Manipulation checks. Using a 7-point scale, a judge, who was blind to our hypothesis, coded each participant’s description of the happy or unhappy event for the extent to which the event induced happiness and the extent to which it induced unhappiness (reverse coded). We averaged these measures (r = .64), and found that the events that happy (vs. unhappy) participants described were indeed more happiness inducing (M = 5.74, SD = 0.60 vs. M = 2.18, SD = 0.99), t(38) = 13.86, p < .01. Additional manipulation check measures indicated that happy and unhappy participants did not report differences in task interest (M = 4.38, SD = 1.56 vs. M = 4.68, SD = 1.45), F < 1, and they did not report differences in the amount of effort they put into the task (M = 2.48, SD = 1.47 vs. M = 3.11, SD = 2.01), F(1, 38) = 1.22, p > .25.
Goal adoption. We averaged the ratings of the two items that measured adoption of the self-improvement goal (r = .69) and the two items that measured adoption of mood management goals (r = .70) to form two independent indexes of goal adoption. We conducted a 2 (mood: happy vs. unhappy) 2 (goal type: self-improvement vs. mood management) 2 (goal frame: adoption vs. rejection) mixed ANOVA, with the level of agreement serving as dependent variable. The results appear in Figure 11. The ANOVA revealed a main effect for goal frame, F(1, 38) = 32.97, p < .01, indicating participants’ overall tendency to agree more with adopting than with rejecting statements (M = 1.51, SD = 1.18 vs. –0.41, SD = 1.53), and a main effect for goal type, F(1, 38) = 29.66, p < .01, indicating participants’ overall tendency to agree more with self-improvement statements than with mood maintenance statements (M = 1.05, SD = .81 vs. M = 0.04, SD = .79). The two-way interaction between goal frame and goal type was also significant, F(1, 38) = 33.13, p < .01, suggesting participants’ tendency to agree more with rejecting self-improvement than mood-management goals, although they were similarly likely to agree with adopting self-improvement and mood-management statements.
However, most important, this analysis yielded the predicted two-way interaction between mood and goal frame, F(1, 38) = 13.75, p < .01, indicating that happy participants were more likely than unhappy participants to agree with adopting statements (M = 2.05, SD = .61 vs. M = 0.90, SD = 1.37), t(38) = 3.15, p < .01, but happy participants were less likely than unhappy participants to agree with rejecting statements (M = –1.08, SD = 1.29 vs. M = 0.31, SD = 1.47), t(38) = 2.11, p < .01.
These results support our hypothesis that positive mood facilitates the adoption of goals and, regardless of the content of the goal. Notably, similar differences did not emerge for agreement with goal-neutral statements (F < 1), which were rated similarly by happy (M = 0.14, SD = 0.87) and unhappy (M = 0.03, SD = 0.90) participants. This allows us to interpret the effects of mood on the goal adoption with more confidence. If happy participants are more likely than unhappy participants to adopt goals, they should increase self-control efforts when a self-improvement goal is made accessible and reduce self-control efforts when a mood maintenance goal is made accessible. We designed the following studies to test for these possibilities.
Study 2: Charity Donations
A charity donation serves the long-term goals of both helping a charity campaign and helping the donor become a better person. However, it also has immediate costs, including giving away cash and exposing oneself to potentially unpleasant information or disturbing images. Therefore, we predict that happy (vs. unhappy) people are more likely to donate if a self-improvement goal is made salient (i.e., they strive to “be better”) than if a mood management goal is made salient (i.e., they strive to “feel better”) or if they are not primed with any particular goal (control condition). We tested these predictions by priming goals either to be better or to feel better (or to remain neutral) in happy and unhappy participants, who were then offered to participate in a charity campaign that promotes protecting young children from injury or death by improving children’s product safety.
Participants. One hundred ninety-four University of Chicago undergraduate students (112 women and 82 men) participated in the study in return for monetary compensation.
Procedure. This study employed a 2 (mood: happy vs. unhappy) 3 (goal: self-improvement vs. control vs. mood management) between-subjects design, in which the dependent variable was donation to a charity. Participants read that they were taking part in series of unrelated studies.
The first task induced a positive or negative mood, using a procedure similar to the one we used in Study 1. Depending on the experimental condition, participants wrote a vivid description of a happy or an unhappy event they could recall from their lives.
