1
Fulfilling potential: the prevalence of emotional appeals in climate change news
Student: Joris Holleman Student number: 11220538
Course: Master’s programme Communication Science Master Thesis at Graduate School of Communication Supervisor: Linda Bos
Wordcount: 7217 (excluding tables and graphs) Date of completion: 23-06-2022
2 Abstract
Emotional appeals (EA) play an important role in attracting attention and potentially engaging people with climate change. Different EA have different effects and are seen as the
‘missing link’ link in effective climate communication (Roeser, 2012). Much research has been done on the effects of EA but the prevalence of EA in climate journalism has received no attention yet. This study will investigate the use of positive/negative words as indicators for EA (study 1) while also zooming in on hope appeals (HA) and fear appeals (FA) (study 2).
The sample consists of articles from The Guardian which discussed climate change. Study 1 uses an automated tone detector while study 2 utilises a constructed codebook manually coding HA/FA. Krippendorff’s alpha of the codebook showed that HA/FA are reliably measured and can therefore be used for other research. Both studies found volatility in the trend lines while actual trends were only found in the use of negative words in headlines (increasing) and positive words in article content (decreasing). These results indicate the use of EA is only moderately influenced by trends over time and might be influenced by specific events such as climate school strikes and the COP21. As a consequence, The Guardian, might not use the full potential of EA to engage people with CC due to the limited use of EA and specifically HA.
3 Introduction
In May 2019, The Guardian decided to change its language surrounding climate
change (CC) wanting to ´update their reporting style according to science´ (Carrington, 2019).
This marked a shift in how issues surrounding CC were communicated to its readers. Most citizens obtain CC information through the media, potentially creating engagement with the topic (Newman et al., 2020; Moser, 2010; Areia, 2019; Weathers & Kendall, 2016). At present, we find ourselves in a narrowing window of opportunity to evade the effects of CC (IPCC, 2022). The IPCC authors state the media should provide information contributing to a just, sustainable, and resilient climate community. The information provision surrounding CC has been the subject of numerous studies, ranging from frame analyses (Kenix, 2008;
Feldman, & Hart, 2021), to the impact of journalistic practices on CC reporting (Boykoff, 2013), to analyses investigating which people are present in CC-related news (Parks, 2020;
Boulliane & Belland, 2019). While all relevant, this study focuses on the use of Emotional Appeals (EA) which are described as the ‘missing link’ surrounding effective CC
communication (Roeser, 2012).
Recent evidence suggests that EA in CC-related news are capable of making complex information more understandable (Höijer, 2010), are essential to create a feeling of urgency amongst readers (Roeser, 2012), and increase pro-environmental behaviour compared to non emotionally laden information (Davidson & Kecinski, 2022). However, different EA have different effects on the recipients. For example, (extreme) fear appeals (FA) attract attention but might also decrease pro-environmental behaviour (O’Neill, & Nicholson-Cole, 2009; Kok et al., 2018). In the case of CC-related reporting (extreme) forms of fear are used, negatively affecting the potential of pro-environmental behaviour (Hulme, 2008; Ballet et al., 2022).
Adversely, hope appeals (HA) are, under specific circumstances, also capable of attracting attention, but can increase engagement with CC and stimulate pro-environmental behaviour
4
(Lorenzoni et al., 2007; Salama & Aboukoura, 2018; Hartmann, 2014). Although the effects of EA on readers are extensively studied (for example Ettinger et al., 2021; Salama &
Aboukoura, 2018; O’Neill, & Nicholson-Cole, 2009), to my knowledge, no research investigated the actual prevalence of EA in CC-related communication. This is crucial to understand given the described effects of EA and their potential to induce pro-environmental behaviour. The analysis over time allows for spotting differences between years but also investigating if new forms of journalism such as the rise of constructive journalism are found in CC-related reporting (Bro, 2019; Lough & McIntyre, 2021). Therefore, the major objective of this study is to investigate the prevalence of EA in CC-related news. This leads to the following research question:
To what extent did the use of EA in reporting surrounding the CC, change over the past ten years in the international news outlet The Guardian?
This study used a dataset consisting of CC-related articles from the news outlet The Guardian ranging from 2011 until 2021. The Guardian is chosen as the subject of analysis due to its freely accessible news, international reach, and 19.8 million monthly readers (The Guardian, n.d.). A quantitative approach was employed to analyse EA using positively and negatively charged words as indicators for EA (Höijer, 2010) (analysed with an automated tone detector) (study 1). In study 2 HA and FA were manually identified using a codebook allowing for a more specific analysis of EA. The codebook constructed in this study might provide useful for other researchers wanting to investigate the use of HA and FA in CC- related news. The choice to analyse HA was made because of its potentially beneficial traits and link to constructive journalism (Bro, 2019; Chadwick, 2015) while FA is chosen due to its historical relation with CC-reporting and, in some instances, contrary traits compared to HA (O’Neill, & Nicholson-Cole, 2009; Lorenzoni et al., 2007; Hulme, 2008). The
combination of both studies allows for the analysis of a large set of articles (study 1) while
5
also benefitting from the precision offered by manual coding (study 2).
This study tries to contribute to research on EA by offering a quantitative answer to the question of how much EA are used. On a societal level, this research offers insights into how CC-related news is presented over time. Differences in the presentation might trigger distinctive effects influencing human behaviour (Davidson & Kecinski, 2022). In effect, the results can contribute to deploying efficient communication about one of the biggest
challenges ever faced by humanity.
Theoretical framework
Climate journalism
The media must inform and warn citizens of immediate/future threats ( Zaller, 2003).
In this context, climate journalism should devote attention to the risks of CC and inform citizens accordingly. Recent research showed that the informing role of the media gained prominence in the digital era, moving from ‘gatekeepers’ to information ‘scouts’ guiding readers through the jungle of (false and misleading) information (Brüggeman, 2017;
Shoemaker & Vos, 2009). The implication of this new role of the media is characterised by the adoption of the scientific consensus surrounding CC moving away from ‘balanced’
reporting which focused on showing ‘both sides’ to a ’weight of evidence’ principle which tries to be in line with the scientific consensus (Brüggemann, 2017).
At present, (academic) debates exist around the role of climate journalism. While most authors seem to agree that CC journalism should strive for public engagement with the topic, debates exist on how to achieve this. On the one hand, authors state that CC science is
communicated inadequately and argue for the inclusion of more CC-related uncertainty bound to the complex nature of CC (McIlwaine, 2013; Nisbet, 2019). In this way readers should be able to estimate the severity of CC more accurately, thereby engaging the public. On the other hand, authors argue that it is time to move beyond scientific doubts and engage readers by
6
describing the political debate on how to reach a carbon-neutral society (Pearce, 2019). This would engage people who are more about other topics than CC. In practice, these
propositions to climate journalism might be compromised by the use of mainstream journalistic norms and increased market pressures (Pepermans & Maeseele, 2017; Painter, 2019).
The role of emotions in journalism
Before explaining the use of EA in CC reporting, it is important to distinguish between emotions and EA. In current literature, some authors use both terms interchangeably but this can be problematic. Emotions are defined as a complex state of feeling which encompasses physiological, behavioural, and cognitive aspects and are experienced on the recipients’ level (Salama & Aboukoura, 2018). Journalists can also seek to induce these feelings by creating an emotionally-laden article through EA (O’Neill, & Nicholson-Cole, 2009; Höijer, 2010; Pligt
& Vliek, 2017). One of the ways to increase the use of EA, is to use more
negatively/positively laden words (Höijer, 2010; Ettinger et al., 2021). For example, a
journalist describes the future as ‘catastrophic’ due to a deteriorating climate. In this way, the journalist tries to induce fear by using alarmist language. However, if and what emotion is experienced by the reader of the EA is dependent upon multiple aspects (Rakow et al., 2015).
