1
The mediating role of emotions:
How episodic and thematic visual framing affect risk perception in climate change communication
by
Vigourt-d’Hennezel, Laora
Student number: 10602526
Master Thesis
Political Communication
2 The mediating role of emotions: How episodic and thematic framing affect risk perception in
climate change communication
Vigourt-d’Hennezel, Laora
University of Amsterdam
Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of Master of Science (Msc. Political Communication) under the supervision of Richard Van Der Wurff, at the Amsterdam Graduate School of Communication.
3 Abstract
This paper explores the indirect effects of visual framing on risk perception through emotional responses in news media in the context of global warming. As different representational styles are used in newspapers, this study investigates whether the effects of episodic and thematic framing differ in terms of stimulating readers’ perceptions of global warming risk. Based on an experimental study (N = 318) conducted on French-speaking participants, this article presents moderated mediation analyses of episodic and thematic visual framing on risk
perception, testing for emotions as a mediator, together with two important moderations: prior concerns and gender. Results revealed that (1) emotions mediate the relationship between visual framing and risk perception, (2) the extent to which this mediation takes effect depends on prior concerns about global warming, and (3) episodic visual framing did affect the more but in a negative way. This paper provides new insights in the field of risk communication regarding news framing effects on risk perception.
Keywords: risk perception, visual framing, emotions, prior concerns, gender, global
4 Global warming is one of the most important challenges our modern society face today (Field, Barros, Mach & Mastrandrea, 2014). Confronted with this global threat, experts, organizations and political actors need to know how to communicate about this topic in order to predict and handle how people perceive this risk (Wahlberg, & Sjoberg, 2000). Newspapers can provide this information to the public and thus, play a major role in shaping their risk perception (Otieno, Spada & Renkl, 2013).
While many studies on public discourses on environment challenges had a
text-focused approach, in the recent years, researchers advise to examine more carefully the role of visuals in framing public perception nowadays (Doyle, 2011; Hansen & Machin, 2009). Indeed, there is a growing coverage of extreme weathers reported with images in the news (Smith & Joffe, 2009), and visuals are now used as “document evidence of its [global
warming] reality and its causes” (Nerlich & Jaspal, 2013, p.254). What is the impact of these images on the readers’ perception of global warming risk is the question that my study will investigate.
More precisely, I will analyze to what extent emotions, triggered by visual framing, impact reader’s risk perception in the context of climate change. Recently, scholars in risk communication advise to look closely at the role that emotions might play in global warming perception (e.g. Roeser, 2012). This research supports this idea by suggesting the mediating role of emotions in the relationship between visuals and risk perception.
Moreover, media coverage on global warming varies in the representational styles in which information is produced (Otieno et al., 2013). It is precisely framing theories that give scientific insights on how information is presented to the public (Lecheler & de Vreese, 2012). In the context of global warming communication, message framing is effective in rising risk perception (Otieno et al., 2013). However, some grey areas remain on the effects of framing in global warming communication (Morton, Rabinovich, Marshall & Bretschneider,
5 2011). Only recently scholars investigate the concept of visual framing (e.g. Brantner,
Lobinger & Westzstein, 2011). Thus, using an experimental design, this study focuses on the impact of visual framing on risk perception, and focuses particularly on episodic and thematic framings. In addition, this study suggests analyzing the possible moderating role of prior concerns about global warming and gender in the relationship between visual framing and risk perception mediated by emotions. Prior concerns mainly because framing effects differ when the issue is considered as important (Lecheler, de Vreese & Slothuus, 2009) Gender mainly because women and men differ in their emotional reaction and risk perception (e.g.
Gustafsod, 1998, Stedman, 2004).
Furthermore, my study extends past research, which focused mostly on English-speaking individuals, by concentrating on French-English-speaking participants. It considers the example of flooding to illustrate a direct impact of global warming for three reasons. Firstly, even if people cannot experience global warming directly, they can experience its effects and consequences (Whitmarsh, 2008), such as flooding. Secondly, there is a new trend in news media to present the threat “closer to us” (Smith & Joffe, 2009, p. 653) and people respond more emotionally to visuals picturing humans in danger (Lester & Cottle, 2009). Thirdly, flooding is one of the most prevalent impacts of global warming that will appear in Europe in the next years (Field et al., 2014).
Therefore, this research presents moderated mediation analyses including prior
concerns and gender as moderators in the context of global warming, with risk perception as a dependent variable. It attempts to bridge the gap between framing effects, emotional
6 Literature Review
In a large perspective and in the context of global warming, this section aims to first shed some light on the relationship between visuals and risk perception, secondly discuss the evidence that emotions might play a mediator role on this relationship, thirdly introduce the concept of visual framing and finally suggest the moderator role of gender and prior concerns.
First of all, risk perception is defined here as “the perceived likelihood of negative consequences to oneself and society from one specific environmental phenomenon”
(O’Connor, Bord & Fisher, 1999, p.462). Moreover, it relates to the intuitive risk judgments (Slovic, 1987) of individuals that differ between people according to the level of information they have (Messner & Meyer, 2006). Media provide information to the public and play a major role in shaping public perception (Otieno et al., 2013; Wahlberg & Sjoberg, 2000). Thus, newspapers can inform the readers about a potential danger and may therefore increase their risk perception.
Furthermore, prior researchers have investigated the effect of visuals and demonstrated their capacity to represent risk to some extent (e.g. Boholm, 1998). Gibson and Zillman (2000) established that images added in news stories affect the perception of the risk discussed in the text. Indeed, they demonstrated that visuals when combined with text provided additional information to the readers that affect significantly their perception of the issue discussed in the text. In parallel, text information is also better acquired by the readers when combined with images portraying the threat addressed in the text (Gibson & Zillman, 2000). Also in the context of global warming information, visuals added to the text affect people and their perception of global warming risk (Gamson, Croteau, Hoynes & Sasson, 1992). Thus, visuals elements added in newspapers’ articles enhance the text information on a possible risk and may therefore strengthen readers’ perception of this risk.
7 Past research suggested various reasons explaining why images are powerful in
newspapers and may affect readers’ risk perception. Two reasons are worth mentioning in this context. Firstly, visuals are considered as an eye-catcher in newspapers since they are quick and easy to perceive (Bucher & Schumacher, 2006). They capture the attention of the viewer in the first place (Iyer & Oldmeadow, 2006), which arouses the likelihood for the reader to read the article (Bucher & Schumacher, 2006; Wanta & Roark, 1993).
Secondly, after catching the eye of the reader, visuals tend to influence how people
understand the text and perceive the information (Brantner et al., 2011). Indeed, images shape the perception of real issues (O’Neill, Boykoff, Niemeyer, & Day, 2013). They help the readers to identify threat and dangers (Smith & Joffe, 2009) and make the readers witnessing the event (Doyle, 2007). Global warming is a quite invisible issue but images bring truth and can make people feel that global warming is really happening (Doyle, 2007; Brantner et al., 2011). For example, a picture of melting glaciers illustrates the risk global warming on earth (Doyle, 2007). By witnessing this risk, viewers can see the alarming impacts of global warming and subsequently shape their perception of this risk (Smith & Joffe, 2009). Thus, adding an image to the text may affect readers’ risk perception.
