Appendix C
Measures and complete list of scale items
Emotion Granularity Trait Measure Scale taken from Kang and Shaver (2004):
Please indicate how much the following statements describe you personally.
[Battery of questions with a scale from 1 = does not describe me very well to 7 = de-scribes me extremely well]
• I am aware of the different nuances or subtleties of a given emotion
• Each emotion has a very distinct and unique meaning to me
• I usually experience a limited range of emotions
• I don’t experience a variety of feelings on an everyday basis
• Feeling good or bad – those terms are sufficient to describe most of my feelings in everyday life
Emotional Granularity Performance-based Measure Scale adjusted from Erbas, Sels, Ceulemans, and Kuppens (2016):
Before we dive in further into the topic, could you please indicate how you are feeling right now, by rating the following emotions?
[multiple items slider scale ranging from 1 = not at all to 100 = very much]
• relaxed
• happy
• satisfied
• excited
• angry
• depressed
• anxious
• sad
Concern about climate change Scale taken taken from Chu and Yang (2019):
When it comes to climate change, people may hold very different opinions. Could you please answer the following questions on how you personally perceive climate change?
[Battery of questions with a scale from 1= Not at all to 7 = Very]
• How concerned are you about climate change?
• How serious are the current impacts of climate change?
Climate change mitigation behaviour
Scale adapted from Skurka, Niederdeppe, Romero-Canyas, and Acup (2018):
Many strategies have been proposed to reduce a person’s impact on climate change and to limit global warming. How likely are you to take each of the following actions in the next few months to reduce your own impact on climate change and in turn reduce the likelihood of a continuous spread of emerging infectious diseases?
[Battery of questions with a scale from 1 = very unlikely to 7 = very likely]
• Turn off or unplug electronic devices when not needed
• Walk, bike, carpool, or use public transportation when possible
• When shopping, use your own bags
• Buy paper and plastic products that are made from recycled materials
• Avoid buying products that have potentially harmful environmental effects
Support for climate change mitigation policies Scale inspired by Feldman and Hart (2018):
In addition to individual efforts that you can take to stop climate change, there are also numer-ous governmental actions that play an important role in mitigation intentions. Some of these strategies are listed below. How likely is it for you to support these policies (e.g., by voting for them or signing a petition) in order to mitigate climate change and thus limit the spread of emerging infectious diseases in the future?
[Battery of questions with a scale from 1 = very unlikely to 7 = very likely]
• regulating carbon emissions and other greenhouse gases as pollutants
• increasing government investment in renewable energy industries like wind and solar to make them more energy efficient
• providing government incentives for people who lead a more climate friendly life
• signing an international treaty that requires countries around the world to collaborate in combatting climate change
Climate-related information seeking Scale based on Clifford and Jerit (2018):
Thinking back to the Instagram-post you saw earlier, how likely is it that you are going to…
[Battery of questions with a scale from 1 = very unlikely to 7 = very likely]
• look up more information about dengue and its connection to climate change in the fu-ture?
• discuss this issue with your friends or family?
Manipulation Check
Now, please think about the Instagram-post again: On a scale from 1 to 7 with "1" being "Not at all" and "7" being "Very much", to what extent did you perceive the following emotions as being transported in the Instagram-post about climate change?
[Battery of questions with a scale from 1 to 7]
• disgust
• anxiety
• guilt
Appendix D
Further information and results for analyses conducted over the performance-based emotional granularity measurement
Further information on performance-based emotional granularity scale
It has already been mentioned in the main part of this study that the variance across positive emotions did not correlate with the variance across negative emotions, r(393) = -.035, p = .485 for the performance-based emotion granularity measurement. This indicates that people who knew how to differentiate between positive emotions did not necessarily have the same skill when it came to negative emotions. This could also be seen on a total of 47 cases who displayed a wide variance across positive emotions, but none for negative emotions, s² = 0.00.
Such a huge discrepancy points towards a relevant difference between positive and negative emotional granularity, which has also been argued for in previous studies (Feldman Barrett et al., 2001; Tugade et al., 2004).
