Master thesis; Environmental Risks: Framing effects of Scope and Responsibility on Policy Support

Av Annika Rødeseikes, fagsjef i Pykologi i CStN.

Abstract:

The nature of global environmental risks (e.g. climate change) are often complex and thereby

difficult to understand through human sensory reception. The perception of these risks will

therefore often come from communications from experts and the public discourse. How

people evaluate this communication, and how they act in associated decision-making

processes are crucial because is leads to environmentally relevant behaviour. The present

study examined the effect of framing (psychological distance and moral responsibility) of

messages about environmental risks on environmentally relevant policy support. Using an

experimental design, 183 participants were randomly assigned into one of four conditions in

two fictitious environmental risk scenarios: a local risk with a focus on collective moral

responsibility (N= 32), a global risk with a focus on collective moral responsibility (N = 53), a

local risk with an individual focus on moral responsibility (N = 52), or a global risk with a

focus on individual moral responsibility (N = 49). The results showed no effect of framing on

policy support, but all the included types of policy supports were predictable from values

(biospheric, altruistic and egoistic) and emotions (ethic related and consequence related). This

indicates that framing of the type used in this experiment, do not affect peoples moral

considerations in relation to environmental risk related policies.

Keywords: environmental risk evaluation, morality, framing, psychological distance,

responsibility

II

Sammendrag:

Globale miljørisikoer er ofte komplekse av natur, og følgelig vanskelige å forstå gjennom

menneskelig persepsjon. Oppfatningen av denne type risiko vil derfor ofte komme fra

kommunikasjon fra eksperter og den offentlige diskurs. Hvordan folk evaluerer denne

kommunikasjonen, og hvordan de responderer i tilknyttede beslutningsprosesser er

avgjørende fordi det fører til ulike typer miljøatferd. Dette studiet undersøkte effekten av

innramming (psykologisk avstand og moralsk ansvar) av budskap om miljørisiko på

miljøpolitisk støtte. Gjennom å bruke et eksperimentelt design, ble 183 deltakere tilfeldig

tildelt én av fire scenario i to ulike fiktive miljørisikoscenarier: en lokal risiko med fokus på

kollektivt moralsk ansvar (N = 32), en global risiko med fokus på kollektivt moralsk ansvar

(N = 53), en lokal risiko med et individuelt fokus på moralsk ansvar (N = 52), eller en global

risiko med fokus på individuelt moralsk ansvar (N = 49). Resultatene viste ingen innvirkning

av innramming på politisk støtte, men alle inkluderte typer miljøpolitisk støtte var mulig å

predikere gjennom verdier (biosfæriske, altruistiske og egoistiske) og emosjoner (etisk

relaterte og konsekvens relaterte). Dette indikerer at innramming av typen som ble brukt i

dette eksperimentet, ikke påvirker folks moralske hensyn i forhold til ulike typer miljøpolitisk

støtte.

Nøkkelord: miljørisiko evaluering, moralitet, innramming, psykologisk distanse, ansvarlighet

III

Acknowledgement

Research on the psychological aspects of environmental global change, has been my

main interest throughout my Master`s degree. I am convinced that the contribution from my

field of psychological science is very important in the transformation to a low carbon society.

Psychological research offers frames for understanding the human mind in ways that will

make the transition faster and easier. To get the chance to immerse in the field of cognitive

psychology with a heavy element of applied value, was very rewarding.

I would like to thank my supervisors Gisela Böhm and Rouven Doran for

professional support throughout the process. I would also like to thank Bergen

Fylkeskommune and the DIGSSCORE community for providing me with master stipends,

which helped to realize the data collection. Further, I would like to thank the Centre for

Climate and Energy Transformation for providing me with an office and an inspirational

interdisciplinary working environment. Last, but not least, I would thank the E.ON Stipend

Funds for the exchange scholarship, and Prof. Rüdiger Pfister with co-workers at the LüneLab

at the Leuphana Universität Lüneburg to have met me with great hospitality at my exchange

stay this November.

Finally, I would like to thank my mother, father, sister, brother and close friends for

endless encouragement. And to my boyfriend, Thomas; thank you for all the love, laughter

and support.

Bergen, 5th of December 2017

Annika Rødeseike

IV

Innholdsfortegnelse

Abstract …………………………………………………………………………………………………………………….. I

Sammendrag ……………………………………………………………………………………………………………. II

Acknowledgement …………………………………………………………………………………………………… III

Table of Contents ……………………………………………………………………………………………………. IV

Appendix …………………………………………………………………………………………………………………. V

Figure List ……………………………………………………………………………………………………………….. V

Table List ………………………………………………………………………………………………………………… V

Introduction ……………………………………………………………………………………………………………… 1

Theoretical and Empirical Foundations ………………………………………………………………………. 4

Morality ……………………………………………………………………………………………………………… 4

Environmental Risk Characteristics ………………………………………………………………………… 6

Environmental Risk Evaluations ………………………………………………………………………….. 10

Responsibility ……………………………………………………………………………………………………. 17

Framing …………………………………………………………………………………………………………….. 20

The Framing of Location and Responsibility in Environmental Risks …………………….. 22

Research Aim ………………………………………………………………………………………………………… 23

Method ……………………………………………………………………………………………………………………. 25

Pilot Study ……………………………………………………………………………………………………………. 25

Main Study……………………………………………………………………………………………………………. 26

Participants ………………………………………………………………………………………………………… 26

Design ………………………………………………………………………………………………………………. 27

Measures …………………………………………………………………………………………………………… 28

Manipulation Check …………………………………………………………………………………………. 28

Emotions ………………………………………………………………………………………………………… 28

Policy Support …………………………………………………………………………………………………. 28

Values ……………………………………………………………………………………………………………. 29

Demographic Items ………………………………………………………………………………………….. 29

V

Procedure …………………………………………………………………………………………………………………. 30

Results ……………………………………………………………………………………………………………………. 32

Manipulation Check ……………………………………………………………………………………………….. 32

Cross Balance ……………………………………………………………………………………………………….. 33

Main Effects ………………………………………………………………………………………………………….. 33

Policy Support ……………………………………………………………………………………………………. 33

Emotions …………………………………………………………………………………………………………… 33

Regressing Policy Support on Emotions and Values …………………………………………………… 38

Values and Emotions in the CCS scenario …………………………………………………………….. 38

Values and Emotions in the Plastic scenario ………………………………………………………….. 40

Discussion ……………………………………………………………………………………………………………….. 44

Manipulation effects ………………………………………………………………………………………………. 44

Exploratory Approach ……………………………………………………………………………………………. 50

Theoretical Implications …………………………………………………………………………………………. 52

Practical implications ……………………………………………………………………………………………… 54

Conclusions and Further Directions …………………………………………………………………………. 56

Ethics ……………………………………………………………………………………………………………………… 58

References ………………………………………………………………………………………………………………. 59

Appendix ………………………………………………………………………………………………………………… 69

Appendix A – Pilot study questions ……………………………………………………………………………… 69

Appendix B – The main study questionnaire ………………………………………………………………… 70

Appendix C – Correlation matrix from regression analysis …………………………………………… 97

Figure List

Fig. 1. Dual-Process model of Risk Evaluation …………………………………………………………….. 12

Table List

Table 1 Group Differences for the Local and Global Condition in the CCS Scenario and

Plastic Scenario ………………………………………………………………………………………………………… 26

Table. 2. Group Differences for the Individual and collective Condition in the CCS Scenario

and Plastic Scenario ………………………………………………………………………………………………….. 26

VI

Table. 3. Two-Way (Location and responsibility) Analysis of Variance for the four

Aggression Related Policy Support measurements in the CCS scenario and Plastic scenario 35

Table. 4. Two-Way (Location and responsibility) Analysis of Variance for the four Help