Nevertheless, in recent years, researchers in the risk communication field advised examining the role of emotions when investigating news frames effects (e.g. Major, 2011) and when investigating global warming risk perception (e.g. Roeser, 2012). Thus, this study investigates the possible mediating role of emotions in the relationship between visual framing and risk perception.
Emotions as Mediator Between Visual Framing And Risk Perception
Emotions and images. Examining the affective effect of newspaper imagery on the public is an increasing area of empirical research. The emotional power of images can be
8 explained by three reasons. Firstly, images engage the viewer affectively and emotionally (Small, Lowenstein & Slovic, 2007) because of the vividness of their content (Brantner et al., 2011; Smith & Joffe, 2009). As well in the context of global warming, images, as symbols, influence how people feel about climate change (Nerlich & Jaspal, 2013).
Secondly, noteworthy to mention the “picture-superiority effect”, based on dual-coding theory (Gibson & Zillmann, 2000). This effect explained that pictures dominate the content because they are direct, concrete (Gibson & Zillmann, 2000), and easier to remember than words (Paivio, Rogers & Smythe, 1968). Indeed, images “stick in the mind” (Joffe, 2008, p.5) of the reader by making the event portrayed more real and more genuine (Detenber & Reeves, 1996). Thus, due to their emotional connotation, readers tend to recall more easily visuals than the text (Paivio et al., 1968).
Thirdly, visuals gave a stronger emotional engagement than texts (Boholm, 1998) and “send people along emotive pathways where textual material leaves them in a more rational, logical and linear pathway of thought” (Joffe, 2008, p.84). Thus, because of its emotional power, images may trigger stronger emotional responses than text and indirectly affect more readers’ risk perception.
The role of emotions in risk perception. Typically, considered irrational, emotions are often excluded from the field of risk communication, especially in the context of global warming (Roeser, 2012). Nevertheless, so far, some studies already demonstrated the major and positive role that emotions play in risk perception (e.g. Sjöberg, 2007). Firstly, Slovic, Finucane, Peters and MacGregor (2007) and subsequent studies found that emotions determine risk perception. Indeed, through the concept of the affect heuristic, Slovic et al., (2007) explain that individuals make use of their affect to guide their judgments of risk. In other words, if people feel negative about something, they tend to judge the risk as high. In the domain of global warming, Roeser (2012) demonstrated the importance of emotions as
9 factors in risk perception and how it is even crucial to consider the effect of emotions when communicating global warming risk. Indeed, emotional pictures portraying flood disaster significantly influence the perception of flood risk (Keller, Siegrist & Gutscher, 2006). Thus, not only a positive relationship between emotions and risk perception is assumed, but it is expected that visuals added to the text will trigger emotional responses, which in turn contribute to higher levels of risk perception.
Concerning the type of emotions investigated, the present study focuses on negative emotions for various reasons. First, because most of the images on climate change tend to have “largely negative emotional meaning” (Nerlich & Jaspal, 2013, p. 212), the study investigates only negative emotions. Also, negative emotions, such as anger, have a greater effect on risk perception than positive emotions (Sjöberg, 2007). Additionally, anger can be an efficient mediator of framing effects (Lecheler, Schuck, & de Vreese, 2013).
Secondly, in the domain of global warming risk communication, Roeser (2012) suggested for further research to consider emotions such as sadness or worry. Also, Nerlich and Jaspal (2013) found that most of the images portraying climate change tend to have negative emotional connotation such as guilt. Readers who feel guilty when reading a story about global warming, may have higher risk estimates. The combination between anger, sadness, anxiety and guilty has successfully been tested by in the context of investigating news frames effects on risk perception (Otieno et al., 2013). Thus, this study examines the effects of sadness, anxiety, worry and guilty in addition of anger.
In addition to the role of emotions in news frames effects, this study investigates the different effects of two different types of frames (i.e. episodic vs. thematic) since the various manners media communicate about global warming exert a strong influence on the public (Otieno et al., 2013). Thus, this study investigates further the effect of visuals on readers’ perception of risk through their emotional responses by adding the concept of visual framing.
10 Frames: Episodic And Thematic Framing on Emotional Responses
A growing body of literature in the field of framing effects looks closely at the significant effect of framing on emotions (e.g. Aarøe, 2011; Druckman & McDermott, 2008; Gross, 2008; Gross & D’Ambrosio, 2004; Major, 2011). Appraisal theory literature (e.g. Lazarus, 1991) revealed that emotions depend on the appraisal of the situation (Siemer & Reisenzein, 2007), which in turn is based on the power of frames (Gross & D’Ambrosio, 2004). People subjectively evaluate a given event or situation through emotional states, such as feeling angry (Lecheler et al., 2013). In this context, emotions act as channels (Aarøe, 2011; Druckman & McDermott, 2008) and as underlying processes (Gross, 2008; Lecheler et al., 2013) that carry out the effects of news frames. Prior studies investigated the role of emotions as moderators of framing effects. For example, Aarøe (2011) demonstrated that the strength of news framing rely upon the intensity of emotional responses that the individuals feel. Nevertheless, recently, some researchers suggested to examine emotions as mediators of framing effects (e.g. Hart, 2010; Gross, 2008). For instance, Lecheler, Schuck, and de Vreese (2013) established that framing effects on opinion are mediated by emotions such as anger.
Since the use of specific frame affects people differently (Gross, 2008), this study investigates two types of frames: episodic and thematic. Iyengar (1994) distinguished between the episodic frame which “focuses on specific events or particular cases” (p.2) and the
thematic frame which “places political issues and events in some general context” (p.2). In terms of episodic and thematic framing, Gross (2008) was a pioneer in investigating the mediator role of emotions and demonstrated how episodic frames affect issue support through the feeling of empathy. He showed that episodic frames play a greater role on emotional responses of the reader than thematic frames. He argued that episodic frames, more vivid, tend to be more compelling, more emotionally engaging, and more persuasive than thematic frames. The reader is more likely to memorize, to pay attention to and to think about the
11 information presented in episodic frames than in thematic frames that tend to be “pallid statistics” and dispassionate (Gross, 2008).
This research goes further in investigating the effects of different news frames by adding the concept of visual framing in the relationship between news frames and risk perception. Visual framing is newly investigated (e.g. Brantner et al., 2011; Coleman, 2010) and refers to “the selection of some aspects of the perceived reality and their accentuation by visual stimuli” (Brantner et al., 2011, p. 4), “in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation for the item described” (Entman, 1993, p. 52). In other words, visual framing is seen as the choice and production of particular photographs or graphs to portray a specific aspect of the reality on a given topic. Besides, in the context of risk perception, Joffe (2008) demonstrated that the effects of risks messages are mediated by emotions which are also stimulated by visuals.
