Abstract
Two experiments were carried out to explore an anger-reducing strategy based on Brehm's emotional intensity theory. According to this theory, anger can be reduced indirectly by interfering with the feeling of anger rather than by dealing directly with the source of anger. One strategy involves providing the angered person with a reason for feeling happy. We predicted that anger intensity would be reduced not only by a large reason for feeling happy, such as a large gift, but also by a small reason, like a tiny gift. A medium-size gift was expected to maintain anger at approximately its instigated level. Both experiments instigated anger by personal insult and then measured the intensity of felt anger and retaliation after either no further treatment, or a small, a moderate, or large irrelevant gift was presented. The results for felt anger and retaliation confirmed our theoretical expectations.
ACKNOWLEDGMENTS
The development of the theory and some of the research reported here were supported by a Senior Research Award from the Alexander von Humboldt Foundation to Jack W. Brehm.
Notes
1After pretesting the procedure, participants were randomly assigned, in blocks of 6, to one of the five conditions: no gift certificate, $1, $2, $3, or $5. Because there was no difference between the $2 and $3 conditions, the two conditions have been combined in to a “moderate deterrence” condition.
Note. Scales ran from 0 (not at all) to 10 (extremely). Row means with different subscripts are reliably different (p ≤ .05). Ns denote the cell sizes.
a Retaliation was measured with the item, “The experimenter is competent” from 1 (not at all) to 5 (very much). Lower scores indicate more retaliation.
2Of the two possible forms of a cubic trend among four means, only the one that conforms to the theory makes sense. If the opposing cubic form were found, we would consider the data meaningless. Because the cubic trend was significant in the direction we predicted (decrease from the control condition to low, increase from low to moderate, and decrease from moderate to high deterrence), all the reported p values for the polynomial contrasts are one-tailed, unless otherwise specified. The F tests for the cubic effects and the statistical tests for the manipulation checks are all two-tailed.
Note. Scales ran from 0 (not at all) to 10 (extremely), except for the “Liking of the issue,” for which the scale runs from − 5 (extremely dislike) to + 5 (extremely like). Row means with different subscripts are reliably different (p ≤ .05). Ns denote the cell sizes.