRST practice test you ever took of the MCAT, and the final score you got on the actual MCAT (after all the studying, 20 or however many practice tests you took, etc.)?

CarlBrannen
Posts: 381
Joined: Mon May 24, 2010 11:34 pm

Re: Discrepancy between different practice test scores

Post by CarlBrannen » Mon Oct 24, 2011 12:26 pm

quizivex wrote:
CarlBrannen wrote:For me, I was much more likely to do type b) mistakes than type a). The chance of making a stupid mistake is probably independent of difficulty, so for me the number of questions I'd get wrong on a PGRE is fixed (independent of difficulty) to first approximation. Therefore, I was much more likely to do well on a "hard" test with a generous curve than an easy test where you needed 85 fora 990.
And I'm the opposite. I can grind out huge numbers of simple calculations without error (or more accurately, with good error checking), so I tend to do best on time pressured tests where you want to get a lot of questions right. I get tripped up on tests where there are too many complicated questions because managing time is very difficult; I hate to leave a question unanswered.

freda88
Posts: 1
Joined: Thu Oct 27, 2011 10:09 am

Re: Discrepancy between different practice test scores

Post by freda88 » Thu Oct 27, 2011 10:20 am

CarlBrannen wrote:And I'm the opposite. I can grind out huge numbers of simple calculations without error (or more accurately, with good error checking), so I tend to do best on time pressured tests where you want to get a lot of questions right. I get tripped up on tests where there are too many complicated questions because managing time is very difficult; I hate to leave a question unanswered.
Fully agree, I support your position phone Card Calling Cards



Post Reply