Hypotheses
Based on the literature discussed above, visuals added to a text are expected to have stronger effects on readers’ emotions and risk perception than the text-only because of their superiority effect (Gibson & Zellman, 2000) and their emotional engagement (Small et al., 2007). In other words, the two visual frames investigated here are expected to arouse more emotions and more risk perception in the readers than when there is no visual frame. In this regard, the following hypotheses are formulated:
H1a: Episodic visual framing gives stronger emotional responses than the text alone. H1b: Thematic visual framing gives stronger emotional responses than text alone. H2a: Episodic visual framing gives higher risk perception than the text alone. H2b: Thematic visual framing gives higher risk perception than the text alone.
12 Different types of visuals may have different effect on the readers, the difference between episodic and thematic visual framings is investigated. Given the vividness of their content, pictures will rather attract the reader's attention than a lackluster graph (Domke, Perlmutter & Spratt, 2002; Smith & Joffe, 2009). Indeed, due to the statistic fatigue phenomenon (Slovic & Peters, 2006), a graph is more likely to be rational and less emotional than a picture (Joffe, 2008). Thus, episodic visual frame are expected to trigger stronger emotional responses and stronger risk perception than thematic frame. The following hypothesis is formulated:
H3a: Episodic visual framing gives stronger emotional responses than thematic visual
framing.
H3b: Episodic visual framing gives higher risk perception than the thematic visual
framing.
Visual framing is expected to indirectly affect risk perception through emotional responses. The following hypothesis is formulated:
H4: Emotions play a mediating role in the relationship between visual framing and
risk perception.
The Moderating Effect of Prior Concerns About Global Warming
Lecheler et al., (2013) investigated the mediating role of emotions in news framing effects. They suggested for further studies to examine these effects in a deeper way by adding a moderator in the analysis, such as prior beliefs. Indeed, when considering a model with mediated framing effects, it is important to deal with moderator variables (Lecheler & de Vreese, 2012). Because individual’s pre-dispositions may influence the effect of news frame on emotions, this current study posits an interaction between visual framing and prior
concerns about global warming. Leiserowitz (2006) explained how people will get the information differently according to the level of importance they gave to the issue. Besides,
13 Lecheler et al., (2009) argued that framing effects differ when the issue is considered as important. The more people care about the issue, the stronger their preexisting ideas on it will be and the less likely the frame is going to have an effect on them (Lecheler et al., 2009). Following this idea, the strength of the relationship between visual framing and emotions will be lower when prior concerns are high. Indeed, the effect of visual framing on emotions may be weaker on individuals who are already concerned about global warming. Thus, prior concerns may moderate the mediation stated above: the mediating role of emotions may exist differently for different conditional values of prior concerns.
Since there is no clear expectation on the interaction between visual framing and prior concerns on emotions, the following research question is formulated:
Research question 1: Does prior concern about global warming moderates the effects of visual framing on emotions?
The Moderating Effect of Gender
This study has the particularity to consider gender differences as an additional
intervening variable that may affect the strength of the relationship between emotions and risk perception. First, gender literature suggested that since women and men differ in experiencing life, gender differences can be seen when investigating the affect (Harriman, 1985, as cited in Russ & McNeilly, 1995). Indeed, research indicated that women and men differ in their emotional reactions (Gustafsod, 1998, Stedman, 2004). Women tend to be more emotional than men in terms of frequency and intensity (Simon & Nath, 2004). For instance, Brebner (2003) investigated the frequency and intensity of emotions experienced in the past months in male and female students. He demonstrated that in general women respond more emotionally than men and experience more negative emotions such as anger (Brebner, 2003). For
14 Secondly, research indicated also that women and men are different in their climate change risk perception (e.g. Gustafsod, 1998, Stedman, 2004). Indeed, women report higher levels of risk perception than men in the context of climate change (Stedman, 2004). To explain this discrepancy, previous scholars have pointed to differences in social roles and everyday activities. Women are more likely to perceive risks related to their home and family (Gustafsod, 1998) because they behave more as nurturer (Davidson, Williamson and Parkins, 2003). On the other hand, men, as income earners (Davidson et al., 2003), tend to perceive risks oriented toward their working life (Gustafon, 1998) and are more likely to whether create or handle it, or even to benefit from it (Flynn, Slovic and Mertz, 1994).
Thirdly, investigating the role of gender as a moderator is scarce in the context of global warming. However, in the field of health risk communication, some studies
investigated gender as a moderator. Indeed, gender differences can appear in the relationship between worry and perceived risk in the context of cancer risk communication (McQueen, Vernon, Meissner, & Rakowski, 2008; Zajac, Klein & McCaul, 2006). Indeed, Zajac, Klein and McCaul (2006) demonstrated that gender had a moderating effect on the relationship between risk perception and worry.
Therefore, according to this line of reasoning, gender differences could be expected in the context of global warming: women may react more emotionally and perceive higher degrees of risk perception. This study aims to see whether emotions are associated differently with perceived risk across the two types of genders. Hypothetically, the strength of the
relationship between emotions and risk perception may be influenced by gender: gender could condition the extent to which emotional responses affect risk perception. Since there is no clear expectation on the effect of the interaction between gender and emotions on risk perception, the following research question is formulated:
15
Research question 2: To what extent do gender differences moderate the relationship between emotions and risk perception?
Method
Sample
The underlying effects of visual framing effects on risk perception were investigated through an online experiment. A total of 318 adults voluntarily participated in the experiment, with a response rate of 73%. Participants ranged in age from 18 to 78 years, with a mean age of 31 years (SD = 14.31). Fifty-six percent of the sample was female (n = 177) and forty-four percent was male (n = 141). As regards level of education, 1% of respondents had no
qualifications, 23% of respondents had attained a high school diploma, 5% obtained a technical qualification, 35% had attained an undergraduate academic degree, 32% had attained a graduate academic degree and 4% obtained a doctoral diploma. Most of the participants were recruited online via a digital advertisement. Some students from the Catholic Institute of Paris volunteered to participate in the study. Only French adults were recruited. Informed consent was obtained from all participants.
Stimulus
Participants read a fabricated news story constructed for this experiment. External validity is relatively high as the information used in the text was taken from different news stories from various French newspapers (e.g. Le Monde, Le Figaro, L’Express). The story presented first the recent flood in Montpellier (France) and its impacts, then, secondly the link between flooding and climate change, and thirdly, how this can have terrible consequences on the human and non-human nature.
16 Two different visuals were added to the text in the two experimental conditions
whereas the control group was composed by the text-alone. The two visuals were different in terms of visual framing: a picture was added to the experimental condition 1 representing episodic visual framing whereas a graph was added to the experiment condition 2 representing thematic visual framing. Indeed, following the previous definitions concerning the distinction between episodic and thematic visual framing, the picture added in the episodic condition portrayed a rescue of a child in a street of a typical French village during a flood. The picture was found online on the website of Météo-Paris which was taken in Rasiguères (France) on the 29th November 2014 by Mr. Monroig for L’Indépendant (Météo-Paris, 2014). In the thematic condition, a graph was added to the text and represented France as whole with different parts exposed to flooding risks accompanied with whether human issue, strong risks or waves flooding. The graph was found online on the website of and made by La chaîne
météo (La-Chaîne-Météo, 2014).