Related Policy Support measurements in the CCS scenario and Plastic scenario ………………. 36

Table. 5. Two-Way (Location and responsibility) Analysis of Variance for the Three Types of

Emotional Reactions in the CCS scenario and the Plastic scenario ………………………………….. 37

Table. 6. Regression analysis Summary for Value and Emotion Variables Predicting

Aggression Related Policy Support ……………………………………………………………………………… 42

Table. 7. Regression analysis Summary for Value and Emotion Variables Predicting

Aggression Related Policy Support Measurements in the CCS scenario ………………………….. 42

Table. 8. Regression analysis Summary for Value and Emotion Variables Predicting

Aggression Related Policy Support Measurements in the Plastic Scenario. ………………………. 43

Table. 9. Regression analysis Summary for Value and Emotion Variables Predicting Help

Policy Related Support Measurements in the Plastic Scenario ………………………………………… 43

Table. 10. Means, Standard Deviations, and Intercorrelations for Support politics that Punish

Polluters and Emotion and Value Predictor Variables in the CCS scenario ………………………. 97

Table. 11. Means, Standard Deviations, and Intercorrelations for Boycott Products and

Services and Emotion and Value Predictor Variables in the CCS scenario ……………………….. 97

Table.12. Means, Standard Deviations, and Intercorrelations for Increase Tax on Fossil Fuels

and Emotion and Value Predictor Variables in the CCS scenario ……………………………………. 98

Table. 13. Means, Standard Deviations, and Intercorrelations for Limit Population Growth

and Emotion and Value Predictor Variables in the CCS scenario ……………………………………. 98

Table. 14. Means, Standard Deviations, and Intercorrelations for Donate Money to

Environmental Organisations and Emotion and Value Predictor Variables in the CCS ……… 99

Table. 15. Means, Standard Deviations, and Intercorrelations for Consume and Buy Less and

Emotion and Value Predictor Variables in the CCS scenario ………………………………………….. 99

Table. 16. Means, Standard Deviations, and Intercorrelations for Promote Environmental

Education and Emotion and Value Predictor Variables in the CCS scenario …………………… 100

Table. 17. Means, Standard Deviations, and Intercorrelations for Replace Fossil Fuels with

Renewables and Emotion Value Predictor Variables in the CCS scenario ………………………. 100

Table. 18. Means, Standard Deviations, and Intercorrelations for Support Politics that punish

Polluters and Emotion and Value Predictor Variables in the Plastic Scenario …………………. 101

Table. 19. Means, Standard Deviations, and Intercorrelations for Boycott Products and

Services and Emotion and Value Predictor Variables in the Plastic Scenario ………………….. 101

VII

Table.20. Means, Standard Deviations, and Intercorrelations for Increase tax on Fossil Fuels

and Emotion and Value Predictor Variables in the Plastic Scenario ………………………………. 102

Table. 21. Means, Standard Deviations, and Intercorrelations for Limit Population Growth

and Emotion and Value Predictor Variables in the Plastic Scenario ………………………………. 102

Table 22. Means, Standard Deviations, and Intercorrelations for Donate Money to

Environmental Organisations and Emotion and Value Predictor Variables in the Plastic

Scenario …………………………………………………………………………………………………………………. 103

Table. 23. Means, Standard Deviations, and Intercorrelations for Consume and Buy Less and

Emotion and Value Predictor Variables in the Plastic Scenario. ……………………………………. 103

Table. 24. Means, Standard Deviations, and Intercorrelations for Promote Environmental

Education and Emotion and Value Predictor Variables in the Plastic Scenario ……………….. 104

Table. 25. Means, Standard Deviations, and Intercorrelations for Replace Fossil Fuels with

Renewables and Emotion and Value Predictor Variables in the Plastic Scenario …………….. 104

1

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

Introduction

Global environmental problems and climate change are some of the biggest threats

humanity is facing. Human impact on the natural environment, such as increased CO2

emissions, challenges our livelihoods (IPCC, 2014). Due to an increase in average

temperatures, sea level rise and extreme drought, consequences for both the ecology,

economy and public health will be severe (National Research Council, 2010). In the Paris

agreement, 175 countries have agreed to aim at keeping the global temperature rise this

century well below 2 degrees Celsius (above pre-industrial levels), preferably further to 1,5

degrees Celsius (Morgan, Dagnet & Tirpak, 2014). During the last two decades, possible

solutions to better mitigate and adapt to environmental risks, have been heavily debated in the

public discourse, as well as in the social and natural sciences.

The need for a transition from fossil energy dependence is clearly present, considering

that it is the biggest source of CO2 emissions on earth today (Metz, Davidson, De Coninck,

Loos, & Meyer, 2005). Policies that are needed to reach the goals of the Paris agreement and

change the energy system are, in addition to research and innovation, dependent on public

support and engagement. It the context of policy support, framing (filters) of communications

concerning environmental risks is an unavoidable reality, as our evaluations and decisions

never are formed or drawn in a vacuum (Nisbet, 2009). A central questions when

communicating environmental risks is: who is causing the risks and who will suffer the

consequences? Within psychology there is literature arguing that people`s moral

considerations, as well as emotional reactions and personal values, are very important in the

evaluation of responsibility in climate and environmental contexts (Stern, Dietz, & Kalof,

1993; Groot & Steg, 2007). In addition, there is evidence showing that the complex structure

of environmental risk may prevent people from detecting the causal structure, and thereby not

2

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

evaluate environmental risks as a moral imperative. Thus, it is reasonable to look closer at the

psychology behind human perception, evaluation and behaviour connected to environmental

risk. Exploring the communicational triggers for specific policy support could in turn

contribute to better the communication of environmental issues.

Research shows that despite an increasing amount of evidence indicating increased

environmental risk caused by anthropogenic environmental changes over the last decades, the

overall public concern and perceived importance of these issues have in many countries

declined (Pidgeon, 2012). This is particularly the case in wealthy western countries (Kohut,

2013). In the context of climate change, some call this the ‘climate paradox’ (Stoknes, 2014;

Nordgaard, 2011), and this could be considered an example of moral failure. The discrepancy

between the increased scientific knowledge and decreased public concern has been

investigated through a large amount of psychological literature (Swim et al., 2011; Sterman,

2008; Weber, 2006; Doherty & Cayton, 2011). Some would claim that environmental risks,

like climate change, can be challenging for our moral judgement systems to fully understand

and engage in. As a result, we might evaluate environmental risks morally different than for

example terror, fraud, or forced marriage (Markowitz & Shariff, 2012). Böhm and Pfister

(2000; 2005; 2017) proposes a model that seeks to investigate how people evaluate

environmental risks. The model includes both moral, cognitive and emotional components,

and forms the basis of a mental model approach. This model is helpful when trying to

understand why people perceive and evaluate environmental risks the way they do, because it

looks at causal evaluation with a step-by-step approach.

Two aspects that have been proposed as potential barriers for the moral activation

when evaluating environmental risks is the lack of communicated risk proximity and a clear

moral responsibility (Markowitz & Shariff, 2012; Markowitz, 2012b; Gardiner 2006;

Jamieson, 2007). This thesis will use experimental methods to explore participant`s risk

3

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

perception and policy evaluation when exposed to different framings of distance and moral

responsibility in environmental risks scenarios. The prediction is that it is possible to trigger

people’s moral thinking in such a way that it is reflected in specific political support. In

addition to this, emotional reactions and personal values will be examined as possible

contributing factors in the evaluation process.

