Hacker News Reader

Comment by gehant | Hacker News Reader

Why do we still defer to IQ? It's a 2D approach at measuring intelligence in a 3D world.

"The scale, properly speaking, does not permit the measure of intelligence, because intellectual qualities are not superposable, and therefore cannot be measured as linear surfaces are measured." -Alfred Binet, 1905

View on HN

Thanks for sharing the quotation from Alfred Binet, developer of the first practical IQ test battery (for the unambitious purpose of efficient school placement of elementary pupils in schools with lock-step age grading). His view was correct then, and it is still correct today, but widely ignored.

More recent authors who criticize the mainstream psychometric concept of intelligence (operationalized as IQ scores) include Keith Stanovich.

http://www.project-syndicate.org/commentary/stanovich1

http://yalepress.yale.edu/yupbooks/book.asp?isbn=97803001238...

Isn't IQ only one dimensional?

Neither. IQ tests are more like a 163 dimensional approach in a 882 dimensional world. They measure a subset of all human capacities.

Right, but those are all projected onto a scalar.

It's important to note that IQ test scores have only ordinal properties--one cannot make interval inferences from them validly, for example the frequently heard assertion that "A child with an IQ of 150 is as different in intelligence from a normal child as a child with an IQ of 50." Such a statement is very hard to verify in the first place, but in any event IQ scores show ORDINAL relationships (subject to a lot of error of estimation) but don't show interval relationships of how far (on the same scale) one score is from another in any valid way.

Reply on news.ycombinator.com