Tuesday, May 8, 2012

Hi everyone!

Here I shall not talk much about myself. You can find some things about me on my web site. Instead, I want to say some things about high range IQ tests and their norms. More precisely, I want to say how I started solving, making and norming IQ tests.

But first, what are high range IQ tests actually? They try to measure intelligence quotient above the ceiling of standardized tests, which is normally IQ=145 or three standard deviations above average. High range IQ tests are regularly untimed because of there structure: they contain some indeed hard items. Besides, I guess that nobody likes time pressure.

One of the pioneer of high range IQ testing is certainly Paul Cooijmans. He started his work in 1995. I started 10 years later.

In 2004, I took Mensa test and joined Mensa Croatia. On Mensa forum I found link to excellent spatial IQ test by Robert Lato: Logima Strictica 36 or popularly LS36. I scored 21/36 which was roughly IQ 162 according to the norm by Paul Cooijmans. That score was in perfect correlation with my performance in mathematical competitions: I was one of the best in my generation in Croatia (among 50.000 pupils) and I achieved a bronze award for Croatia in the 35th International Mathematical Olympiad (Hong Kong, July 1994).

On Mensa forum I also met Mislav Predavec who scored extremely high on LS36: 27/36 in the first attempt. Nevertheless, his IQ score was only a few points higher than mine. We both concluded that Paul's norm is not realistic above IQ 160 and we started to think on how to improve it.

My score on LS36 was among top 10 percent and so my first assumption was that top 10% percent of scores on each high range IQ test should be at IQ 160 level, or higher. The second assuption was actually a well known fact that average scorer on high range tests has approximately IQ 145. Finally, my last assumption was that interest for taking high range IQ tests increases with IQ.

Then I modeled my assumptions with particular exponential and logarithmic functions and new LS36 norm appeared. You can find the final norm report here.

After taking several more IQ tests I started to create my own high range tests. One of them, Numerus Light, was best suited to test my assumptions. Even with that pretty easy IQ test I didn't receive more submissions from "average" persons and scores correlated well with scores on other IQ tests. Of course, my assumptions are not "the Bible" of high range norming and slight deflections are expected from test to test. So, I regularly ask for scores on other tests and calculate various statistics to predict such deflections.

Conclusion:
Today my norms seem pretty stable, and I'm offering norming services for other tests too.

10 comments:

  1. Very interesting Ivan! What is the average IQ score on your tests? Also 145?

    ReplyDelete
  2. Yes. Actually +/- few points.

    ReplyDelete
  3. Hello Ivan, I find this topic very interesting.
    A question about the average iq of scorers in hrt: some months ago I calculated the average score in some tests by Hoeflin and I remember it was about 138-139 sd15. The number of scorers was very high (close to 1000 testees for one or those tests), so data should be reliable. What could explain this difference of about 6.5 points between that average value and the value (145) you mentioned?
    Flynn effect? Different groups or testees? .....I'm just guessing.....
    Thanks

    ReplyDelete
  4. People guess the average IQ based on different tests.
    Someone would say 138, the others 140, 143, 145, ...
    I think that 145 is a good guess because it is the ceiling of most of standardized tests.

    ReplyDelete
    Replies
    1. (M) That doesn't make sense. That would only make sense if nearly everyone who takes high-range tests does so because he's maxed out on "the ceiling of most standardized tests." I don't believe that 145 ceiling is actually the ceiling of most tests anyway.

      If that's your reason for saying 145, then it's just guesswork.

      And I recognized those estimates from Paul Cooijman's page.

      Making wild guesses leads to wild scores. Inflation, they call it. (W)

      Delete
    2. SD 15 means that the score deviates (+/-) 15 points... 160-15=145...........

      Delete
  5. Thank you for your comment. It is nice to know someone is reading.

    intelligence test quiz
    intelligence tests online

    ReplyDelete
  6. This comment has been removed by a blog administrator.

    ReplyDelete
  7. Hai Ivan!
    I want to find the high range iq test, but only one problem. I'm still 14 years old so i fill your iq quiz is so hard for me. Is there an accurate iq test to find kid's iq like me?

    ReplyDelete
  8. No kids allowed!

    Now go solve the logic of Hop Scotch squares or something!

    ReplyDelete