4

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

Theoretical and Empirical Foundations

This section will give an overview of the existing literature that is relevant for this

thesis. It will start by looking at why morality is central to environmental risk perception and

evaluation. This is followed by an overview of environmental risk characteristics, which will

clarify the psychological complexity of environmental risks and how this relates to distance

and morality. Further, the process of environmental risk perception and evaluation will be

explored through the mental model approach. Finally, this is followed by an introduction to

human values, and its importance in understanding moral responsibility in conjunction with

risk perception and evaluation. In this thesis, climate and environmental issues will be used

without major differentiation. Most of the literature about the psychological aspects of climate

change is applicable when talking about environmental risks in general.

Morality.

Haidt (2001) states that morality is the driver to human (social) behaviour, and that the

way we interpret and evaluate potential moral issues in conjunction with environmental risks

are crucial (Haidt, 2001; Sjöberg, 2000; Feinberg & Willer, 2013; Böhm & Pfister, 2000,

2005). For several moral philosophers, environmental issues, like climate change, are to be

considered a fundamentally moral issue (Jamieson, 2010; Singer, 2006; Gardiner 2006). This

is because of the negative outcomes climate change will have for humans and animals, and

because the earth`s atmosphere, that provides us with ‘life sustaining services’ and therefore

considered a public good, has limited resources (Singer, 2006). In addition to this, Jamieson

(2010) highlights the moral aspect of injustice, stating that the rich take more of the global

public goods than the poor, and harm the poor additionally by contributing to global change

(which in the main will affect the poorest parts of the world). Haidt (2001) supports the

assumption that environmental issues are a morally laden problem, by emphasizing how

5

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

moral intuitions and emotions become intertwined when attitudes are anchored in moral

thinking. This is because humans seem to have an innate disposition to care deeply about right

and wrong, and thereby about other people`s intentions. The visceral responses that often

occur when faced with attitudes that challenge one’s own view in conjunction with moral

judgements, further supports this (Damasio, as sited in Forgas, 2012; Greene & Haidt, 2002).

It has been shown that morality influences political attitudes (Emler, 2003), but also

people`s attitudes and behaviour connected to climate change (Stern, Dietz, Abel, Guagnano,

& Kalof, 1999; Markowitz, 2012b). Studies that empirically combine these assumptions find

that individuals that consider the ethical implications in environmental risks show greater

support for pro-environmental policies (Shwom, Bidwell, Dan, & Dietz, 2010; Skitka, 2010,

Markowitz, 2010a). There is also evidence from neuropsychological studies using FMRI

showing that moral judgements correlate with different patterns of neural activity in

emotionally related brain areas and therefore to the characteristics of the situation that people

evaluate (Greene, Sommerville, Nystrom, Darley, & Cohen, 2001). Böhm and Pfister (2001)

suggests that evaluation of risks that includes a consideration of potential harm to others, is

highly relevant when talking about cognitive evaluation of environmental risks. These

judgements clearly involve subsequent emotional reactions, which is something Böhm and

Pfister point out as a very important factor in their work connected to environmental risk

evaluation (Böhm & Pfister, 2001; Böhm 2005).

From the above findings, I derive that there seems to be a connection between the

perception of climate change as caused by humans, and corresponding ethical considerations.

The mapping of what or who is causing a risk, and what or who suffers the consequences, that

some researchers call ‘the causal structure’ (Böhm & Pfister, 2001; Bostrom, 2017), appears

somewhat to be a key factor for human ethical evaluation for environmental risks. However, it

is important to have in mind that environmental risks are highly complex by nature, among

6

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

other things because of its social dilemma structure and aggregated causation (Böhm and

Pfister, 2000).

Environmental risk characteristics.

In the following section, I will describe the difference between a risk and a perceived

risk, followed by some selected characteristics of environmental risks. The highlighted

characteristics are relevant for this thesis, because they are empirically shown to affect the

perception and evaluation of moral responsibility of environmental risks. The term risk is

traditionally used to describe an event, situation or activity that involves (a) a degree of loss

(of something humans value) and (b) a degree of uncertainty of an outcome (Slovic, 1997).

Risk perception, on the other hand, is the subjective evaluation of risk, which involves a

personal assessment of the severity and characteristics of a risk. Supporting evidence shows

that while risk is related to beneficial outcomes (e.g. financial decisions) in the world, in

people`s minds and judgements, a risk is related to negative outcomes (e.g. low risk is

associated with high benefits and vice versa) (Slovic & Peters, 2006). Research within the

field of cognitive psychology also shows that risk perception is influenced by heuristics and

biases, like the affect heuristic (Zajonc, 1980). This means that people not only judge a risk

based on what they think about it, but also how they feel about it (Finucane, Alhakami,

Slovic, & Johnson, 2000). These finding show that the term risk often used by laypeople as

something dangerous and harmful, largely is due to social factors (such as social norms), the

media (Böhm & Pfister, 2008), but also emotional reactions (Slovic & Peters, 2006).

According to the field of risk perception, it seems as if people judge problems that

they perceive to possess an immediate effect on their everyday life, as more severe than for

long-term problems that happen far away (Koger & Winter, 2011). Additionally, a study by

Böhm and Pfister (2001) showed that lay people associate global environmental risks with

7

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

negative consequences (loss) for humans, which also was evaluated as worse than negative

consequences for nature (Böhm & Pfister, 2005). To get a better understanding of why these

judgements occur, one could look closer at the structure of environmental risks.

Environmental risks tend to be complex by nature, and not easily comparable with

other risks. A subject of interest for risk researchers within psychology, has been the scope of

environmental risks (Pawlik, 1991; Klöckner, 2011; Koger & Winter, 2011). Climate change

is an example of an environmental risk with a large scope, and because of the extraordinary

character and complexity, people lack experience in dealing with it (Nordgaard, 2011). As a

result, people may have the same numbingexperience of dealing with climate change in the

same way as is described about nuclear power: “being haunted by something we cannot see or

even imagine” (Lifton, 1982). The discrepancy between personal resources (both emotional

and cognitive) and the scope of this risk, is large and hard for people to deal with. This may

lead to emotional reactions like the feeling of hopelessness or helplessness, or even anger and

fear (Markowitz and Sheriff, 2012). This very same mechanism can also be used to explain

why some people are in denial of environmental risks like climate change (Nordgaard, 2011).

Because of the scope and complexity, environmental risks may provoke a self-defensive bias

(Moser, 2010). This bias could be due to the public discourse that tends to tell people that

their consumption and way of living is what is causing environmental damage, and may

further provoke the feeling of guilt (or other negative emotions). As a consequence, this might

lead to non-ameliorative reactions like the focus of costs of mitigation (Markowitz & Shariff,

2012; Doherty & Cayton, 2011). In fact, recent findings suggest that those most responsible

for a great share of the harmful global effects are the people that would actively try to avoid

feeling responsible for causing climate change. They do this in part by blaming others for

their contributions and inaction to the problem (Stoll-Kleemann, O’Riordan, & Jaeger, 2001;

Nordgaard, 2011; Markowitz & Shariff, 2012). This is problematic because it might hinder

8

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

pro-environmental action. Research focusing on scope and the defensive bias response are

relevant in the context of the environment and morality, as it contributes to the scientific

understanding of both the climate paradox, rejection of climate change, and nonenvironmentally

friendly actions (e.g. increased personal consumption).

Another feature characterizing environmental risks is the social dilemma structure

(Vlek, 1996). A social dilemma structure, entails that while individual members of a group

may have an incentive to follow an immediate personal interest (e.g. drive a car), it might not

be beneficial for the group as a whole (humanity) in the long run (e.g. increased CO2

emissions resulting in extreme weather). At the same time, if all cooperate, then all will

benefit (Dawes and Messick, 2000). Based on the social dilemma structure, an environmental

risk will, in some way, require a solution that does not necessarily satisfy the individual (e.g.

stop driving a car), but that would be best to do based on ethical considerations (Böhm &

Pfister, 2000). This idea also indicates that social belongingness is central in the context of

risk perception, as in being close to or far away from where the environmental risk exhibits its

consequences.