Besides, to ensure that respondents paid attention to the visuals, both visuals took a salient part in the article. Research indicated that the size of the image can prompt different effects: the larger the image, the more people tend to feel in control (Detenber & Reeves, 1996). Thus, both visuals were quite large, placed above the text and their size was equal to the size of the text. The three different versions of the article are included in appendices A, B and C. To establish the validity of the experiment, an online pilot study has been conducted: the images turned out to be effective in terms of episodic/thematic framing and the quality of the text has been assessed.
Procedure
Participants were randomly exposed to one of three conditions: an episodic or thematic visual framing with a text if they were assigned to one of the experimental condition, or only a text without any image if they were in the control group. The text presented was exactly the
17 same in the three conditions. They were first queried to answer questions on how concerns they feel toward societal and political issues, such as terrorism for example, then, secondly, read an article, and thirdly, fill out a questionnaire containing the experimental measures. Both articles and the questionnaire were in French.
Manipulation Check
To ensure that the manipulation was effective, respondents of the first two conditions were asked to rate on a scale from 1 (not at all) to 7 (extremely) four statements designed to measure the two different frames investigated in this study (see table 1). The two visuals significantly differed in terms of framing: it revealed successful manipulation. On one hand, on average, respondents assigned to the episodic visual framing found more that the image illustrates a particular case of flooding (M = 4.00, SD = 2.07) than respondents assigned to the thematic visual framing condition (M = 3.30, SD = 1.78) (t(211) = 2.63, p = .009, 95% CI [0.18, 1.22]). On average, respondents of the episodic visual framing condition found more that the image shows a concrete example of flooding (M = 5.53, SD = 1.54) than respondents of the thematic visual framing condition (M = 4.35, SD = 1.84) (t(211) = 5.10, p < .001, 95%
CI [0.73, 1.64]). On the other hand, on average, respondents of the thematic visual framing
condition found more that the image provides general information about flooding (M = 4.44,
SD = 1.74) than respondents of the episodic visual framing condition (M = 3.64, SD = 1.90)
(t(211) = -3.20, p = .002, 95% CI [-1.29, -0.31]). Thus, the picture representing the episodic framing illustrates a particular case and a concrete example of framing whereas the graph symbolizing the thematic framing described general information.
Risk perception rating. Global warming risk perception was determined by a set of nine items adapted from Leiserowitz’s study (2006) measured on a 7-point scale ranging from 1 (not at all) to 7 (extremely). In Leiserowitz’s study (2006), these items were directly
18 combined together (α = .94). However, in this study, a principal component analysis revealed two factors among global warming risk perception (see Appendix D). Here, only the second factor will be taken into account. The first factor (eigenvalue 4.64, five factor loadings > .71, 51% of variance explained, α = .86) can be best interpreted as the likelihood of some global warming impacts to happen in the future. In the second factor (eigenvalue 1.27, 14% of variance explained), the item “water shortages in general” was positively correlated with both factors and was not connected with the other items that seemed to represent global warming risks that are actually happening. It was therefore deleted and the three other items were grouped together (factor loadings > .69, α = .79). Only the combined measure of the three inter-correlated items constituting the factor 2 (e.g. likelihood that the standard of living will decrease) was taken into account for analysis and termed global warming risk perception. Only factor 2 has been analyzed because in the context of the study, it made more sense to investigate risk perception of actual global warming risks to see how people actually perceive global warming risks today. The fact of having visuals in the experiment can make people realize what the consequences of global warming are nowadays. One may think that it is easier for people to answer questions on how they perceive the current risks than the future risks of global warming.
Emotions ratings. I used a self-response format to measure emotional response. Respondents were asked to rate a limited number of emotions that best describe the nature of the state they feel. This variable appeared first in the questionnaire after reading the text to test the direct emotional response of respondents. The strength of the feeling of a particular emotion is interesting to consider (Gross, 2004) and is relevant given the hypotheses
mentioned above. Based on prior successful approaches (Gross & D’Ambrosio, 2004; Gross, 2008; Otieno et al., 2013), respondents were asked to rate (from 1 = not at all to 7 =
19 worried/guilty”. A principal component analysis conducted on the 5 items suggested one factor among these negative emotions (eigenvalue 2.89, factor loadings > .67, 58% of variance explained). For analysis, a combined measure of these 5 inter-correlated items was used and termed emotional responses (α = .81). Those items have been combined and used successfully in previous studies (e.g. Otieno et al., 2013) except from the items affected which was replaced by worried in this study.
Prior concerns about global warming. This research has the novelty of asking, before any manipulation, how concerned respondents are toward global warming in order to clear up the impact of a possible moderating effect of prior concerns on visual imagery and emotions. Before reading the news item, respondents were asked to estimate how much they are concerned about global warming on a scale from 1 (not at all) to 7 (extremely). Items relative to diverse political and social issues, such as unemployment, were included in order to disguise the research interest. No significant differences were found between conditions concerning prior concerns of people about global warming, F (2, 315) = 1.88, p = .154.
Gender. The questionnaire contains an item on gender. There was no significant difference between groups for the demographic of gender χ2 (2, N = 318) = .02, p = .993.
Analyses
Firstly, to test the first three hypotheses, two analyses of variance (ANOVA) were conducted to identify whether the values of the mediator and dependent variables significantly differed by conditions. The two ANOVA’s allowed me to compare the different respondents’ emotional involvement and their perception of global warming risk.
Secondly, to test the hypothesis 4 and the two research questions, I examined the indirect effects between the three groups and global warming risk perception with a
20 even if basic mediation analyses are based on the causal-steps approach recommended by Baron and Kenny (1986), recent researches tend to consider the bootstrapping analyses (Preacher & Hayes, 2008) when investigating frames effects and dealing with mediators (Hart, 2010; Lecheler & de Vreese, 2012; Lecheler et al., 2013). Moderated mediation can be seen “when the strength of an indirect effect depends on the level of some variable”
(Preacher, Rucker & Hayes, 2007, p. 193). Significant indirect effects are determined by the BCa 95% confidence interval (CI) which does not include a zero.
Following this bootstrapping method, three bootstrapped moderated mediation models were run investigating the indirect effects of episodic and thematic framing on risk perception through emotions, which in turn is moderated by prior concerns and gender differences. 95% bias-corrected accelerated confidence intervals (95% bca CI) have been based on the 5,000 bootstrapping samples for specific indirect effects (Preacher & Hayes, 2008). Process software of Hayes in SPSS was used to conduct the analyses.
So, after conducting the ANOVA’s, I looked at the conditional indirect effects with moderated mediation: Preacher et al., (2007) defined a conditional indirect effect as “the magnitude of an indirect effect at a particular value of a moderator (or at particular values of more than one moderator)” (p. 186).
They also explained that because the a-path and b-path quantify the strength of a simple mediation effect, then “any moderation of this quantity by a moderator, by definition, results in an indirect effect that is conditional on some other variable” (Preacher et al., 2007, p. 195). Preacher et al., (2007) explained that there are some difficulties to obtain regions of
significance in a model with more than one moderator. To overcome this difficulty, conditional values of one moderator, prior concerns, should be chosen, and the regions of significance for the other variable, gender, will be obtain at those conditional values (the ones of prior concerns) (Preacher et al., 2007). Following their bootstrapping model, the indirect
21 effects are tested with one standard deviation above, around and below the mean of the
moderator (in that case prior concerns).
