ABSTRACT
Of the many recurring misconceptions about human abilities, one of the most harmful for educational researchers is the belief that they can adequately represent ability in their research by some measure of g. This belief is abetted in part by the fact that g is the single best predictor of many important educational and social criteria, especially those that are also averages over diverse performances. However, more specific criteria (e.g., grades in a particular class rather than GPA) are better predicted when ability constructs other than g are added to the mix (Gustafsson & Balke, 1993; Wittmann & SiiB, 1999). Indeed, as behavior becomes more specific and contextualized, measures of g become less useful and measures of more specific knowledge, skills, and abilities become more useful for prediction and explanation. Educational researchers who include ability constructs in their studies need to represent abilities of different levels of breadth or generality. But to do this well requires knowledge of the abilities at different levels in the ability hierarchy and of the tests that can be used to estimate each. Therefore, we briefly summarize the key features of the major contemporary model of human abilities.