Even though climate change is the direct result of goal-directed behaviour (because

nearly all activities that emit greenhouse gases are due to consumption or production of goods

and services requested by humans), studies show that people often perceive these actions as

unintentional (Markowitz & Shariff, 2012). This is possibly because (1) it is hard to detect a

single agent (or even several) who may be responsible for the risky development and (2)

people don`t judge others to hold a lifestyle that causes harm intentionally (Pawlik, 1991;

Markowitz & Shariff, 2012). Since unintentionally caused harms are judged less harshly than

intentional ones, this might weaken the moral judgement of these types of risks (Markowitz &

Shariff, 2012; Guglielmo, Monroe, & Malle, 2009). These mechanisms have been explained

on the basis of that environmental risks are often a result of the aggregated actions of many

9

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

people (Böhm & Pfister, 2001). The numbing and non-engaging notion of not knowing who is

responsible for pollution or contamination can be prevented by clarifying who has done what

and with what consequences.

Another feature of environmental risks, is psychological distance. A recognizable

argument is that many people express a sense of distance to cause and/or consequences in

connection with environmental risks (Spence, Poortinga, & Pidgeon, 2012). According to the

Construal Level Theory (CLT), developed by Liberman and Trope (2008), there are four

types of psychological distance: a geographical distance to the problem, a social distance

(hard to culturally identifying with the people who suffer the consequences, because they

often live far away) and a temporal distance (the long time-horizons, e.g. future temperature

rise that will affect future generations). The CLT proposes that psychological distance is

mentally represented in people`s minds in a way that is directly linked with the psychological

distance to an object or event. Distant events or objects are mentally represented with abstract,

decontextualized, high-level construals, while proximal events or objects are represented with

low-level, concrete, and detailed construals. Furthermore, the theory imposes that the

psychologically proximal and distant objects (or events) are represented in the similar mental

space in people`s minds. This means that because each dimension of distance in interrelated,

impact on one aspect of distance will influence the other (Liberman and Trope, 2008).

Experimental studies show that when you ask people to focus on stimuli that is congruent

(e.g. temporal uncertainty) with psychological distance (e.g. geographical distance), this will

facilitate the processing of information given about the psychologically distant stimuli

because they are cognitively associated (Bar-Anan, Liberman, Trope, & Algom, 2007). The

decontextualized representation of psychological distance also influences the ability of

performing abstraction tasks (Förster, Friedman, & Liberman, 2004), in the same way that the

focus on psychological proximity improves the performance of tasks that requires focus on

10

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

specific details (Wakslak, Trope, Liberman, & Alony, 2006).

For geographical distance, environmental risk studies have found that by highlighting

a local focus to an environmental risk, emotional and cognitive engagement will arise

(Lorenzoni & Pidgeon, 2006). This is explained as being due to the increased experience of

salience (Lorenzoni et al., 2007). Research also shows that people who experience phenomena

(e.g. floods) that they attribute to climate change, show increased perception of personal and

local risk from climate change, as well as higher levels of concern and worry (Reser, Bradley,

Glendon, Elul, & Callaghan, 2012; Akerlof, Maibach, Fitzgerald, Cedeno, & Neuman, 2013)

Based on the presented literature, I draw the assumptions that there are characteristics

about environmental risks that are important to consider when trying to understand people`s

emotional reactions, evaluations, and behaviours in relation to them. I interpret two

components as being of special importance: (1) perceived risk severity (what is at stake), and

(2) the ethicality (who is responsible), which is supported by cognitive risk researchers like

Böhm and Pfister (2001). In conjunction with the literature on psychological distance, I see a

need for communicating environmental risks at a more local level to reduce the perception of

scope (and thereby the social distance), and thereby increase a sense of severity and urgency.

This will in turn promote moral considerations when evaluating environmental risks. My

conclusion further indicates that the manipulation and framing of these components might be

crucial.

Environmental risk evaluation.

To better understand the process of perception and evaluation of global environmental

risks, Böhm and Pfister (2000; 2005) suggests a mental model approach. This type of

approach is very helpful when looking at risk evaluation and moral responsibility, because it

tells us what people see as cause and effect (which lays the foundation for moral judgements)

11

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

and could potentially determine people`s action tendencies. This approach is also a useful tool

when looking for framing effects in environmental risk evaluation, because it provides a

framework for looking at causal evaluation with a step-by-step approach.

The model from Böhm and Pfister (2000; 2005) includes both moral, cognitive and

emotional components. As Figure 1 shows, the model assumes that the starting point of an

environmental risk, is a mental representation (mental model) of the risk situation. A mental

model is a representation or a set of causal beliefs which occurs when people perceive the

surrounding world (Bostrom, 2017). A person`s mental model can influence how the person

learns, reacts to information, defines a problem, and makes decisions (Gentner & Stevens,

2014). Previous research on mental models and environmental risk perception suggests that

the way people perceive and understand things like the climate system, shapes their beliefs

and evaluations of environmental risk (Böhm & Pfister, 2001; Morgan, 2002; Sterman, 2008;

Bostrom, 2017). The relationship between smoking and cancer has been used as an analogy to

the phenomena of mental models (Newell & Pitman, 2010). Many would probably agree that

it would be hard to explain the relationship with all the medical technical steps. At the same

time, the fragmented knowledge about the relationship is sufficient to represent the risk in our

minds. This is similar to the fragmented knowledge about the relationship between the

increasing atmospheric COleading to global temperature rise, and the threat this temperature

rise will impose.

12

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

Fig. 1. Dual-Process model of risk evaluation (Böhm & Pfister, 2017).

Norman (as sited in Bostrom, 2017) proposes four elements of mental model research

that, put in the context of environmental risk perception, looks like this: (1) the target system

(in this context that would be an environmental risk like climate change), (2) a conceptual

model of the target system (a representation of the system of different concepts that are

involved in the process of climate change), (3) the user`s mental model of this target system,

and the (4) the researcher`s conceptualization of the user`s mental model. This framework

shows how the mental models of lay people are subject to interpretation from the researcher,

due to their abstract nature. Yet, an international study focusing on mental models showed

that perceived risk and causality of climate change corresponds with the support of different

policy alternatives (Bostrom et.al., 2012). For example, people who think that carbon

emissions are the cause of environmental harm, tend to support policies that focus on reducing

carbon emissions, because they think of this as the most effective policy. Despite the

researcher’s defining role, this study shows the importance of studying mental models because

it proves that people support what they think is efficient by relying on a perceived causal

13

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

structure of an environmental risk. This is very important to consider in a climate policy

context because it shows that even though climate change is overall a perceived risk, it is the

causal thinking that guides the support for the various policies.

Through their model, Böhm and Pfister (2000; 2005) argue that when forming a

mental representation of a risk, there are two evaluative aspects that are relevant:

deontological evaluations and consequentialist evaluations. These two aspects involve

specific cognitive evaluations, emotions, and different types of action tendencies (Böhm &

Pfister, 2001) (See Fig.1). The consequentialist way of evaluation refers to consequences of

potential loss, where the seriousness and uncertain negative consequences that might occur,

will be weighed. Note that this type of focus also includes the evaluation of experienced

outcomes, that refers to the ongoing processes of pollution and destruction, and negative as

well as positive consequences (Böhm, 2003). The deontological way of evaluation, on the

other hand, is about the ethical considerations of the actions themselves being more important

than the consequences of actions. This mode of evaluation focuses more upon the potential

violation of moral principles, and the focus therefore lies on the actors and the actions. This is

related to what Baron and Spranca (1997) would call protected values. Their research

indicates that people evaluate some actions to violate values that can`t be traded off. For

example, people will not let natural resources be destroyed or let people die for monetary

gains. This is considered taboo and will elicit emotions like anger and rage (Böhm & Pfister,

2009).