Results
Firstly, I predicted that episodic and thematic framing gives stronger emotional responses and higher risk perception than the text alone. In addition, it was assumed that episodic visual framing gives stronger emotional responses and higher risk perception than thematic visual framing. To test the first three hypotheses, I conducted two analyses of variance (ANOVA):
The first ANOVA examined if the experimental conditions (episodic, thematic and control) differed in respondents’ emotional responses. There was a significant weak effect among respondents in the three conditions on levels of emotions, F(2, 315) = 3.58, p = .029. Subjects who had been exposed to the episodic visual framing feel the least emotions (M = 2.95, SD = 1.12) while those who have been exposed to the text only feel the most emotions
(M = 3.39, SD = 1.45). Subjects who saw the thematic visual framing scored in between (M =
3.33, SD = 1.30). A post-hoc test indicated that the only significant difference found was that between subjects who were exposed to the episodic condition and those who were exposed to the control condition (Mdifference = -0.44, p =.043). No significant difference between subjects who were exposed to the thematic condition and the episodic condition, and no significant differences between thematic condition and the control condition were established.
Secondly, a second ANOVA was conducted to determine visual framing effects on risk perception. There was no significant effect among the three different conditions with perception of global warming risk, F(2, 315) = .41, p = .663.
Therefore, H1a, H1b, H3a are rejected which means that both (1) episodic visual framing and (2) thematic visual framing does not lead to higher emotional responses than the
22 text and, 3) episodic visual framing does not give stronger emotional responses than thematic visual framing. Respondents assigned to the text-only condition had stronger emotional responses to the article they were exposed to than the ones assigned in the other conditions. Also, H2a, H2b, H3b are rejected. No significant difference has been found between the three groups in terms of global warming risk perception. This means that (1) episodic visual
framing and (2) thematic visual framing does not give higher risk perception than the text, and (3) episodic visual framing does not give higher risk perception than thematic visual framing.
Moderated mediations
After comparing groups in ANOVA’s, I decided to conduct further analyses to test the possible moderated mediation effects of prior concerns and emotions, together with the effect of gender, on the relationship between visual framing and risk perception. I assumed that visual framing contributes to more emotional responses (a-path), which are expected to affect subsequently risk perception (b-path) (see figure 1). In addition, I suspected that prior
concerns about global warming moderate the effect of visual framing on emotions (a3i-path) and that gender moderates the link between emotions and risk perception (b3i-path). As explained previously, three bootstrapping models have been run.
Moderated mediation model 1 (episodic vs. text). The first mediation test looked at the difference between the episodic visual framing condition and the text-alone (control) condition (thematic condition was excluded). As explained previously, I tested the conditional indirect effects with moderated mediation. The indirect effects should be
investigated by estimating conditional indirect effects at values of the moderator (Preacher et al., 2007).
The conditional indirect effects were tested at three values of the moderator: at one standard deviation below the mean (M = 3.21; “low prior concerns”), around the mean (M =
23 4.89, “medium prior concerns”) and one standard deviation above the mean (M = 6.56, “high prior concerns”) based on 5,000 bootstrapping samples. The indirect effects estimates
revealed that (1) prior concerns moderated the mediating effect of emotions only with individuals with high prior concerns and (2) gender differences appear in the moderated mediation and are significant only for individuals with high prior concerns (see table 2). Indeed, when looking at individuals with high prior concerns, the indirect effect of episodic visual framing on global warming risk perception was negative and quite strong (for men (b = -.48, SE = .16, 95% BCa CI 0.82; -0.19]) and women (b = -.38, SE = .12, 95% BCa CI [-0.62, -0.16]) – see table 2).
Thus, in this model, a moderated mediation occurred since the strength of the indirect effect depended on the level of the variable prior concerns. Indeed, this moderated mediation happened only when the respondents had high prior concerns about global warming. Results revealed that the effect of the mediator on risk perception is stronger for men than women. So, prior concerns together with emotions have a moderated mediation effect on the relationship between episodic visual framing and risk perception, and this moderated mediation effect is stronger for men than women.
Moderated mediation model 2 (thematic vs. text). A second mediation test looked at the difference between the thematic visual framing condition and the text-alone (control) condition (episodic condition was excluded). The same method of the precedent model 1 has been used here. In the same way, the conditional indirect effects were tested at three values of the moderator: at one standard deviation below the mean (M = 3.24; “low prior concerns”), around the mean (M = 4.93, “medium prior concerns”) and one standard deviation above the mean (M = 6.63, “high prior concerns”).
24 However, unlike the model 1, here, the indirect effect estimates revealed that prior concerns moderated the mediating effect of emotions only with individuals with low prior concerns. Indeed, when looking at individuals with low prior concerns, the indirect effect of thematic visual framing on global warming risk perception was positive and moderately weak (for men (b = .26, SE = .10, 95% BCa CI [0.09; 0.50]) and women (b = .26, SE = .09, 95% BCa CI [0.10, 0.45]) – see table 3). Gender does not differ in this moderated mediation and is significant only for individuals with low prior concerns (see table 3).
Thus, in this model also a moderated mediation occurred but unlike the previous one, here, it happened only when the respondents had low prior concerns about global warming. Here, the effect of the mediator on risk perception is not different for men and women. Prior concerns together with emotions have a moderated mediation effect on the relationship between thematic visual framing and risk perception, and this moderated mediation effect is more or less the same for men and women. Another difference with the previous model is that here the indirect effects are positive and not negative which showed that episodic visual framing and thematic visual framing are different in affecting global warming risk perception. Given the results of the two previous models, it made sense to probe the moderated mediation when comparing directly the two frames (excluding the control group).
Moderated mediation model 3 (episodic vs. thematic). The third mediation test looked at the difference between episodic and thematic conditions (control condition was excluded). The same method of the two precedent models has been used here also. In the same way, the conditional indirect effects were tested at three values of the moderator: at one standard deviation below the below the mean (M = 2.96; “low prior concerns”), around the mean (M = 4.72, “medium prior concerns”) and one standard deviation above the mean (M = 6.48, “high prior concerns”).
25 As in the model 1, the indirect effects estimates revealed that (1) prior concerns
moderated the mediating effect of emotions with individuals with high prior concerns and (2) gender differences appear in the moderated mediation. When looking at individuals with high prior concerns, this indirect effect is also negative but moderately weak (for men (b = -.22, SE = .13, 95% BCa CI [0.52; 0.00]), and women (b = .24, SE = .12, 95% BCa CI [0.50, -0.01]) – see table 4).
However, in this model, significant indirect effects were also seen with respondents with medium prior concerns. Indeed, when looking at individuals with medium prior concerns, the indirect effect of episodic visual framing on global warming risk perception is relatively weak and negative (for men (b = -.17, SE = .08, 95% BCa CI [-0.37, -0.03]), and women (b = -.18,
SE = .08, 95% BCa CI [-0.36, -0.04]) – see table 4). Also, here, gender differences are
significant for individuals with medium and high prior concerns. It was noted that unlike the previous models, here, the moderated mediation is slightly stronger for women than men.