The link between EA and journalism is a tricky one since journalism traditionally departed from the norm of objectivity (Coward, 2013). In practice, this resulted in a detached form of journalism minimising the use of EA in news (Maras, 2013). However, over the past years, the use of EA attracted attention implying a wider acceptance of EA in reporting (Wahl-Jorgensen, 2016). The shift towards more affective journalism can be explained by the rise of the internet and social media. These technologies altered the role of the audience from a mere receiver of information to active participants influencing journalists and consequently
7
news (Papacharissi, 2015). Moreover, citizens were now capable of sharing content
themselves which was often personal and emotional consequently shaping journalistic norms (Wahl-Jorgensen, 2016). Nonetheless, journalists still adhere to the traditional norm of objectivity and therefore might not accept the use of EA entirely (Wahl-Jorgensen, 2016;
Aitamurto & Varma, 2018).
A second explanation for the use of EA in CC reporting is their beneficial impact on readers. EA can familiarise and simplify complex information thereby increasing readability (Höijer, 2010). It must be noted that the research of Höijer (2010) is a content study and has not measured the effect of EA at the readers’ level. Other studies, relying on an experimental design, investigated the effect of EA on the recipient level and concluded that EA increases engagement, increases support for CC mitigating policies, reduces the distance to the topic, and can mobilise people into pro-environmental behaviour (O’Neill, & Nicholson-Cole, 2009;
Cole et al., 2013). However, a word of caution is necessary since the effects of different EA differ among different forms of EA (Cole et al., 2013). Furthermore, the extremity of EA influences the consequences of the message in multiple and sometimes adverse ways (Roberto et al., 2018). For example, a message containing extreme words of fear can have significantly different consequences compared to a message of moderate fearful language. Other research showed that every person reacts differently to different stimuli based on their historical social experiences and therefore EA might have different and potentially adverse effects on different people (Davidson & Kecinski, 2022).
Nonetheless, the traits attributed to EA can be beneficial for news outlets since they operate increasingly in an attention economy where media options are abundant and attention is a scarce resource. Attention is defined as an investment by the consumer and is essential to compete in a media system (Nielsen, 2020: Boykoff, 2011). Research has shown that
journalists are aware of these advantages and actively apply EA to reap their benefits (Pantti,
8 2010).
Based on the evidence presented above it is difficult to predict how the use of EA changed over the years. On the one hand, journalists recognise the benefits of EA while on the other hand, they still adhere to objectivity steering clear of EA. RQ1 contributes to a
quantitative answer on if EA in/decreases over time. By quantifying the use of EA, one can (indirectly) see the potential influence of CC news has had on recipients over the years.
RQ1: How did the use of EA change over the last ten years?
Hope
Hope is a positive emotion that is elicited when humans expect good things to happen in the future (Höijer, 2010). It is also related to the perceived capability to undertake action to effectively mitigate a threat (Bandura, 1997). Chadwick (2015 p. 598) states that hope is an emotion that is driven by behaviour alterations that encourage ‘taking advantage of future
‘opportunities'. These conceptualisations are translated to individual responsibility where individuals can take micro-actions towards a better climate (Höijer, 2010). HA are
characterised by four components namely, that a message contains a future outcome that is 1) achievable, 2) important, 3) aligned with the goals of the recipient, and 4) creates a better future (Chadwick, 2015).
The reason HA are used in reporting can be split into two arguments 1) the rise of constructive journalism and 2) it attracts attention and increases news value (Bro, 2019).
Discussions exist about the birth time of constructive journalism, while Bro (2019) traces its beginning back to 2008 others find first forms as far back as 1959 (Aitamurto & Varma, 2018). In addition, constructive journalism attracted increased (academic) attention since 2014/2015 showing an increase in academic publications about constructive journalism (Lough & McIntyre, 2021). Key traits of constructive journalism are a focus on solutions for
9
social problems, an ideal for social change, reporting on hopeful practices, and journalists as
‘change agents’ (Aitamurto & Varma, 2018; Bro, 2019; Haagerup et al., 2017)1. Constructive journalism is also defined as an ‘umbrella term’ enveloping other forms of journalism such as impact and future-focused journalism sharing the described traits of constructive journalism (McIntyre & Sobel, 2018). However, constructive journalism should not be confused with positive journalism since it tries to show both positive and negative sides but does not aim for positivity as the end goal (Haagerup et al., 2017).
A second reason for the use of HA are the benefits they offer in the light of the earlier described attention economy. First, HA, like other EA, are capable of familiarising complex information increasing readability. Second, HA are only capable of attracting attention when the message was perceived as unimportant and the proposed hopeful solution was highly probable. Unimportance and the high probability of a solution are not seen as a characteristic of CC news and might trigger a surprise reaction attracting attention (Chadwick, 2015).
In addition, while HA are different from good news, they do use aspects of good news and could, partially, comply with the news value of good news increasing newsworthiness (Haagerup et al., 2017). These traits are beneficial for increasing the competitiveness of a news outlet in the attention economy of the media environment. Due to increased attention towards constructive journalism (Lough & McIntyre, 2021) and the beneficial traits of HA, I expect that its use has increased in the past 10 years. The answer to this hypothesis might also serve as an indication for the spread of constructive journalism in CC-related news.
H1: The use of HA has increased over the past 10 years.
1 For essential articles defining and describing constructive journalism read Haagerup, 2017 & Gyldensted, 2015.
These are coined as the two main proponents of constructive journalism.
10 Fear
Fear is a negative emotion that is generated by the idea that bad things happen in the future (Höijer, 2010). Present threats can also induce fear and the key common ground is that the outcome of a situation is seriously bad. Fear is a warning system for humans and serves the purpose to change one’s behaviour to adapt to the situation (Ballet et al., 2022). FA are characterised by two aspects, 1) an attempt is made to arouse fear by posing a threat and 2) the posed threat is relevant to the recipient (Ruiter et al., 2014). In the light of CC reporting, FA come in the form of threatening information about the climate, such as ‘droughts,’ and relate to the recipient by stating that droughts may cause famine in the future. This example explains why droughts are relevant for humans which makes it easier to relate to the threat while it also aims to induce fear.
The use of FA to explain CC-related topics dates back to as early as 1963 when the first scientists started to raise the alarm about potential climatic dangers (Hulme, 2008). Early CC news used fearful terms but also attenuated these terms due to the lack of scientific consensus surrounding the existence and severity of CC. The mid-late 1980s were
characterised by a scientific consensus surrounding the dangers of CC, and terms of fear, such as ‘chaos´ and ‘collapse’, were increasingly used to explain the urgency of CC (Hulme, 2008). In recent years, the language surrounding CC has evolved into an apocalyptic
discourse describing CC in more extreme terms of collapse showing the apparent link between CC and FA (Ballet et al., 2022).