As Figure 1 also shows, the model also includes an emotional aspect. Frijda (1986)

claims that emotions have a guiding effect on action, and different psychological theorists

have tried to clarify the role of affect in environmental risk perception (Nerb & Spada, 2001;

Böhm & Pfister, 2000; Swim et al., 2011). It is an old assumption that emotions have a

negative impact on decisions (Baumeister, Vohs, & Tice, 2012), and even though there is

14

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

conflicting evidence, emotions seem nevertheless to be helpful for decision making (e.g.

Damasio, as sited in Forgas, 2012). Some would claim that it is separate systems that work

independently (Zajonc, 1984), while others would say that emotions include some sort of

cognitive appraisal (Lazarus, 1982), which is the assumption of the presented model of Böhm

and Pfister (2000; 2005). Either way, because a perceived risk is associated with negative

emotional reactions, the valence of emotional stimuli is important for how we further

experience and evaluate risk (Finucane, Alhakami, Slovic, & Johnson, 2000). This was shown

in a study where people were induced with negative emotions, whereupon the overall

experience of risk would increase (Johnson & Tversky, 1983). In addition, Meijnders,

Midden, and Wilke (2001) showed that by inducing fear through a short emotional film about

climate change, participants were more willing to lower their energy consumption.

Nevertheless, Böhm & Pfister`s (2008) research supports a much broader and

multifunctional view on emotions. As outlined in their model, different emotions with the

same valence can have different functions in a decision-making process. This highlights the

importance of nuance when looking for effects of (or on) emotions. Böhm and Pfister`s model

(2000; 2005) implies that cognitions precede different emotional responses, and that these

emotions in turn will affect the person`s behavioural tendencies in an environmental risk

context. In a study from 2003, Böhm analysed the emotional reactions to different

environmental risks, using the model by Böhm and Pfister (2000; 2005). Participants were

presented with environmental risks scenarios, which afterwards had to indicate how strongly

they experienced different emotions. The result confirmed the model`s distinction between the

two different types of emotions: ethic-based and consequence-based. The first type, ethicbased

emotions, includes emotions like disgust, anger, disappointment, guilt or shame. These

emotions are motivated seemingly by the participants judgement that there have been

violations of ethical principles. The consequence-based emotions are motivated by evaluating

15

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

(past or future) consequences. Supporting evidence show that anger and fear are both negative

emotions, but while fear tends to result in helping behaviour, anger will cause a more

aggression related behaviour (Böhm & Pfister 2000; Nerb & Spada, 2001).

When considering the role of emotions in an environmental risk context, Böhm (2003)

further differentiate between two types of the suggested emotional modes: (1) prospective

(anticipated), (2) retrospective (experienced) consequence-based emotions, (3) other- and (4)

self-related ethics-based emotions. Her research shows that people seem to experience more

of emotions like fear and worry when they think about things that might happen (1), and

emotions like sadness or sympathy when evaluating events that already had taken place (2).

The latter distinctions (3, 4) indicate that based on who is responsible for the risk (the

individual or the collective), people experience different emotions. If one feels self-blame,

emotions like shame and guilt arise, while emotions like anger and outrage occur if somebody

else seems guilty of causing the risk. Based on an emotion intensity rating, results from the

study by Böhm (2003), showed that prospective consequence-based emotions were rated to be

the strongest, while ethic-based self-directed emotions were the weakest. Another study, by

Harth, Leach, and Kessler (2013) also show that the feeling of anger and guilt would be

elicited when participants is being told that they had the personal responsibility for

environmental damage. The feeling of guilt would further predict behavioural intentions that

concern the repairing of environmental damage, whereas anger would predict intentions

involved around punishment.

As Figure 1 shows, the way in which the evaluative focus triggers both emotional

reactions and actions tendencies is consistent with the mode of the evaluation. A

deontological evaluation (e.g. oil spill) will trigger moral judgements (e.g. a company is to

blame), that also trigger morality-oriented emotions (e.g. outrage). This may result in agent

related behaviour (e.g. vote for a party who will punish companies who pollute). Bostrom

16

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

(2017) further supports the assumption that causality is linked to the presentation of

characteristics of risks, by referring to a study by McDaniels, Axelrod, and Slovic (1996). The

results showed that deforestation was evaluated as more risky than global warming, and that

energy production was viewed as less risky than both global warming and energy production.

This shows that separate human activities are perceived as less risky than the actual

consequences (pollution and emissions). This was further supported through a study by Böhm

and Pfister (2005) that investigated the foundations for their dual-process model. By using a

distinction between consequences for humans at the one hand, and consequences for the

natural environment on the other hand, they found that risk types that involve negative

consequences for humans were perceived riskier than risk types that affect only nature.

What I specifically draw from review on emotion is that the emotional reactions in

some risk literature might lack nuance (Böhm, 2003), and therefore needs to be investigated

more thoroughly by using different emotions of the same valence. That is because the

different emotions are considered important factors in risk judgement and behaviour, that

again are closely connected to moral consideration and behaviour (Böhm, 2003; Nerb &

Spada, 2001; Harth, Leach, & Kessler, 2013). The assumption that there are two emotion

types (consequence based and ethic based emotions) that show different types of action

tendencies, emphasizes this connection even further (Böhm & Pfister, 2000). Another

conclusion would be that the specific emotions that could play a motivating role in getting

people to think of environmental risk as something that is threatening to themselves and that

they are responsible for causing (e.g. guilt), might seem hard to activate (Böhm, 2003). I

would argue that this is connected to the perceived direction of the relationship between

environmental risk and certain emotions. This perception seems to depend on the persons

knowledge about the risk. In addition to this, the distinction of consequence based and ethic

based emotions probably would play an important role in environmental risk evaluation when

17

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

exposed to certain framings, as previously shown by Harth, Leach, and Kessler (2013). I

interpret the above findings to be closely related to specific contextual factors, including

framing, which is possible to manipulate for communicational purposes. An interesting

question to ask when considering morality, framing effects, and the structure of mental

models, is whether people will respond to environmental risks differently when risk is framed

as something caused by one as an individual or the world community. It is plausible that

manipulated information about agency will result in different outcome measures, when the

risk is being presented with a clear causal structure indicating a detectable moral

responsibility.

Responsibility.

In the further search for risk research that can explain what triggers the feeling of

personal moral responsibility when evaluating an environmental risk, frameworks within the

value theory domain offers useful input. Personal values are shown to be indirectly related to

pro-environmental behaviour (Stern, 2000). Schwartz defines a value as “a desirable transsituational

goal varying in importance, which serves as a guiding principle in the life of a

person or other social entity” (1992, p. 21). Schwartz’s conceptualization of values is a good

way at looking at broad subdivisions of different values connected to pro-environmental

attitudes and actions. His 56 universal values can be placed into a two-dimensional space,

where the values that are close to each other in the circumplex are compatible. The two

dimensions are: self-transcendence (which includes altruism, forgiveness, loyalty) vs. selfenhancement

(which includes power, ambition and hedonism), and openness to change

(which includes self-direction and stimulation) vs. conservation (which includes security,

conformity and tradition).

Despite the strong position that Schwartz has in conceptualizing human values in

18

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

social psychology (Corner, Marowitz, & Pidgeon, 2014), there are other researchers that have

developed scales which have proven useful in an environmental risk context. There are three

types of values that seem to play an important role when looking for environment relevant

behaviour tendencies (Stern, Dietz & Kalof, 1993; De Groot & Steg, 2007): altruistic values,

biospheric values and egoistic values. Based on Schwartz`s values system, the values reflect

the distinction between self-transcendence and self-enhancement dimension. The altruistic

and biospheric values are represented in the self-transcendence dimension (e.g. universalism)

and the egoistic values in the self-enhancement dimension (e.g. power). Even though altruistic

and biospheric values, unsurprisingly enough, are highly correlated, the difference between

altruistic values and biospheric values is that the first reflects a special concern for human

welfare, while the latter one reflects a concern for the nature and environment. Egoistic values

in this context reflects the self-interest connected to environmental protection. A well-known

example here is the NIMBY (“not in my backyard”) statement, where environmental concern

increases when threat to one self or one’s family is recognized (Stern, Dietz, & Kalof, 1993).

