University of Southern California Dissertations and Theses

What does climate change look like to you? The role of internal and external representations in facilitating conceptual change about the weather and climate distinction

(USC Thesis Other)

What does climate change look like to you? The role of internal and external representations in facilitating conceptual change about the weather and climate distinction

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
WHAT DOES CLIMATE CHANGE LOOK LIKE TO YOU? THE ROLE OF INTERNAL
AND EXTERNAL REPRESENTATIONS IN FACILITAING CONCEPTUAL CHANGE
ABOUT THE WEATHER AND CLIMATE DISTINCTION

by

Neil G. Jacobson

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfilment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(URBAN EDUCATION POLICY)

May 2022

Copyright 2022 Neil G. Jacobson

ii
DEDICATION

I dedicate this to you Eryka Mateja Ueltzen. You are my everything, and you’ve been
everything I’ve ever needed. There is no possible way I could’ve accomplished this without you.
I don’t know how many thousands of baked goods you’ve made over these long years, but I
know they kept me going. I thank you for these kindnesses, but I dedicate this document and
what its completion means to the sacrifices you’ve made to support me. Ich liebe dich meine
Schatzi für immer und ewig.

iii
ACKNOWLEDGEMENTS

I first want to thank my parents for teaching me to value education, and my sister for
paving the way and making it easier for me to pursue higher education.
I want to thank Dr. Gale M. Sinatra for without her I would not be writing these words.
She invested her time and energy, supported me through incredibly difficult times, and helped
me grow into the individual I am today.
I also want to thank Dr. Neil Schwartz for opening my eyes to the world of research and
providing me opportunities to develop as a researcher.
I would also like to thank Dr. Erika Patall for her constant support and guidance. I feel
incredibly lucky and grateful to have been able to learn from you.
Finally, I would like to extend my gratitude to Christine Politte, Ava Montano, and Mansi
Gaur for their assistance and insight on coding measures in this study.

iv
TABLE OF CONTENTS
DEDICATION ii
ACKNOWLEDGEMENTS iii
LIST OF TABLES v
LIST OF FIGURES vi
ABSTRACT vii
Chapter I: Introduction 1
Chapter II: Literature Review 20
Chapter III: Method 511
Chapter IV: Results 69
Chapter V: Discussion 86
References 105
Appendix A: Rationale for Data Cleaning 117
Appendix B: Distinctions Between Weather and Climate Measure 121
Appendix C: Weather and Climate Assessment 122
Appendix D: Attitudes Toward Global Waming Measure 124
Appendix E: Climate Policy Preferences 125
Appendix F: Cognitive Construal Task 126
Appendix G: Epistemically Related Emotions Scale 128
Appendix H: Political Ideology and Partisanship Measures 129
Appendix I: Demographics Questionnaire and Attention Checks 130
Appendix J: Experimental Control Text 131
Appendix K: Experimental Refutation Text 132

v
LIST OF TABLES

Table 1 Hypotheses for Research Questions 1(a), 1(b), and 1(c)

12
Table 2 Hypotheses for Research Questions 2(a), 2(b), and 2(c)

13
Table 3 Hypotheses for Research Questions 3(a), 3(b), and 3(c)

16
Table 4 Hypotheses for Research Question 4(a)

17
Table 5 Hypotheses for Research Questions 4(b), and 4(c)

19
Table 6 Analysis of Polar Bear Imagery and Framing Effects from Born
(2019)

29
Table 7 Accurate Representations of Climate Change according to
Lewandowsky and Whitmarsh (2018)

31
Table 8 Examples of External Representations Serving Different Relational
Reasoning Functions

38
Table 9 Breakdown of Text and Graphic Conflict Awareness from Mason et
al., (2017)

41
Table 10 Summary Chart of Relevant Literature According to Topic,
Population, Manipulation, and Outcomes

47
Table 11 Descriptive Statistics for Participant Characteristics of Final Sample
N = 371 Adults

52
Table 12 External Representations in the 2-Thematic Frame (weather, polar) x
2-Cognitive Alignment (aligned, misaligned) Experimental Design
with Norming Data from Lehman et al., (2019)

65
Table 13 Descriptive Statistics Overall and by Thematic Frame and
Alignment Condition

70
Table 14 Descriptive Statistics for all Emotion Variables Overall and by
Condition

78
Table 15 Hypothesized and Obtained Results for Research Questions 3(a),
3(b), and 3(c)

92

vi
LIST OF FIGURES

Figure 1 Results from Leiserowitz et al., (2006) Study on Mental Images of
Climate Change and Affective Rating

32
Figure 2 Frequency of Mental Image for Climate Change by Opinion on
Climate Change (Leviston et al., 2014)

35
Figure 3 Interaction of Thematic Frame, Alignment, and Time on Climate
Policy Preferences

77
Figure 4 Interaction of Thematic Frame, Alignment, and Time on Climate
Tax Policy Preferences
77

vii
ABSTRACT

What does Climate Change Look Like to You? The Role of Internal and External

Representations in Facilitating Conceptual Change About the Weather and Climate

Distinction

by

Neil G. Jacobson

Communicating scientifically accurate information about climate change is difficult, partly due
to characteristics of the general public (inaccurate knowledge, negative attitudes, negative
emotions) and partly due to characteristics of climate change messages (inaccurate, disengaging).
The purpose of this dissertation study was to investigate the impact of a climate change message
paired with a thematically framed external representation (i.e., polar, weather) that depicted
climate change impacts accurately (aligned) or inaccurately (misaligned) on climate change
understanding, attitudes, policy preferences, and emotions. Results indicated an online sample of
adults (N = 371) increased their understanding and attitude toward climate change. However,
thematically framed ERs resulted in greater climate understanding compared to control
(specifically the temporal aspect of climate change) and results suggested a beneficial role of
ERs overall. These effects were qualified by thematic frame and alignment interactions on policy
preferences where support for tax and general climate policies increased only for the misaligned
weather ER. Finally, for emotions, results indicated that a polar compared to weather frame
resulted in more focus on the nature, people, and animals impacted by climate change. Together
results provide support for the importance of external representations in climate communication,
for these choices about theme and alignment can differentially influence object focus,
understanding of climate, and climate policy preferences, although everyone learned.
1
Chapter 1

Introduction

The global climate is changing with devastating effects and these impacts are distilled
into news frames, depicted in television and film, discussed by politicians, and contested by the
public (Capstick et al., 2015; Maier et al., 2014; Rebich-Hespanha et al., 2015). Research
indicates these climate change messages can sometimes do more harm than good by reinforcing
misconceptions (Marlon et al., 2019), undermining self-efficacy, and incorrectly describing and
depicting climate change itself (Lewandowsky & Whitmarsh, 2018). For instance the frames
‘global warming’ or ‘climate change’, have fueled misconceptions (e.g., it’s only supposed to be
getting hotter), negative attitudes (e.g., they are making it up to control us), and politicization
(e.g., senator showing a snowball to deny climate change). Chong (1993, p. 870) articulates the
importance of framing stating, “public opinion formation in general lies in the distillation or
sorting out of frames of reference on political issues.” Chong (2000) also explicates how framing
effects occur, “…that some representations of an issue are more persuasive than others, so that
attitudes and opinions will be swayed in predicable ways if attention can be concentrated on
those representations” (p. 143).
However, for all the media coverage and framing, Leiserowitz et al. (2020) reported that
66% of those sampled in a nationally representative sample thought global warming was
affecting the weather. While it is true that climate change is impacting the weather, this finding
belies the difficulty the general public has in understanding the temporal and spatial relationships
needed to determine climate. Content analyses of thematic frames and external, visual
representations (Rebich-Hesphana et al., 2015) found that external representations, “render
nearer aspects and perspectives more visible and more distant places nearly out of sight”
2
(O’Neill et al., 2013, p. 18). This study only speaks to spatial aspects of climate; however, the
temporal aspect is key. In this dissertation I explore the construction of climate change messages,
specifically the framing and depiction of climate and weather, and how these design choices
influence climate change knowledge, attitudes, emotions and policy preferences.
Organization
In this chapter (Chapter 1), I discuss the importance of effectively communicating
information about climate change to the general public. I then describe the theoretical lenses
employed in this investigation. Finally, I conclude the chapter by proposing my research
questions and associated hypotheses. In Chapter 2, I review the literatures relevant to my
study,—specifically research addressing climate communication and its impacts on knowledge,
attitude, emotion, and policy preferences relative to learner and message characteristics. This
literature review aims to illuminate strengths of previous investigations and highlight
shortcomings that are addressed in my investigation. I conclude this review by highlighting how
this investigation builds on prior research and its unique contribution to the field. I next present
the methodology for this dissertation in Chapter 3, and analytical results in in Chapter 4. Finally,
in Chapter 5, I conclude the dissertation by reviewing the overall study findings as well as the
theoretical, and methodological limitations of the investigation. I conclude by discussing relative
contributions to research on conceptual change, text and external representation comprehension,
and emotional thought, before concluding with implications for instruction and future research.
Summary of Theoretical Frameworks

The ‘Warming’ Trend: Cognitive Reconstruction of Knowledge Model (CRKM)

The warming trend describes a shift in the study of conceptual change from one that
foregrounded cognition only to one that acknowledges the role of warm constructs including
3
motivation, emotion, and social context (Pintrich et al., 1993; Sinatra, 2005). Dole and Sinatra
(1998) synthesized perspectives on conceptual change (i.e., cognitive psychology, science
education), and attitude and belief change (i.e., social psychology) to propose the Cognitive
Reconstruction of Knowledge Model (CRKM). The CRKM views conceptual change as a
dynamic interaction between a learner and an incoming message and attends to characteristics of
each. Foundationally, the CRKM is the only model of conceptual change to account for the
linkages between and among scientific knowledge and scientific attitudes, specifically
acceptance (Sinatra & Seyranian, 2016). Dole and Sinatra (1998) describe how an individual’s
existing conception can be assessed via strength, coherence, and commitment. Strength describes
the richness of the existing conception, coherence the explanatory power of the existing
conception, and commitment describes how invested an individual is in their current conception
and arises from sources which include social group membership, sensory experiences, and
political affiliation (Ballew, Pearson, et al., 2020; Ballew, Rosenthal, et al., 2020). They also
describe motivational learner characteristics including dissatisfaction (Muis et al., 2018; Muis et
al., 2020; Pekrun et al., 2017; Thacker et al., 2020), personal relevance (Gill et al., 2020), need
for cognition, and social context (Lombardi et al., 2014; Trevors et al., 2016). Dissatisfaction
arises when an existing conception conflicts with an incoming message (Brun & Kuenzle, 2008;
Pekrun et al., 2017). Motivation also arises from personal relevance, including a stake in the
outcome, emotional involvement with the topic, or high self-efficacy (Linnenbrink-Garcia &
Patall, 2015; Linnenbrink-Garcia et al., 2016; Sinatra et al., 2014). Another element, need for
cognition, (Cacioppo & Petty, 1982) describes a disposition wherein one enjoys considering
many sides of an issue. The final motivation, social context, describes how interactions with
members of a community or peer group, such as political affiliation, can motivate message
4
processing. According to the CRKM these characteristics are in dynamic interaction with
characteristics of a message. Next, I describe knowledge attitude links before describing message
characteristics.
Knowledge Attitude Link
Sinatra and Seyranian (2016) extended the CRKM to describe how science knowledge
(accurate, inaccurate) and attitude (pro, con) often shift together. For instance, controversial
issues like GMFs or climate change often show a pattern of increased accurate scientific
knowledge being associated with increased acceptance of the science, resulting in a more pro
position (Heddy et al., 2016; Thacker et al., 2020). Ranney and Clark (2016) documented a
similar phenomenon terming it justified climate acceptance. This linking of knowledge and
attitude differs from the stasis view proposed by Kahan (2013) which argues scientific
information cannot be used to resolve attitudinal disagreements.
However, the CRKM does not consider visual message formats (e.g., photographs,
graphs, diagrams, and models). However, this would be inaccurate for O’Neill (2017) argues that
ERs of climate change can provide an emotional connection, serve as an analogy, promote
knowledge building, and serve as an index (i.e., literal similarity with the concept or topic).
Additionally ERs including statistical graphics (Isberner et al., 2013; Nyhan & Reifler, 2019),
analogies and science diagrams (Danielson et al., 2016; Mason et al., 2017) influence conceptual
change either by providing additional information beyond the text or allowing for comparison
with their mental image (Damasio, 2004; Schnotz, 2002). Therefore, I next describe the
Integrative Theory of Text and Picture Comprehension (ITPC) (Schnotz, 2005, 2014) as it
provides two cognitive mechanisms to explain how a climate change message featuring a text
5
and an external representation may influence conceptual change for the distinctions between
weather and climate.
Integrative Theory of Text and Picture Comprehension (ITPC)