A second reason for the use of FA in CC reporting relates to the potential benefits it offers in the light of the attention economy. FA are effective in grabbing readers’ attention (O’Neill, & Nicholson-Cole, 2009). However, these results must be approached with caution since the intensity of the FA used in experimental studies shapes the effects of FA (Pligt &
Vliek, 2017; Roberto et al., 2018; Ballet et al., 2022). Therefore, some authors argue that a
11
more nuanced, less extreme view of CC would be fitting (Nisbet, 2019) while others state that these extremes are a scientific reality and should therefore be communicated (Reser &
Bradley, 2017). The second benefit of FA is that they comply with the news value ‘negativity’
increasing the newsworthiness of CC news (Molek-Kozakowska, 2017; Harcup, & O’Neill, 2017; Boykoff, 2011).
Based on the concepts discussed above one can get an image of why FA are used in CC communication. FA are historically bound to CC-related news (Hulme, 2008) while also offering benefits in the light of the attention economy such as attracting attention and
increasing newsworthiness (O’Neill, & Nicholson-Cole, 2009; Molek-Kozakowska, 2017).
However, to my knowledge, no research has investigated the prevalence of FA and for this reason, RQ2 is constructed. The answer to RQ2 will fill this knowledge gap providing insights into how CC is communicated.
RQ2: To what extent did the use of FA change in the last 10 years?
Second, Based on the historical bond between FA and CC, and the beneficial traits of FA I expect that FA will be more prevalent when compared with HA (H2).
H2: FA are more prevalent than HA.
Study 1
In this first study, I discuss the trends surrounding positively and negatively laden words in news surrounding CC. Emotions can be expressed through the use of
positive/negative words and function as an indicator for EA (Hancock et al., 2007). For example, Ettinger et al. (2021) use alarmist language by including negative words as a
stimulus to trigger FA while using positive words and language to trigger HA. This study also found that negative words in article content were significant predictors for FA while positive words were significant predictors for HA (see Tables 6 and 7 in the appendix). This is by no
12
means a perfect measure of EA but it does allow for automated analysis of thousands of articles. The percentage of positive and negative words will be examined in both the headlines and the actual content of the article. This constitutes the first answer to research RQ1, which stated:
How did the use of EA change over the last ten years?
Method
Sample:
The sample consisted of articles from The Guardian which was chosen as the subject of analysis due to its (international) reach of 19.8 million readers per month (The Guardian, n.d.). Only articles that discuss or mention CC were included. The sample stretched from 2011 to 2021 to capture a trend over multiple years while also including the rise of attention towards constructive journalism which started around 2013-2014 (Bro, 2019; Lough &
McIntyre, 2021). This sample was retrieved by using the open API of The Guardian. Only articles from the ’environment’ section which mentioned ‘climate change’ or were tagged as
‘climate change’ were included. This was done to increase the number of relevant articles.
Videos, photo items, and liveblogs were excluded from the sample since these were not applicable for analysis. This resulted in a final sample of 12110 articles which contained different forms of articles such as blog posts, reader comments, analyses, interviews, and news articles (N=12110).
Data preparation
To analyse the tone of the headlines and content of the articles, the python
module lexicon VADER was used. This tool calculates the percentage of negative, neutral, and
13
positive words in a given line of text returning a continuous score from 0 to 1. The VADER module is as good as or better in predicting tone compared to human coders and as good as or better than other sentiment analysis modules (Hutto & Gilbert, 2014).
Analysis
To investigate RQ1 two line plots are made to analyse potential trends over time. This results in four different trend lines (positive and negative words in the headline, positive and negative words in content). These visualisations provide insight into how the average use of positive and negative words might have changed over the years. To statistically investigate the trendlines an Augmented Dickey-Fuller (ADF) test is performed to test if a trendline is stationary (no trend present) or non-stationary (trend present). When a trend is non-stationary, the visualisation can be used to show in which direction a trend moves.
Results
Headlines
On average, negative words made up 12% (M= 0.12, SD= 0.02) of the total used words in headlines compared to 8% of positive words (M= 0.08, SD= 0.02). However, Graph 1 shows a wide variety in the use of positive words reaching a minimum of 4% and a maximum of 14%. An ADF test is used to analyse if a trend can be found in the average use of positive words and shows that there was a significant result for ADF= -6.43, p<.01. This implies that the line is stationary meaning there is no trend and time is not a predictor of positive words in headlines.
Second, negative word use in headlines varied widely over the months measuring a minimum of 7% and a maximum of 19%. An ADF test showed that the trend line for negative words in headlines was non-stationary implying that a trend is present for ADF= -2.05, p=.26.
14
When viewing Graph 1 the line suggests there is an upward trend with an interesting peak around 2019 (see discussion study 1 for further elaboration).
Graph 1: This graph shows the average use of positive and negative words in headlines of articles reporting on CC.
Article content
On average, positive words (M=0.09, SD= 0.01) made up a larger share of article
content than negative words (M= 0.06, SD<0.01). However, the percentage of positive words varies among different months measuring a minimum of 9% and a maximum of 11%. An ADF test was conducted which showed that the trend line of positive words in article content was non-stationary for ADF= -1.38, p= .59. When looking at Graph 2 the trend of positive words in article content seems to be a downward one. The use of negative words in article content is also characterised by volatility measuring a minimum of 5% and a maximum of
15
7%. An ADF test showed that the use of negative words in article content was stationary, indicating the lack of a trend for ADF= -3.17, p= .02.
Graph 1: This graph shows the average use of positive and negative words in article content of articles reporting on CC.
Discussion
Both Graph 1 and 2 showed high volatility therefore I would like to highlight the year 2019 as an example and provide context. While the method of this study cannot prove if the events described below affected the use of positive and negative words it can provide an insight in how events might shape the use of EA. The year 2019 was characterised by climate school strikes, extinction rebellion campaigns, large burnings of the Amazon, and the
prominence of Greta Thunberg in the media. These topics lend themselves to the use of negative words to create urgency which might be reflected in the use of negative words in the
16
headlines. This shows how certain events might shape the use of EA in climate reporting rather than a trend over time.
Concerning RQ1 the use of negative words in article content and positive words in headlines are remain roughly the same indicating there is no trend present. This study only found evidence for increased use of EA in negative words in headlines, while positive words in article content showed a decline. In addition, Graphs 1 and 2 showed a high volatility which might indicate that factors aside from time influence the use of EA. To investigate the use of EA more precisely study 2 is conducted.
Study 2
The second study zooms in on specific EA namely, HA and FA. This allows for a
more precise analysis of EA beyond the use of negative/positive words. Study 2 allows for answering the following questions and hypotheses.
RQ2: To what extent did the use of FA change in the last 10 years?
H1: The use of HA has increased over the past 10 years.
H2: the use of FA is more prevalent compared to the use of HA.
Adding to this study 2 will contribute to RQ1 by searching trend lines for two specific EA.
Method
Sample
The dataset of the study 2 consisted of a manually coded sample (N=311) of the larger dataset described in study 1. The sample was drawn from the years 2011, 2013, 2015, 2017, 2019, and 2021 and contained 50 articles each year. These articles were spread over the entire
17
year to reflect the news surrounding accordingly. In practice, this meant that each year was sorted on publication date and had a certain percentage of the articles that were coded to achieve 50 articles. For example, in 2011 each 11th article was coded resulting in a spread sample spanning across all the months of a year.
Operationalization
To establish whether HA/FA were present in an article, a codebook was constructed (see appendix for full codebook) containing four questions about HA and FA each. The questions represented different forms of HA/FA.