Several studies support the use of and the distinction between altruistic values, biospheric

values, and egoistic values by showing that pro-environmental attitudes and actions often are

higher for people that show higher scores on self-transcendence oriented values, compared to

self-enhancement oriented values (Stern & Dietz, 1994; Nordlund & Garvill, 2002; Bardi &

Schwartz, 2003).

An example of a theory that explores this relationship between these values types and

environmental behaviour, is the Value-Belief-Norm Theory (VBN), which is an extension of

The Norm Activation Theory (NAT) by Schwartz (1977). Put simply, NAT proposes that proenvironmental

actions follow from the activation of personal norms because it reflects the

feeling of moral obligation to act in a certain way. This activation is due to the following

situational factors: (1) the awareness of the problem (what are the consequences of not

19

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

acting), (2) the feeling of responsibility as a result of being aware of the negative

consequences, (3) the identification of actions to reduce environmental problems and (4)

one`s ability to contribute to hinder the negative consequences. The VBN theory (Stern et al.

1999; Stern, 2000) serves as an extension to the NAT theory, by assuming that these

situational factors additionally are dependent on personal values, which include biospheric

values, altruistic values and egoistic values. This means that these values are activated in

people who believe that environmental issues pose a threat to the biosphere, to people and

species, and one self. Thus, the theory implies that the strength of this activation will

determine further assessment of moral responsibility concerning pro-environmental actions.

Empirical evidence shows that every variable in the VBN model is significantly related to the

next variable in the causal chain. Only the biospheric values were directly related to the sense

of obligation to act pro-environmental, when other variables were controlled for. This implies

that biospheric values have great explanatory power in the context of environmental risk

evaluation. Supporting evidence for the VBN theory comes from studies that have focused on

a variety of general pro-environmental actions (e.g. Nordlund & Garvill, 2002), and som more

specific, like explaining environmental citizenship (Stern, Dietz, Abel, Guagnano, & Kalof,

1999), acceptability of various energy policies influencing households (Steg, Dreijerink ,&

Abrahamse, 2005), willingness to reduce car use (Nordlund & Garvill, 2003), and policy

acceptability (Eriksson, Garvill & Nordlund, 2006, 2008).

Further studies show that people may react negatively when asked to make choices

that includes moral considerations, such as “putting a price” on nature (Tetlock, 2003). This is

most likely due to the individual evaluation that some values are more important than others.

Our values seem to be organized in a system where competing choices are based on the most

important values (Keeney & Raiffa, 1976). Both biospheric values and altruistic values tend

to be positively related to pro-environmental behaviour, but when people are forced to choose

20

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

between the two, the difference between altruistically and biospherically oriented people

becomes apparent (De Groot & Steg, 2008; Stern, 2000). As mentioned when describing the

VBN theory, different situations can activate specific values when the situation is relevant for

a value that is central to our self-concept. This means that situations can trigger specific

values by, for instance, enhancing one`s self focus (ask people which values matters the most)

or to provide cognitive support to activate the value system (ask people to provide a reason

for their values) (Verplanken & Holland, 2002).

From the above literature preview, I draw the assumption that the VBN theory offers a

good contribution to the explanation on where morality is coming from, and how one could

explain the process of individual evaluation on environmental risk. As shown in the VBN

theory (Stern et al. 1999; Stern, 2000), the individual`s moral consideration would originate

from his/her personal value system. Since the activation of biospheric, altruistic, and egoistic

values are dependent on situational triggers that are linked to a person’s self-concept and

supported by cognitive reasons (e.g. damage to the environment or people, or saving money

by using switching to solar power), these triggers are a subject of interest. Despite this, the

activation of values might overrun the effect of framing or the perceived causal structure of a

risk (situational factors), and show of as higher levels of ethic related policy support. This

would especially be the case for biospheric values (Stern, 2000; De Groot & Steg, 2008).

Such findings would support the assumption that personal values are crucial in terms of the

activation of moral responsibility in climate and environmental contexts (Groot and Steg,

2007).

Framing.

Communication is powerful in the way that it can alter the impact on a recipient`s

decisions, depending on how the message is framed. Hulme (2009) argues that it is impossible

21

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

to present information about climate change without some sort of context, thus making

framing paramount. Framing, as a concept or area of research, concerns several social science

disciplines. Frames are “interpretive storylines that set a specific train of thought in motion,

communicating why an issue might be a problem, who or what might be responsible for it,

and what should be done about it” (Nisbet, 2009, pp.15). Framing is often used with the aim

to “trim” information in a way that gives greater weight to certain aspects and elements than

others, but this is not to be mistaken with telling a lie, or leaving out important information

(Nisbet, 2009).

In the context of climate change, there are many types of empirically different frames

that are being used (for review see Levin, Schneider, & Gaeth, 1998). Examples of framing

could be outcome framing (based on the prospect theory by Kahneman and Tversky, 1979) or

attribute framing. The last one implies focusing on a specific aspect, which is commonly used

in political debates. One example of this is how Republican supporters often emphasize the

aspect of uncertainty when they talk about climate change (Nisbet & Mooney, 2009).

Communicators have been using frames like national security, health, and economic

wellbeing to reach the public awareness about environmental risks, and more recent ly, as a

moral issue (Moser, 2010; Wardekker, Petersen, & Van Der Sluijs, 2009). Al Gore`s movie

‘An inconvenient truth’ or a campaign called ‘What Would Jesus Drive?’ (The Guardian,

2002) are both examples on framings that aim at motivating people to think about the moral

aspects of global environmental change. This exemplifies how frames link two concepts (e.g.

morality and religion) so that people, after exposure to this linkage, accepts this connection

and use this as a basis for further evaluations and decisions. At the same time, Nisbet (2009)

stresses that this type of specific frame will be ineffective if it`s not relevant for people`s preexisting

interpretations. In connection to the example of morality and religion, this probably

would not be relevant for people who weren`t religious.

22

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

The framing of location and responsibility in environmental risks. As previously

mentioned, people see climate change as a risk that is distant and that have geographically and

temporally distant impacts for people and places (Leizerowitz, 2005; Räthzel & Uzzell, 2009).

Swim et al. (2011) argue that the exposure most people tend to have to climate change has

been very much impersonal, which means that people only have virtual representation through

movies, documentaries and news media of what may seem like a “remote” area of the world.

Spence and Pidgeon (2010) use the attribute of “distance” as means to increase personal

relevance, by arguing that risk communicating should focus on making environmental risk

“closer”. This includes framing climate change as a proximal and relevant “here-and-now”

event. When a local focus is highlighted, both an emotional and cognitive engagement will

arise due to the increased experience of salience (Lorenzoni & Pidgeon, 2006; Lorenzoni et

al., 2007). Rayner and Malone (1997) supports this by claiming that by highlighting local

impacts of climate change, actions to mitigate it becomes more tangible. The same way in

which location of a risk is shown to affect risk evaluations, the different framings of

responsibility are also relevant. The mental model approach by Böhm and Pfister (2001;

Böhm, 2003) implies that when people evaluate risks to be moral blameworthy (with

associated feelings and behavioural tendencies), this could be due to framing effects. In an

experiment done based on their model, the evaluative focus (attention to morality of actions)

was shown to co-vary with the risk type. When a risk was framed as human caused, (instead

of naturally) the persuasiveness of morally-based arguments increases (Böhm & Pfister,

2017).

