Scoring System on the CAT

Even if you knew the answer to this question... would you spend the time during the test to guess whether the question is experimental or not, and then answer correctly/incorrectly accordingly? THAT would be something to see.

Since you can't do anything about it anyway, just try to answer all questions to the best of your ability. I don't think purposely trying to answer experimental questions wrong in order to get easier questions is worth it.

for the experimental questions to be properly tested and to be statistically sound and with reliable results they are always thrown at you randomly. see my upcoming debrief. as a matter of fact, if there are n experimental questions, it is possible for your first n questions to be experimental by chance alone.


but honestly, do not worry about the algorithim. i can't remember my first quant question today... but i might have gotten it wrong, because question 2 was easy. i don't remember... but dont worry about it. i got a 48Q today. my last gmatprep i got #2 and #3 wrong in Q , but i still got a 46Q

This is just something I was curious about. Obviously, everyone tries to answer every question correctly. Just wanted to know whether anyone had experimented with this while taking the GMAT/Power Prep and figured it out . It wasn't meant to be a very serious question.

why would there be experimental questions on the GMAT/Power prep?

well, they were probably just "non-scoring" questions. but if you look at PowerPrep answers, they only show you 30 answers... so you never know if you got the other 7 or 11 experimental questions right or wrong... kinda annoying.

They have them there to simulate the full test. That's my guess.

I think powerprep had experiemental questions. I think I remember after the test, when i went to check on my results, only seeing like 29 Quant quetsions and 30 something Verbals. I could be wrong... PowerPrep was a weird piece of software for me.

I think that if the practice CATs use the exact same algorithim as the actual test, the practice CATs would most likely have "experimental" questions to keep everything accurate. The questtions just aren't graded.

If someone were good with mathematical models, they could probably take the GMATPrep test hundreds and hundreds (if not thousands or millions) of times, record the results, and get an estimate of what the algorithm is like... but it would still be an estimate , and it would be a pretty fruitless endeavor. If you can perdorm such a feat, you would surely do well on hte Quant section ) But remember, GMAT percentiles and scores are a dynamic thing, so an 750 level question from 5 years ago might be a 700 level now.

i think it'd be easier to disassemble the java bytecode of GMAT Prep and reverse engineer the algorithm...i used to do things like that many years ago

It's tougj