The A-10 was an early experimental association test, and the very first test to be administered to I.Q. society members in 1995. The A-1 through A-9 have never been administered.
I was shocked at the time that the scores were so low, and thought I was doing something wrong, that the problems were too far-fetched. I had expected much higher scores intuitively. It was the first time I saw scores on a test created by me. In hindsight I know this kind of score pattern is entirely normal on a hard test, and that this very first test, despite having many bad items which have never been solved, had ruthless discriminative power in the high range right from the start on.
Later I added easier items to the tests to make them doable for more people, realizing that my intuitive notion of item difficulty was tremendously off on the low side. I also learnt that this underestimation of difficulty — or overestimation of others' intelligence — is typical of highly intelligent individuals; it is the naivety of assuming one's own intelligence level in others. I learnt that creating I.Q. tests is not so much a matter of "how can I make hard enough problems so they will not get everything correct", but rather of "how low do I have to go to to be understood" and "how easy do I have to make it to avoid getting a score pattern concentrated at or just above zero".
|Compound of all available scores||17||0.62|
Weighted average of correlations: 0.62 (N = 17, weighted sum = 10.56)
Estimated g factor loading: 0.79
Because there are no other tests with enough scores to compute correlations per test, the correlation with the compound variable comprising all known other scores has been computed.
|Year of birth||4||-0.92|
|Age class||n||Median score|
|60 to 64||1||8.0|
|45 to 49||1||3.0|
|35 to 39||1||4.0|
|30 to 34||1||3.0|
|22 to 24||1||0.0|
N = 5
|Year taken||n||median score|