Hope appeals
HA come in different forms but, in general, relate to an achievable better future for the individual or humanity as a whole. A key aspect is the description of a better future, in other words, how something could be better for humans (Höijer, 2010). For this reason, only outcomes beneficial for humans are seen as eliciting hope. For example, increased job opportunities are seen as hope while the reduction of CO2 emissions without an explanation of why this reduction is beneficial to humans is no form of hope. The benefits could be on many distinct levels such as individual opportunities but also for humanity as a whole. Based on the literature, four forms of HA were constructed: 1) individual achievability 2) collective achievability 3) a better future 4) positive language and trends.
Individual achievability describes actions an individual could take to achieve a better future, for example, proposing to eat less meat (Höijer, 2010). Collective achievability focuses on actions towards a better future which humans could take as a collective such as a society or humanity as a whole (Chadwick, 2015). For example, a country that states it has potential for a modern economy by transitioning to a low carbon economy. The better future appeal is
18
based on the notion of ‘opportunities’ which states that hope is also elicited once humans think they can benefit from outcomes in the future (Chadwick, 2015). For example, if an article describes how solar energy might reduce energy costs in 10 years. This appeal differs from the appeal of collective achievability by not requiring specific actions to achieve a better future but explaining the opportunities of this better future. At last, positive language and trends relates to the notion of Ettinger et al. (2021) who state that hope is characterised by the capability to mitigate the negative effects. Positive language and trends are seen as proof that effective mitigation is possible and describe a form of HA.
Fear appeals
FA are formed when describing a threat which elicits or tries to elicit fear. An essential part of eliciting fear is that the message should be relevant to the recipient. For this reason, I chose to define only articles that contained a clear threat to humans as FA. For example, extreme weather was seen as FA while the extinction of animals was not defined as FA if the article did not describe why this extinction was threatening to humans. Immediate
threat focuses on threats that are already present such as the description of extreme weather happening across the world (Ballet et al., 2022). Future threat describes threats that lie in the future and are predicted to happen (Höijer, 2010). For example, threatening situations
described in CC-related prediction reports such as the IPCC. Threat due to inaction describes a threatening situation if no action is undertaken. Threat due to inaction differs from Future threat since it focuses explicitly on the lack of human actions instead of predicted threats (Ettinger et al., 2021). At last failure of proposed solutions describes how proposed solutions will fail humanity and thereby increase CC-related threats (Ettinger et al. 2021). For example, when an article describes how solar panels cause more extreme weather.
Intercoder reliability
19
To test if the codebook was reliable, 31 randomly selected articles were coded by a second coder (10% of the entire sample). Before coding extensive training and adjustments to the codebook were necessary. The Krippendorff’s alpha for FA was 0.79 and for HA of 0.76.
For the Krippendorff’s Alpha’s of the other appeals see table 1.
Table 1:
Krippendorff’s Alpha per appeal
Appeals Krippendorff’s Alpha
Individual achievability 0.67
Collective achievability 0.72
Better future 0.71
Positive language and trends 0.78
Immediate threat 0.76
Future threat 0.64
Threat due to inaction 0.71
Doubts surrounding solutions 0.78 Note. Krippendorff’s Alpha per separate appeal.
Analyses
The second study consists of multiple parts analysing the average use of HA and FA.
The first step is to visualise the average trend of both appeals over the years and statistically test this trend using logistic regressions, regressing the presence of EA on year controlled for article length and type. The average scores of HA and FA were compared to see which one more prevalent, using multiple sets of t.tests. At last, the different forms of HA/FA will be visualised and analysed to spot differences between specific appeals over time. Multiple Kruskal Wallis tests are conducted to spot statistically significant differences between years
20
and specific FA/HA. At last, I will offer some context to show how events might influence the use of appeals.
Results
Hope and fear appeals trends
Graph 3 visualizes the use of HA (M= 0.2, SD= 0.17) over the years which does not seem to show a trend in the use of HA. However, to test if time predicted the use of HA a logistic regression is conducted with article length and type functioning as control variables.
Table 1 shows that, as predicted by viewing Graph 3, there is no significant effect of the trend variable which is a variable for different years in the dataset for 0.04, p=.64. This indicates there is no significant difference in the use of HA between different years. However, the type of article did seem to influence the use of HA since ‘blog’ (1.06, p = .02) and ‘comment’
(1.58, p<.01) have a significantly higher probability to contain HA compared to the reference type ‘news article’. It must be noted that the explained variance of this model is low with Pseudo-R2= 0.06.
21
Graph 3: Average use of HA in years. The grey area surrounding the line represents the differences in averages measured per month.
Table 2:
Regression table HA
95% confidence interval
β SE Z-score p-value Lower
limit
Upper limit
Trend 0.04 0.1 0.43 .67 -0.15 0.23
Length <0.01 <0.01 0.54 .59 -0.00 0.01
Blog 1 0.44 2.3 .02 0.15 1.87
Letter 1.09 0.6 1.83 .07 -0.08 2.25
Comment 1.58 0.54 2.94 <.01 0.52 2.65
Analysis -0.76 1.07 -0.72 .47 -2.85 1.33
Feature 0.02 0.54 0.04 .97 -1.05 1.09
22
Other -1.28 1.06 -1.2 .23 -3.35 0.8
Constant -1.94 0.47 -4.16 <.01 -2.86 -1.03
Pseudo-R2= 0.06, Log-Likelihood= -146.19
Note. The regression table of HA, significance p<.05.
second step is to analyse the use of FA (M= 0.43, SD=0.5) as shown in Graph 4. When looking at the graph there is no apparent trend present. However, this is also tested using a logistic regression which controls for article length and type (see Table 2). The regression analysis showed no over time significant differences in the probability that an article contained FA. Based on these results one cannot find a trend or difference in the use of HA over the years. Adding to this the explained variance was low pseudo-R2= 0.01 - indicating that other factors besides time, article type, and article length influence the use of FA.
Graph 4: use of FA (annual average) Table 3:
Regression table FA
95% confidence interval
β SE Z-score p-value Lower
limit
Upper limit
Trend 0.04 0.07 0.6 .55 -0.01 0.19
Length <0.01 <0.01 0.62 .54 -0.00 0.01
Blog 0.17 0.39 0.44 .66 -0.59 0.92
Letter 0.25 0.56 0.45 .66 -0.84 1.34
23
Comment 0.64 0.52 1.22 0.22 -0.39 1.66
Analysis -0.48 0.63 -0.76 .45 -1.72 0.76
Feature -0.58 0.43 -1.34 .18 -1.43 0.27
Other -0.56 0.52 -1.08 .28 -1.59 0.46
Constant -0.51 0.35 -1.46 .14 -1.2 0.18
Pseudo-R2= 0.01, Log-Likelihood= -210.44
Note. The regression table of FA, significance p<.05.
To see how HA and FA relate to each other multiple t.tests are conducted to see if one is used significantly more in reporting surrounding CC. A t.test showed that, overall, FA are significantly more prevalent for H(1)=34.84, p<.01. A t.test showed that in 2011 there was no significant difference between the use of hope and FA for H(1)= 2.93, p= .08. The same situation occurred in 2015 when the t.test found no significant difference between the use of fear and HA for H(1)= 0.59, p=.44. In 2013 FA were significantly more prevalent for H(1)= - 5.16, p< .01. A t.test showed that in 2017 FA were more prevalent than HA for H(1)=-3.1, p<.01. The same situation occurred in 2019 where FA were more prevalent than HA for H(1)=-3.47, p<.01. At last, in 2021 FA were more prevalent than HA for H(1)= -2.29, p= .03.
Concerning H3 these results can conclude that, in general, FA are more prevalent.