To conduct analyses over the performance-based emotional granularity scale, the fo-cus lay on the negative emotion differentiation skill rather than the positive one, since this studies’ overall emphasis is on negative emotional appeals in climate change communication.
Also, it has been argued that decreased ability to differentiate negative emotions is more con-sequential (Feldman Barrett et al., 2001). Like in the first model, respondents were divided into a low granularity (𝑛1 = 197) and a high granularity group (𝑛2 = 196), based on median variance of emotion ratings across negative emotions in the sample (Mdn = 129.00, min. = .00, max. = 2417.58), to explore the moderating effect of emotional granularity as a skill.
Influence of emotional appeals on concern about climate change
The first analysis focused on whether exposure to climate change mitigation campaigns framed in a disgust, anxiety, or guilt-inducing way, would lead to higher levels of CC than ex-posure to a campaign framed with a rational appeal. To do so, and to reveal a potential inter-action effect between emotional appeal and emotional granularity, a two-way ANOVA was conducted, with emotional appeal (anxiety vs. guilt vs. disgust vs. rational) and emotional granularity (low vs. high level of granularity) as independent variables, and CC as a depend-ent variable.
Looking at the results, there was no statistically significant main effect of emotional appeal on CC, F(3, 385) = 1.06, p = .368, η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2 = .01. This means that neither the anxiety (M = 6.48, SD = 0.76, p = 1.000), nor the guilt (M = 6.26, SD = 1.08, p = 1.000), or disgust (M = 6.34, SD = 0.93, p = 1.000) appeal led to significantly more CC compared to the rational appeal condition (M = 6.41, SD= 0.96). On the contrary, looking at the means, it can be ob-served that there seems to be a tendency for guilt and disgust to predict even lower levels of CC compared to the rational condition. Additionally, a marginally significant main effect of emotional granularity on CC could be shown, F(1, 385) = 2.87, p = .091, η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2 = .01. Par-ticipants with a higher level of emotional granularity (M = 6.46, SD = 0.85) had significantly more CC than respondents with a lower level of emotional granularity (M = 6.30, SD = 1.00).
Still, there was no significant interaction effect between emotional appeal and emotional gran-ularity F(3, 385) = 0.44, p = .722, η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2 = .003.
Therefore, H1a-c must be rejected for the performance-based emotional granularity measurement as well. This means that exposure to climate change mitigation campaigns, framed in an emotion-inducing way, did not lead to higher levels of CC than exposure to a
campaign framed with a rational appeal. Also, H5 has to be rejected for CC as emotional granularity did neither strengthen, nor weaken the effect of emotional appeals on CC.
Influence of emotional appeals on support for climate change mitigation policies
Turning now to assessing the effect of emotional appeals on climate change mitigation policy support: it was expected that emotional appeals in climate campaigns would lead to higher levels of SMP than rational appeals. Again, a two-way ANOVA was conducted to reveal both the main effect of emotional appeal on SMP, as well as a potential moderation of emotional granularity.
Looking at the results, there was no significant main effect of emotional appeal, F(3, 385) = 1.12, p = .339, η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2 = .01, meaning that neither the anxiety appeal (M = 6.27, SD = 1.00, p = 1.000), nor the guilt (M = 6.02, SD = 1.16, p = 1.000), or disgust (M = 6.15, SD = 0.91, p = 1.000) appeal led to significantly more SMP compared to the rational appeal condition (M = 6.19, SD = 1.04). Based on the means, there is furthermore no clear tendency in the expected direction that emotional appeals would lead to more SMP. Furthermore, there was no significant main effect of emotional granularity, F(1, 385) = 2.50, p = .115, η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2 = .01 on SMP. Respondents with a higher level of emotional granularity (M = 6.24, SD = 1.01) did not indicate significantly higher levels of SMP compared to less granular participants (M
= 6.09, SD = 1.05). Looking at the moderation, a marginally significant interaction effect was found, F(3, 385) = 2.19, p = .088, η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2 = .02.
