Help needed with practice score statistics

[shanser.be]

Hi folks,

Firstly before i begin with my question, i want to thank MGMAT for everything...

Back to my question, :-) . I have registered for the GMAT on the 5th of April 2012 and have been studying since December 2011. I seem to be making progress, but have a few doubts.

Stats:
No practice
Dec 1st MGMAT 1:- 540 Q(37) V(28) {rc- 6/12: cr- 8/14: sc-5/15}

Worked out first 100 PS and DS sums from OG12
Jan 1st MGMAT 2:- 610 Q(46) V(29) {rc- 5/12: cr- 8/14: sc-8/15}

MGMAT SC guide basics, Solved first 100 SC's from OG12
Jan 29th MGMAT 3:-640 Q(44) V(34) {rc- 5/12: cr- 9/14: sc-9/15}

Worked out first 60 CR's from OG12
Feb 4th GMATPrep V1 test 1(the old executible, 2005-2006 version)
650 Q(49) V(29) {rc- 8/14: cr- 7/11: sc-9/16}

Now im wondering why did my verbal score plummet... i got more rc's correct, the cr's seem constant & sc's took a tiny beating(i was exhausted by the 10th verbal question for some reason...)

1.)Is the MGMAT test's verbal score algorithm more liberal? i fear that im being mislead
2.)Also is my latest quant score of 49 not representative of the real thing, cause its an old version of GMATPrep?
3.)Also, the number of sc's cr's and rc's are constant on MGMAT, but different in the GMATPrep V1. Is this because of the old GMATPrep software?

Can you help me figuring out this discrepency?, i know its asking too much, but i really appreciate your help.

Further stats:
Feb 4th GMATPrep V1 test 1 Quant : timing was perfect, also sums were less tideous than the MGMAT ones.

Feb 4th GMATPrep V1 test 1 Verbal section.
Last 13mins had 9 questions to go!
1-CR- CORRECT
2-SC- WRONG
3-CR- CORRECT
4-RC- WRONG
5-RC- WRONG
6-RC- WRONG
7-RC- WRONG
8-CR- CORRECT
9-CR- CORRECT
10-RC- CORRECT
11-RC- CORRECT
12-RC- CORRECT
13-CR- CORRECT
14-SC- CORRECT
15-RC- WRONG
16-RC- CORRECT
17-RC- CORRECT
18-RC- CORRECT
19-SC- WRONG
20-SC- WRONG
21-SC- WRONG
22-CR- WRONG
23-SC- CORRECT
24-SC- CORRECT
25-RC- WRONG
26-RC- CORRECT
27-RC- CORRECT
28-SC- CORRECT
29-CR- WRONG
30-SC- WRONG
31-SC- CORRECT
32-SC- WRONG
33-CR- CORRECT
34-SC- CORRECT
35-SC- WRONG
36-CR- WRONG
37-SC- CORRECT
38-SC- CORRECT
39-SC- CORRECT
40-CR- CORRECT
41-CR- WRONG

i cant seem to figure out what affected (or effected :-) ) my score from MGMAT 3 to this GMATPrep V1 old test...

Regards,
Shanser

PS: i took the GMATPrep V1 test 1(the old executible, 2005-2006 version) cause someone told me the UI of the real test was blue in color and i was getting used to the MGMAT grey color... dont want a color surprise coming in the way of my hopeful 700...

[StaceyKoprince]

It's hard to compare two different practice tests (from different sources). There is already a standard deviation for any test just within that test itself (on the real test, the SD is about 30 points). Then add in the variability that you have between two different sources... it's not unusual to see what you would think is a very large difference, but it's actually not a huge difference from a statistical standpoint.

Plus, there are any number of other reasons why your score might fluctuate. Timing problems are a big source - if you're having to rush, then you're going to make more careless mistakes, and that introduces a great deal of uncertainty into your test scores. Mental fatigue can also play a large role. If you skipped the essays on one test but took them on the other, if you took longer breaks than allowed - basically, if you give yourself an artificial advantage, then you might wind up with an artificially inflated score.

But you can't add up the number right or wrong and tell - nor can you even just look at the list of which were right and wrong. The algorithm is way more complex. It can't be processed by a human brain (and, even if it could, we'd need to know a lot more data, such as the specific difficulty curves for each item).

I think that the older version of GMATPrep was built on the same algorithm, so it's accurate in that sense, but the scoring requirements (what it takes to achieve a certain score) have changed over time. There are more and more people taking the exam who are more highly educated, and in particular the increase in the number of non-US students taking the test has increased the performance on the quant side of things. So the knowledge and skill it took to get a 49 in 1995 will probably get you a lower score today. The old test, though, will still give you the score on the old scale because it doesn't know that things have changed.

Short answer: yes, it's possible that the 49 on the old GMATPrep is inflated.

You can take a new GMATPrep instead or you can also do GMAT Focus (www.gmatfocus.com) to see whether you really can hit that scoring range.

Next, you can use this article to analyze your MGMAT tests:
http://www.manhattangmat.com/blog/index ... ice-tests/

And if you're concerned about a score drop, you can use this article to help you figure out what might have happened:
http://www.manhattangmat.com/blog/index ... ent-wrong/

Note that usually we only use that last article for very substantial score drops - equivalent to 80-100+ points in the overall score. But the general ideas apply and may give you some ideas about why your score is fluctuating, whether your top scores are truly representative or whether they might be artificially inflated, and so on.

Finally, nice work so far! Keep up the great work!

[shanser.be]

Hi Stacey,

Thank you so much(not countable :-) ) for your post. I think i will stop searching for the algorithm and just focus on how to improve my weakness.

I gave MGMAT 4 today and got 640 Q 46 V 32 {rc- 5/12: cr- 8/14: sc-9/15}.

On review, found that i wasted 7 mins and 4 mins for 2 quant sums. (Hotel California (2) & Circular Track (2) )... and got them both wrong !!!. Ended up guessing the last 6 questions. Will take the tennis game advice from now on.

Verbal, i think the next time i should be more relaxed and give it my best shot. With 75 mins of pure concentration i should be able to make fewer mistakes. I still have not practiced RC's, so thats room for improvement.

Again, thanks a lot for your post, i really appreciate it.

Regards,
Shanser

[StaceyKoprince]

I think i will stop searching for the algorithm

You can't find it anyway, no matter how hard you search. It's not publicly available and is, in fact, a very closely guarded secret. :)

Good luck - let us know how it goes!