Different appeals over time
The last step is to investigate how different forms of HA and FA relate to each other
over time. Both FA and HA will be analysed separately to show the differences between these two different appeals. First, the most used HA is Collective achievability (M= 0.15, SD=
0.16) while Individual achievability was used the least (M= 0.04, SD= 0.09). However, the use of these appeals does differ over time as shown in Graph 5. To test if the differences are significant, multiple Kruskal Wallis tests have been conducted to test differences per year and
24
between different HA. Table 3 shows where these differences occur. When comparing years only Better future differed significantly where 2015 used significantly more of the appeal compared to 2017 and 2019 (also see graph 5). Overall, Individual achievability and Positive language and trends were used significantly less compared to Collective achievability and Better future. However, when zooming in, these differences vary are not the same each year (see Table 4).
Table 4:
Different HA over time and their differences
Different HA
Years Individual
achievability
Collective achievability
Better future Positive language and trends
Graph 5: The use of different hope appeals over time (annual average)
25 All years
average
0.04 (0.09) a 0.15 (0.16) b 0.13 (0.16) b 0.05 (0.12) a
2011 0.02 (0.07) a 0.18 (0.21) b c 0.14 (0.17) b c x y
0.04 (0.14) a c
2013 0.06 (0.12) a 0.15 (0.14)a b 0.14 (0.14)a b, x, y
0.02 (0.07)a c
2015 0.04 (0.11)a 0.2 (0.18) b c 0.24 (0.21) b c y
0.11 (0.17)a c
2017 0.03 (0.08) 0.11 (0.15) 0.08 (0.11) x 0.08 (0.14) 2019 0.05 (0.09) a 0.16 (0.13) b 0.07 (0.11)a b x 0.02 (0.07) a 2021 0.04 (0.1)a b 0.12 (0.16) a 0.09 (0.14)a b x
y
0.02 (0.06) b
Note. The letters a,b and c indicate a difference between different for of HA tested through a test with a significance level of p< .05.
The letter x, y tested a difference between specific appeals per year for p<.05
The first number in the cells are the mean and the second numbers are the standard deviations.
Example: the top row shows that Individual achievability differs significantly from Collective achievability and Better future and not from Positive language and trend while Positive language and trend are not significantly different from Collective achievability and Better future.
Second, the different use of FA are analysed showing that the most used FA was
Future threat (M= 0.34, SD= 0.26) while describing Failure of proposed solutions was used the least (M= 0.08, SD= 0.14). It must be noted that the use of different FA differs over time which is shown in Graph 6. To calculate if differences between years occur several Kruskal Wallis tests are conducted which showed that Future threat was used significantly less when comparing 2013 with 2015. The use of Threat due to inaction also differed among years
26
showing that 2011, 2015, 2017, and 2019 differed significantly from 2013 (see table 4). To test differences between different FA multiple Kruskal Wallis tests were conducted which showed that, on average over all the years, Failure of proposed solutions and Future threat differed significantly from Immediate threat and Threat due to inaction. However, when zooming in this effect differed per year (see table 5).
Table 5
Different FA over time and their differences
Different FA
Years Immediate
threat
Future threat Threat due to inaction
Failure of proposed solutions All years
average
0.17 (0.18) a 0.34 (0.26) b 0.2 (0.21) a 0.08 (0.14) c
2011 0.14 (0.14) a b 0.24 (0.2) a x y 0.1 (0.16) a b x 0.04 (0.1) b 2013 0.15 (0.2) a 0.45 (0.25) b x 0.3 (0.23) a, b y 0.18 (0.23) a
Graph 6: The use of different FA over time (annual average)
27
2015 0.13 (0.14) a b 0.23 (0.13) b y 0.18 (0.17) a b x y
0.08 (0.11) a
2017 0.23 (0.22)a c 0.43 (0.32)c x y 0.12 (0.17)a b x y
0.03 (0.08) b
2019 0.21 (0.17) a 0.4 (0.26) b x y 0.25 (0.23) a b x y
0.07 (0.11) c
2021 0.18 (0.19) 0.3 (0.29) x y 0.26 (0.24) x y 0.1 (0.14) Note. The letters a,b and c indicate a difference between different for of FA tested through a Kruskal Wallis test with a significance level of p< .05
The letters x and y indicate a difference between different years tested through multiple Kruskal Wallis tests with a significance level of p< .05
The first number is the mean and the value between brackets is the standard deviation.
Example: The bottom row shows that ‘immediate threat’ is not significantly different from the other appeals while ‘future threat’ is significantly different from ‘failing proposed solutions’.
Discussion
In this section, I would like to offer some context to Graphs 3 and 4. This is done to
explain what could cause the volatility in HA/FA use between different years since a clear trend is absent in the data. It is beyond the scope of this research to 1) explain what topics might explain a change in the use of EA and 2) describe the context for each year present in the dataset. Based on these reasons, I will offer context on the year 2015 since the use of FA was lower (see Graphs 4 and 6) while the use of HA was higher (see Graphs 3 and 5). CC news of 2015 was characterised by the Paris agreements and the months leading up to the conference. A possible explanation for the increase in HA could be that action groups tried to raise attention for alternative energy production in the months leading up to the climate agreement. An example of this was the major attention to divesting from fossil fuels offering multiple advantages of a better future according to a Guardian-initiated action group. Another explanation could be that prominent politicians and figures stated more hopeful messages surrounding the major CC convention in Paris lowering their use of fearful language. This
28
context shows how other factors than time might influence the use of HA/FA.
Based on the results presented above, RQ2 can be answered. The use of FA has not changed significantly over time showing no (statistically significant) signs of an
increase/decrease in its use while the Graphs show some volatility. Based on this study one cannot conclude that the use of HA/FA has increased.
It was hypothesized that the use of HA increased over time (H1). This hypothesis is rejected based since no statistically significant trend was found. At last, the results indicate that FA were significantly more prevalent over time, therefore, accepting H2. However, the difference was not present in the years 2011 and 2015. Concerning RQ1 one cannot conclude that there is a different use in EA use over time due to the absence of significant trends over time.
Conclusion
This research set out to investigate the prevalence of EA, in specific HA and FA, in CC-related news. Study 1 used an automated tone module which allowed the analysis of large numbers of articles while Study 2 zoomed into specific HA and FA. Based on the two studies one can conclude that there is no clear rise in EA in CC-related articles. The tone analysis showed an increase in negative words in headlines but found a decrease in positive word use in article content. Study 2 showed that there was no significant trend over time although this was expected for HA. The analysis also showed that FA were more prevalent compared to HA, as expected. Study 2 also showed that the codebook used to identify HA/FA was reliable and could be used for further research. Both studies showed volatility in the use of
positive/negative words and HA/FA indicating that factors other than time influenced the use of EA.
Study 1 showed that negative words in headlines made up a larger share of the
29
headline compared to positive words. Interestingly, in the content of the article positive words made up a larger share of the words compared to negative words. There are several possible explanations for this result. The use of negative words is actively promoted and seen as newsworthy compared to positive news which could explain the larger use of negative words in headlines (Parks, 2019). This effect is strengthened once messages have to be shortened, as is the case in writing headlines, negative words tend to survive the cut being regarded as more
‘successful’ (Gligorić et al., 2019). Graphs 1 and 2 also showed very volatile lines indicating that other factors influence CC news such as earlier described Extinction Rebellion campaigns and school strikes (Richardson, 2020). The context sketched in the discussion, paired with the low explained variance, showed that newsworthy events potentially shape the use of EA.