Thus, in this model too, a moderated mediation occurred since the strength of the indirect effect depended on the level of the variable prior concerns. But here, this moderated mediation happened with both respondents with medium and respondents with high prior concerns about global warming. Another difference with the previous models was that here the effect of the mediator on risk perception is stronger for women than men. So, prior concerns together with emotions have a moderated mediation effect on the relationship between episodic visual framing and risk perception, and this moderated mediation effect is stronger for women than men.
So, all in all, in the three models, when considering all the variables a moderated mediation occurred. First, these moderated mediation analyses revealed that there is a significant indirect effect of visual framing on risk perception through emotional responses. Hypothesis 4 that stated the mediating role of emotions in the relationship between visual
26 framing and risk perception is confirmed. Second, the moderated mediation is significantly moderated by prior concerns about global warming. Third, even if similarities were seen in the three models, the moderated mediation effect of prior concerns was different in the three models when considering the level of this variable (low vs. medium vs. high prior concerns). It seemed relevant to investigate the different effects of the two frames when compared with the control group or between them.
Discussion
In this study, the relationship between visual framing and risk perception has been explored in the context of global warming, taking flooding as an illustrative example. The dynamics of this relationship has been investigated by means of an experimental design. I examined the various exposures of French respondents to two different visual framing, namely episodic and thematic framing, in relation with emotional responses and risk perception.
Perhaps the most significant result to emerge from this study is the moderating effect of prior concerns on the relationship between visual framing and emotional responses. Indeed, as expected, the results revealed that prior concerns about global warming moderate the mediation effect of emotions on the relationship between visual framing and risk perception. In other words, the relationship between visual framing and risk perception through the mediating effect of emotions was significant depending on the level of prior concerns about global warming that respondents had.
In terms of specific framing, episodic visual framing did affect risk perception through emotional responses, but in a negative way. When compared to the control group, episodic visual framing has an indirect and negative effect on emotions through the moderating role of prior concerns. When compared to thematic visual framing, episodic visual framing still affect
27 significantly and negatively emotions when people have high prior concerns about global warming but also when people have medium prior concerns.
Thus, when looking at the episodic visual framing, the more the reader is already concerned about global warming before the manipulation, the more the frame will negatively affect his emotions. This confirms the findings of Lecheler et al., (2009) that framing effects differ when the issue is considered as important. However, it does not confirm entirely that the more people care about the issue, the stronger their preexisting ideas on it will be and the less likely the frame is going to have an effect on them (Lecheler et al., 2009). Here, the more people care about global warming, the more likely the episodic visual frame is going to have an effect on them even if the effect if negative.
On the contrary, in terms of thematic framing, this last statement is reversed and results confirmed the findings of Lecheler et al. (2009). Indeed, when compared to the control group, thematic visual framing has an indirect and positive effect on emotions through the moderating role of prior concerns. This frame affect significantly and positively emotions when people have low prior concerns about global warming. So, as Lecheler et al., (2009) found, here, the less people care about the issue, the more likely thematic visual framing will have an effect on them. Adding a graph to a text in newspapers in the context of global warming give stronger emotional responses and risk perception of the issue only for people who are not basically or slightly concerned about the issue presented (global warming in this case).
Thus, overall, evidence suggested that it is important to take into account prior concerns of people about global warming when investigating the effects of visual framing on risk perception through emotions. As Lecheler and de Vreese (2012) suggested, further studies should include moderation in their mediated model, such as prior concerns.
28 Moreover, the evidence on the effect of thematic visual framing on risk perception is in line with previous studies that found positive results in the relationship between visuals and risk perception: reports with no visuals generate the lowest risk estimates (e.g. Gibson & Zillman, 2000). However, on the contrary, the results found on the effect of episodic visual framing on risk perception is not in line with these previous studies. This can be explained by the fact that their approach was different when investigating the effects of pictures on risk perception. For instance, Leiserowitz (2006) looked at affective images (i.e. asking people the first image that comes to their mind when thinking about climate change) and discovered that affective images and negative affect did influence global warming risk perception. Here, people are directly exposed to an image and this might affect them differently and may explain why no direct relationship existed between imagery and risk perception.
Also, here, the picture added to the article captured the reader, but in a negative way. Indeed, reading a text with episodic visual framing causes weaker emotional responses than the text only. This may be due to the quality of the picture since different types of images do not have the same strength in stimulating readers’ emotions (Brantner et al., 2011). From the manipulation check, I know that the picture and the graph selected were efficient in terms of episodic and thematic visual framing. Evidently the readers paid attention to the visual they were exposed to since they were able to answer more or less correctly to the manipulation check questions. Thus, visuals indeed captured the readers.
However, it seems that the text alone was more efficient. One reason explaining this unexpected result might be that when people only focused on one element, which is textual here, they are fully focusing on the information and are more likely to be affected. On the contrary, when people have both image and text, they might have divided their attention between the visual element and the textual element, and thus, be less affected by the details. According to the pilot study, text was quite informative. If people paid more attention to the
29 information when exposed to the text alone than the text with visuals, this might explain the unexpected findings. One of the limitations of my study is that it was conducted online and the exact time and attention that people took to read the article was not assessed. Thus, further research should pay attention to how long and in which conditions people are reading the articles.
Furthermore, another important finding that emerged from this study is the significant mediating role of emotions on the relationship between visual framing and risk perception. My study was in line with previous studies investigating the role of emotions in news frames effect (e.g. Aaroe, 2011; Gross, 2008) by testing how visual framing involves the reader emotionally and how these emotional responses to the message influence risk perception. The results revealed that visual framing significantly affect emotions, which in turn is significantly associated with risk perception. This brings my research into accordance with previous
research demonstrating the significant mediating effect of emotions on news frames (e.g. Lecheler & de Vreese, 2012). For instance, Otieno et al., (2013) demonstrate how human-interest frame strongly affect emotionally the reader and his perception of risk. This study demonstrated that also visual framing affect indirectly risk perception through emotional responses. It supports the idea that the emotional involvement of the readers is valuable to take into account for successful risk communication.
Besides, when considering the mediating role of emotions, another limitation of this study is seen: specific effect of the different negative emotions has not been examined due to a lack of time. Further research should investigate this relationship in a multiple mediation model which examines independently different emotions that may mediate visual framing effects. For instance, Lecheler et al., (2013) investigates the effect of enthusiasm,
contentment, anger and fear in mediating the impact of news frames on political opinion. They found that only anger and enthusiasm, as mobilizing emotions, had mediating effect on
30 the relationship between news frame and political opinion. Thus, even if anger has been investigated in this study, its mediating effect alone on risk perception needs more investigation.
On a final note, when considering gender in the moderated mediation model, differences between men and women have been seen. When comparing episodic visual framing to the text-alone, the effect of the moderated mediation is stronger for men than women. However, when comparing the two different frames, it appears that the moderated mediation effect is stronger for women than men. When comparing thematic visual framing to the text-alone, this effect is the same for women and men.