Participants then moved on to a second, supposedly unrelated, task, which induced either a self-improvement goal (be better), a mood management goal (feel better), or a no-goal (control) condition. Participants read that we were interested in undergraduate students’ activities. We randomly selected a third of the participants to list activities they do to become better people. The instructions read, “Different people choose to pursue different activities in order to improve themselves. In this survey, we are interested to learn about the type of activities that you usually pursue in order to become a better person. Please list everything that you do in order to become a better person.” We provided another third of the participants with similar instructions, but we asked them to list activities they pursued to maintain a positive mood. We asked the remainder of the participants in the control condition to list routine activities they intended to pursue that day. A space was provided to list up to five activities. Participants in the be-better condition listed activities such as “be friendly,” “show people around campus,” “read the newspaper,” and “pray.” Participants in the feel-better condition listed activities such as “listen to music,” “hangout with friends,” “draw,” and “take a walk.”
The third task measured charity donation. Participant learned that the experiment was over and they received their compensation. To solicit donations, participants received half of their compensation ($1) in quarters. The experimenter then informed them that the researchers were “working with a local charity, ‘Kids in Danger,’ to publicize their efforts.” The experimenter added that they could read some details about the charity and donate a portion of their earnings during the experimental session toward this charity.
We took the details on the charity from the Kids in Danger published brochures (see http://www.kidsindanger.org); these details contained negatively-valenced information. On the front page, the brochure stated that “Kids in Danger was established following the death of 3-year old Danny, in Chicago,” and that “Danny was asleep in his crib at his daycare when the crib collapsed, killing him. The company was aware of a defect in the manufacture of the crib and had recalled the product from its stores, but had not alerted consumers to the dangers of this product.” Participants then read, “We are now collecting money to increase consumer awareness about defective products and to protect other kids in danger and would like to ask for your help.”
“Kids in Danger” provided the rest of the charity materials, including a newsletter, an annual report that highlighted shortfalls in revenues, a short summary of the dangers to young children from malfunctioning consumer products, and an envelope in which participants could (supposedly anonymously) place any money they wished to donate to the charity. We selected this specific charity because it contained negatively-valenced information that can potentially adversely affect mood. All donations were forwarded to the charity organization. After participants completed this part of the experiment they were debriefed and dismissed. None of the participants identified the connection between the different parts of the experiment.
Results and Discussion
Manipulation check. Two independent judges, who were unaware of the hypothesis or the mood condition, coded the list of activities that participants provided as part of the goal induction manipulation for the total number of statements each participant listed (r = 0.96). A 2 (mood: happy vs. unhappy) 3 (goal: self-improvement vs. control vs. mood management) ANOVA conducted on the number of distinct goal-related statements that each participant listed revealed the expected main effect of mood, F(1, 188) = 8.74, p < .01, showing that happy participants listed more statements endorsing the goal than did unhappy participants (M = 5.31, SD = 1.73 vs. M =4.64, SD = 1.37), regardless of what the goal was. In addition, the main effect of goal was significant, F(2, 188) = 12.15, p < .01, showing that participants listed fewer statements in the be-better (self-improvement) condition than in the control or feel-better (mood-management) condition (M = 4.22, SD = 1.37 vs. M = 5.37, SD = 1.31 vs. M = 5.33, SD = 1.79). No other effect emerged in this analysis.
The two independent judges further coded responses of participants who listed be-better versus feel-better goals for the extent to which participants adopted the goal with which they were provided (7-point scale, anchored by 1 = not at all, 7 = very much). We averaged the ratings of the two judges (r = .72) to form an index of goal adoption. A 2 (mood: happy vs. unhappy) 2 (goal: self-improvement vs. mood management) ANOVA conducted on the level of each participant’s goal adoption revealed the expected main effect of mood, F(1, 126) = 14.64, p < .01, showing that happy participants adopted the goals more strongly than unhappy participants (M = 4.12, SD = 0.96 vs. M = 3.48, SD = 0.91), regardless of whether the goal was a be-better or feel-better one. In addition, the main effect of the goal was significant, F(1, 126) = 5.86, p < .05, showing that participants were less likely to adopt the be-better goal than the feel-better goal (M = 3.59, SD = 1.00 vs. M = 4.01, SD = 0.93). No other effect emerged in this analysis. Taken together, these two measures of goal adoption support our hypothesis that happy (vs. unhappy) mood increases goal adoption and they are consistent with the results of Study 1.