23

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

Research aim

The aim of the thesis is to use experimental methods to investigate how people`s

environmental risk perception and following evaluations vary when exposed to different

frames about the risk. This will be done by using an experimental design to manipulate

contextual framings in a fictitious environmental risk scenario and thereby measure the

potential effect on different types of environmentally related policy support. In addition to

this, a measure of scenario-specific emotions will be included in order to test whether they

will mediate the relationship between perceptions, cognitive judgements, and behavioural

tendencies. Furthermore, the use of three distinct value types (biospheric, altruistic, and

egoistic) will be measured to look at the effect of people`s personal value dispositions on

policy support.

The contextual framings will be manipulated using two types of dimensions: risk

location and moral responsibility focus. The experiment will manipulate the level location of

a potential risk, using either: a local or a global focus in a fictitious risk scenario, assuming

this will induce the feeling of high or low severity. Moral responsibility will be manipulated

using to types of moral focus: either an individual or a collective moral focus. The policy

support measurements will differentiate between four different aggression related and four

different help related policy supports. Here, aggression related policies correspond to morality

oriented behavioural tendencies, and help related policy supports correspond to consequence

related behavioural tendencies. This distinction is adopted from the mental model approach by

Böhm and Pfister (2000; 2005). (See Fig.1).

It is reasonable to believe that these conditions will show that a risk scenario framed as

local with a personal moral responsibility, will elicit morality oriented (aggression related)

policy support. This expectation is based on previous research showing that a when a risk is

perceived as a proximal, salient and severe risk (Lorenzoni & Pidgeon, 2006; Lorenzoni et al.,

24

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

2007), and to have a clear structure of cause and effect indicating an individual moral

responsibility (Böhm & Pfister, 2017), this will give rise to more moral thinking. In addition

to this, it is also rational to believe that morality-oriented emotions will mediate this

relationship, as it has appeared to be a very strong predictor in environmental risk perception

and evaluations (Böhm, 2003). Lastly, people`s value dispositions are predicted to influence

when evaluating policy support.

25

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

Method

Pilot study

A pre-test with a student dominated sample (N = 10) was conducted to clarify whether

the scenarios that were made would be considered plausible and an appropriate manipulation

for risk severity and moral responsibility in environmental risk scenarios. Participants were

randomly assigned to one of the treatments. These treatments were identical to the ones that

were used in the main study. The only difference was that the participants had to answer 11

questions after every scenario, aimed at identifying the trustworthiness and appropriateness of

the scenarios. Examples of questions with open text boxes would be: “What was the text

about?” and “Do you experience what you just read to be a problem/something risky? (If

yes/no; why?)”. Examples of questions with a scale ranging from 1 (not at all) to 7 (to a large

extent), are: “While reading the text, I could imagine what was described.”, “The story

affected me emotionally”, and “I became engaged while reading the text”. Examples of

questions measuring the manipulation were: “Were does this risk take place? (1 locally to 7 –

globally), “If anyone, who is responsible for this risk? (The individual (you and me) the

community/world`s population no one)”, and one example with a forced choice question: “If

you had to choose, who would you say were responsible? 1 (the individual) 2 (the

community/the world`s population).A complete list of all the questions is attached in

Appendix A. Results of the pre-study revealed that the content of both scenarios, and the

additional questions seemed appropriate to use in a main study. Table 1 shows that those

individuals who were in the local conditions judged the scenario to happen more on a local

level than a global level, and vice versa. Table 2 shows that those who got the individual

condition judged the moral responsibility to be more on the individual than the collective, and

vice versa.

26

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

Table 1.

Group Differences for the Local and Global Condition in the CCS Scenario and Plastic Scenario

CCS Plast

Location M SD t(4) p M SD t(4) p

Local 5.00 2.74 4.08 .015 5.20 1.30 8.91 .001

Global 5.80 1.09 11.84 .000 5.80 1.64 7.89 .001

Table 2

Group Differences for the Individual and collective Condition in the CCS Scenario and Plastic Scenario

CCS Plast

Responsibility M SD t(4) p M SD t(4) p

Individual 1.40 0.54 5.71 .005 1.40 0.54 5.71 .005

Collective 1.80 0.45 9.00 .001 1.60 0.54 6.53 .003

Main Study

Participants.

The sample consisted of 183 participants, with 63,9% (N = 117) female and 36,1% (N

= 66) men. 90,7 % (N = 166) of the participants were full-time students and were aged

between 18 and 42 years, with a mean age of 24 years (SD = 3.3). In the sample, there were

8,8% (N = 16) who had a full-time job and 17,1% (N = 31) who did not work (either full-time

or part-time). 43,1% (N = 79) had a high school degree, while 42,6% (N = 78) held a

Bachelor’s degree, and 13,7% (N = 25) with a Master’s degree. Nearly 75% of the

participants responded between 1-6 on a 12-point scale, with 0 indicating ‘left wing’ and 12

‘right wing’. (M = 4.98, SD = 2.37, Range 10)

The recruitment of participants was made using an existing pool held by DIGSSCORE

(The Digital Social Science Core Facility, an infrastructure for social science data collection

at the University of Bergen), Facebook and personal appeal (mainly at the Faculty of

Psychology). The DIGSSCORE-pool consisted of about one thousand participants who were

27

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

mainly students, but with a broad range of study affiliations and educational degrees. As one

might expect from the recruitment, there are some deviations from the general population, in

respect to gender, age, and education: young people, females and persons with higher

educations are overrepresented.

Design.

The two independent variables, risk location and moral focus, were manipulated using

a 2 x 2 scenario-based design. The scenarios manipulated for (i) risk location, varied on a

local and global level and in (ii) moral focus, with the two levels: individual and collective.

This resulted in four different scenarios: 1. A local risk with a focus on collective moral

responsibility (N= 32), 2. global risk with a focus on collective moral responsibility (N = 53);

3. A local risk with an individual focus on moral responsibility (N = 52); 4. A global risk with

a focus on individual moral responsibility (N = 49).

Two fictitious scenarios were made, with inspiration from issues that had been

mentioned in both national, as well as international media. The first scenario was about

carbon capture and storage (CCS). The text described what CCS is, what the risks associated

with taking advantage of this technology would be, and what other consequences that could

occur; both if we used it and if we didn`t. The other scenario was about the problems with

plastic. It described the great risk of increased plastic in the sea, as well as the danger with

micro plastic. The two scenarios will be referred to as the CCS scenario and the plastic

scenario. The two independent variables location and moral focus were varied in the two

scenarios using the words local/global and individual/collective, but additional adjustments of

the text were made to make the scenarios coherent.

28

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

Measures.

Manipulation check.

After reading the scenario, the participants were told to answer where the risk was

taking place, and who had the moral responsibility. The response category was a forced

choice between: a local or global level and the individual or the world`s population. This was

measured twice (once for every scenario exposure), similar to the next two variables

Emotions.

Emotions were measured using a list of 11 emotions. Four of them were ethic-based

emotions (anger, contempt, rage, indignation), and five of them were consequence-based

emotions (sympathy, sadness, fear, worry, sorrow). The last two were resignation-based

emotions (helplessness, hopelessness). All the emotions were selected based on a factor

analysis by Böhm and Pfister (2005), as also supported by other studies (Ortony et al., 1988;

Böhm & Pfister, 2000, Harth, Leach, & Kessler, 2013). The question asked was: “When you

think about the scenario you just read, how intensely do you feel…?” The rating scale went

from 1 (not at all) to 7 (very strongly).

Policy support.

