FROM
The Social Construction of What? by Ian Hacking
Their authors [Alfred Binet and Lewis Terman] were committed to the idea that biological characteristics should be displayed upon a Gaussian or Normal probability curve. . . . Binet devised questions which his subjects answered in such a way that scores shaped up on the familiar bell-shaped curve [just an example]. The trick was to get a set of questions which, when answered, had this property. Terman, with his able female assistants who administered most of the tests, discovered that women did better on his IQ tests than men. Since women "couldn't" be more intelligent than men, this meant that the questions were wrong. Some of the questions that women answered better than men had to be deleted and replaced by ones on which men did better." [Italics added]