Charity donation. A 2 (mood: happy vs. unhappy) 3 (goal: self-improvement vs. control vs. mood management) ANOVA conducted on the amount of earnings participants donated to charity revealed the predicted interaction between mood and goal, F(2, 188) = 3.23, p < .05, and no main effects, Fs < 1. The results appear in Figure 2. Planned contrasts designed to investigate the source of this interaction revealed that when primed with a goal to be better, happy participants donated more money to the charity than did unhappy participants (M = $0.37, SD = 0.50 vs. M = $0.15, SD = 0.36), t(188) = 2.02, p < .05. However, in the no-goal (control) condition, happy and unhappy participants donated similar amounts (M = $0.17, SD = 0.37 vs. M = $0.28, SD = 0.43), t(188) = 1.01, ns. Finally, when primed with a feel-better goal, happy participants donated no more and somewhat less money than did unhappy participants (M = $0.18, SD = 0.38 vs. M = $0.32, SD = 0.45), t(188) = 1.40, p = .16.
To control for the possibility that a few extreme donations had a disproportionate impact on the statistical analysis, we conducted a similar analysis on participants’ decisions to donate or not. We coded the response of each participant who donated money as 1 and the response of each participant who did not donate money as 0, and ran a binary logistic regression on this measure, including mood, goal, and mood goal as predictors. This revealed a main effect of mood (b = –1.03), t(188) = 3.85, p < .05; a marginal effect of goal (b = –.53), t(188) = 3.29, p = .07; and a significant mood goal interaction (b = .87), t(188) = 4.36, p < .05. This interaction indicates that the proportion of participants who donated followed a similar pattern to the average donations.
These results support our hypothesis that positive mood facilitates helping behavior when the accessible goal refers to self-improvement rather than to mood management. Thus, happy participants were more likely to adhere to the be-better goal by increasing their charity donations, but they abstained from donations when their overriding goal was incompatible with the task, that is, they were striving to feel better.
Consistent with our analysis, we also observed a tendency among unhappy participants to increase charity donations when primed to feel better rather than to be better (M = $0.32, SD = .45 vs. M = $0.15, SD = .36), t(188) = 1.76, p = .08. This suggests that unhappy participants who rejected the accessible be-better goal were less likely to donate money to charity than unhappy participants who rejected the accessible feel-better goal. We also observed a similar pattern in the no-goal and the mood management conditions, which suggests that in this particular situation, the default goal that participants held was probably more closely associated with feeling better, and therefore most of the movement (relative to no-goal condition) was elicited by priming a self-improvement goal.
Study 2 examined the effect of mood and accessible goals on exposure to emotionally negative information that had immediate affective costs. However, it is possible that our specific mood manipulation of recalling life events influenced goal adoption directly, rather than by inducing mood, for example, by increasing perceived self-efficacy after recalling past success in the happy conditions. Although increased self-efficacy is mainly relevant for the adoption of self-improvement (vs. mood management) goals, in order to rule out this alternative, Study 2 employed a different mood manipulation that does not potentially involve recalling of previous successful goal pursuits. In addition, it is still unclear whether positive mood further enhances performance relative to neutral (rather than negative) mood and whether positive mood enhances performance on tasks that are cognitively or physically draining rather than emotionally taxing. To test these hypotheses, we asked participants in our next study to squeeze a handgrip device for as long as they could. We predicted that priming a self-improvement goal would lead to improved performance among happy (vs. neutral mood) participants on this physical endurance task.
Study 3: Physical Endurance
This study tested whether happy participants with an accessible health goal perform better on a physical endurance task (squeezing a handgrip) than neutral participants with a similar goal. We adopted this task from previous self-control studies, and it has been shown to require self-control (e.g., Muraven, Tice, & Baumeister, 1998). We further expected that happy participants with a mood management goal would perform no differently from control participants with a similar goal, because both groups should be deterred from performing a task that is incongruent with the goal prime. Because Study 2 already established the null effect of a no-goal condition, we did not include this condition in following studies.