In study 2 I expected to find a rise in HA due to the increased interest in constructive journalism. However, no evidence was found to support this hypothesis. A possible
explanation for the absence of a trend is that constructive journalism attracted attention in academic spheres more than in news outlets (Lough & McIntyre, 2021). Another possibility is that constructive journalism is found in specific alternative news sites and not, or to a lower degree, in more mainstream news outlets such as The Guardian (Pepermans & Maeseele, 2017). In addition, study 2 showed that FA were used more often compared to HA. This difference can be explained by the historic link between FA and CC (Hulme, 2008; Ballet et al., 2022). Interestingly, in 2015, there was no significant difference in the use of FA and HA which could be explained by the wide use of the sustainability frame which focused on solutions and opportunities used in communication surrounding COP21 (Pan et al., 2019). A second explanation could be that surrounding COP21 NGOs and protesters attracted media attention while sending messages of climate justice and other solutions which could
potentially have influenced the different use of appeals (Wozniak et al., 2017; Wahlström et al., 2013). These results suggest, as in study 1, that events seem to shape the use of EA instead
30 of a trend over time.
These results suggest that the media does not use its full potential as a mobiliser of people to take CC mitigation actions. First, positively framed news increases engagement with CC at the recipients' level (Baden et al., 2019). However, at present, the use of positive words is in decline in article content which might mean that potential engagement with CC decreases. Second, behavioural change is more probable when individuals perceive
behavioural change as an effective measure to mitigate a threat (Pligt & Vliek, 2017). HA, and consequently constructive journalism, can show people effective solutions to mitigate the threat of CC increasing the possibility of a behavioural change towards CC mitigation. This effect is possibly even stronger when HA and FA are both present in an article (Hartmann et al., 2014). The lack of HA indicates that the potential of media as a mobiliser is not fully used. This is also seen in the scarce use of Individual achievability which is seen as an effective manner to induce hope and consequently empower people to CC action (Chadwick, 2015). Based on this study, no evidence is found to support a rise in constructive journalistic practices in reporting surrounding CC in The Guardian.
The limitations of this study lie in the use of The Guardian as the only analysed news outlet. First, The Guardian is seen as a left-leaning news outlet which could affect the
frequency and topics discussed compared to more right-leaning newspapers such as The Times and The Telegraph (Saunders et al., 2018). Second, The Guardian linked extreme heatwaves (62%) way more to CC compared to other right-leaning (British) newspapers (14%-19%) affecting the tone of the articles (Painter et al., 2021). Linking CC to extreme weather events increases the use of FA which might skew the results and lowering
generalisability (Ballet et al., 2022). However, the focus on The Guardian also increased the probability to find a trend of EA compared to a more right-leaning news outlet due to their lower frequency of writing CC-related news. Third, The Guardian is a news outlet from the
31
global north which, in comparison to news outlets from the global south, dedicates less attention to the impact of CC on humans which is an essential part of HA/FA lowering the generalisability of the results over other news outlets (Hase et al., 2021; Chadwick., 2015;
Ruiter et al., 2014). Other news outlets were excluded due to a lack of data access and
language barriers. Furthermore, the design was limited to only investigating positive/ negative words and the use of HA/FA. This is not the most optimal way of measuring EA since 1) positive and negative word use is not the only way EA are formed 2) more forms of EA exist and are not manually coded. However, the results of both studies point in the same direction which indicates the robustness of the results.
Further research should investigate which specific EA are used to create a more complete image of the prevalence of EA in CC reporting. Soon, this might be possible by using the text2emotion module designed by Diaz (2018). Although hope is not yet included in the module, and was therefore not fit for this research, five other emotions are. The research of Hase et al. (2021) performed similar research focusing on topics instead of EA and could serve as a template. A second proposal is to investigate which articles reach the most people.
This can be investigated in multiple ways but one way is to analyse the spread of the story on social media and relate it to EA. Murayama et al. (2021) investigated this with fake news and could serve as an example. This form of analysis could prove the effectiveness of different EA in practice.
Overall, this research found limited evidence for an increase in EA and even a decrease in the positive language in articles discussing CC. While FA are more prevalent compared to HA no sign of in/decrease has been found. Only negative words in article headlines increased while the negative language in article content remained the same over time. An interesting finding was that the use of positive/negative words and the use of HA/FA were (highly) volatile indicating that other factors such as newsworthy events might influence
32
the use of EA. In addition, one can conclude that the media might not use its full potential in mobilising people into CC mitigation behaviour. The IPCC stated that multiple actors should take responsibility for effectively addressing the climate crisis and changing the use of EA might be a way for the media to do their part.
Reference list
Aitamurto, T. & Varma, A. (2018). The Constructive Role of Journalism: Contentious metadiscourse on constructive journalism and solutions journalism. Journalism Practice, 12(6), 695–713. https://doi.org/10.1080/17512786.2018.1473041
Areia, N., P., Intrigliolo, D., Tavares, A., Mendes, J. M., & Sequeira, M. D. (2019). The role of media between expert and lay knowledge: A study of Iberian media coverage on climate change. The Science of the Total Environment, 682, 291–300.
https://doi.org/10.1016/j.scitotenv.2019.05.191
Baden, D. McIntyre, K., & Homberg, F. (2019). The Impact of Constructive News on
Affective and Behavioural Responses. Journalism Studies (London, England), 20(13), 1940–1959. https://doi.org/10.1080/1461670X.2018.1545599
Ballet, J., Bazin, D., & Petit, E. (2022). The Ecology of Fear and Climate Change: A Pragmatist Point of View. Environmental Values.
https://doi.org/10.3197/096327122X16452897197711
Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change.
Psychological Review, 84(2), 191–215. https://doi.org/10.1037/0033-295X.84.2.191 Boulliane, S. & Belland S. (2019) Who Matters in Climate Change Discourse in Alberta?
Pinto, Gutsche, R. E., & Prado, P. (Eds.), Climate Change, Media and Culture:
Critical Issues in Global Environmental Communication. Emerald Publishing Limited.
33
Boykoff, M. T. (2013). Public Enemy No. 1?: Understanding Media Representations of Outlier Views on Climate Change. The American Behavioral Scientist (Beverly Hills), 57(6), 796–817. https://doi.org/10.1177/0002764213476846
Boykoff, M., T. (2011). Who Speaks for the Climate?: Making Sense of Media Reporting on Climate Change. Cambridge University Press.
https://doi.org/10.1017/CBO9780511978586
Bro, P. (2019). Constructive journalism: Proponents, precedents, and principles. Journalism, 20(4), 504–519. https://doi.org/10.1177/1464884918770523
Brüggemann, M. (2017). Shifting roles of science journalists covering climate change.
In Oxford Research Encyclopedia of Climate Science.
Carrington, D. (2019). Why the Guardian is putting global CO2 levels in the weather forecast.
The Guardian, April 5. Retrieved from:
https://www.theguardian.com/environment/2019/apr/05/why-the-guardian-is-putting- global-co2-levels-in-the-weather-forecast , Accessed on April 29, 2022
Chadwick, A. (2015). Toward a Theory of Persuasive Hope: Effects of Cognitive Appraisals, Hope Appeals, and Hope in the Context of Climate Change. Health Communication, 30(6), 598–611. https://doi.org/10.1080/10410236.2014.916777
Cole, S., Balcetis, E., & Dunning, D. (2013). Affective Signals of Threat Increase Perceived Proximity. Psychological Science, 24(1), 34–40.
https://doi.org/10.1177/0956797612446953
Coward, R. (2013). Speaking personally: The rise of subjective and confessional journalism.
