A preliminary analysis of the construct validity of Kerr & Jermier's ‘Substitutes for Leadership’ Scales

AbstractDespite the intuitive appeal of Kerr & Jermier's (1978) ‘substitutes for leadership’ constructs, recent studies have provided only mixed support for this conceptualization of leadership. Perhaps the most compelling reason for the mixed findings is the lack of empirical construct validity of the substitutes subscales themselves. Internal consistency reliabilities are presented from 11 samples which indicate that over two-thirds (nine) of the 13 subscales have a median reliability below 0.62. Following this, preliminary factor analyses of the Substitutes for Leadership Scales were conducted in five samples, and an iterative ‘item trimming’ procedure was then undertaken to improve the reliability and conceptual integrity of the original subscales. Even though a total of eight items was dropped from four of the subscales, the reliabilities of those scales which were modified generally stayed the same, and the factor structure was reduced from 16–18 factors for the original scale to 14–15 factors for the reduced scale. In addition, the discriminant validity of the revised version of the substitutes measures and their relationships with Crowne & Marlowe's (1964) social desirability measure were assessed. Suggestions for further improving the construct validity of the Substitutes Scale are then discussed.