Louis Thurstone pioneered the concept of multiple intelligences, in opposition to the idea of Spearman's g-factor. After conducting his tests, he used a statistical fallacy called "restriction of range" to hide correlations between the abilities he was studying. Basically, he held IQ constant by only comparing, say, people who scored 100 IQ overall, and found that overall IQ score had no predictive power (which is true by definition). Later scientists found the expected g-factor in his data using factor analysis on the full dataset. Whether this was incompetence on his part or disingenuousness is a matter of speculation, but I believe it was almost certainly disingenuous.