Simple effects analyses were performed that compared emotional appeals separately for people with low and people with high levels of emotional granularity. The simple effect analysis revealed that when people described themselves as being less granular, neither the anxiety appeal (M = 6.27, SD = 1.00, p = 1.000), nor the guilt (M = 6.10, SD = 1.04, p =
1.000), or disgust (M = 5.87, SD = 1.02, p = 1.000) appeal led to significantly more CC com-pared to the rational appeal condition (M = 6.06, SD = 1.15).
When people indicated to have high levels of emotional granularity, the anxiety (M = 6.27, SD = 1.00, p = 1.000), the guilt (M = 5.93, SD = 1.31, p = .398), and disgust (M = 6.43, SD = 0.69, p = 1.000) appeal again, did not lead to significantly more CC com-pared to the rational appeal condition (M = 6.33, SD = 0.90).
Hence, H2a-c have to be rejected, meaning that exposure to climate change mitigation campaigns framed in a disgust, anxiety, or guilt-inducing way, did not lead to higher levels of SMP than exposure to a campaign framed with a rational appeal. Despite the marginally sig-nificant interaction effect, simple effects analyses did not show any sigsig-nificant moderation ef-fect of emotional granularity, wherefore H5 still has to be rejected in regard of SMP.
Influence of emotional appeals on individual climate change mitigation behaviour
Then it was tested whether campaigns framed in a disgust-inducing way would lead to less climate change mitigation behaviour than exposure to climate change mitigation campaigns framed in an anxiety- or guilt-inducing way. Just as in the first two analyses, a two-way ANOVA was run, this time with IMB as dependent variable.
The analysis revealed that there was once more no significant main effect of emotional appeal, F(3, 385) = 0.93, p = .426, η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2 = .01, indicating that neither the anxiety appeal (M = 6.08, SD = 0.66, p = 1.000), nor the guilt (M = 5.93, SD = 0.85, p = .837), or the rational appeal condition (M = 6.01, SD = 1.02, p = 1.000), led to significantly more IMB than the dis-gust appeal (M = 6.11, SD = 0.81). Looking at the absolute means did not reveal any tenden-cies in terms of groups differences in the expected direction.
Again, no significant main effect of emotional granularity, F(1, 385) = 1.19, p = .277, η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2 = .003 was found. Highly granular people (M = 6.08, SD = 0.77) did not indicate sig-nificantly more willingness to engage in IMB, compared to participants with lower levels of emotional granularity (M = 5.99, SD = 0.91). Also, no significant interaction effect was found, F(3, 385) = 0.82, p = .482, η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2 = .01.
Therefore, H3a and H3b have to be rejected as well, indicating that exposure to a cli-mate change mitigation campaign framed in a disgust-inducing way does not lead to signifi-cantly less climate change mitigation behaviour than exposure to climate change mitigation campaigns framed in an anxiety- or guilt-inducing way. Just like before, emotional granularity did not interact with the effect of emotional appeals on IMB in a significant way, wherefore H5 must be rejected in regard of IMB.
Influence of emotional appeals on climate-related information seeking
The last assumption that was tested, using a two-way ANOVA again, was whether exposure to a climate change mitigation campaign framed in a disgust-inducing way would lead to less CIS than exposure to a climate campaign framed in an anxiety- or guilt-inducing way.
This time, a significant main effect of emotional appeals on CIS was found, F(3, 385)
= 2.86, p = .037, η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2 = .02. A Bonferroni post-hoc test revealed a marginally significant difference between exposure to the disgust (M = 3.74, SD = 1.81) and the anxiety (M = 4.35, SD = 1.71, p = .093) appeal, which indicated that people in the disgust condition were slightly less likely to engage in CIS compared to people in the anxiety condition. For the two other conditions, group comparisons were not statistically significant, but looking at the means, it may be that the disgust appeal also led to minimally less CIS compared to the guilt (M = 3.81, SD = 1.78, p = 1.000) or the rational (M = 3.90, SD = 1.64, p = 1.000) appeal condition.