Models and theories of text and graphic comprehension are surprisingly rare given the
ubiquity of images and graphs in scientific media (Maier et al., 2014) and climate change
communication (Lehman et al., 2019; Rebich-Hespanha et al., 2015). Additionally, no theory of
text and graphic comprehension exists for conceptual change learning although recent
integrations exist (Danielson & Sinatra, 2017; Kendeou et al., 2017). Only Schnotz (2005, 2014)
describes how an individual comprehends and integrates verbal (text or words) and visual
information into a mental model. In the following sections I briefly describe Schnotz’ analysis of
internal and external representations before specifying the cognitive processes that drive text and
graphic comprehension.
External Representations
Schnotz (2002) defines two types of external representations (ERs): descriptive and
depictive. Descriptive representations are comprised of symbols (e.g., natural language, or
mathematical formulae) such as “climate change,” or “y = mx + b. ” Here, ‘climate’ is a symbol
used to describe entities or nouns, while ‘change’ describes relation. Key here is that descriptive
representations have little or no physical similarity with reality, and instead rely on convention.
Depictive representations, however, are composed of iconic signs that have specific inherent
structural features. For example, an image of a polar bear on an ice sheet is depictive as it uses
no symbols (i.e., words). Second, by depicting an object (e.g., polar bear) in a specific instance
(i.e., ice has specific thickness, sun is in specific position) an individual can read off relational
6
and structural information (Danielson & Sinatra, 2017; Hegarty, 2011; O’Neill, 2017) although
Schnotz only addresses analogical relations.
Internal Representations
So far, I discussed descriptive (i.e., symbols) and depictive (i.e., iconic) external
representations (ERs) though Schnotz also describes internal representations (i.e., visual images
and mental models). In the ITPC a mental model is not bound to a specific modality and
integrates information from the text, ER, and prior knowledge (Johnson-Laird, 1980). Thus, an
individual’s mental model contains less information than the ER, but also more information
given the role of prior knowledge (Leiserowitz, 2006; Lorenzoni et al., 2006).
Cognitive Processes. Schnotz describes an iterative comprehension process where the
individual first selects relevant information from the text and ER and organizes the information
in working memory. Selection and organization are influenced by prior knowledge, attitudes, and
emotions and influence later integration. In addition to selection, organization, and integration
Schnotz specified processes of thematic selection and analogical structure mapping. Thematic
selection is a neglected process in Schnotz’ model, however, it provides a mechanism to
understand how thematically framed ERs create an “emotion object focus,” (Pekrun & Stephens,
2012) and influence thinking by activating different considerations (Chong, 2000; Damasio,
2004). For example, a hungry polar bear may yield a thematic focus impact on polar animals and
induce compassion for pain resulting in congruently sad thoughts (Immordino-Yang et al., 2009).
Thematic selection may impact knowledge simply by depicting different objects because they
occur at different time scales and places (e.g., hungry polar bears on sea-ice vs. extreme weather
event from last week). Analogical structure-mapping (Gentner, 1983; Gentner, 1989) describes
the comparative process of assessing the alignment between an ER, and mental model. Schnotz
7
foregrounds similarity and analogy described by Gentner (1983) as comparisons that are
determined by the overlap between object-attributes (i.e., glacier is accurate index of climate),
and inter-object alignment (i.e., when, where, and what). Lewandowsky and Whitmarsh (2018)
argue that a glacier recession over time ER depicts literal similarity with climate because there is
alignment of object-attributes (i.e., glacier as accurate index of climate) and inter-object overlap
(alignment) because climate has temporal (when: 20-30 years) and spatial dimensions (where:
local, regional, global; what: glacier).
The ITPC accounts for learning from text and external representations by focusing on
how external representations (ERs) communicate thematic and analogical (i.e., similarity,
analogy) information. It is only by addressing both of these elements, and their association with
emotions that we can address Lewandowsky and Whitmarsh’s (2018) question, “how can we
avoid misrepresentation and find legitimate affective triggers?” Unfortunately, the CRKM and
ITPC do not acknowledge the role of emotions directly, and therefore require a framework for
their integration.
Emotional Thought
Immordino-Yang and Damasio (2007) proposed Emotional Thought which provides a
logical extension of the ‘warming’ trend in conceptual change research. Emotional thought is
argued to provide, “the platform for learning, memory, decision-making, and creativity, both in
social and non-social contexts” (Immordino-Yang & Damasio, 2007, p. 8). The authors embrace
an emotional rudder metaphor where our brains respond to emotions and, “direct our reasoning
into the sector of knowledge that is relevant to the current situation or problem” (p. 8).
According to Damasio (2004) these responses primarily influence attention, mental image
production, and associative memory similar to descriptions of framing manipulations and
8
thematic selection processes. Fundamentally, they direct attention and persuade in line with their
object focus by differentially activating considerations (i.e., beliefs, attitudes, emotion) that
impact decision-making (Chong, 2000).
Emotions
Damasio (2004) defines emotions as, “bioregulatory reactions that promote physiological
states to ensure survival and promote well-being,” (p. 50) and provide an immediate reaction
(conscious or not) to specific challenges and opportunities in the environment. While debate
exists regarding the conceptualization and measurement of emotions (Immordino-Yang, 2010;
Pekrun, 2006), there is agreement that emotions are object-focused (i.e., focused on an
emotionally competent trigger). Climate change is a socio-scientific issue (Sadler, 2004)
meaning it has both social and scientific considerations and objects of focus. For instance,
individuals can have topic emotions (e.g., scared and concerned about climate change), social
emotions (e.g., compassion for those displaced by climate change, anger at politicians for
inaction), and epistemic emotions (e.g., surprise about climate being measured over 20-30 years).
Pekrun and Stephens (2012) state these emotions and others can and do arise from instructional
content, although many researchers ignore the role of external representations, calling them
decorational (Carney & Levin, 2002), or labelling them seductive details (Harp & Mayer, 1997).
Individuals are often unaware of the objects or triggers that give rise to emotion or the
feeling of emotion and therefore self-report belies the richness of their emotional experience and
the relationships between emotions, their object focus, and congruent thinking. Given the
difficulty in disentangling social and topic emotions (Immordino-Yang, 2010; Pekrun, 2006;
Pekrun et al., 2017) and my interest in thematically framed representations and their role in
conceptual change for the climate and weather distinction Emotional Thought provides the
9
platform to address how individuals think, feel, and ultimately make decisions about how to
address climate change, or not.
Conclusion
I argue that to understand the role of external representations (e.g., photographs,
illustrations, graphs) in climate communication we must first assess the cognitive alignment
between the targeted concept and the ER (e.g., discussing climate but depicting an extreme
weather event). Secondly, we must assess the ERs theme and associated object focus to interpret
the source and therefore type of emotional response (Pekrun & Stephens, 2012). For example, if
communicating about the climate and weather distinction using either a polar bear photo or
temperature change graph, a cognitive alignment analysis would indicate the temperature change
graph accurately depicts time and space (i.e., global change over 30 years) whereas the polar
bear photo only accurately depicts space (i.e., polar ice sheet) but not time (i.e., seasonal ice
melt). While seasonal and 20–30-year ice changes are both processes, the role of time is
markedly different and may impact policy preferences for its mitigation. It may also focus
emotions on an object that is not an accurate representation of climate change. This is not to say
seasonal ice melt is not emotional. Regarding object focus alignment, the polar bear serves as the
thematic frame and may activate emotional considerations (compassion or empathy) or
attitudinal and political considerations (anger and frustration), given bolar bears are a political
trope (Born, 2019; Harvey et al., 2018).
Only after specification of the concept’s alignment with the text and ER (or
misalignment) and specification of frame theme and its associated object focus can we
meaningfully understand how ERs activate considerations and what considerations impact,
attitudinal, and policy preference change. Currently, few studies assess the role of external
10
representations on climate change through the lens of knowledge, attitude, and emotions,
although researchers acknowledge the need for their integration (Chapman et al., 2016;
Lewandowsky & Whitmarsh, 2018). In the next section, I provide an overview of the study
before stating research questions and associated hypotheses.
Purpose and Overview of the Study
I argue that to facilitate the general publics’ understanding of climate change
(specifically, weather and climate) communication should leverage thematically framed and
emotionally engaging external representations (ERs) that accurately depict the spatial and
temporal aspects of climate change. Currently, little to no research addresses the process of
conceptual change for the climate and weather distinction relative to an ERs thematic frame, and
alignment with the concept. Additionally, no study has explored how an ER designed to
communicate the temporal and spatial aspects of climate differentially impact understanding of
climate change, attitude, policy preferences, emotions and feelings of emotion for climate
change.
The purpose of this study was to test whether external representations (ERs) used to
thematically frame climate change impacts (i.e., polar, weather, science communication) and
their alignment in depicting temporal and spatial aspects of climate (i.e., aligned or misaligned)
influence shifts in understanding of the climate and weather distinction, attitudes toward climate
change, and climate policy preferences. I also examined how the ERs influenced self-reported
emotions and their focus in addition to the construal of the situation specifically, affect, and the
construal’s relationship to the communication message. My research contributes to the literature
by addressing how climate change communications can leverage two dimensions of external
representations (i.e., thematic frame and alignment) to support the public’s understanding of the
11
distinctions between weather and climate, the emotions experienced after reading and how they
relate to attitude toward climate change, and policy preferences to mitigate it (Lewandowsky &
Whitmarsh, 2018; Chapman et al., 2016).
Research Questions and Hypotheses

My investigation is guided by four research questions:
1. To what extent does an external representation’s (ERs) thematic frame influence
changes in climate change knowledge (1a), attitudes (1b), and policy preferences to
mitigate climate change (1c)?
2. To what extent does an external representation’s (ERs) cognitive alignment, with the
concept climate, influence changes in climate change knowledge (2a), attitudes (2b),
and policy preferences to mitigate climate change (2c)?
3. To what extent do an external representation’s (ERs) thematic frame and cognitive
alignment interact to influence changes in climate change knowledge (3a), attitudes
(3b), and policy preferences to mitigate climate change (3c)?
4. To what extent do an external representation’s (ERs) thematic frame and cognitive
alignment interact, to influence the intensity of self-reported epistemic emotions (4a),
the number of emotion objects for a given emotion (4b), the total number of emotion
objects across emotions (4c), and cognitive construal of the situation (4d)?
Hypotheses
In response to Research Question 1a, I predict the polar and weather impact thematic
messages will result in the largest degree of conceptual change compared to the scientific
communication messages because they provide an external representation known to facilitate
comprehension via the multimedia principle (Mayer, 2005) and recommended by science
12
educators (Lombardi & Sinatra, 2012; Nussbaum et al., 2017). The ER provides an additional
information source that may alert the individual to conflict between their current understanding
and the to be learned information (Mason et al., 2017). Additionally, ERs facilitate interest, and
may increase situational interest that results in deeper processing that drives conceptual change
(Stenseth et al., 2016; Thomas & Kirby, 2020). Also, the ERs used in this study were normed
assuring the images were perceived as relevant, related to climate change, and activated negative
emotions (Lehman et al., 2019).
Table 1
Hypotheses for Research Questions 1(a), 1(b), and 1(c)
(1a) Conceptual
Change
(1b) Attitude
Change
(1c) Policy Preference
Change
Thematic
Frame
Hypothesis Hypothesis Hypothesis
Polar and Weather
impacts
>
Science communication
Polar and Weather
impacts
>
Science communication
None

For Research Question 1b, I predict the scientific communication frame will produce less
attitudinal change compared to the polar or weather impact frames, in line with hypothesis 1a—
specifically, that messages with an ER will result in the greatest conceptual change. Invoking the
knowledge-attitude link identified by Sinatra and Seyranian (2016) I expected increased accurate
science knowledge to positively relate with attitudes consistent with the scientific consensus on
the issue (Heddy et al., 2016; Lombardi et al., 2014; Ranney & Clark, 2016; Sinatra et al., 2012;
Thacker et al., 2020). While research indicates certain types of climate change knowledge (i.e.,
polar regions) and attitudes can influenced by partisan identity (Hamilton, 2020), Ranney and
Clark (2016) demonstrated that mechanistic climate change knowledge is not sensitive to
political polarization, and increased what they termed ‘justified climate acceptance’.
13
However, for Research Question 1c, I offer no a priori hypothesis because climate change
policy preferences are only minimally influenced by knowledge (van Valkengoed & Steg, 2019),
but instead social variables such as political identity (Ballew et al., 2020), negative affect
(Leiserowitz, 2006), and attitude (Ranney & Clark, 2016). However, Ranney and Clark (2016)
documented that the publics’ understanding of climate change is so far from the scientific
consensus (Marlon et al., 2019) (~60% think the climate changes yearly) that increasing accurate
climate knowledge may influence policy preferences.
For Research Question 2a, I predict that an ER cognitively aligned with the concept climate
(i.e., accurate representation of climate along temporal and spatial dimensions) will produce
significantly greater conceptual change than an ER that inaccurately represents climate along
temporal and spatial dimensions. Given the goal here is to increase accurate scientific
understanding of climate, I anticipate viewing an ER that accurately depicts climate will produce
the largest conceptual change, in line with recommendations by Nussbaum et al., (2017). The
aligned ERs would best facilitate analogical structure mapping and the search for similar and
dissimilar information among their prior knowledge, the text, and the ER (Danielson et al., 2016;
Mason et al., 2017).
Table 2
Hypotheses for Research Questions 2(a), 2(b), and 2(c)

(2a) Conceptual
Change
(2b) Attitude
Change
(2c) Policy Preference
Change
Alignment
Hypothesis Hypothesis Hypothesis
Aligned
>
Misaligned
Aligned
>
Misaligned
Aligned
>
Misaligned

