re: When does the PoliBoard start posting their IQ scores?

quote:

---

In addition, I don't know of a single individually administered IQ test that hasn't gone to a uniform distribution of standard scores with a mean of 100 and standard deviation of 15. The Stanford-Binet changed from 16 to 15 in 2003.

---

Yeah, I was already out of the scene by 2003, so I am working with knowledge that's over 20 years old.

quote:

---

The still have a fairly high G-loadings, and I'm not sure if they are any more or less than previous version. I would imagine that the issue is that they have fundamentally different samples that form the distribution (college and/or graduate bound), and the residuals are more likely to be non-random, systematic error because they are not intended to be intelligence tests.

---

All I know is Mensa stopped accepting the tests after they were "recentered" in the late 90's and early 2000's. Mensa has a full time PhD psychologist in charge of their test admissions policy and she made this decision evidently due to changes that were made to the tests. It's been years since I've read up on it, so I don't remember the full details of what changes were made. I don't know whether it was due to some scoring or sampling changes as you suggest or whether psychometricians determined that the tests weren't as g loaded. Perhaps it was both. In any case, after Mensa made this decision, other groups began doing the same.

More interestingly, some of the IQ groups created their own tests. Ron Hoeflin (one of the founders of the "higher" IQ groups) created the Mega test and published it in Omni magazine in 1985. Various well known people took the test and submitted their answers; Marilyn Vos Savant and John Sununu among them. Hoeflin claimed that a score of 40 or 41 (out of 48 questions) put a person at the 1 in a million level. John Sununu scored 44. Marilyn vos Savant scored 45 or 46. Chris Langan (the bar bouncer) scored 42 on his first try and 47 on his second try.

Various people have "normed" the Mega test and you can find all of their statistical work online. Most of the norming was done by "anchoring" (i.e. taking a person's known IQ from the WAIS or the SAT and then correlating it to a Mega Test score. I believe the technical term is equipercentile equating). After testing 150 people, they came up with the following score distribution:

10 correct = 98th percentile (Mensa)
14 correct = top 1%
24 correct = 99.9th percentile (Triple Nine)
36 correct = 99.997th percentile (Prometheus)
43 correct = 99.9999th percentile (Mega Society)

Anything above about 4 S.D.s is completely theoretical as no conventional test is accurate beyond that (and many would argue that even a 160 score on a conventional test is probably questionable). The extreme right-tail of the bell curve might not be a perfectly smooth. However, the Onni "norming" did show that members of Prometheus scored higher, on average, than members of TNS and members of Mega scored the highest. So, perhaps there is some extreme right-tail validity to tests like this. Unfortunately, this is not an area that most professional psychometricians want to study (it's just not that important to society really), so the best we have are data from "amateurs" like Ron Hoeflin and Keving Langdon who do their own normings using traditional statistical techniques.