Investigating gender differences is meaningful because different interventions messages can be required when influencing risk perception of women and men (McQueen et al., 2008). However, considering that gender may have a moderating role in the mediation effect of emotions on global warming risk perception appeared as another limitation of this study. It seemed that gender actually should be taken into account as a control variable. Further research should whether (1) analyze the possible effect of gender on another path of a moderated mediation model or (2) analyze it as a control variable. Thus, the specific role of gender in risk communication needs more investigation.
So, all in all, this research gave insights on the effect of images in the understanding of how risk perception is shaped in the context of global warming. Emotional responses have a significant mediating role on the relationship between visual framing and risk perception. When investigating this relationship and the mediating role of emotional responses, prior concerns about the risk presented in the newspaper’s article should be taken into account. Future studies along these lines can advise journalists, experts, and interested parties in informing and communicating adequately with the public through newspapers about issues such as global warming. For instance, when a journalist informs the public about future global
31 warming risk, he/she should consider the emotional involvement of his/her readers. Another example would be that if interested parties want to communicate on global warming risks, it is important that they take into account how people are basically concerned about this topic in order to anticipate their reaction. Professionals who work with visuals in French newspapers for instance and want to raise people’s risk perception might add a graph to the text in order to improve the information and influence people’s risk perception.
Hopefully, these findings will inspire future studies to investigate further the role of visuals in the field of news framing effects.
32 References
Aarøe, L. (2011). Investigating frame strength: The case of episodic and thematic
frames. Political Communication, 28(2), 207-226.
Bahrick, L. E., Parker, J. F., Fivush, R., & Levitt, M. (1998). The effects of stress on young
children's memory for a natural disaster. Journal of Experimental Psychology:
Applied, 4(4), 308.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social
psychological research: Conceptual, strategic, and statistical considerations. Journal of
personality and social psychology, 51(6), 1173.
Böhm, G. (2003). Emotional reactions to environmental risks: Consequentialist versus ethical
evaluation. Journal of environmental psychology, 23(2), 199-212.
Boholm, A. (1998). Comparative studies of risk perception: a review of twenty years of
research. Journal of risk research, 1(2), 135-163.
Brantner, C., Lobinger, K., & Wetzstein, I. (2011). Effects of visual framing on emotional
responses and evaluations of news stories about the Gaza conflict 2009. Journalism &
Mass Communication Quarterly, 88(3), 523-540.
Brebner, J. (2003). Gender and emotions. Personality and individual differences, 34(3), 387
-394.
33 content. An eye-tracking study on attention patterns in the reception of print and
online media. Communications, 31(3), 347-368.
Coleman, R. (2010). Framing the pictures in our heads. Doing News Framing Analysis.
Routledge: New York, 233-261.
Davidson, D. J., Williamson, T., & Parkins, J. R. (2003). Understanding climate change risk,
and vulnerability in northern forest-based communities. Canadian Journal of Forest
Research, 33(11), 2252-2261.
Detenber, B. H., & Reeves, B. (1996). A bio‐informational theory of emotion: Motion and
image size effects on viewers. Journal of Communication, 46(3), 66-84.
Domke, D., Perlmutter, D., & Spratt, M. (2002). The primes of our times? An examination of the ‘power’of visual images. Journalism, 3(2), 131-159.
Doyle, J. (2007). Picturing the clima (c) tic: Greenpeace and the representational politics of
climate change communication. Science as culture,16(2), 129-150.
Doyle, J. (2011). Mediating climate change. Ashgate Publishing, Ltd..
Druckman, J. N., & McDermott, R. (2008). Emotion and the framing of risky choice. Political
Behavior, 30(3), 297-321.
Entman, R. M. (1993). Framing: Toward clarification of a fractured paradigm.Journal of
communication, 43(4), 51-58.
34 impacts, adaptation, and vulnerability. Contribution of Working Group II to the Fifth
Assessment Report of the Intergovernmental Panel on Climate Change.
Flynn, J., Slovic, P., & Mertz, C. K. (1994). Gender, race, and perception of environmental
health risks. Risk analysis, 14(6), 1101-1108.
Galea, S., Nandi, A., & Vlahov, D. (2005). The epidemiology of post-traumatic stress
disorder after disasters. Epidemiologic Reviews, 27(1), 78-91.
Gamson, W. A., Croteau, D., Hoynes, W., & Sasson, T. (1992). Media images and the social
construction of reality. Annual review of sociology, 373-393.
Gibson, R., & Zillmann, D. (2000). Reading between the photographs: The influence of
incidental pictorial information on issue perception. Journalism & Mass
Communication Quarterly, 77(2), 355-366.
Gross, K. (2008). Framing persuasive appeals: Episodic and thematic framing, emotional
response, and policy opinion. Political Psychology, 29(2), 169-192. Gross, K., & D’Ambrosio, L. (2004). Framing emotional response. Political
Psychology, 25(1), 1-29.
Gustafsod, P. E. (1998). Gender Differences in risk perception: Theoretical and
methodological erspectives. Risk analysis, 18(6), 805-811.
Hansen, A., & Machin, D. (2013). Researching visual environmental
35
Culture, 7(2), 151-168.
Harriman, A. (1996). Women/men/management. Greenwood Publishing Group.
Hart, P. S. (2010). One or many? The influence of episodic and thematic climate change
frames on policy preferences and individual behavior change.Science Communication.
Iyengar, S. (1994). Is anyone responsible?: How television frames political issues. University
of Chicago Press.
Iyer, A., & Oldmeadow, J. (2006). Picture this: Emotional and political responses to
photographs of the Kenneth Bigley kidnapping. European journal of social
psychology, 36(5), 635-647.
Joffe, H. (2008). The power of visual material: Persuasion, emotion and
identification. Diogenes, 55(1), 84-93.
Kamhawi, R., & Grabe, M. E. (2008). Engaging the female audience: An evolutionary
psychology perspective on gendered responses to news valence frames. Journal of
Broadcasting & Electronic Media, 52(1), 33-51.
Keller, C., Siegrist, M., & Gutscher, H. (2006). The role of the affect and availability
heuristics in risk communication. Risk Analysis, 26(3), 631-639.
La-Chaîne-Météo. (2014). Retrieved from http://actualite.lachainemeteo.com/actualite
meteo/2014-12-01-07h10/le-risque-inondation-en-france---dossier-24027.php
36 Lecheler, S., & de Vreese, C. H. (2012). News Framing and Public Opinion A Mediation
Analysis of Framing Effects on Political Attitudes. Journalism & Mass
Communication Quarterly, 89(2), 185-204.
Lecheler, S., de Vreese, C., & Slothuus, R. (2009). Issue importance as a moderator of
framing effects. Communication Research, 36(3), 400-425.
Lecheler, S., Schuck, A. R., & de Vreese, C. H. (2013). Dealing with feelings: Positive and
negative discrete emotions as mediators of news framing effects.
Leiserowitz, A. (2006). Climate change risk perception and policy preferences: the role of
affect, imagery, and values. Climatic change, 77(1-2), 45-72.