For Research Question 2b, I predict that the aligned ERs will yield larger shifts in
positive attitude toward climate change in line with the knowledge-attitude link identified by
14
Sinatra and Seyranian (2016), and the justified climate acceptance articulated by Ranney and
Clark (2016). Research indicates that some climate change knowledge, specifically that human
activity is directly related to anthropogenic climate change is subject to partisan influences
(Pfirman et al., 2021). However, my focus is on foundational scientific definitions, ones that
individuals of all political affiliations get wrong (Marlon et al., 2019). For this reason, I
anticipate all individuals, regardless of political identity and partisanship, will increase their
understanding of key definitions that will result in justified climate acceptance, in line with
knowledge-attitude links.
Finally, for 2c I hypothesize that climate policy preference shifts will be largest for the
aligned ERs for they accurately depict temporal, and spatial elements of climate change. I
anticipate that the knowledge-attitude link proposed by Sinatra and Seyranian (2016) will extend
to policy preferences similar to how the justified climate acceptance documented by Ranney and
Clark (2016) extended to self-reported willingness to sacrifice. Specifically, I anticipate viewing
an ER that accurately represents climate to facilitate the correct scientific understanding of
climate, which relates to climate acceptance attitudes, and will extend to support of consensus
mitigation policy preferences. Drawing from van Volkengoed and Steg’s (2019) meta-analysis
on climate adaptation, knowledge about climate change and its hazards (r = .14), risk perception
(r = .20), belief in climate change reality (r = .23), perceived responsibility (r = .25), and
negative affect (r = .29) all drive adaptive behaviors. Given overall climate knowledge is low,
and there is not consensus among the public about the reality of climate change and its associated
risks, increasing scientific knowledge will likely increase acceptance attitudes (Sinatra &
Seyranian, 2016) or justified climate acceptance (Ranney & Clark, 2016) that may extend to
scientifically backed mitigation strategies.
15
Research Question 3 focuses on the interaction between an ERs thematic frame and
cognitive alignment on indices of change. For Research Question 3a, similar to 2a I hypothesize
that the ERs aligned with climate as opposed to weather will produce greater conceptual change
for the messages were designed to facilitate the scientific understanding of climate, not weather,
although they are deeply intertwined and related (Lewandowsky & Whitmarsh, 2018).
Additionally, only the messages featuring an aligned ER, provided temporal information about
climate twice. I anticipated the repetition of the key temporal information in the ER would
facilitate conceptual change by providing a secondary means of determining conflict among prior
knowledge, the text, and the ER (Danielson et al., 2016; Mason et al., 2017). Relatedly for (3b), I
hypothesize that the cognitively aligned ERs will produce the largest shifts in climate acceptance
attitudes, based on the knowledge-attitude link (Sinatra & Seyranian, 2016) and justified climate
acceptance (Ranney & Clark, 2016) findings discussed above.
For Research Question 3c, I predict that the aligned ERs will yield larger shifts in attitude
toward climate change acceptance in line with the knowledge-attitude link identified by Sinatra
and Seyranian (2016), and justified climate acceptance articulated by Ranney and Clark (2016).
Again, in line with findings from Ranney and Clark (2016) I anticipate that increased climate
knowledge (i.e., climate is 20-30 years) and increased climate acceptance will also relate to
increased support for policy preferences in line with the prevailing scientific consensus.
I also hypothesize an interaction between alignment and thematic frame, specifically that
the misaligned weather frame, which depicts impacts of extreme weather on humans, will
increase support for climate change policies because it activates personal risk perception,
negative affect, experience with natural disasters, and place attachment whereas the misaligned
polar ER depicts impacts on distant animals (Leiserowitz, 2006; van Valkengoed & Steg, 2019).
16
Table 3
Hypotheses for Research Questions 3(a), 3(b), and 3(c)
(3a) Conceptual
Change
(3b) Attitude
Change
(3c) Policy Preference
Change
Alignment Hypothesis Hypothesis Hypothesis
Aligned
>
Misaligned
Aligned
>
Misaligned
Aligned
>
Misaligned
Misaligned Hypothesis Hypothesis Hypothesis
None None
Weather
>
Polar

Research Question 4 pertains to the influence of an ERs thematic frame and cognitive
alignment on indices of emotion. For Research Question 4a (i.e., self-reported intensity) I
anticipate an effect of alignment where epistemic emotions associated with conceptual change
specifically, surprise, curiosity, and interest (Pekrun et al., 2017; Vogl et al., 2019; Muis et al.,
2015) and negative topic emotions (i.e., anxious, hopeless) (Heddy et al., 2016) would be more
intense for aligned compared to misaligned ERs. Above I argued how an aligned ER would best
facilitate analogical structure mapping, and the detection of conflicting information. Detection of
these conflicts will drive greater intensity for surprise, curiosity, and interest, and this greater
awareness of conflict associated conceptual change will result in more intense anxiety and
hopelessness for these emotions often arise when learning accurate climate change information.
For the misaligned ERs I predict more intense epistemic emotions not associated with change
(i.e., confusion, frustration, boredom, anger) and more positive topic emotions (i.e., enjoyment,
hopeful). Next, I predict the misaligned polar ER (i.e., polar bears) will produce more intense
ratings of hopelessness, not due to the information, but due to compassion for the emotion object
compared to the aligned ERs. Finally, I predict more intense anxiety, and hopelessness for the
17
misaligned weather ER compared to the misaligned polar ER, again as it depicts impacts near
and on humans.
Table 4
Hypotheses for Research Question 4(a)

(4a) Conceptual Change Emotions (4a) Topic/Social Emotions
Associated with
change
Not associated
with change
Positive Negative

Surprise,
Curiosity,
Interest
Confusion,
Frustration,
Boredom,
Anger
Enjoyment,
Hopeful
Anxious,
Hopeless
Alignment
Hypothesis Hypothesis Hypothesis Hypothesis
Aligned
>
Misaligned
Misaligned
>
Aligned
Misaligned
>
Aligned
Aligned
>
Misaligned
Misaligned
Polar
Hypothesis Hypothesis Hypothesis Hypothesis
None None None
Polar
>
Both Aligned
Misaligned
Weather
Hypothesis Hypothesis Hypothesis Hypothesis
None None None
Weather
>
Polar

For RQ4b, I anticipated the selection of more emotion objects (i.e., when asked about
surprise how many of the six objects were selected) when viewing an aligned compared to
misaligned ER. I predict the aligned ERs will best facilitate awareness of knowledge conflict,
regardless of thematic frame ((Danielson, 2017; Danielson & Sinatra, 2017; Mason et al., 2017)
and would be reflected by the number of emotion objects selected. Additionally, I anticipate no
main effect of thematic frame, although I predict the misaligned weather ER that depicts weather
impacts on a human city to have a greater number of emotion objects compared to the aligned
18
weather ER which depicts global temperature change as the prior activates personal risk
perceptions, experience with a natural disaster, and negative affect while the latter does not.
Research question 4c addresses how an ERs thematic frame and alignment influence the
repeated identification of specific emotion objects (e.g., how many times ‘the fact that climate
change is happening’ was selected across all emotions). Here, I hypothesize that individuals who
view an aligned ER will select the emotion objects ‘the fact that climate change is happening’
and ‘ideas in the text’ significantly more than the misaligned ERs. Directly related to my
hypotheses about ER alignment on conceptual change, the aligned ERs will best facilitate
awareness of conflicts between and among knowledge, and this awareness will be captured by
individuals repeatedly identifying the sources of their emotions.
Additionally, I hypothesize the emotion object ‘people impacted by climate change’ will
occur significantly more in the misaligned weather ER condition as it is the only ER that depicts
humans. For the emotion object, ‘animals impacted by climate change’ I predict the greatest
frequency in the misaligned polar ER as it is the only ER to depict animals. For the emotion
object ‘nature impacted by climate change’ I do not propose a directional hypothesis for all ERs
depict an environment as both themes relate to nature.
For the final emotion object ‘politicians’, I hypothesize the aligned ERs will result in more
selections of ‘politicians’ as emotion sources for only the aligned ERs depict the temporal aspect
of climate as occurring over 30-years. I anticipate the aligned ERs to best facilitate the accurate
scientific understanding of climate along temporal and spatial dimensions. Given this hypothesis,
I anticipate politicians to be identified more in this condition because these ERs will best
highlight the tension between political inaction and the scientific understanding of climate.
19
Table 5
Hypotheses for Research Questions 4(b), and 4(c)

Finally, research question 4d addresses the influence on an ERs thematic frame and
alignment on one’s cognitive construal, specifically affect and concreteness. Here, I hypothesize
more affectively negative construals for the aligned ERs, for they will best facilitate an accurate
understanding of climate which will be construed negatively. For concreteness of construal
(defined as the explicit mention of the experimental material) I predict the aligned ERs will yield
more concrete construals for the individuals will be indicating how the content of the
experimental materials impacted their emotions and thinking about climate change.

Fact that
climate
change is
happening
Ideas in the
text
People
impacted by
climate
change
Animals
impacted by
climate
change
Politicians
Alignment
Hypothesis Hypothesis Hypothesis Hypothesis Hypothesis
Aligned
>
Misaligned
Aligned
>
Misaligned
None None
Aligned
>
Misaligned
Interaction
Hypothesis Hypothesis Hypothesis Hypothesis Hypothesis
None None
Misaligned
Weather
>
Else
Misaligned
Polar
>
Else
None
20
Chapter 2

Literature Review

Climate change is arguably the greatest issue of our time, but is often perceived as
distant, not my problem, or simply not real (O’Neill et al., 2013). Much of this is attributable to
science communication being distilled and framed by media outlets (Maier et al., 2014) and
political entities (Rebich-Hesphana et al., 2015) resulting in persistent misconceptions
(Leiserowitz et al., 2020; Leiserowitz et al., 2021; Lombardi & Sinatra, 2012; Read et al., 1994;
Reynolds et al., 2010). Chong (2000, p. 151) states framing, “assumes that some representations
of an issue are more persuasive than others, so that attitudes and opinions will be swayed in
predictable ways if attention can be concentrated on those representations.” Gamson and
Modigliani (1989) extend the definition of frame to include, metaphors, catchphrases, visual
images (i.e., external representations), moral appeals, and other symbolic devices. I adopt this
broader definition of framing to encompass text and external representations (ERs) to answer
Lewandowsky and Whitmarsh’s (2018) call to seek accurate and emotionally engaging
representations and Chapman et al. (2016) call to investigate how ERs differentially influence
individuals’ responses to frames.
Fortunately, the Cognitive Reconstruction of Knowledge Model (CRKM) and its focus
on text-based message characteristics can be extended by the Integrative Theory of Text and
Picture Comprehension’s (ITPC) focus on text and external representations. As discussed in the
previous chapter, neither the CRKM nor the ITPC attend to the role of emotion directly. Dole
and Sinatra (1998) describe how emotional content in a message may provide a peripheral cue,
and (Schnotz, 2005, 2014) acknowledges visual thematic selection however neither foreground
emotion.
21
In what follows, I first synthesize the extant literature on climate change communication
for text-based messages. Second, I review external representations (e.g., images, maps, graphs,
illustrations), before finally reviewing text and ER messages. For each communication method I
highlight major findings, open questions, gaps, and methodological limitations. Much of the
empirical literature assesses only two of the following three aspects: conceptual change, external
representations, or emotions. Therefore, my review builds toward an experimental study
designed to facilitate conceptual change for the weather and climate distinction, and attitudes
toward climate change by manipulating two aspects of climate ERs: thematic frame and
alignment to understand how ERs and their associated emotions impact change processes.
Text-Based Climate Communication
Text-based climate communication typically aims to influence knowledge, attitudes, or
policy preferences and therefore leverage different considerations (Maier et al., 2014). A focus
on knowledge leads to reliance on scientific sources (e.g., International Panel for Climate
Change [IPCC]) in hopes of establishing source credibility and explanation plausibility
(Lombardi, 2019; Lombardi & Danielson, 2021). A focus on attitudes or policy preferences
leverage persuasive arguments and moral appeals (Thacker et al., 2020), personal beliefs (Duchi
et al., 2020), or political cues (Chong, 1993). Recent studies provide empirical support for
Sinatra and Seyranian’s (2016) prediction that conceptual change often co-occurs with attitude
change (Heddy et al., 2016; Lombardi et al., 2013; Ranney & Clark, 2016; Thacker et al., 2020).
Leiserowitz et al. (2020) provide conceptual evidence for these effects by documenting that 74%
or respondents viewed climate change as a scientific issue and 60% as a political issue, firmly
making this a socio-scientific issue. In the following section, I describe empirical studies of text-
22
based communication for socioscientific topics in general and climate change in particular. I
begin by describing impacts on knowledge and attitudes before discussing impacts on emotion.
Empirical Studies of Text Impacts on Change
Knowledge and Attitude
Many research traditions are interested in promoting accurate understandings for socio-
scientific issues; however, there are differences in approach, specifically message format. Some
implement “de-bunking” texts (Cook & Lewandowsky, 2011) others endorse “corrective texts”
(Nyhan & Reifler, 2010) while many including myself use a “refutation text” to facilitate
conceptual change (Guzzetti et al., 1993; Tippett, 2010). A core assumption underlies the choice
among approaches, mine being that a misconception is not “replaced” but instead must compete
for activation with the new scientific conception in line the Knowledge Revision Components
Framework (KReC) (Kendeou & O’Brien, 2014). The following studies relate to socio-scientific
communication and while some do not feature a refutation text, they are included for their focus
on other study aspects (e.g., measures, samples, or design).
Lombardi and Sinatra (2012) investigated college students’ understanding of the weather
and climate distinction over a semester of instruction. Undergraduates (aged 18 to 66) were
recruited from two thematic science courses, one global warming, and the other physical
geography. Participants completed a pretest and posttest using the Distinctions Between Weather
and Climate Measure (DWCM), a measure of geological deep-time understanding, and
plausibility perceptions about climate change. The DWCM is a 13-item dichotomous choice
measure where participants categorize a statement as weather or climate (i.e., categorical shift).
The measure was derived from survey data documenting persistent misconceptions among
students, educators, and the general public (Gowda et al., 1997; Leiserowitz et al., 2020;
23
Papadimitriou, 2004). Results indicated that regardless of thematic focus, all students increased
their understanding of the weather and climate distinction, and that knowledge of deep time
related to performance on the DWCM. Despite the learning gains, the authors argued that an
instructional scheme (e.g., external representations) that promotes organization and reduces the
complexity of deep-time may facilitate future understanding, a recommendation extended by
Resnick et al. (2017) focus on relational reasoning and external representations.
Given the difficulty in understanding climate change Nussbaum et al. (2017) designed
refutation texts to facilitate conceptual change for several misconceptions including the weather
and climate distinction, ozone depletion, and the enhanced greenhouse effect. Participants were
146 undergraduates (age range 17 to 58) who read either a refutation text or an unrelated
expository text and completed the multiple response format Weather and Climate Assessment.
Methodologically, the use of an unrelated expository severely limits interpretation however, their
investigation is informative regardless. For instance, the Weather and Climate Assessment only
showed learning differences for the explanation questions (mental model level) and not the
selected response questions (belief revision). Additionally, compared to the other climate
misconceptions, knowledge of the weather and climate distinction regressed over their two-week
delay providing evidence for the intractability of this misconception. Similar to Lombardi and
Sinatra’s (2012) call for an instructional scheme to organize and reduce complexity Nussbaum
and colleagues suggest supplementing texts with strong visual models and images.
Although studies call for the use of external representations to facilitate conceptual
change, here I review text-based message characteristics, specifically peripheral cues. Lombardi
et al. (2014) were interested in how a persuasive message discussing climate information and
mitigation policies from an unknown author (e.g., scientist or politician) influenced perceptions
24
of plausibility and attitudes. This study is foundational given its focus on a competitively framed
socio-scientific issue varied on peripheral cues for adults (age range 18 to 38) in an informal
environment. Results indicated that messages perceived as more trustworthy and certain were
evaluated as more plausible, independent of prior knowledge. However, the authors did not
analyze attitude change and only alluded to the potential role of topic emotions, a relationship
borne out in recent investigations (Stenseth et al., 2016; Thacker et al., 2020; Van Boekel et al.,
2017). Kaplan et al. (2016) also investigated how credibility cues related to belief change for
strongly held political and non-political beliefs. I discuss this study in more detail below,
however, results indicated that perceptions of credibility were positively associated with belief
change for both political and non-political statements.
Emotion
Text-based climate messages also influence emotion. Muis et al. (2015) conducted two
studies to assess how epistemic emotions related to learning strategies, and ultimately learning
outcomes. In study one, 439 undergraduate students (mean age 22) from three countries (Canada,
Germany, and the US) read conflicting texts about climate change discussing positive and
negative consequences. Participants completed a measure of prior knowledge, a measure of
epistemic emotions (Pekrun & Meier, 2011), and study strategies. The authors highlighted the
importance of object focus, and explicitly worded the epistemic emotions measure to target the
knowledge generating activities. Overall, results indicated that epistemic emotions predicted
study strategies, metacognitive self-regulation, and elaboration. Interestingly, and potentially
related to the socio-scientific nature of the issue, surprise negatively related to critical thinking,
while anxiety was positively related.

