HIGH-END PROBLEMS
Rick Rosner

(Editor's comments on the Quest Test in the form of an article)

One problem with getting Mega members to take IQ-type tests is that we've already done just about as well as we possibly can. I personally don't want to spend hours and days on a test just to be informed that I'm stupider than I thought and claimed (as happened the last time I took a Langdon test, scoring 158). As Christopher Harding said in the previous issue what you tend to get is regression to the mean.

Another related problem is with the sheer number of problems it takes most tests to reach an interestingly-high ceiling. Most high-end tests have over 40 problems. One's performance on a series of such tests might first be a measure of persistence followed by a measure of capacity for careless errors as one becomes overconfident or impatient.

I reallv liked Chris Harding's 10-item multiple choice Multimax Test, with a ceiling he claims is 211. I might tend to agree with Ron Hoeflin that such a test might be psychometrically suspect, but I still like to claim that my IQ is the one given by my score on Harding's test.

I would guess that among high-IQ test composers, Hoeflin takes norming to the furthest extreme. He knows exactly which problems correlate with the highest scores on other aptitude tests. I'd guess that other test makers have at least an approximate idea of which of their test problems are the strongest indicators of high scores on other tests.

It'd be good to know which problems these are before attempting a test. It'd also be good to know what kind of ceiling we can shoot for. Mr. Inman, if you have this info, please send it to me.

I know what I want in new tests--fewer problems and higher ceilings. This isn't very fair to test makers, but I have little to gain except knowledge and pleasure gained through intellectual pursuit, and unfortunately my life isn't structured to allow much of that.

What I'm proposing is similar to what I've suggested before in a different context--that .abridged versions of high-end tests be made available with high ceilings and high basements. We need small sets of selected very hard problems. We need credit given for having already done well on other tests. (When it was suggested that credit NOT be given to some members, did y'all get pissed or what?) If credit turns out to have been unjustified and we don't make it out of the high basement, it's not that big a deal to test makers or test takers. I'm sure that each of us has achieved test scores we're not proud of.

To use a concrete example: the Titan has 48 problems with a basement of around 120? and a ceiling of 190+. Aren't there 10 or 12 problems which could be selected to make a short test with a basement of 150 and a ceiling of 185+? Could The Quest Test be similarly abridged?

I hope members will try the analogy tests that frequently appear here because they're full and challenging. I hope test makers will offer us condensed tests with high ceilings.