Macmillan International Higher Education. ISBN: 1137368519
34
Davidson, D., J. & Kecinski, M. (2022). Emotional pathways to climate change responses.
Wiley Interdisciplinary Reviews. Climate Change, 13(2).
https://doi.org/10.1002/wcc.751
Diaz, Shaik, JMM, Santofimi JCG, Quintero MCG (2018) Intelligent execution of behaviours in a nao robot exposed to audiovisual stimulus. In IEEE 2nd Colombian Conference on Robotics and Automation (CCRA), IEEE, Barranquilla, Atla´ntico, Colombia, 1–3 November, pp. 1–6
Ettinger, J., Walton, P., Painter, J., & DiBlasi, T. (2021). Climate of hope or doom and gloom? Testing the climate change hope vs. fear communications debate through online videos. Climatic Change, 164(1-2). https://doi.org/10.1007/s10584-021-02975- 8
Feldman, L. & Hart, P. S. (2021). Upping the ante? The effects of “emergency” and “crisis”
framing in climate change news. Climatic Change, 169(1-2).
https://doi.org/10.1007/s10584-021-03219-5
Gligorić, K., Anderson, A., & West, R. (2019). Causal Effects of Brevity on Style and Success in Social Media. Proceedings of the ACM on Human-Computer Interaction, 3(CSCW), 1–23. https://doi.org/10.1145/3359147
Guardian, the (n.d.) Why us? The Guardian, retrieved 11-06-2022 from https://advertising.theguardian.com/advertising/why-us
Gyldensted, C. (2015). From mirrors to movers: Five elements of positive psychology in constructive journalism. GGroup Publishing, ISBN: 1514777495
35
Haagerup, U., Schmidt, H., & Møller, M. (2017). Constructive news : how to save the media and democracy with journalism of tomorrow. (2nd ed.) Aarhus University Press.
https://doi.org/10.2307/j.ctv62hhq4
Hancock, J., T., Landrigan, C., & Silver, C. (2007). Expressing emotion in text-based
communication. Conference on Human Factors in Computing Systems: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems; 28 Apr.-03 May 2007, 929–932. https://doi.org/10.1145/1240624.1240764
Harcup, T. & O’Neill, D. (2017). What is News?: News values revisited (again). Journalism Studies (London, England), 18(12), 1470–1488.
https://doi.org/10.1080/1461670X.2016.1150193
Hartmann, P., Apaolaza, V., D’Souza, C., Barrutia, J. M., & Echebarria, C. (2014).
Environmental threat appeals in green advertising: The role of fear arousal and coping efficacy. International Journal of Advertising, 33(4), 741–765.
https://doi.org/10.2501/IJA-33-4-741-765
Hase, V., Mahl, D., Schäfer, M. S., & Keller, T. R. (2021). Climate change in news media across the globe: An automated analysis of issue attention and themes in climate change coverage in 10 countries (2006–2018). Global Environmental Change, 70, 102353–. https://doi.org/10.1016/j.gloenvcha.2021.102353
Höijer, B. (2010). Emotional anchoring and objectification in the media reporting on climate change. Public Understanding of Science (Bristol, England), 19(6), 717–731.
https://doi.org/10.1177/0963662509348863
Hutto, C., J. & Gilbert, E (2015) VADER: A Parsimonious Rule-based Model for Sentiment Analysis of Social Media Text. Association for the Advancement of Artificial
Intelligence,
36
https://www.researchgate.net/publication/275828927_VADER_A_Parsimonious_Rule -based_Model_for_Sentiment_Analysis_of_Social_Media_Text
IPCC (2022) Climate change 2022Impacts, Adaptation and Vulnerability, Summary for Policymakers. IPCC,
https://www.ipcc.ch/report/ar6/wg2/downloads/report/IPCC_AR6_WGII_SummaryFo rPolicymakers.pdf
Kenix, L., J. (2008). Framing Science: Climate Change in the Mainstream and Alternative News of New Zealand. Political Science, 60(1), 117–132.
https://doi.org/10.1177/003231870806000110
Kok, G., Peters, G.-J. Y., Kessels, L. T. E., ten Hoor, G. A., & Ruiter, R. A. C. (2018).
Ignoring theory and misinterpreting evidence: the false belief in fear appeals. Health Psychology Review, 12(2), 111–125. https://doi.org/10.1080/17437199.2017.1415767 Lorenzoni, I., Nicholson-Cole, S., & Whitmarsh, L. (2007). Barriers perceived to engaging
with climate change among the UK public and their policy implications. Global Environmental Change, 17(3), 445–459.
https://doi.org/10.1016/j.gloenvcha.2007.01.004
Lough, K., & McIntyre, K. (2021). A systematic review of constructive and solutions journalism research. Journalism. https://doi.org/10.1177/14648849211044559 Maras, S. (2013). Objectivity in journalism. Polity Press.
McIlwaine, S. (2013). Journalism, climate science and the public: Towards better practices.
International Journal of Media and Cultural Politics, 9(1), 47–58.
https://doi.org/10.1386/macp.9.1.47_1
37
McIntyre, K. & Sobel, M. (2018). Reconstructing Rwanda: How Rwandan reporters use constructive journalism to promote peace. Journalism Studies (London, England), 19(14), 2126–2147. https://doi.org/10.1080/1461670X.2017.1326834
Molek-Kozakowska, K. (2017). Communicating environmental science beyond academia:
Stylistic patterns of newsworthiness in popular science journalism. Discourse &
Communication, 11(1), 69–88. https://doi.org/10.1177/1750481316683294
Moser, S., C. (2010). Communicating climate change: history, challenges, process and future directions. Wiley Interdisciplinary Reviews. Climate Change, 1(1), 31–53.
https://doi.org/10.1002/wcc.11
Murayama, T., Wakamiya, S., Aramaki, E., & Kobayashi, R. (2021). Modeling the spread of fake news on Twitter. PloS One, 16(4), e0250419–e0250419.
https://doi.org/10.1371/journal.pone.0250419
Newman, N., Fletcher, R., Nielsen, R., Andi, S. & Schulz, A. (2020). Reuters Institute Digital News Report 2020. Reuters Institute for the Study of Journalism.
https://www.digitalnewsreport.org/survey/2020/
Nielsen, R. (2020) Economic contexts of journalism, Wahl-Jorgensen, & Hanitzsch, T.
(Eds.). The handbook of journalism studies (2nd edition.). Routledge/Taylor & Francis Group.