25

Online emotion measures.
Given Muis and colleagues use of self-reported epistemic emotions the authors conducted
a second study employing a think-aloud protocol to capture epistemic emotions and study
strategies. Participants were 56 Canadian undergraduates (mean age 22) who engaged in a
similar procedure to Study 1. Participants generated 143 think-aloud statements with the
following object foci and frequencies, 83% epistemic, 15% topic, and 2% achievement
demonstrating the importance of object focus as one in seven thoughts related to the topic.
Kendeou et al. (2019) extended our understanding of cognitive processes and associated
emotional responses when learning from refutation texts in three experiments. Results indicated
that refutation compared to control texts resulted in significantly fewer associations (i.e.,
comments providing information activated from prior knowledge but not helpful for building
coherence), significantly more statements of cognitive conflict (i.e., comments that express
conflict between individuals’ prior knowledge and the information in the current text) and more
monitoring comprehension (i.e., comments that reflect individuals’ understanding or lack
thereof).
Sources of Emotion. Climate change is a controversial issue in the United States and is
therefore informed by partisan beliefs and attitudes (Capstick et al., 2015). This combination of
national level cultural attitudes and within country politics may explain why Muis et al. (2015)
found that surprise negatively related to critical thinking while anxiety positively related (Chan
et al., 2017; Cook & Lewandowsky, 2011; Kaplan et al., 2016). Contextualizing the role of
emotion in socio-scientific conceptual and attitudinal change is difficult because conflict
communicates challenge and threat (Gill et al., 2020; Kaplan et al., 2016; Nyhan & Reifler,
26
2019). However, challenge can be perceived at two levels, the self or group (e.g., attack on
identity), or the knowledge (e.g. “wow, I didn’t think that.”)
Kaplan et al. (2016) investigated belief change for political and non-political statements
from a neurobiological perspective attending to belief strength, perceived challenge, and brain
activity. Participants were 40 self-identified liberal adults (age range 18 to 39) from Southern
California who held strong political and nonpolitical beliefs. First, results indicated that strongly
held political beliefs could be changed by counterevidence but were more resistant than strongly
held non-political beliefs. Additionally, belief resistant individuals exhibited significantly greater
activation in the amygdala and dorsal anterior insular cortex, regions associated with responding
to threatening stimuli, and the generation of emotions and feelings respectively (Damasio &
Carvalho, 2013). Second, evidence that countered political beliefs was processed more slowly
than non-political challenges, providing evidence of additional processing (e.g., elaboration,
critical thinking, counterarguing) in line with previous studies (Chan et al., 2017; Cook &
Lewandowsky, 2011). The authors found differential activation in the default mode network
(DMN) an interconnected set of structures associated with disengagement from external reality
and an inward focus on self (Immordino-Yang et al., 2012) and provides evidence for the role of
motivated processing for strongly held beliefs (Dole & Sinatra, 1998). In sum, the study by
Kaplan et al. (2016) provides evidence that the social context of beliefs relate to one’s emotional
reaction, thoughts, and feelings about the event in a way that bridges methodologies across
disciplines (Damasio, 2004; Immordino-Yang et al., 2009).
Given the role of threat and challenge in conceptual and attitudinal change, I discuss two
studies that employ refutation texts to facilitate conceptual change for another socio-scientific
issue, genetically modified foods (GMFs). Rampant misconceptions abound for GMFs around
27
issues of cloning, hormone injection, and it being an unnatural process (Heddy et al., 2016).
Genetically modified foods are also a socio-scientific issue given their origins in science and
their intersection with the general public. Therefore, I review several studies about GMFs
together given their shared topic, base text, emotion measure, and knowledge assessment.
Muis et al. (2018) investigated the roles of task value, epistemic emotions, and self-
reported learning strategies on conceptual change for 120 undergraduate students in Canada
(mean age 22). Thacker et al. (2020) however, addressed the relationship between and among
knowledge, attitudes, and epistemic emotions for 424 undergraduates from three countries (mean
age 22). Finally, Jacobson et al. (2021) investigated the impact of explanations in refutation texts
focusing on knowledge, attitudes, and emotions. Despite differences in focus and methodology
their combined results indicated that knowledge and motivational variables (e.g., task value,
attitudes) are moderated by message characteristics that influence conceptual and attitudinal
change and that this moderation is mediated by epistemic emotions (Pekrun, 2006; Pekrun &
Perry, 2014).
In sum, text-based message characteristics for socio-scientific issues (e.g., text structure,
peripheral cues) are documented to influence perceptions of source credibility and explanation
plausibility (Lombardi et al., 2014), brain activity in threat detection areas (Kaplan et al., 2016)
self-reported epistemic emotions (Kendeou et al., 2019; Trevors et al., 2021; Trevors &
Kendeou, 2019) and conceptual change (Heddy et al., 2013; Heddy et al., 2016). Most studies
reviewed here investigated conceptual change for younger adults in controlled learning
environments without political implications limiting generalizability to the public. It remains to
be seen if larger nationally representative samples with voting age adults demonstrate the
28
patterns documented above, specifically Sinatra and Seyranian (2016), Thacker et al. (2020), and
Jacobson et al. (2021).
External Representation Based Climate Communication
External representations (ERs) are used to visually frame climate change and influence
knowledge, attitudes and, emotions, by activating different considerations (Chong, 2000;
Leiserowitz, 2006; Leiserowitz et al., 2009; O’Neill & Smith, 2014). Unfortunately, ERs can
also create conflict among these considerations. For example, climate impact ERs (e.g., melting
ice sheets) are viewed as salient however they decrease perceptions of self-efficacy (O’Neill et
al., 2013). Compare this with climate solution ERs (e.g., solar panels) that promote self-efficacy
but reduce issue salience (O’Neill & Nicholson-Cole, 2009). This conflict was borne out in a
meta-analysis (N = 64,511) conducted by van Valkengoed and Steg (2019) on behavioral
adaptations for climate change finding strong positive effects for self-efficacy (r = .26) and
negative affect (r = .29). In the following section, I describe empirical studies that assessed the
impacts of climate change ERs on knowledge, attitudes, emotions, and policy preferences.
Empirical Studies of External Representations and Climate Change
Rebich-Hesphana et al. (2015) conducted a systematic content analysis of US newspaper
images from 1969 through 2009 distilling 350 images into a set of 15 common frames. Two of
the most common frames “government and politics” (34%) and “science and the people who do
it” (21%) provide evidence for the socio-scientific nature of the topic and the political nature of
visual frames (Born, 2019; Leiserowitz, 2006; O’Neill, 2017; O’Neill et al., 2013). Two related
themes “monitoring and quantifying” (21%) and “temperature record” (15%) typically employ
graphs and charts to communicate quantitative data (e.g., hockey stick graph). The authors also
identified environmental impact themes including: “impacts on polar animals and landscapes”
29
(6%), “changing landscapes” (9%), and “enjoyment of nature at risk” (7%). Together these
studies provide evidence that specific external representations (i.e., polar bear images and
hockey stick graph) are value-laden political symbols that impact knowledge, attitudes,
emotions, and policy preferences (Born, 2019; O’Neill, 2017; O’Neill & Smith, 2014). See Table
6 for an example of how a single thematic frame can elicit different considerations.
Table 6
Analysis of Polar Bear Imagery and Framing Effects from Born (2019)
Label
Anthropomorphized
Polar Bear
Polar Bears in
Context
Polar Bears in
Danger
External
Representation

Explanation
“…polar bears are
depicted in an
anthropomorphized
manner, allowing
readers to both identify
with and emotionally
relate to them…” Born
(2019, p. 655)
“While the family
theme is still
predominant, the
topic of protecting
and sheltering
emerges, as can also
be seen by the title
“Refuge in White.”)
Born (2019, p. 655)
“…images now
connotatively or
explicitly display the
danger the bears are in.”
Born (2019, p. 656)

Given the social, scientific, and political considerations inherent in climate change
O’Neill and Smith (2014) identified three themes in climate imagery: time, truth, and power. I
have already discussed the conceptual difficulty in understanding scale and time (Lombardi &
Sinatra, 2012; Nussbaum et al., 2017; Resnick et al., 2017) so it is important this pattern emerged
from additional analyses. The theme of truth can be interpreted through the lens of
Lewandowsky and Whitmarsh’s (2018) call to avoid inaccurate scientific representations. The
30
final theme, power, acknowledges the difference in political power and efficacy to affect change.
For instance, van Volkengoed and Steg (2019) found that perceived norms were large and
positive predictors of climate adaptation behaviors, and that perceptions of self and outcome
efficacy were the largest predictors. Additionally, the meta-analysis indicated that trust in
government was a weak predictor and that trust in government implementations was a non-
significant factor.
Returning to the power theme indirect support is provided by Leiserowitz (2006) analysis
of mental images (internal representations) of climate change and Rebich-Hesphana et al. (2015)
analysis of ER frames. Leiserowitz (2006) conducted a national (US) survey of 673 individuals
(47% older than 55) to understand what images came to mind when thinking about global
warming and how negatively they were perceived. Regarding mental images Leiserowitz found
10 images covered 95% of climate mental images. Only 8% of participants had a mental image
that corresponded to politics, 6% of which described climate skeptics. Compare this with
Rebich-Hesphana and colleagues conclusion that 34% of ERs framed climate change as politics,
government, and negotiation, and 22% depicted regular (sometimes vulnerable) people but only
8% focused on citizen leaders. Put simply, climate change is framed as an issue that impacts
regular people but is decided by political elites, not citizens.
Time and Truth —The Roles of Accuracy
Lewandowsky and Whitmarsh (2018) addressed the importance of accuracy, which nests
under truth, and accords with the thematic analysis. Lewandowsky and Whitmarsh provide
examples of what they consider to be accurate representations of climate change (see Table 7).
31
Table 7
Accurate Representations of Climate Change according to Lewandowsky and Whitmarsh (2018)
Caption
Retreating glacier on
Svalbard
Coastal salination
from sea level rise in
Bangladesh
The Alaskan village of
Kivalina is at risk
from climate change
External
Representation

Explanation
“It follows that because
nearly all glaciers
worldwide are
retreating, images of
their retreat (as in Fig
1), which shows
glaciers on the
Norwegian island of
Svalbard) capture the
long-term global
warming trend and are
legitimate illustrations
of climate change.”
(Lewandowsky &
Whitmarsh, 2018, p. 3)
“Stories or images that
relate to the
consequences of sea
level rise—such as
changes in soil salinity
in coastal Bangladesh
—are legitimate
illustrations of the
consequences of
climate change, as in
Fig 2.” (Lewandowsky
& Whitmarsh, 2018, p.
4)
“Similarly, stories
about the village of
Kivalina in Alaska,
which may have to
relocate because of the
changes resulting from
climate change, are
legitimate illustrations
and trigger points for
affective engagement
(Fig 3).”
(Lewandowsky &
Whitmarsh, 2018, p. 4)

Leiserowitz (2006) found that 17% of respondents imagined “heat” or some form of
temperature rise when thinking about climate change, which may or may not accord with the
accurate conception. Key here is that a ‘truthful’ representation must accurately depict or
describe time. For example, if the individual thinks of ‘30-year temperature changes’ their
mental image is accurate, however if they think ‘temperatures have increased this year’, the
mental image is inaccurate. This assessment of mental images is further contextualized by
Rebich-Hesphana and colleagues finding that 15% of visual frames described the temperature
record (although an analysis of ER accuracy was not conducted), indicating the frames that are
32
used to communicate climate change are used by individuals when thinking about the issue.
Crucially related to truth and time, Leiserowitz (2006) found that 14% of individuals lacked a
mental image of climate change. Most striking was the conclusions that a majority of individuals
lacked an accurate, vivid, concrete, and personally relevant affective image for climate change.
Despite holding inaccurate or no mental images, individuals employed them to appraise risk, and
select policy preferences. Additionally, both studies found that all mental images were evaluated
negatively (see Figure 1).
Policy Preference Impacts
Leiserowitz (2006) hypothesized that an individual’s mental image of climate change
likely impacted affect, risk perception, and climate policy preferences for tax, and international
policies on a 4-point Likert scale. Results indicated that if one rated their mental image as
negative, they perceived the risk as greater independent of political ideology. A similar pattern
emerged for policy preferences with negative affect positively related to policy support, although
political ideology remained a significant predictor.

