Supplemental Material
Happy and Sad Results
Study 1b
To examine how emotionally evocative the clips were, we averaged each subject’s happiness and sadness ratings across the entire clip (see Table S1). These data were submitted to a 2 (clip: control, bittersweet) x 2 (emotion: happiness, sadness) mixed-model ANOVA, where clip was a between-subjects variable. The only significant effect was a clip x emotion interaction, partial η2 = .51. Follow-up between-subjects t-tests indicated that the bittersweet clip elicited less happiness than the control clip, d = 1.14, but more sadness, d = 1.05.
Table S1
Happiness and sadness ratings in Study 1b
Study 2
We averaged each subject’s happiness and sadness ratings across the entire clip (see Table S2) and submitted these data to a 2 (condition: no-hypothesis/control clip, bipolarity hypothesis/bittersweet clip) x 2 (emotion: happiness, sadness) mixed-model ANOVA, where condition was a between-subjects variable. The only significant effect was a condition x emotion interaction, partial η2 = .28. Follow-up between-subjects t-tests indicated that subjects in the bipolarity hypothesis/bittersweet clip group experienced more sadness than those in the no-hypothesis/control clip group, d = .54; the two groups experienced comparable levels of happiness, p > .1.
Table S2
Happiness and sadness ratings in Study 2
Study 3
As in Study 1a, we computed the total amount of time that subjects pressed each button. These data, which are shown in Table S3, were submitted to a 2 (condition: control clip/endorsers, bittersweet clip/non-endorsers) x 2 (emotion: happiness, sadness) mixed-model ANOVA, where condition was a between-subjects variable. The only significant effect was a main effect of emotion, such that subjects experienced more sadness than happiness, partial η2 = .08.
Table S3
Number of minutes that Study 3’s subjects reported happiness and reported sadness
Study 5
The percentage of subjects in each condition who reported happiness and sadness is shown in Table S4. A chi-square test on the happiness data revealed an effect of condition and a pair of follow-up Fisher’s exact tests revealed that both groups of subjects who watched the bittersweet clip were more likely to report happiness than those who watched the control clip. A chi-square test on the sadness data revealed no effect of condition, p > .7.
Table S4
Percentage of subjects in Study 5 who reported happiness and sadness
Study 6
The percentages of subjects in each condition who reported happiness and sadness are shown in Table S5. A chi-square test on the happiness data revealed an effect of condition and a pair of follow-up Fisher’s exact tests revealed that both groups of subjects who watched the bittersweet clip were more likely to report happiness than those who watched the control clip. A chi-square test on the sadness data revealed effect of condition, p > .3.
Table S5
Percentage of subjects in Study 6 who reported happiness and who reported sadness