Similar results were obtained by Keating (1978) using a different set of social-cognition measures. Keating administered the Social Insight Test (Chapin, 1967), which asks individuals to read about problem situations and to select the best from among four alternative interpretations of the situation; the Defining Issues Test (Rest, 1975), based on Kohlberg's (1963) theory of moral development; and the Social Maturity Index (Gough, 1966), which is a self-report measure of effective social functioning. Keating failed to find substantial intercorrelations among the social-cognition measures, and found no evidence, from either a multitrait-multimethod analysis or a factor analysis, that social cognition was distinct from academic cognition. All of Keating's measures, like those of Riggio et al. (1991), were verbal, which may have contributed to the inability to discriminate between abstract and social cognition.

2.1.2 Behavioral approaches to measuring social cognition

As a result of frustrations in trying to distinguish social from academic cognition, many researchers returned to Thorndike's (1920) definition and considered the behavioral as well as cognitive dimension of the construct. These researchers (e.g., Ford and Tisak, 1983; Frederickson, Carlson, and Ward, 1984) proposed that cognitive aspects of social cognition might expectedly be more closely associated with abstract cognition, whereas behavioral aspects would represent a more distinct construct.

A second set of tests to those of O'Sullivan et al. (1965) emerged from Guilford's (1967) Structure of Intellect model. These tests focused on behavioral rather than cognitive skills and defined social cognition as the skill to cope with people (Hendricks, Guilford, and Hoepfner, 1969). Hendricks et al. administered their tests to 252 high-schools students. Through principal-components analysis they identified factors that readily were interpretable as divergent production skills, and found that these factors were independent of behavioral cognition. These findings were later confirmed by Chen and Michael (1993).

A study by Ford and Tisak (1983) took the next step by distinguishing a behavioral measure of social cognition from academic cognition. The investigators conducted their study with more than 600 high-school students. Their measure of social cognition included self, peer, and teacher ratings of social competence, Hogan's (1969) empathy test, and a judgment of social competence from an individual interview. In addition, they obtained measures of verbal and math skill from school grades and standardized test scores. The measures of academic and social cognition were found to load on separate factors. They further found that the ratings of social competence and scores on the empathy scale were more predictive of interview ratings than were the measures of verbal and math skill. Ford and Tisak suggested that the difference between their findings and those of Keating (1978), reviewed earlier, were attributable to using a behavioral rather than a cognitive measure of social cognition.

A number of subsequent studies obtained findings consistent with Ford and Tisak (1983). Marlow (1986), for example, found that scores on several self-report measures of social cognition were unrelated to scores on measures of verbal and abstract cognition. Similarly, Frederickson et al. (1984) did not find significant correlations between ratings of interview behavior and measures of scholastic aptitude, achievement, or problem solving. However, Stricker and Rock (1990) did find a correlation between verbal skill and participants' skill to judge accurately a person and a situation from a videotaped interview.