Figure 1
Results from Leiserowitz et al., (2006) Study on Mental Images of Climate Change and Affective
Rating
33
Power —The Roles of Salience and Efficacy
I now discuss two mixed method investigations addressing perceptions of salience and
efficacy. O’Neill and Nicholson-Cole (2009) and O’Neill et al. (2013) conducted sorting tasks
and interviews with adults in the U.S., U.K., and Australia to understand how climate
representations are interpreted and understood by the public. Participants first sorted images into
three salience categories: “makes me feel climate change is not important,” “I am undecided,”
and “makes me feel climate change is important” before performing a second sort to rank order
images from most to least salient. Across countries individuals agreed the most salient images
related to the climate impacts theme including aerial flood view, melting ice sheets,
deforestation, and polar bears which made climate change salient by making the threat visible
and personal, linking it to personal experience, or evoking emotion (Lorenzoni et al., 2006). The
temperature graph was also evaluated as making climate change salient by providing evidence
for rapid change that induced fear and fright (O’Neill & Nicholson, 2009; O’Neill, et al., 2013).
In addition to the quantitative approach individuals participated in interviews to discuss their
most salient images. For example, Americans rated the aerial flood view as making climate
change appear very salient, with one individual stating, “the weather has been messed up. It’s
been raining here for like three weeks and it never rains here, so something is going on”
(O’Neill, 2013, p. 5). This quote illustrates the intermingling of power (e.g., salience, efficacy),
time and accuracy. While the flood aerial view makes climate change salient, the individual’s
response shows that depicting climate impact events fails to depict time accurately. However, if
the person talked about how during their lifetime, they had never seen that much rain their
representation may be more accurate.
34
The sorting and interview procedures were repeated for perceptions of efficacy and the
prompt, “this image makes me feel that I can do something about climate change.” Again,
climate impact events communicated salience however they undermined efficacy with one
individual saying, “they all feel like we’re hopeless to really do anything about those things, if
they happen, they just happen when they do” (O’Neill, 2013, p. 6). Here we see that the emotion
(i.e., hopeless) is directed toward the impact event, and fails to describe the process of climate
change.
O’Neill and Nicholson-Cole (2009) also documented that the temperature change graph
undermined self-efficacy, perhaps due to the fear induced by increased salience. The most
efficacious, and least salient climate representations related to the energy futures theme and
included solar panels, electric cars, and home insulation by proposing routes to personal action,
and co-benefits. This mirrors van Valkengoed and Steg’s (2019) findings that self and outcome
efficacy were among the largest drivers of climate change adaptation.
Climate communication relies on representations whether internal or external that frame
perceptions of salience, efficacy, emotion, emotional thoughts, and policy preferences. Focusing
attention on specific visual objects and relationships, “render nearer aspects and perspectives
more visible and more distant places out of sight” (Leviston et al., 2014; Rebich-Hespanha et al.,
2015). These representations also provide more information than they contain due to prior
knowledge, attitudes, emotions, and political ideology (Gaines et al., 2007; O’Neill, 2017;
O’Neill & Smith, 2014).
In reviewing the literature on climate communication and external representations, little
research addressed the relationship between ERs and knowledge about climate change, let alone
conceptual change. Leviston et al. (2014) surveyed a nationally representative sample of
35
Australians (N = 2,502) documenting the relationship between climate change imagery and
belief in the reality of climate change. For those who believed climate change was happening,
either human induced (52%) or natural (40%), the top three images were rising sea level,
drought, and melting ice caps. For those who claimed they, “don’t know” if climate change was
happening (3.5%), the top image was “don’t know,” followed by rising sea levels, drought, and
melting ice caps providing evidence that image associations are similar to those who believe in
the reality of climate change (see Figure 2).
Most interesting, and relevant to my investigation of climate change misconceptions are
the images selected by the 3.5% of individuals who, “do not believe climate change is
happening.” Their top six images, in order, were, “don’t’ know,” “no such thing,” “scam,” and
“hot sun” followed by the two images identified by all other groups, “drought” and “rising sea
levels.” The consistency in images for those who believe in or don’t know about the reality of
climate change is striking when compared to the images for those who do not believe climate
change is happening. This also supports Lombardi and Sinatra’s (2012) and Nussbaum and
colleague’s (2017) call to supplement climate change instruction with strong visual models that
organize and simplify these difficult concepts. It also reinforces Lewandowsky and Whitmarsh’s
(2018) call for accurate and emotionally engaging climate representations. Therefore, in the final
Figure 2
Frequency of Mental Image for Climate Change by Opinion on Climate Change (Leviston et al.,
2014)

36
section I review the empirical literature on climate communication using a combination of text-
based messages and external representations to facilitate conceptual change for socio-scientific
issues.
Text and External Representation Based Climate Communication
Communication in general, and science communication particularly, rely on text and
external representations to facilitate understanding (Maier et al., 2014: Rebich-Hesphana et al.,
2015). In their analysis of media frames and image themes, Rebich-Hesphana and colleagues
documented graphs and charts were used to communicate “monitoring and quantifying data”
illustrations to communicate “visions of the impacts on climates and sea level rise,” photos to
communicate “government, politics, and negotiation,” and maps to communicate “water-related
impacts.” The thematic combinations of text and ERs vary regarding their cognitive alignment
and thematic object focus. Carney and Levin (2002) identify five types of alignment between a
text and external representation: decoration, representation, organization, interpretation, and
transformation. A decorational ER, according to the authors, has little or no alignment with the
text content (e.g., text on climate change, image of roads) while representational ERs mirror part
or all text content (e.g., text on sea level rise, graph of rise over time). Organizational alignment
provides structural information and communicates information not in the text content (e.g., text
on greenhouse effect, map of Arctic). Interpretational alignment provides additional information
beyond the text to clarify difficult text content (e.g., text on climate change, diagram of evidence
strength). The final alignment, transformational, seeks to increase recall of text information by
making it concrete and meaningful (e.g., text on seasonal change, inaccurate ER of seasonal
change). Recall that many individuals do not have a concrete, accurate, or personally relevant
mental image of climate change (Lorenzoni, 2006). Despite the utility of cognitive alignment, I
37
argue this conceptualization is too narrow as the authors ignore the role of affect in learning.
Additionally, recent evidence suggests that the affective charge of an ER in addition to text-ER
alignment together influence learning (Schneider et al., 2018). In the remainder of this section I
review text and ER based studies on socio-scientific issue communication and impacts on
conceptual and attitudinal change.
Empirical Studies of Text and External Representations on Climate Change
Climate change requires accurate communication of time (i.e., past, present, and future)
and relies on photographs, statistical graphs, illustrations, and diagrams to distill the complex
information (Rebich-Hesphana, 2015; Lewandowsky & Whitmarsh, 2018; O’Neill & Smith,
2014). In nuancing the discussion of alignment I invoke the construct of relational reasoning
(Dumas et al., 2013) which describes how learning can occur via analogy, anomaly, antimony, or
antithesis. Analogy describes a case of alignment, where the conceptual or perceptual properties
of one object are mapped onto another (Danielson et al., 2017; Gentner, 1983; Kendeou et al.,
2017). Anomalies, antimony, and antithesis all describe relations based on misalignment (See
Table 8 for examples of relational reasoning ERs). Anomalies describe encountering an event,
data or occurrence that does not fit with ones’ expectations (Chinn & Brewer, 1993) while
antimony is the process of learning what something is by ascertaining what it is not. The final
relational misalignment is antithesis and describes interaction with two representations that are in
oppositional relation (e.g., reading a flat earth text next to an image of the globe from space). In
the following sections, I first review studies that explore how aligned ERs impact conceptual
change before turning to studies that explored misalignment of ERs.
38
Table 8
Examples of External Representations Serving Different Relational Reasoning Functions
Relational
Reasoning
Analogy Anomaly
ER

Source Danielson et al., (2015) Danielson et al. (2016)
Relational
Reasoning
Antinomy Antithesis
ER

Source Danielson and Sinatra (2017) Mason et al., (2017)

Cognitively Aligned ERs
Danielson et al. (2016) investigated the influences of a refutation text, analogy, and
statistical graph on facilitating conceptual change about the enhanced greenhouse effect.
Participants (mean age 20) read one of four text and graphic combinations and completed an
open-ended knowledge measure at three time points (pre, immediate post, and one-week delayed
39
post). Results indicated that exposure to any of the refutation texts resulted in significant
increases in knowledge at immediate post; however, no differences emerged between conditions.
Although not a significant difference, individuals who viewed the text with analogy and graphic
demonstrated a more accurate understanding that may be attributable to the graph serving as a
peripheral (plausibility) cue that facilitated a more robust situation model (Isberner et al., 2013)
or that the text, analogy, and graph together provided more information that facilitated
competing activation (Kendeou & O’Brien, 2014). However, after a one week-delay knowledge
was significantly higher for those who read the refutation text with an analogy alone or the
analogy and graphic. To interpret their results, the authors created a causal network
representation to assess text and ER alignment concluding the graph was likely representational
in nature (i.e., mirrored content in text) and was not necessary to answer the open-ended
questions. The different patterns at immediate post and delay provide evidence that the graph
served as a plausibility cue immediately, but that the analogy served to bolster the competing
activation necessary to revise beliefs at delay, regardless of the presence of the graph.
Cognitively Misaligned ERs
Thus far, I discussed how maximizing ER alignment can facilitate conceptual change.
However, Mason et al. (2017) investigated the use of intentionally misaligned ERs in two studies
on seasonal change. In Study 1, Italian high school students (N = 78; mean age 18) completed a
knowledge measure and were randomly assigned to one of four text and graphic combinations
for seasonal change (i.e., refutation text + aligned ER, refutation text + misaligned ER, standard
text + aligned ER, standard text + misaligned ER). Participants also completed a measure of
metacognitive awareness and the same knowledge measure at immediate post and again after a
one-week delay. Results indicated that all text and ER combinations facilitated conceptual
40
change at immediate posttest, providing evidence that any of these text and graphic combinations
successfully facilitated conceptual change. However, at delay only those who read the refutation
text with either the aligned or misaligned ER (Table 9) maintained their conceptual change and
was attributed to differential awareness of conflicts created by the text or the ER. Given that the
aligned and misaligned ERs potentially invoked different relational reasoning (i.e., analogy vs.
antithesis vs. antimony) the authors designed a second study to further assess the impacts of the
various graphics.
Study 2 assessed if modifying task instructions could prompt inspection of the ER that
would increase metacognitive awareness of conflict and facilitate conceptual change (Mason &
Gava, 2007). A new sample of Italian students (N = 98; mean age 18) engaged in the same
procedure as Study 1 aside from modified instructions, and the addition of a probe to the
metacognitive awareness questions. Results of Study 2 differed from Study 1 in that only those
who saw the standard text and misaligned ER performed better than the standard text and aligned
ER at immediate posttest. At delay the pattern shifted in support of the refutation texts paired
with either ER. These results were partially explained by increased metacognitive awareness of
text conflict for individuals who read a refutation text compared to the standard text.
Interestingly, despite explicit instruction to process the ER there were no differences in
awareness of ER conflict across conditions. However, awareness of conceptual change overall
was significantly greater for those who viewed a refutation text and either the aligned (60.9%) or
misaligned (73.9%) ER compared to the standard text with either ER (30.7% and 46.1%
respectively) indicating that misalignment may facilitate conceptual change via different
relational reasoning approaches (Danielson et al., 2017; Kendeou et al., 2017).
41
Table 9
Breakdown of Text and Graphic Conflict Awareness from Mason et al., (2017)
Text Type Graphic Type
Study 1: No graphic
instruction
Study 2: With Graphic
instruction
Text
conflict
Graphic
conflict
Text
conflict
Graphic
conflict
Standard Standard 15.7% 5.2% 11.5% 3.8%
Standard Refutation 40.0% 10.0% 34.6% 11.5%
Refutational Standard 60.0% 15.0% 47.8% 13.0%
Refutational Refutation 31.5% 5.2% 39.1% 17.4%

