The influence of leader humor on relationships between leader behavior and follower outcomes

Journal of Managerial Issues, Summer, 2009 by Robert P. Vecchio, Joseph E. Justin, Craig L. Pearce

Analytic Approach

To test the moderating effect of humor on leader behavior for the dependent variable of performance, a hierarchical regression approach was employed following mean-centering of the predictor variables (Aiken and West, 1991). The hierarchical regression analyses consisted of creating three equations. In the first equation, the variables of years of teaching and job satisfaction were entered as control variables. In the second equation, the mean-centered predictor variables of (1) leader humor and (2) either leader contingent personal reward or leader integrity were added. Finally, in the third equation, a multiplicative term (composed of the variables entered in the second equation) was added to represent the interaction of the predictors in accounting for unique criterion variance. The increment in accountable criterion variance (i.e., the increase in BY between equations 2 and 3) was then tested for statistical significance.

RESULTS

In order to establish the independence of the constructs that were assessed via self-report, a confirmatory factor analysis (CFA) was conducted, wherein a single-factor model (representing the view that only general affect is manifest in the self-report data) was contrasted with a multi-factor model (representing the view that the proposed distinct factors underlie the self-report data). The results of the CFA, which are presented in Table 1, reveal that the multi-factor model provided a superior representation of the responses in that the change in chi-square indicated a significant improvement for the proposed multi-factor model over the single-factor model. The various fit indices also pointed to a superior fit for the multi-factor model. Specifically, all of the multi-factor fit indices exceeded .95 and the RMSEA was less than .08 (Bentler, 1990; Browne and Cudeck, 1993).

In addition to these analyses, internal consistency and average variance extracted values were calculated from the factor analytic results. These values (provided in Table 2) further underscore the appropriateness of treating the proposed measures as useful indices of the proposed constructs (Fornell and Larcker, 1981).

Table 3 presents the descriptive statistics, reliabilities, and intercorrelations for the outcomes, predictors, and control variables. The performance assessment given by the principals was positively associated with the teacher's assessments of the principal's leadership behavior and the teacher's level of job satisfaction. Use of humor and the other leader behavior variables were positively correlated. More importantly, the correlation of the leader's use of humor with teacher performance was comparable to that previously reported by Avolio et al. (1999) in a for-profit organizational setting (i.e., current r = .23, p < .01, versus prior r = .24, p < .05, for Avolio et al. (1999)), thereby supporting Hypothesis 1. Leader's use of humor was correlated positively with teacher job satisfaction (r = .25, p < .01), replicating Decker's (1987) earlier finding.