Manipulation check. At the end of the study, participants were asked to share their thoughts about what they felt the purpose of the study was. None of the participants identified the interruption by the confederate as a planned, strategic part of the study, nor were they able to identify the confeder- ate’s role in the study.
w2 analyses, one-way analysis of variance (ANO- VA), and multilevel modeling were used to assess the hypotheses proposed for Study 1. Given the number of analyses conducted, Po.01 was used as an indicator of statistical significance.
Examining participants’ global negative reactions (bitchy or not bitchy) to the confederate we found striking differences by condition (w2(2)5 51.71; Po.0001; see Table I). Specifically, when women were rated as bitchy, all but 2 were in the sexy condition. And of those rated as not bitchy, most were in the conservative condition. The remaining eight women’s reactions were coded as unsure (two in the conservative condition and six in sexy condition).
When we examined how bitchy the women’s reactions were on a scale from 0 to 10, we found a statistically significant main effect for condition (sexy vs. conservative; F(1, 82)5 64.00, Po.0001), a marginal effect for type of dyad (friends vs. strangers; F(1, 82)5 4.10, Po.05), and no interac- tion between conditions (F(1, 82)5 1.67, P5 .20). Women in the sexy condition were rated as being bitchier (Mean7SE5 3.4070.28) than women in the conservative condition (Mean7SE5 0.2070.27) and women in friendship dyads were rated as being slightly bitchier (Mean7SE5 2.2070.28) than women in stranger dyads (Mean7SE5 1.4070.28) (Fig. 2). The effect size for the sexy condition was large (Cohen’s d5 1.74) and for the dyad condition small (Cohen’s d5 0.34). Because participants’ reactions were nested within
dyads, interactions between dyad and condition were also examined using multilevel modeling with participants’ global negative reactions (level of bitchiness) as the outcome. Excluding those in the unsure group and controlling for the type of dyad,
TABLE I. Relative Frequency of ‘‘Bitchy’’ Classification by Experimental Condition
Bitchy 2 (5.0%) 34 (74%)
Not bitchy 36 (90%) 6 (13%)
Unsure 2 (5.0%) 6 (13%)
572 Vaillancourt and Sharma
we found that again, condition (sexy vs. conserva- tive) was a statistically significant correlate of outcome (OR5 exp (4.651)5 104.69, SE72.60; Po.001). Unfortunately, we could not test the initial hypothesis that women would behave more poorly when with a friend in the sexy condition (i.e., the interaction between condition and type of dyad) because the model did not converge.
Examining emotions coded with EmFACS across the sexy and conservative conditions we found that of the 11.6% (n5 10) of the total sample that displayed true happiness all were assigned to the conservative condition (w2(1)5 13.01, Po.001). Moreover, of the 12.8% (n5 11) of the total sample that displayed anger, all were assigned to the sexy condition (w2(1)5 10.97, Po.001). There were no other statistically significant findings for EmFACS identified reactions by conditions (friends vs. stranger or sexy vs. conservative) and no interactions between conditions.