O’Neill, S. & Nicholson-Cole, S. (2009). Fear Won’t Do It: Promoting Positive Engagement With Climate Change Through Visual and Iconic Representations. Science
Communication, 30(3), 355–379. https://doi.org/10.1177/1075547008329201 Painter, J. (2019). Climate Change Journalism: Time to Adapt. Environmental
Communication, 13(3), 424–429. https://doi.org/10.1080/17524032.2019.1573561
38
Painter, J., Ettinger, J., Doutreix, M.-N., Strauß, N., Wonneberger, A., & Walton, P. (2021). Is it climate change? Coverage by online news sites of the 2019 European summer heatwaves in France, Germany, the Netherlands, and the UK.
https://doi.org/10.1007/s10584-021-03222-w
Pan, Y., Opgenhaffen, M., & Van Gorp, B. (2019). Negotiating climate change: A frame analysis of COP21 in British, American, and Chinese news media. Public
Understanding of Science (Bristol, England), 28(5), 519–533.
https://doi.org/10.1177/0963662518823969
Pantti, M. (2010). The value of emotion: An examination of television journalists’ notions on emotionality. European Journal of Communication, 25(2), 168–181.
https://doi.org/10.1177/0267323110363653
Papacharissi, Z. (2015). Toward New Journalism(s): Affective news, hybridity, and liminal spaces. Journalism Studies (London, England), 16(1), 27–40.
https://doi.org/10.1080/1461670X.2014.890328
Parks, P. (2019). Naturalizing negativity: how journalism textbooks justify crime, conflict, and “bad” news. Critical Studies in Media Communication, 36(1), 75–91.
https://doi.org/10.1080/15295036.2018.1533990
Parks, P. (2020). Is Climate Change a Crisis - And Who Says So? An Analysis of Climate Characterization in Major U.S. News Media. Environmental Communication, 14(1), 82–96. https://doi.org/10.1080/17524032.2019.1611614
Pearce, W. (2019). TROUBLES IN CLIMATE JOURNALISM. Issues in Science and Technology, 36(1), 12–13.
39
Pepermans, Y. & Maeseele, P. (2017). Climate Change Journalism: From Agony to Agonistic Debate. Desenvolvimento e Meio Ambiente, 40.
https://doi.org/10.5380/dma.v40i0.49257
Pligt, J. v.d. & Vliek, M. (2017). The psychology of influence : theory, research and practice.
Routledge, an imprint of the Taylor & Francis Group.
Rakow, T., Heard, C. L., & Newell, B. R. (2015). Meeting Three Challenges in Risk Communication: Phenomena, Numbers, and Emotions. Policy Insights from the Behavioral and Brain Sciences, 2(1), 147–156.
https://doi.org/10.1177/2372732215601442
Reser, J. P., & Bradley, G. L. (2017). Fear appeals in climate change communication. In Oxford research encyclopedia of climate science. Oxford University Press.
https://doi.org/10.1093/acrefore/9780190228620.013.386
Richardson, B., J. (2020). From student strikes to extinction rebellion : new protest movements shaping our future. Edward Elgar Publishing Limited.
Roberto, A., J., Mongeau, P. A., & Liu, Y. (2018). A (Re)defining moment for fear appeals: a comment on Kok et al. (2018). Health Psychology Review, 12(2), 144–146.
https://doi.org/10.1080/17437199.2018.1445546
Roeser, S. (2012). Risk Communication, Public Engagement, and Climate Change: A Role for Emotions: Risk Communication, Public Engagement, and Climate Change. Risk Analysis, 32(6), 1033–1040. https://doi.org/10.1111/j.1539-6924.2012.01812.x Ruiter, R. A. C, Kessels, L. T. E., Peters, G.-J. Y., & Kok, G. (2014). Sixty years of fear
appeal research: Current state of the evidence. International Journal of Psychology, 49(2), 63–70. https://doi.org/10.1002/ijop.12042
40
Salama, S. & Aboukoura, K. (2018) Role of emotions in climate change communication, Leal Filho, Manolas, E., Azul, A. M., Azeiteiro, U. M., & McGhie, H. (Eds.) Handbook of Climate Change Communication: Vol. 1 : Theory of Climate Change Communication, Springer International Publishing AG,
https://link.springer.com/book/10.1007/978-3-319-69838-0
Saunders, C., Grasso, M. T., & Hedges, C. (2018). Attention to climate change in British newspapers in three attention cycles (1997–2017). Geoforum, 94, 94–102.
https://doi.org/10.1016/j.geoforum.2018.05.024
Schäfer, M, S. & Painter, J. (2021). Climate journalism in a changing media ecosystem:
Assessing the production of climate change‐related news around the world. Wiley Interdisciplinary Reviews. Climate Change, 12(1). https://doi.org/10.1002/wcc.675 Schmidt, A., Ivanova, A., & Schäfer, M. S. (2013). Media attention for climate change around
the world: A comparative analysis of newspaper coverage in 27 countries. Global Environmental Change, 23(5), 1233–1248.
https://doi.org/10.1016/j.gloenvcha.2013.07.020
Shoemaker, P. J. & Vos, T. P. (2009). Gatekeeping theory. Routledge.
https://doi.org/10.4324/9780203931653
Wahl-Jorgensen, K. (2016). Emotion and journalism. The SAGE handbook of digital journalism, 128-143.
Wahlström, M., Wennerhag, M., & Rootes, C. (2013). Framing “The Climate Issue”: Patterns of Participation and Prognostic Frames among Climate Summit Protesters. Global Environmental Politics, 13(4), 101–122. https://doi.org/10.1162/GLEP_a_00200
41
Weathers, M. R. & Kendall, B. E. (2016). Developments in the Framing of Climate Change as a Public Health Issue in US Newspapers. Environmental Communication, 10(5), 593–
611. https://doi.org/10.1080/17524032.2015.1050436
Wessler, H., Wozniak, A., Hofer, L., & Lück, J. (2016). Global Multimodal News Frames on Climate Change: A Comparison of Five Democracies around the World. The
International Journal of Press/politics, 21(4), 423–445.
https://doi.org/10.1177/1940161216661848
Willig, I., Blach-Ørsten, M., & Burkal, R. (2022). What is “Good” Climate Journalism?
Public Perceptions of Climate Journalism in Denmark. Journalism Practice, 16(2-3), 520–539. https://doi.org/10.1080/17512786.2021.2016069
Wozniak, A., Wessler, H., & Lück, J. (2017). Who Prevails in the Visual Framing Contest about the United Nations Climate Change Conferences? Journalism Studies (London, England), 18(11), 1433–1452. https://doi.org/10.1080/1461670X.2015.1131129 Zaller, J. (2003). A New Standard of News Quality: Burglar Alarms for the Monitorial
Citizen. Political Communication, 20(2), 109–130.
https://doi.org/10.1080/10584600390211136
Appendix Table 6:
Regression HA and percentage of positive/negative words
95% confidence interval
β SE Z-score p-value Lower
limit
Upper limit Positive
words
34.76 5.63 6.18 <.01 23.73 45.79
42 article
content Negative article content
2.75 5.61 0.49 .62 -8.24 13.74
Positive words headlines
0.58 1.35 0.43 .67 -2.01 3.23
Negative words headlines
-0.15 1.11 -0.14 .89 -2.32 2.02
Constant -4.84 0.76 -6.74 <.01 -6.26 -3.44
Note. This table shows that the higher percentage positive words an article contains the higher the probability is that that article contains HA. Pseudo- R2= 0.17, Log-Likelihood = -129.48
Table 7:
Regression FA and percentage of positive/negative words.
95% confidence interval
β SE Z-score p-value Lower
limit
Upper limit Positive
words article content
4.66 4.04 1.15 0.25 -3.26 12.58
Negative article content
24.71 4.74 5.22 <.01 15.43 34
43 Positive
words headlines
-1.28 1.12 1.13 .26 -3.49 0.93
Negative words headlines
-1.24 0.88 -1.42 .16 -2.97 0.48
Constant -0.51 0.35 -1.46 .14 -1.2 0.18
Note. This table shows that the higher percentage negative words an article contains the higher the probability is that that article contains FA. Pseudo- R2= 0.08, Log-Likelihood = -196.70.
