Inspection of correlations within conditions indicated that participants in the Control condition who selected Person B were significantly more likely to estimate Person B’s age as younger (r=-.40, p=.009). In contrast, participants in the Subordinate condition who selected Candidate B were significantly more likely to estimate Candidate B’s age as being older (r = .38, p = .007). In the Supervisee condition,there was no significantcorrelation(r=.04,p=.790).This suggestedan interactive effect of condition and perceived age on hiring decisions. To test this possibility, we dummy coded conditions and created interaction terms between the Control condition and estimates of Candidate B’s age, and between the Subordinate condition and Candidate B’s estimated age. We then conducted a regression analysis to test the effects of participants’ age and gender, Control
114 Abrams, Swift, and Drury
condition and Subordinate condition, age estimate of Candidate B, and the two interaction terms on whether participants selected Candidate B.