Supervisor ratings. In addition, I tested hypotheses 2a-2h using the supervisor
ratings of behaviors. The correlations between supervisor and coworker ratings of the
behaviors of the focal person are displayed in table 15. While help and achievement
displayed some degree of statistically significant convergence across the different sources
(r= .22, p< .05 and r= .32, p< .01), the correlations between the supervisor and coworker
ratings of innovation and compliance were lower and not statistically significant (r=.12,
n.s. and r=.13, n.s., respectively). These low correlations suggest differences between the
perspectives of supervisors and coworkers with respect to the focal person’s behaviors.
Table 16 describes the inter-correlations, means, standard deviations, and
reliabilities of the variables used to test the proposed relationships between roles and
supervisor ratings of behaviors.
Hypothesis 2a through 2h, which maintained that roles would lead to behaviors,
were not supported using the supervisor ratings. Specifically, only one of the control
variables had a significant effect on two of the outcome variables. Female individuals
were viewed as more helpful and more compliant than males by their supervisors ( =
.27, p< .05, R2 = .07 and = .26, p< .05, R2 = .07, respectively). Consequently, this
control variable was retained as a control for both help and compliant behavior. Table 17
provides the regression results for hypotheses 2a through 2h.