As in the previous analyses, results indicated no significant main effect of emotional granularity, F(1, 385) = 0.04, p = .851, η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2 < .001. Highly granular people (M = 4.00, SD = 1.75) did not show a significantly higher willingness to search for further climate-re-lated information than participants with lower levels of emotional granularity (M = 3.96, SD = 1.74). Other than that, like before, no significant interaction effect was found, F(3, 385) = 0.91, p = .435, η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2 = .01.
Still, H4a can be accepted, while H4b has to be rejected. Participants who were ex-posed to a climate change mitigation campaign framed in a disgust-inducing way did report marginally significantly less willingness to engage in CIS compared to respondents who were exposed to a climate change mitigation campaign framed in an anxiety-inducing way. This difference did not prove statistically significant when looking at the guilt condition. Again, emotional granularity did not affect this relationship in a significant way, rejecting H5 also for the last dependent variable under investigation.
Considering that emotional granularity as a performance-based measure did not affect any of the relationships previously investigated, it can be concluded that the effect of emo-tional appeals in climate change mitigation campaigns on CC, SMP, climate change mitiga-tion behaviour, and CIS is neither stronger nor weaker for people with higher levels of emo-tional granularity compared to lower granular people, thus requiring to reject H5a and H5b entirely.
Appendix E
Randomization check for gender and educational background
Table 1
Distribution of gender and educational background across emotional appeal conditions rational
condition
anxiety condition
guilt condition
disgust condition
gender male = 24.5%
female = 75.5%
male = 22.8%
female = 77.2%
male = 26.8%
female = 73.2%
male = 22.2%
female = 77.8%
χ2(3, 386) = 0.66, p = .884
educational background
1 = 13.4%
2 = 4.1%
3 = 82.5%
1 = 12.8%
2 = 6.8%
3 = 80.3%
1 = 9.3%
2 = 6.2%
3 = 84.5%
1 = 12.2%
2 = 4.9%
3 = 82.9%
χ2(6, 393) = 1.78, p = .939
Total n = 97 n = 117 n = 97 n = 82 N = 393
*Note. 2 people who identified as non-binary and 5 people who didn’t want to disclose their gender were excluded from the gender randomization check to ensure sufficient statistical power
**Note. For educational background, number 1 refers to people who at least finished high school, num-ber 2 comprises participants who completed some post-obligatory technical education, and numnum-ber 3 represents university graduates and PhDs within the respective condition
Appendix F
Exploratory mediation analyses
Mediation analyses were performed using the PROCESS macro (version 3.5.3) by Andrew Hayes, which uses ordinary least squares regression, yielding unstandardized path coefficients for total, direct, and indirect effects. Bootstrapping with 10,000 samples together with hetero-scedasticity consistent standard errors were employed to compute the confidence intervals and inferential statistics. Effects were deemed significant when the confidence interval did not in-clude zero.
A simple mediation was performed to analyse whether emotional granularity predicted CC and whether the direct path would be mediated by political ideology. An effect of emo-tional granularity on CC was observed, B = 0,114, p = .017. After entering the mediator into the model, emotional granularity significantly predicted political ideology, B = -0.405, p <
.001, which in turn predicted CC in a statistically significant way, B = -0.178, p < .001. Fur-ther investigation of the data revealed that the relationship between emotional granularity and CC was fully mediated by political ideology, indirect effect ab = 0.072, 95%-CI[0.036, 0.114].
A second simple mediation investigated whether emotional granularity predicted SMP and whether political ideology would mediate this direct path. An effect of emotional granu-larity on policy support was observed, B = 0,158, p = .004. As it could be shown in the previ-ous analysis, after entering the mediator into the model, emotional granularity significantly predicted political ideology, B = -0.405, p < .001. Political ideology in turn significantly pre-dicted SMP, B = -0.216, p < .001. Moreover, it was found that the effect of emotional granu-larity on climate change mitigation policy support was fully mediated by political ideology, indirect effect ab = 0.088, 95%-CI[0.047, 0.136].