Lester, E. A., & Cottle, S. (2009). Visualizing climate change: Television news and ecological
citizenship. International journal of communication, 3, 920-936.
Major, L. H. (2011). The mediating role of emotions in the relationship between frames and
attribution of responsibility for health problems. Journalism & Mass Communication
Quarterly, 88(3), 502-522.
McQueen, A., Vernon, S. W., Meissner, H. I., & Rakowski, W. (2008). Risk perceptions and
worry about cancer: does gender make a difference?. Journal of health
communication, 13(1), 56-79.
Messner, F., & Meyer, V. (2006). Flood damage, vulnerability and risk perception-challenges
37 Météo-Paris. (2014). Retrieved from http://www.meteo-paris.com/actualites/crue-et
inondation-sur-l-aude-et-les-pyrenees-orientales-30-novembre-2014.html
Miceli, R., Sotgiu, I., & Settanni, M. (2008). Disaster preparedness and perception of flood
risk: A study in an alpine valley in Italy. Journal of Environmental Psychology, 28(2),
164-173.
Morton, T. A., Rabinovich, A., Marshall, D., & Bretschneider, P. (2011). The future that may
(or may not) come: How framing changes responses to uncertainty in climate change
communications. Global Environmental Change,21(1), 103-109.
Nerlich, B., & Jaspal, R. (2013). Images of Extreme Weather: Symbolising Human Responses
to Climate Change. Science as Culture, (ahead-of-print), 1-24.
O'Connor, R. E., Bord, R. J., & Fisher, A. (1999). Risk perceptions, general environmental
beliefs, and willingness to address climate change. Risk analysis, 19(3), 461-471. O’Neill, S. J., Boykoff, M., Niemeyer, S., & Day, S. A. (2013). On the use of imagery for
climate change engagement. Global environmental change, 23(2), 413-421.
Otieno, C., Spada, H., & Renkl, A. (2013). Effects of News Frames on Perceived Risk,
Emotions, and Learning. PloS one, 8(11), e79696.
Paivio, A., Rogers, T. B., & Smythe, P. C. (1968). Why are pictures easier to recall than
words?. Psychonomic Science, 11(4), 137-138.
38 and comparing indirect effects in multiple mediator models.Behavior research
methods, 40(3), 879-891.
Preacher, K. J., Rucker, D. D., & Hayes, A. F. (2007). Addressing moderated mediation
hypotheses: Theory, methods, and prescriptions. Multivariate behavioral
research, 42(1), 185-227.
Roeser, S. (2012). Risk communication, public engagement, and climate change: a role for
emotions. Risk Analysis, 32(6), 1033-1040.
Russ, F. A., & McNeilly, K. M. (1995). Links among satisfaction, commitment, and
turnover intentions: The moderating effect of experience, gender, and performance.
Journal of Business Research, 34(1), 57-65.
Siemer, M., Mauss, I., & Gross, J. J. (2007). Same situation--different emotions: how
appraisals shape our emotions. Emotion, 7(3), 592.
Simon, R. W., & Nath, L. E. (2004). Gender and Emotion in the United States: Do Men and Women Differ in Self‐Reports of Feelings and Expressive Behavior? 1. American
journal of sociology, 109(5), 1137-1176.
Smith, N. W., & Joffe, H. (2009). Climate change in the British press: the role of the
visual. Journal of Risk Research, 12(5), 647-663.
Sjöberg, L. (2007). Emotions and risk perception. Risk Management, 9(4), 223-237.
39 Slovic, P., Finucane, M. L., Peters, E., & MacGregor, D. G. (2007). The affect
heuristic. European Journal of Operational Research, 177(3), 1333-1352.
Slovic, P., & Peters, E. (2006). Risk perception and affect. Current directions in
psychological science, 15(6), 322-325.
Small, D. A., Loewenstein, G., & Slovic, P. (2007). Sympathy and callousness: The impact of
deliberative thought on donations to identifiable and statistical victims. Organizational
Behavior and Human Decision Processes, 102(2), 143-153.
Sotgiu, I., & Galati, D. (2007). Long-term memory for traumatic events: experiences and
emotional reactions during the 2000 flood in Italy. The Journal of psychology, 141(1),
91-108.
Stedman, R. C. (2004). Risk and climate change: perceptions of key policy actors in
Canada. Risk Analysis, 24(5), 1395-1406.
Wahlberg, A. A., & Sjoberg, L. (2000). Risk perception and the media. Journal of risk
research, 3(1), 31-50.
Wanta, W., & Roark, V. (1993). Cognitive and Affective Responses to Newspaper
Photographs.
Whitmarsh, L. (2008). Are flood victims more concerned about climate change than other
people? The role of direct experience in risk perception and behavioural
40 Zajac, L. E., Klein, W. M., & McCaul, K. D. (2006). Absolute and comparative risk
perceptions as predictors of cancer worry: Moderating effects of gender and
41 Table 1
Manipulation check results: “In your opinion, the image”: Episodic n = 107 Thematic n = 106 Significance M SD M SD Illustrates a particular case of flooding 4.00 2.07 3.30 1.78 p < .01 Shows a concrete example of flooding 5.53 1.54 4.35 1.84 p < .001
Shows large trends concerning flooding risks 4.41 1.73 4.80 1.61 p = .09 Provides general information about flooding 3.64 1.90 4.44 1.74 p < .01
42 Table 2
Model 1: Conditional indirect effect of the episodic visual framing on global warming risk perception through emotional responses (thematic condition excluded)
Note. Prior concerns about global warming -1SD from mean (3.21); Prior concerns about global warming: mean (4.89); Prior concerns about
global warming: +1SD from mean (6.56). Unstandardized effect sizes. Independent variable coded as 1=episodic visual framing, 0=control
group, Syst. miss. = thematic visual framing. Significant indirect results are presented in bold.
BC 1000 BOOT Prior concerns Gender Mediator Point estimate
(b)
SE LL95 UL95
Emotional responses
Low prior concerns Women .1473 .0850 -.0131 .3255
Low prior concerns Men .1881 .1124 -.0115 .4316
Medium prior concerns Women -.1162 .0684 -.2546 .0180
Medium prior concerns Men -.1483 .0880 -.3304 .0209
High prior concerns Women -.3796 .1179 -.6211 -.1576
43 Table 3
Model 2: Conditional indirect effect of the thematic visual framing on global warming risk perception through emotional responses (episodic condition excluded)
Note. Prior concerns about global warming -1SD from mean (3.24); Prior concerns about global warming: mean (4.93); Prior concerns about
global warming: +1SD from mean (6.63). Unstandardized effect sizes. Independent variable coded as 1= thematic visual framing, 0=control
group, Syst. miss. = episodic visual framing. Significant indirect results are presented in bold.
BC 1000 BOOT Prior concerns Gender Mediator Point
estimate
SE LL95 UL95
Emotional responses
Low prior concerns Women .2559 .0899 .0954 .4515
Low prior concerns Men .2607 .1041 .0891 .5026
Medium prior concerns Women .0447 .0700 -.0877 .1869
Medium prior concerns Men .0455 .0728 -.0854 .2035
High prior concerns Women -.1665 .1130 -.3929 .0516