Emotion Objects and Aligned ERs
My review thus far has focused on science communication using graphs and illustrations;
however, climate is frequently communicated with photographs of people and places that frame
the issue and make salient or render invisible certain considerations (Rebich-Hesphana et al.,
2015). The impacts of climate ERs on emotions are well studied (Lehman et al., 2019; Lorenzoni
et al., 2006; Leiserowitz, 2006) however, few studies have assessed the impact of climate ER’s
emotional impact on change processes. Therefore, I describe studies on a range of topics and
studies highlighting the relevant study aspects for my investigation.
Schneider et al. (2018) provided the strongest evidence for the role of cognitive and
emotional alignment documenting effects on emotions, task relevant thoughts, and learning,
although not conceptual change. Across three studies the authors explored how young adults (age
range 17 to 25) learned from text and ER combinations about non-controversial issues. ERs were
photographs of people, food, and animals as the authors thought images of baby animals or
endangered animals may trigger emotions (Lang et al., 2008). Across studies, individuals agreed
with pilot study results for emotional alignment, providing evidence that even ‘decorational’ ERs
communicate information (Carney & Levin, 2002) likely due to their specific object focus
(Pekrun & Perry, 2012). The authors also found that negative ERs yielded more task-irrelevant
42
thoughts, although they did not investigate the object focus of the emotion nor did they assess
congruency of the thoughts to the emotional state, although possible (Damasio, 2004;
Immordino-Yang et al., 2009; Yang et al., 2018). For learning outcomes, the authors’ results
confirm the importance of ERs emotional alignment as positive ERs facilitated retention and
transfer compared to negative ERs. Text and ER alignment also related to learning outcomes
with strongly connected text and ERs facilitating learning whereas negative weakly connected
ERs decreased learning outcomes.
Hart and Nisbet (2012) conducted a study to understand how framing climate impacts as
near (e.g., in the state) or far (e.g., in other country) influenced identification with those impacted
and support for climate policies. Here 240 adults (mean age 38, range 18 to 80) were randomly
assigned to one of three conditions, non-political control text, low social distance text, or high
social distance text. Social distance was manipulated by changing the name and location (i.e.,
New York or France) of farmers who were impacted by climate change. Said differently, they
were interested in how the ERs object focus (and its alignment with climate change) interacted
with political ideology to influence decision-making.
Overall, results indicated that social distance manipulations influenced policy decisions
but were largely contingent on political ideology with strong Democrats indicating the greatest
support, Independents indicating moderate support, and strong Republicans indicating the
weakest support. The authors however found no support for the role of climate or scientific
knowledge on policy decisions and called attention to the limitations of the deficit model of
science communication, or the belief that all people need is information. Given their use of
human images and the focus on climate impact events, it is possible that compassion was evoked
by the depiction of personal loss (Immordino-Yang et al., 2009; Pekrun, 2012).
43
Critiques of Studies on Text and ER Alignment
The studies conducted by Danielson et al. (2016) and Mason et al. (2017) advance our
understanding of how external representations influence conceptual change however they are
limited in several ways. First, both studies investigated conceptual change for young adults, and
ignored the socio-scientific considerations for the issue (i.e., attitudes, emotions, political
affiliation). Second, ERs used in these studies were unintentionally flawed (see Table 8). For
instance, the climate temperature graph used in Danielson et al. (2016) is difficult to
comprehend. The y-axis uses Celsius to communicate temperature to Americans, and the x-axis
represents time using whole numbers (i.e., 400) that must be multiplied by “thousands of years”
to calculate the accurate time scale (a known area of difficulty) (Lombardi & Sinatra, 2012;
Resnick et al., 2017). Additionally, both the aligned and misaligned ERs used to depict seasonal
change inaccurately distorted the elliptical pattern of the earth, simply because a three-
dimensional representation was transposed on a two-dimensional space indicating that the ERs
could not be analogically mapped as their features were in opposition.
Regarding emotional alignment among ERs, several studies (Hart & Nisbet, 2012;
Lorenzoni et al., 2006; O’Neill & Nicholson-Cole, 2009) assessed the relationship of ERs and
emotions on climate relevant thinking; however, no study to date has explored how ERs can
facilitate conceptual, attitudinal, and policy preference change. This is partly due to previous
theorizing that ‘decorational’ ERs do not contribute to learning, however Schneider et al. (2018)
provide strong evidence that the cognitive and emotional alignment between a text and ER can
impact comprehension. Combined with Rebich-Hesphana and colleagues finding that visual
frames make salient or render invisible aspects of climate change, it is logical to ask, “how do
44
climate change ERs depicting different emotion objects (themes) impact conceptual, attitudinal,
and policy preference change, and are these impacts attributable to emotional responses?”
Conclusion
This literature review aimed to motivate my investigation into the impact of external
representations on climate change understanding and decision-making. I have attempted to
indicate that the climate change communication research is fragmented regarding the goals for
any given study. I began by reviewing literature on text-based communications on knowledge,
attitude, and emotion as these are the primary goals and research questions from the educational
psychology perspective (Lombardi & Sinatra, 2012; Muis et al., 2015; Nussbaum et al., 2017).
Here, I highlighted the recurrent finding that individuals would benefit from concrete depictions
of climate change, and that when learning about climate change, emotions influenced not only
conceptual change outcomes, but also directed focus toward different emotion objects that
influenced strategy use.
I next reviewed the literature on thematic visual frames in climate communication, as
they are under attended to in the educational psychology literature. I reviewed this literature
separately for I sought to connect it directly to a limitation of the CRKM (little consideration of
visual message characteristics) and tie it to the process of thematic selection in the ITPC.
Therefore, I reviewed studies that employed survey, interview, and sorting tasks to determine
what mental images individuals possess for climate change and how these images relate to affect
and decision-making (O’Neill & Nicholson-Cole, 2009; Lorenzoni et al., 2006, Leiserowitz,
2006). This step was necessary given I am interested in conceptual change, and therefore must
acknowledge the role of mental images in prior knowledge, and their possible connection to
identity-based processing and thematic selection (i.e., attitudes and political ideology). I also
45
reviewed content analyses (Born, 2019; Lehman et al., 2019; Rebich-Hesphana et al., 2015) of
thematic frames employed in public communications about climate change in order to select
external representations and themes that are known to be used when communicating with the
general public. Here, I detailed the common thematic frames, and the modalities of their
presentation (i.e., images, diagrams, graphics) documenting influences on perceptions of
relevance and salience. I also reviewed a large-scale meta-analysis documenting what influences
climate change adaptation, highlighting the role of negative affect, and self-efficacy given
imagery influences both variables (van Volkengoed & Steg, 2019).
Within this section I made note of the fact that many individuals have no mental image of
climate change, or the ones they have are inaccurate (i.e. sea ice melt). I focused on accuracy in
climate change imagery as a leading scholar recently challenged the field to locate images that
avoid misrepresentation and provide legitimate affective engagement. This focus on
misrepresentation dove-tailed with the analogical structure-mapping process explicated by
Schnotz in the Integrative Theory of Text and Picture Comprehension. In this section I reviewed
literature that manipulated text and external representations when communicating about
scientific issues. I focused primarily on accurate analogies, as these are discussed in the ITPC,
however, I also covered recent work that challenges models of text and graphic comprehension
to explore alternative functions that ERs may serve (Danielson & Sinatra, 2017). From here, I
discussed research on accurate and inaccurate external representations. Theoretically these
studies are foundational for they provide evidence that conflict can be detected either in a text or
external representation, and that there are instances where aligned or misaligned ERs benefit
learning. I also highlighted limitations of these investigations, specifically the operationalization
of the misaligned ERs, learners age, topic lacking controversiality, and the ignoring of emotions.
46
It was only by reviewing the literature on themes in climate communication, and the role
of alignment in conceptual change learning that I could review their interaction. Currently, no
study that investigates the impact of climate communications on conceptual change learning
attempts to identify the source of emotions related to the thematic frame. The CRKM cannot
account for the role of ERs in conceptual change learning but does acknowledge the importance
of emotions in the process. The ITPC cannot speak directly to the process of conceptual change
learning but explicates two mechanisms regarding how ERs may influence the process.
Therefore, the final section of the literature review discussed how aligned and misaligned
external representations depict different emotion objects. I leveraged recent studies on the
relationship between an ERs content and its affective charge to provide a rationale for their
investigation (Schneider et al., 2018). Climate communication research has investigated what is
already in the publics’ minds about climate change and how this drives current decision-making.
Research has also investigated how to educate the public or influence their policy preferences
with extant research indicating it is often difficult to address both aspects as they are facilitated
by different emotions and motivations (van Volkengoed & Steg, 2019). There are component
pieces of my research questions spread across disciplines, but no study has sought to address
their interactions when facilitating conceptual change for the issue of climate change (see Table
10). This study builds upon prior work in these disciplines to assess how choices about the
depiction of climate change impacts and their associated emotion objects relate to conceptual,
attitudinal, and policy preference changes.

47
Table 10
Summary Chart of Relevant Literature According to Topic, Population, Manipulation, and
Outcomes
Study Topic Population Manipulation Outcome(s)
Heddy et
al. (2016)
Genetically
Modified
Foods
N = 322
US College students

Ideology
20% conservative
14% moderate
44% liberal
20% libertarian
Refutation text
vs.
no
intervention
Knowledge

Attitude

Emotions
Lombardi
et al.
(2013)
Human-
induced
climate
change

N = 169
US Middle school
students
Critical
evaluation
instruction
(comparison
activity)
vs.
traditional
course
materials
Knowledge

Plausibility
judgements
Ranney
and Clark
(2016)
Climate
Change

Exp 1: N = 270
US adults

39.3% Democrat
Survey to
address
knowledge,
attitude, and
behavior
Knowledge

Attitude

Willingness to
sacrifice
Exp 2: N = 85
US college students
2-Time: pre vs
post
X
2-Intervention:
pretest vs no-
pretest

400-word text
on mechanistic
understanding
Knowledge

Attitude

Surprise
Exp 3: N = 80
US college students
(online Qualtrics)
3-Time: pre,
post, delay
X
400-word text
on mechanistic
understanding
Knowledge

Attitude
Exp 4: N = 28 3-Time: pre,
post, delay
Knowledge

48
US college students
(online Amazon
MTurk)

58% democratic
X
400-word text
on mechanistic
understanding
Attitude
Exp 5: N = 63
US high school
chemistry students
3-Time: pre,
post, delay
X
2-Intervention:
400-word
mechanistic
text + 6 key
statistics
Vs. 400-word
mechanistic
text + 6
unrelated
statistics
Knowledge

Attitude

Environmental
behavioral
Intentions
Exp 6: N = 38
US Amazon
Mechanical Turk
workers

45% Democrats
3-Time: pre,
post, delay
X
7 statistics
only
Acceptance

Surprise
Exp 7: N = 104
UC Berkley students

39% Democrats
2-Time: pre vs
post
X
2-Misleading
statistics
Intervention: 2
item group vs
8-item group

400-word text
on mechanistic
understanding
Acceptance

Policy funding

Lombardi
and
Sinatra
(2012)
Distinctions
between
weather and
climate
N = Two college
undergraduate
courses
Age range: 18-66

Semester
Thematic
instruction:
global
warming vs.
physical
geography
Knowledge:
Categorization
accuracy

Nussbaum
et al.
(2017)
Distinctions
between
N = 146
Undergraduates
Age range: 17-58
2-Text:
Refutation text
Knowledge:
Weather and
Climate
49
weather and
climate

vs. Expository
text
X
3-Time: pre,
post delay
Assessment
(WCA)
Lombardi
et al.
(2014)
Climate
Change
N = 271 USC
undergraduates

Age range: 18-38

Partisanship
59% Democrat
20% Republican
14% Independent
5% Other

2-906 word
text:
Scientist
editorial
vs
Politician
editorial
Knowledge

Attitude

Perceptions of
certainty

Perceptions of
trustworthiness

Perceptions of
author
expertise

Plausibility
perceptions
Stenseth
et al.
(2016)
Climate
Change
N = 153
Norwegian students
Mean age = 17
Knowledge

Attitude

Interest
Thacker
et al.
(2020)
Genetically
Modified
Foods
N = 424
Canadian, US, and
German
undergraduates

Mean age = 22

Knowledge

Attitude

Emotions
Muis et
al. (2015)
Climate
Change
Experiment 1
N = 439
Canadian, US, and
German
undergraduates

Mean age= 22

Experiment 2
N = 59
Canadian
undergraduates
Conflicting
texts on
positive and
negative
impacts
Knowledge

Emotions

Study
strategies

Emotion
Object
50
Muis et
al. (2018)
Genetically
Modified
Foods
N = 120 Canadian
undergraduate

Mean age =22

Knowledge

Epistemic
emotions

Task value

Learning
strategies

51
Chapter 3

Methods

Participants and Context
Data for this study was collected using Amazon’s Mechanical Turk, an online platform
where individuals are paid to complete tasks. Individuals were qualified to participate if 18 years
or older, lived in the U.S. and had completed at least one prior Mechanical Turk task. The initial
dataset contained 1024 responses, 735 were 100% completed by individuals who consented to
participate. Examination of the open-response items to the Weather and Climate Assessment
(WCA) revealed a large number of identical or copy and pasted responses indicative of bots
completing the survey. I reviewed all six pre-intervention responses and assigned a value of zero
if the response was unique and one if the response was copy/pasted and identical to other
responses. I then summed the number of issues for each participants’ responses and used this
categorical variable to conduct a series of one-way ANOVAs on the primary study outcomes
(i.e., knowledge, attitude, policy preferences, emotions). See Appendix A for descriptive
statistics of key variables by response issue and table notes for explanations of findings. Overall,
respondents with zero open-response issues (N = 373) read the experimental texts for
significantly longer, learned significantly more, increased their positive attitude, and experienced
less intense emotions compared to the average of all other response issues. Finally, two
individuals’ data were removed given they failed more than one attention check resulting in a
final dataset of 371.
I chose an online sample of adults given my interest in the publics’ understanding of
science and the importance of informal climate communication (Leiserowitz et al., 2020). I
wanted to understand how learning about climate and weather in your own home influenced
52
short-term decision making about climate policy. Additionally, given the ubiquity of ‘doing your
own research’ the availability of government materials (i.e., NOAA control text), and recent
investigations of informal learning from science websites (Roser-Renouf et al., 2020)
understanding how learning occurs in an informal setting is key.
Table 11
Descriptive Statistics for Participant Characteristics of Final Sample N = 371 Adults
Variable Mean or % SD Min Max
Age 41.23 13.61 18 77
Gender
Female 48.2%
Male 50.7%
Other/prefer not to state 1.1%
Political Ideology
Very to lean liberal 51.8%
Centrist 14.8%
Very to lean conservative 31.2%
Other/prefer not to state 2.2%
Partisanship
Strong to lean Democrat 51.2%
Independent 20.5%
Strong to lean Republican 25.6%
Other/prefer not to state 2.6%
Ethnicity
African American/Black 5.66%
American Indian/Alaska Native 1.08%
Asian American/Asian 8.36%
Native Hawaiian/Pacific Islander 0.54%
Mexican American 0.54%
Puerto Rican 0.54%
53
Other Latino 0.54%
White/Caucasian 76.28%
Prefer not to state 0.81%
Other 1.35%
Identified as Multi-racial 4.31%
Education
Some high school 0.8%
High school diploma/GED 7.5%
Associate degree 9.4%
Some college 16.4%
Bachelor’s degree 40.7%
Master’s degree 20.5%
Doctoral degree 4.0%
Prefer not to state 0.5%

