Expeimental Questions

[Levent-g]

Hi All,

is there anything to know about experimental questions, besides that they won't affect your score? Like they don't appear in the first or last questions etc.?

Thanks

Levent

[StaceyKoprince]

They can appear anywhere. They have nothing at all to do with your score or with your performance in the section. You may be scoring in the 95th percentile and get a question that will eventually be rated 5th percentile or vice versa.

You will not be able to recognize them or tell which ones are experimental.

There's some limit to the # that you can get in a row, or clustered at the beginning or end of the test, but we don't know what the limit is (that information has not been published).

If you run out of time and do not answer all of the questions, then ALL remaining "blank" questions will count against you, even the ones that would have been experimental (and so counted for nothing) had you answered them. So never leave a question blank, even if you have to race and pick randomly at the end. (Though, of course, ideally you won't be in that position at all.)

[Levent-g]

Hi Stacey,
thank you very much for your detailed answer. Much appreciated.

Levent

[StaceyKoprince]

you're welcome!

[Levent-g]

Hi Stacey,

sorry for opening this discussion again - I may get on your nervs :-)

Considering experimental questions, I heard that it should be quite logical that they are not in the first questions, because the algorithm creates a profile of you, and also not in the last questions, because a lot of people have time issues and would not have enough time to solve them. So logically assumed, they should be in the middle part.

Therefore you could use the middle part to gain some time and throw some answers, which you have categorized as loosers, away.

Can you share your expert opinion to that?

Thanks in advance.

Levent

[StaceyKoprince]

I have actually discussed this with the Chief Psychometrician (yes, that's a real title) at GMAC.

Experimentals can appear anywhere from question 1 to question 37 / 41.

Experimentals can absolutely show up in the last 5 questions - we discussed this explicitly.

They do have some kind of mechanism that prevents someone from having a cluster of all-experimental questions at the end of the test, because I asked what would happen if someone's last 5 questions were all experimental and that person ran out of time and just guessed. That person would have an artificial advantage over someone whose last 5 questions counted.

Larry (Dr. Lawrence Rudner) told me that nobody will see 5 experimentals for the last 5 questions but that you can still have experimentals among those questions. He couldn't provide more detail (the algorithm is proprietary, of course).

Basically, anything he tells us is absolutely true, but there are a lot of details he can't actually share.

So don't assume that you will see experimentals clustered in any particular area - they could come from beginning to end.

In fact, while taking the test, you should not be thinking at all about whether this question might be an experimental, or how you're doing, or whether you got the last question right, or anything else. The only thing you should be thinking is, "Can I do the question on screen, right now, accurately and efficiently? If yes, how? If no, can I narrow down the answers before I guess?"

That's it!

[Levent-g]

Like always, many thanks Stacey!!!!!!

[StaceyKoprince]

you're very welcome!