A sample of eight policies were presented for the participants, and for each of them

they had to indicate to what extent they supported these policies. This was done by using a

scale ranging from 1 (not at all) to 7 (very strongly). The two types of policy support

measurements were aggression related (e.g. ‘I would boycott products /services involved in

this issue’) and help related (e.g. ‘To a large extent replace fossil fuels with renewable

energy’). The aggression related policies correspond to ethical related emotions and

behavioural tendencies (See Fig. 1), while help related policy support correspond to

consequence related emotion and behavioural tendencies. The eight policy support

29

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

measurements were used in both scenarios, with adjustments to fit the context. A complete list

is included in the questionnaire in Appendix B. The list of policy support measurements was

selected on basis of the theoretical foundation of moral versus consequence-based outcomes,

established by Böhm and Pfister (2000), and of material used by Bostrom et al. (2012).

However, adjustments were made to fit the context.

The following variables were only measured once, after the manipulation exposure

and the measures of policy support and emotions1.

Values.

This measure was meant to represent people`s value orientations. The value scales that

were used were adopted from De Groot and Steg (2007). Their scale is based on the original

scale from Schwatz (1992), but with to extra biospheric value items included (because of

underrepresentation in Schwartz`s original scale). The scale is used to measure three different

value orientations: egoistic (social power, wealth, authority, influence), altruistic (equality,

world peace, social justice, helpfulness), and biospheric (preventing pollution, respecting the

earth, unity with nature, protecting the environment) value orientations. The respondents had

to indicate on a 9-point scale ranging from -1 (opposed to my values), 0 (not important) to 7

(extremely important), where they had to consider to what extent each value was “a guiding

principle in your life” (De Grot & Steg, 2008). In the description (as in the work of Schwartz;

1977) they were asked to vary their responses, and not to rate more than two values as

extremely important. The word ‘values’ was not mentioned.

Demographic items.

1 Global citizenship (Reysen, Pierce, Katzarska-Miller & Nesbit, 2013) and moral environmental concern

(Steentjes et.al, 2017) were also measured, but not further processed in this thesis.

30

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

The participants were asked to complete six items regarding their age, gender, student

status, employment status, and marital status, their highest acquired degree of education, as

well as their political orientation. Age was answered with an open field, gender had the option

“man,” “women,” and “other” with the latter including an open field to write in. The student

status was answered by clicking either “Yes, fulltime,” “Yes, part time” or “No.The

following answer options were given to describe their employment status: “Fulltime,” “Part

time,” “Self-employed (fulltime),” “Self-employed (part time),” “Extra help/call substitute,

“Other forms of paid work,” “Currently unemployed,” or “Disability benefits.The following

answer options were given to describe their marital status: “Single,” “Boy/girlfriend,

Cohabitant,” “Partnership,” “Married,” “Separated,” “Divorced,” or “Widow/Widower.” To

answer the question about their highest acquired degree of education, they were given the

options: “Primary school,” “High school (general specialisation),” “High school

(occupational),” “Bachelor`s degree,” “Master`s degree,” or “Doctor`s degree.” The last

demographic measure was meant to give an indication of what political “wing” participants

sympathised the most with. The question was: “In politics you often hear people talk about

the left wingand the right wing.’ Below is a scale where 0 represents those who stand to

the far political left, and 10 represent those who stand to the far political right. How would

you place yourself on such a scale?” The scale ranged from 0 (left) to 10 (right), and was

translated from the Norwegian Citizen Panel, Wave 7 (2016).

Procedure

The study was run in the DIGSSCORE lab at the university, with groups consisting of

approximately ten to thirty people. Each participant was randomly assigned to a personal desk

with a computer, placed in a cubicle that were separated by partition walls placed on the sides

and the front of each desk. The order of the scenarios was cross balanced. After being

31

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

presented with general information (how long it would take, that there are no right and wrong

answers, etc.) from the experiment leader, the participants were presented with the two

scenarios: either first the plastic scenario and then CCS scenario (N = 93), or the vice versa (N

= 93).

In the introduction, the participants were told to imagine reading the text in a paper,

and were encouraged to imagine the situations as vivid as possible. After each scenario, the

manipulations check, emotions, and policy support were measured. The final part of the

questionnaire consisted of measures of values, global citizenship, moral concern, political

orientation, and demographic variables. All the dependent variables were randomly presented

for each participant, and the two scenarios (CCS and plastic) belonged to the same condition

with respect to both the independent variables ‘location’ and ‘moral focus.The reason for

using two scenarios, was to increase reliability and generalizability.

32

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

Results

This section will give an overview of the results from the data analyses.

Manipulation check

To test whether the manipulations in the experiment worked, the four groups (local

and individual, local and collective, global and individual, or global and collective) were split

into two dichotomous variables: location and moral focus. An independent samples t-test was

conducted to compare the level of experienced moral responsibility in the individual and

collective conditions. The same was done to compare the level of experienced location in the

global and local conditions. The test was conducted both for the CCS scenario and the plastic

scenario. The significance threshold was set at .01.

The t-test was found to be statistically significant t(182) = -3.5, p < .001; d = 0.52. The

effect size for this analysis (d = 0.52) corresponded to Cohen`s convention for medium effect

(d = .50) (Cohen, 1992). The results indicate that participants in the individual group

(M=1.33, SD=0.47) judged the moral responsibility to be more on the individual than the

collective group (M=1.58, SD=0.50), and vice versa. The results also indicate that individuals

in the local group (M=1.72, SD=0.45) judged the risk to take place on a more local level than

global level, compared to the global group (M=1.92, SD=0.27). The t-test was found

statistically significant t(182) = -3.7, p < .001; d = 0.53. The effect size for this analysis (d =

0.53) was found to correspond to Cohen`s convention for a medium effect (d = .50).

For the other scenario, the t-test was also found to be statistically significant, t(182)= –

3.8, p = .001; d = 0.57. The effect size for this analysis (d = 0.57) represents a medium-sized

effect. These results suggest that the individual group (M=1.56, SD=0.50) also judged the

moral responsibility to be more on the individual than the collective group (M=1.82,

SD=0.387). The test also revealed that the local group (M=1.56, SD=0.50) judged the risk to

33

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

take place on a more local level than global level, compared to the global group (M=1.82,

SD=0.387). The test was also found to be statistically significant, t(182)= -3.9, p < .001; d =

0.58. The effect size for this analysis (d = 0.58) represents a medium-sized effect.

Cross balance.

To control for order effects, an independent-samples t-test was conducted. There was

no significant difference in the scores for experienced moral responsibility and the level of

experienced location, t(182)=958, p > .001, when comparing the (i.) first CCS then plastic (M

= 1.79, SD = 0.41), and (ii.) first plastic then CCS (M = 1.85, SD = 0.36) conditions. These

results suggest that the order the participants read the scenarios did not matter for the

outcome.

Manipulation effects

Policy support.

A two-way ANOVA was conducted to examine the effect of the focus of

responsibility and the level of location on policy support. All eight policy support

measurements were entered as dependent variables, while location and moral focus were

entered as fixed factors. There was found no significant interaction effect between the two

independent variables (responsibility and location) in either the plastic scenario nor the CCS

scenario. There were also no significant simple main effects to be found. See Table 3 and

Table 4.

Emotions.

Like with the latter analysis of policy support, a two-way ANOVA was conducted to

34

ENVIRONMENTAL RISK EVALUATION, FRAMING EFFECTS

examine the effect of the manipulation of location and moral focus on emotional reactions.

The three types of emotions (ethical, consequence and resignation based) were entered as

dependent variables, while location and moral focus were entered as fixed factors. There was

no significant effect of the two independent variables, nor an interaction effect (See Table 5).

This gave no reason to proceed looking for a mediation effects.

 

Legg igjen en kommentar

Din e-postadresse vil ikke bli publisert. Obligatoriske felt er merket med *