Instruments
Mental Image of Climate Change Question. Given the relationship between mental
imagery of climate change and its impacts on perceptions of risk and affect (Lehman et al., 2019;
O’Neill et al., 2013), and climate policy preferences (Leiserowitz, 2006), I included a single
open-response item, “When you think about climate change does a specific image come to
mind?”
Categorization of Statements as Weather or Climate. Understanding of climate
change, specifically the ability to categorize statements as being about weather or climate is a
primary outcome and was measured using the Distinctions between Weather and Climate
Measure (DWCM) (Lombardi & Sinatra, 2012). The DWCM is a 13-item forced choice test and
asks learners to, “Decide if the statement best fits into the category of weather or climate.” (see
Appendix B). Lombardi and Sinatra constructed the measure to represent misconceptions
54
documented in prior research about climate and weather (Gowda, 1997; Papadimitriou, 2004),
and allows measurement of another facet of conceptual change, category mistakes (Chi, 2008).
Expert science educators reviewed the measure for content validity however the authors reported
low internal consistency was low (KR-20 = .48 to .52) which was also found here (KR-20 = .44
to .57). Accurate responses were summed to create two variables (pre, post) categorization
accuracy, such that higher scores indicated more correct categorizations.
Definitions of Climate Change Variables. Understanding of climate change,
specifically the distinctions between weather and climate is a primary outcome in this study and
was measured using a modified version of Nussbaum et al., (2017) Weather and Climate
Assessment (WCA) (Appendix C). The modified WCA presents 6 open-ended questions that
measure beliefs by asking participants to define climate, weather, the length of time to determine
climate, evidence that supports climate change, and the impact of large bodies of water on the
climate. Nussbaum et al. (2017) provided evidence of content validity by having content experts
review the questions and demonstrated acceptable inter-rater reliability (.86). The scoring rubric
has been modified from the original proposed by Nussbaum and colleagues and is therefore
subject to additional tests of validity moving forward. Here results are only presented for the first
WCA question, ‘what is climate’ and were analyzed separately along three dimensions (where,
what, when), and overall.
Climate Change Attitude. Climate change attitude was a primary outcome in this
investigation and was measured using the Attitudes about Global Warming Measure (Sinatra
(Sinatra et al., 2012). The measure presents six items and asks individuals to rate their agreement
on a 5-point Likert scale from 1 (strongly disagree), 3 (unsure), to 5 (strongly agree), such that
change can be described as increased agreement. One item is reversed scored, “the speed with
55
which the melting ice caps may raise sea level is uncertain” so that increased uncertainty, was
not in the direction of increased agreement. Internal consistency for the scale was adequate
(α
𝑝𝑟𝑒 = .86; α
𝑝𝑜𝑠𝑡 = .86), and results were analyzed by averaging responses to all questions, such
that a higher value indicated increased acceptance attitude toward climate change (see Appendix
D).
Climate Policy Preferences. Policy preferences are a primary outcome in this study and
were measured using the Climate Policy Preferences measure created by Leiserowitz (2006) and
modified by Leiserowitz et al. (2009) (see Appendix E). This version includes 13-items and asks
participants to indicate their support or opposition for two types of policy decisions (i.e., general
and tax specific) on a 4-point Likert scale with no neutral midpoint ranging from 1 (strongly
support) to 4 (strongly oppose), although a do not know option was added to allow individuals
the ability to change from no position to a specific position. Leiserowitz (2006) provided
psychometric support for the two sub-scales, with both general policy preferences and tax policy
preferences demonstrating acceptable internal consistency (α > .78). Sub-scale reliabilities in this
investigation were similar: general policy preferences (α
𝑝𝑟𝑒 = .77; α
𝑝𝑜𝑠𝑡 = .78) and tax policy
preferences (α
𝑝𝑟𝑒 = .85; α
𝑝𝑜𝑠𝑡 = .89). Here, variables were calculated as average support or
opposition (after recoding do not know responses as missing data), such that higher values
indicated more support.
Emotion Measures. Given my interest in an individual’s emotional experience, I follow
Immordino-Yang’s (2010) call for interdisciplinary methodologies that extend our understanding
of how emotions impact decision-making. Therefore, I adapt Immordino-Yang et al. (2009) and
Yang et al. (2018) cognitive construal task, and Pekrun and colleagues’ (2017) epistemically
related emotions scale (EES) to bridge conceptual and methodological differences.
56
Cognitive Construal of Task. Immordino-Yang et al. (2009) developed and later
validated (Yang et al., 2018) a multi-stage interview protocol to assess how individuals make
sense of and construe an emotional event (i.e., listening to narratives and viewing photos that
arouse social emotions). The approach relies on three techniques: 1) verbal interviews (recorded
and scored on cognitive and behavioral dimensions), 2) neuro-imaging techniques, and 3) self-
report measures. Given the online context of my study, verbal interviews were not feasible
(although considered), and my selected frameworks are incompatible with offering hypotheses
based on neuroimaging techniques. However, I adapted the verbal interview phase so that
individuals first read an informational text about weather and climate distinctions paired with an
external representation varied on thematic frame (i.e., polar impacts, weather impacts) and
alignment (i.e., aligned, misaligned) before immediately responding to the question, “How does
this information about climate change make you feel?”
A construal synthesizes the social, affective, cognitive, and episodic considerations
available during learning into a short-term and context specific representation. Immordino-Yang
et al., (2009) and Yang et al., (2019) identified three dichotomous dimensions of social emotion
construals: 1) positive affect toward a human emotion object (e.g., “she is amazing”), 2)
descriptions of general beliefs, values, and attitudes (e.g., “she gives me hope for humanity”)
termed abstract, and 3) descriptions contrasting characteristics of self to the human emotion
object (e.g., “I wish I could do that”).
In this study, cognitive construals were coded along three dimensions: affect, abstraction,
and concreteness although modifications occurred to each dimension (see Appendix F). First, the
positive affect dimension was extended to include neutral, negative, and combinations of all
affect codes as many construals were mixed, and climate change is not a positive topic. Neutral
57
was applied to codes that explicitly stated, they didn’t feel any different, or didn’t feel about the
topic or responses that referred to some aspect of learning, feeling informed or understanding
(i.e., epistemic emotions). Positive codes were applied to statements where a positive emotion or
feeling word was indicated, although most positive statements referred directly to how the
information, they read validated, supported, or increased personal confidence and was described
as good. It was also necessary to specify the treatment of responses referring to decreases in
negative affect (e.g., “much less stressed,” or “relieved”) which I ultimately decided to treat as
negative, for it preserves the beginning and starting point of the feeling. It may have decreased,
but it appeared inaccurate to code this as positive. Negative codes were applied to responses that
described a negative emotion (e.g., “it makes me feel anxious,” “sad,” “afraid,” “scared,”
“irritated”) or feeling for example “depressed,” “more concerned,” or “worried.” Finally, four
mixed codes were necessary (i.e., neutral and positive, neutral and negative, positive and
negative, and neutral, positive, and negative). Neutral and positive applied to responses that
referred to learning but were paired with a positive emotion not focused on the information
directly (e.g., “informed, and hopeful we can do something”). Neutral and negative applied to
responses that referred to learning, but also described a negative emotion or feeling not focused
on the information (e.g., “ I don’t feel any way…maybe annoyed because I don’t believe it is
caused by man”). Positive and negative codes applied to responses such as, “concerned the
climate is changing but hopeful we help slow or stop it”.
The second and third dimensions of cognitive construals were abstract and concrete
respectively. Initially, the abstract dimension was sensitive to general beliefs (e.g., “it is
happening,” “it is a hoax”), values (“we need to protect future generations,” “this is important”),
and attitudes (e.g., “climate change isn’t caused by humans”). The third dimension, concreteness
58
was defined narrowly, such that it was only applied when individuals explicitly (i.e., concretely)
referred to the text (e.g., “It gives me more clarification about what affects climate”) but not
when the text was absent from the construal (e.g., “feel so anxious about the future”). The second
and third dimensions of the construal were collapsed approximately halfway through coding, as
both independent coders realized the abstract and concrete codes (as defined) were inversely
related. More specifically, regardless of the affect code, a construal was coded as abstract if it
didn’t connect the general belief, value, or attitude concretely to the text. Both codes could be
applied independently but the only difference between the codes was their concrete connection to
the text or not. For this reason, the abstract code was dropped as it was best captured by the
concrete code.
Two researchers completed the coding process. First, the coders met for an introduction
to the coding scheme and took turns applying codes and discussing code boundaries. Next, both
coders independently coded a training set of 15 responses and again met to discuss, resolve, and
formalize code boundaries. Finally, both coders independently coded a trial set of 25 responses
to assess coder agreement (> 90%.) These same two coders then independently coded all
responses (including those in the initial trial sets) and coder agreement for all dimensions (affect
= 89.9%, and concrete = 98.3%). The authors codes were used for all analyses presented.
Epistemically Related Emotions Scale (EES). The emotions experienced while learning
about climate and weather were of primary interest and were also measured in line with
educational psychology perspectives on emotions, specifically epistemically related emotions.
Here I adopted the Epistemically related Emotions Scale (EES; Pekrun et al., 2017) to measure
the self-reported intensity of emotions experienced after the task, in a procedure similar to
Immordino-Yang et al. (2009) construal task. After reading, participants were asked to indicate
59
the intensity of each emotion they experienced while reading and thinking about the weather and
climate distinction on a 5-point Likert scale anchored at 1 (not at all), 3 (moderate), and 5 (very
strong). The EES addresses seven emotions: surprise, curiosity, enjoyment, confusion, anxiety,
frustration, and boredom and was supplemented with 3 additional terms identified in previous
studies on socio-scientific topics: hopeful, and hopeless (Heddy et al., 2016), and interest which
some consider an emotion (Stenseth et al., 2016; Thomas & Kirby, 2020) for a total of 11
epistemically related emotions (see Appendix G). Here, each emotion variable was treated
separately as I was interested in exploring discrete emotions.
Emotion Object Focus Measure. While the EES asks participants to indicate the
intensity of several emotions on a simple Likert scale, it is difficult to understand the source of a
given emotion. For example, if one selected ‘hopeful’ at a moderate level the researcher has little
sense of the emotion’s source. To capture this information, object focus questions were asked
when individuals selected an intensity greater than one (not at all), for it seemed inappropriate to
inquire about the source of an emotion, when they reported no emotional intensity. However, if
an individual selected any intensity greater than one, they were asked, “Please indicate the
source of your anger. (select all that apply).” Options included 1) ideas in the text, 2) people
impacted by climate change, 3) animals impacted by climate change, 4) nature impacted by
climate change, 5) fact that climate change is happening, 6) politicians, 7) something else (and
contained a text entry box) (see Appendix G). These emotion object questions served two
functions. First, they provided an indirect manipulation check, for thematic frames (polar
impacts, weather impacts) are likely to influence the emotion objects identified (i.e., more
selections of animals in the polar impacts frame; more selections of humans in the weather
impacts), and alignment (i.e., ideas in the text and fact that climate change is happening).
60
Second, they provide a bridge between studies of social emotions, and studies of topic and
epistemic emotions when learning from verbal and visual information about climate change.
Here, two variables were calculated: 1) summed number of emotion objects for a given emotion
(e.g., how many objects were selected for surprise), 2) summed number of emotion objects
selected across emotions (i.e., how many times ideas in the text were selected).
Political Partisanship and Ideology
Political affiliation and ideology are included given their documented relationship to
policy preferences (Hart & Nisbet, 2012), attitudes toward climate change (Leiserowitz, 2006;
O’Neill & Nicholson-Cole, 2009), and resistance to correcting misconceptions (Kaplan et al.,
2016; Nyhan & Reifler, 2010). Here, I adopt Nyhan and Reifler’s (2010) measures of
partisanship anchored at 1 (very Democrat), 4 (Independent), 7 (very Republican), and political
ideology 1 (very liberal), 4 (centrist), 7 (very conservative) (see Appendix H).
Demographic Questionnaire. Individuals were asked to provide information about
several variables and were given options to provide a description of their self-identification or
prefer not to state. Demographics included age, gender, ethnicity, and educational attainment
(see Appendix I).
Procedures
The design for this study was a 2-Time (pre, post) x 3-Thematic frame (polar impacts,
weather impacts, science communication) x 2-Cognitive alignment (aligned, misaligned) mixed
experimental design with one within subject factor (time), and two between subjects’ factors
(thematic frame and cognitive alignment). All participants completed three stages: 1) pre-
intervention (i.e., mental image of climate change, climate change definitions (WCA), weather
and climate categorization accuracy (DWCM), attitude toward climate change, and climate
61
policy preferences (CPP)), 2) random assignment to one of six climate communication messages
varied on thematic frame and alignment, 3) post-intervention (i.e., cognitive construal of emotion
task, the epistemically related emotions scale (EES) with emotion object questions, the WCA
and DWCM, attitude toward climate change, CPP, and the demographic questionnaire Appendix
K). After submitting their survey individuals were shown a debriefing message and provided
their Mechanical Turk payment code. In all, the study took an average of 25 minutes to
complete.
Experimental Conditions
Climate communication messages typically include textual and visual information in the
form of an expository of informational text however, refutational texts are more effective in
communicating accurate scientific and socio-scientific information (Guzzetti, 2000; Guzzetti et
al., 1993; Kendeou et al., 2014; Kendeou et al., 2016; Tippett, 2010). In this study, individuals
were randomly assigned to one of six conditions. I first describe the traditional expository format
and connect it with existing approaches to communicate weather and climate. Then, I describe
the refutational text for the distinctions between weather and climate. Finally, I describe the
external representations that were paired with the refutational text.
NOAA Infographic Control Text
The first control is a ‘naturally occurring’ infographic created by the National Oceanic
and Atmospheric Administration (NOAA), entitled “What’s the difference between weather and
climate?” (Appendix J). Given the text was not constructed by educational psychologists it
deviates from best practice (i.e., expository text format, small text size, violations of spatial
contiguity among others). The infographic serves as an ecologically valid control as it was
created by an established scientific agency (NOAA) to address the weather and climate
62
misconception and does not feature a refutational text structure. Specifically, it neither activates
the misconceptions, nor does it provide a cue that incoming information may conflict with prior
beliefs (both are present in a refutational text structure). Instead it features only the scientific
explanation, which is not as effective as a refutational text structure (Guzzetti et al., 1993;
Kendeou et al., 2014) in facilitating conceptual change.
The infographic contains icons that redundantly depict written words (e.g., above the
word “rain” are three water droplets suspended inside of a hexagon) and is therefore unlikely to
impact conceptual change, although situational interest is beneficial to conceptual change
(Thomas & Kirby, 2020). However, the infographic features a timeline that is color coded to
correspond with the descriptions of weather and climate and may facilitate conceptual change in
line with recommendations of Lombardi and Sinatra (2012). Finally, the infographic does not
contain any thematic overlap with the experimental ERs, however, it and the refutation text only
condition can safely be categorized thematically as science communications. Importantly, while
content coverage between the NOAA infographic and the refutation text differs (i.e., NOAA
focuses on seasonal change, whereas the refutation text attends more to the role of oceans) both
formats contain the information necessary to correctly answer all questions.
Refutation Text Only
I adopted Nussbaum et al. (2017) refutation text titled “What is the Difference Between
Weather and Climate?” that was prepared and reviewed by a secondary geoscience teacher
(Appendix K). The following excerpt from Nussbaum and colleagues text demonstrates the
three-part refutation text structure and can be contrasted with the NOAA infographic (Tippett,
2010).
“Did you know that about 60% of people think that climate changes from year to year?
However, this is not the case. Climate is the long-term (30 years or greater) conditions in
63
the atmosphere, as well as in the ocean and in the sea and land ice. Climate varies by
region when there are long-term differences in these conditions.”

