Abstract
This research examined how strongly appraisals can differentiate positive emotions and how they differentiate positive emotions. Thirteen positive emotions were examined, namely, amusement, awe, challenge, compassion, contentment, gratitude, hope, interest, joy, pride, relief, romantic love and serenity. Participants from Singapore and the USA recalled an experience of each emotion and thereafter rated their appraisals of the experience. In general, the appraisals accurately classified the positive emotions at rates above chance levels, and the appraisal–emotion relationships conformed to predictions. Also, the appraisals were largely judged by participants as relevant to their positive emotion experiences, and the appraisal–emotion relationships were largely consistent across the two countries.
I thank Phoebe Ellsworth, Barbara Fredrickson, Norbert Schwarz and Stephen Taylor and many others who have read previous versions of this paper.
Notes
1 This study is part of a larger study in which other measures (e.g., self-regard and social comparison) were also rated. Participants also described a humble experience and rated the same items. Complete data on this larger study and all measures are available on request
2 These NR responses could be useful for examining issues regarding the perceived relevance of certain emotions in particular emotional experiences (e.g., do people think that contentment is something they would feel when they feel pride). Across all emotions, the percentage of emotion NR responses was 20.95%, which meant that only about 80% of the emotion scores could be analysed. Despite this concern, the manipulation check results support expectations of which emotion should be most strongly reported within each emotion condition. These results also present a potentially important implication that respondents generally find a sizeable number of items in an emotion measure not relevant or meaningful to their experiences. This possibility requires further scrutiny but in a different paper.
3 Further support was found in the NR data. There were generally fewer NR responses for the targeted emotion compared to other emotions in each condition, and fewer NR responses for the targeted emotion in the manipulating condition compared to other conditions.
4 In other analyses, two or more emotions were reported in 97.9% of the 1170 emotion episodes (90 participants × 13 emotion conditions); 71.1% of the episodes comprised between 5 and 11 emotions. Hence, a large number of positive emotion blends were found. However, the targeted emotion remained the most strongly reported in these blends.
5 No method has been developed for testing the interaction effect between a between-participant variable (culture) and a within-participant variable (emotion) on a nominal dependent variable.
6 The accuracy rate could be inflated by compassion because of its negative valance. However, when the same analyses (across cultures) were repeated but minus compassion, the overall classification rate increased but by a mere 0.9%; leave-one-out classification rate increased by only 0.2%. Accuracy rates within individual cultures also changed little (data can be obtained from the author).
7 Similar factor structures were found separately in the two cultures.