Here, the first sentence activates the misconception, the second provides a cue to
potential incongruity between the current belief and incoming information, while the third
sentence states and supports the scientific conception. The refutation text is 711 words, eight
paragraphs and has a grade level of 9.9. Timing data was captured to assess how long
participants spent reading each experimental condition and was used to assess the impacts of
message exposure on indices of change.
The text addresses three misconceptions related to the distinctions between weather and
climate. The first misconception, weather as climate, is decomposable into three dimensions:
time (when), space (what), and spacetime (where). Weather is the variable condition of the
atmosphere (what) over a short period of time (when) in a small area (where), whereas climate is
the average condition of the atmosphere, ocean, land or sea ice (what) over a longer period of
time (when) for a larger area (where). Related to the first misconception is the use of weather
information (a cold, cloudy, rainy, humid, or windy day) as evidence against climate change. The
third misconception relates to the first as individuals struggle to identify what influences the
climate and tend to focus on air, wind, and sunlight (weather) instead of the ocean and its
currents. This refutation text was paired with all external representations in the 2-Thematic
Frame (polar impacts, weather impacts) x 2-Alignment (aligned, misaligned) design.
External Representations
Lewandowsky and Whitmarsh (2018) ask, “how can we legitimately use the images that
we, as humans, find so alluring and convincing without risking scientific inaccuracies?” (p. 2). I
operationalized the message format to appear similar to a New York Times article (although
there is no reference to any news organization). Instead, all the external representations were
64
structured similarly (see Table 12). The text title was center aligned at the top of the page, with
the external representation below. Directly under the ER was a caption that reinforced the ERs
alignment (e.g., “Changes in Arctic sea ice from 1980-2012. Image from NOAA.”). The body
text appeared below. Finally, before the summary paragraph the same ER was presented again to
support spatial contiguity (Mayer, 2005).
Given my use of a NOAA infographic, and the aligned weather impact ER coming from
the NOAA, I decided to attribute the aligned polar ER to the NOAA as well. The aligned
weather impact ER had as caption reading, “Changes in global average surface temperature from
1998-2018. Image from NOAA.”). Here the only imbalance is a difference in the years that
determine the climate interval (1980 to 2012 vs 1998 to 2018). Captions for the misaligned ERs
were identical aside from the emotion objects depicted. The misaligned polar ER had the caption,
“Impact of climate change on polar bears searching for food in the Arctic 2018. Photo obtained
from Betsy Lehman.” Here the photos were attributed to Betsy Lehman, the first author of
Lehman et al. (2019) and provider of the climate ERs used in this investigation. Finally, the
misaligned weather ER was captioned, “Impact of a 2018 hurricane on infrastructure in a United
States town. Photo obtained from Betsy Lehman.”)

65
Table 12
External Representations in the 2-Thematic Frame (weather, polar) x 2-Cognitive Alignment
(aligned, misaligned) Experimental Design with Norming Data from Lehman et al., (2019)
Cognitive Alignment
Cognitively Aligned Cognitively Misaligned
Thematic Frame
Polar
Impacts
Changes in Arctic sea ice from
1980-2012. Image from NOAA.
Impact of climate change on polar bears
searching for food in the Arctic 2018.
Photo obtained from Betsy Lehman.
Relevance1:
7.99
Arousal2:
5.34
Valence3:
2.72
Relevance:
7.33
Arousal:
5.03
Valence:
3.48
Weather
Impacts
Changes in global average surface
temperature from 1998-2018. Image
from NOAA.

Impact of a 2018 hurricane on
infrastructure in a United States town.
Photo obtained from Betsy Lehman.
Relevance:
7.674
Arousal:
5.31
Valence:
2.78
Relevance:
7.45
Arousal:
5.40
Valence:
1.96

1 Relevance ratings were measured on a 9-point Likert scale 1(not relevant) to 9 (relevant).
2 Arousal ratings were measured on a 9-point Likert scale 1(calm) to 9 (excited).
3 Valence ratings were measured on a 9-point Likert scale 1(negative) to 9 (positive).
4 Reported values for this image do not apply specifically to this ER but to another that depicts
the same data, but the scale is in Celsius. I decided to include an ER that aligns with the US
system of measurement to avoid unnecessary confusion.
66
Thematic Frames
My primary focus was on avoiding scientific inaccuracies, and therefore manipulated
cognitive alignment (i.e., aligned or misaligned representation of temporal and spatial
dimensions of climate) in an external representation (ER) for the weather and climate distinction.
Related to the primacy focus (scientific accuracy) is the issue of ‘alluring and convincing
representations’, as they denote attraction and persuasion. Here, I leveraged prior work on
thematic frames in climate communication (Rebich-Hesphana et al., 2015) to identify two
specific thematic frames,— polar impacts (i.e., impacts on polar bears and impacts on polar ice)
and weather impacts (i.e., impacts of extreme weather and impacts on global temperature
change), and one general frame (i.e., science communication). For all thematic frames except the
refutation text only and NOAA infographic conditions, I selected ERs that had similar ratings of
relevance, arousal, and valence based on data collected by Lehman et al. (2019).
Polar Impacts. An underattended issue in climate change communication is the fact that
individuals rate a polar bear on sea ice as equally relevant to climate change as land ice recession
over time, again showing the misrepresentation of space and time in climate communication and
understanding. Therefore, one thematic frame is polar impacts, specifically how Arctic sea ice,
or polar bears are impacted by climate change. Both ERs were scored similarly for relevance to
climate change (8.21 vs. 7.99), arousal (5.30 vs. 5.34), and valence (2.25 vs. 2.72) providing
evidence these ERs depict objects that are perceived as highly relevant to climate change, are
slightly more exciting than calming, and are generally rated as negative. The polar impacts frame
depicted emotion objects (i.e., polar bears) and may induce social (e.g., compassion for physical
or psychological pain; Immordino-Yang et al., 2009) or topic emotions (e.g., political anger;
67
Born, 2019) while the polar ice ER may evoke topic (e.g., sadness) or epistemic (e.g., surprised
at the speed of ice loss) emotions.
Weather Impacts. The second set of ERs focused thematically on weather impacts (i.e.,
global temperature change, extreme weather event) and received similar ratings for relevance to
climate change (7.67 vs. 7.57), arousal (5.31 vs. 5.27), and valence (2.78 vs. 2.51). Here the
object focus was directed at global temperature change or hurricane damage. Again, while
evaluations were similar, one cannot be sure what emotion objects produced this evaluation, nor
how these emotional reactions relate to knowledge.
Analytical Approach
The goal for this research was to explore the influence of external representations when
communicating to the general public about weather and climate. Here I manipulated two aspects
of an ER, its thematic frame and the frame’s accuracy in depicting temporal and spatial aspects
of the concept climate to understand their individual or combined impacts on conceptual,
attitudinal, and policy preference change. Prior to the main analysis I conducted preliminary
analyses to determine equivalence of key study variables across groups. To address my first two
research questions (i.e., does an ERs thematic frame or cognitive alignment impact variables of
interest) I conducted repeated-measures ANOVAs and followed up significant interactions with
simple effects tests using a Bonferroni correction. To answer my third research question, I
conducted several 2-by-2-by-2 repeated measures ANOVAs to address the interactions between
thematic frame and cognitive alignment on indices of climate change understanding, again
decomposing interactions using simple effects tests and a Bonferroni correction to adjust for
alpha inflation. Finally, to answer my fourth research question, I conducted classic 3-by-2-by-2
ANOVAs on each epistemic emotion to assess the influence of thematic frame and alignment on
68
self-reported emotions, taking into consideration message exposure. Additionally, 2-by-2
ANOVAS were used to answer research questions 4b which related to the selection of emotion
objects, and 4c which analyzed the affect and concreteness of one’s construal.

69
Chapter 4

Results
Preliminary Analyses
All variables were examined for skewness and kurtosis. For knowledge, attitudes, policy
preferences, and climate definitions skewness ranged from -1.02 to -0.89, and kurtosis from -
0.67 to 0.98. For the epistemically related emotions skewness ranged from -.45 to 1.95 and
kurtosis from -1.27 to 2.93 (Tabachnick & Fidell, 2013).
To examine equivalence pre-intervention, I tested for differences in prior knowledge
(categorization accuracy DWCM, climate change definitions WCA), attitude, and climate policy
preferences (policy, tax policy) (see Table 13 for descriptive statistics). Results indicated a
significant interaction between alignment and theme at pre-test (p = .047) for the distinctions
between weather and climate measure (DWCM), however no simple effects tests reached
statistical significance (p values > .057). For defining climate, results indicated a main effect of
alignment (p = .026) such that individuals assigned to a misaligned ER (M = 2.11, SD = 1.49)
had a more complete definition of climate compared those randomly assigned to the aligned ERs
(M = 1.73, SD = 1.38). For equivalence of attitude and policy preferences at pre-test results
indicated no main effects (p values > .577) or interactions (p values > .318).

70
Table 13
Descriptive Statistics Overall and by Thematic Frame and Alignment Condition

Full Sample
(n = 257)
Aligned Polar
ER
(n = 64)
Misaligned
Polar ER
(n =60)
Aligned
Weather ER
(n = 72)
Misaligned
Weather ER
(n = 61)
Mean SD Mean SD Mean SD Mean SD Mean SD
Pre DWCM 6.37 1.90 6.28 1.79 6.60 1.87 6.60 1.86 5.97 2.07
Post DWCM 6.72 1.69 6.69 1.51 7.02 1.69 6.69 1.78 6.49 1.73
Pre WCA1 1.91 1.39 1.63 1.34 2.13 1.26 1.82 1.41 2.09 1.49
Post WCA1 3.42 1.62 3.49 1.62 3.52 1.54 3.33 1.77 3.72 1.55
Pre-Attitude 3.66 0.87 3.71 0.79 3.62 0.95 3.68 0.86 3.65 0.88
Post Attitude 3.74 0.87 3.75 0.80 3.69 0.95 3.74 0.91 3.78 0.85
Pre-Policy
Preferences
3.07 0.66 3.12 0.71 3.02 0.74 3.09 0.62 3.03 0.59
Post Policy
Preferences
3.10 0.69 3.16 0.71 3.02 0.72 3.08 0.70 3.13 0.65
Pre-Tax
Preferences
3.02 0.83 3.00 0.87 2.91 0.88 3.12 0.81 3.04 0.77
Post Tax
Preferences
3.05 0.88 3.06 0.93 2.95 0.92 3.05 0.85 3.16 0.81

Impact of Thematic Frame on Knowledge, Attitude, and Policy Preferences
To answer my first research question, To what extent does an external representation’s
thematic frame influence knowledge, attitudes, and policy preferences? I conducted eight two-
way mixed design ANOVAs, five for knowledge (i.e., weather and climate categorization,
overall climate definition, climate definition dimensions) one for attitude, and two to assess
policy preferences (i.e., governmental climate policies, tax policy preferences). For all analyses,
the within-subjects variable was time, the between-subjects variable thematic frame, and all
follow-up tests used a Bonferroni correction.

71

Changes in Knowledge of the Distinctions Between Climate and Weather
To explore the extent to which an external representations thematic frame influenced
knowledge of weather and climate distinctions (categorizations, definitions) (RQ1a), I conducted
two 2-Time x 3-Thematic frame (polar, weather, science communication) repeated-measures
mixed ANOVAs with time as a within-subjects factor and thematic frame as a between-subjects
factor. For both knowledge measures, results indicated significant effects of time (p < .001), such
that all individuals regardless of thematic frame increased their accuracy in categorizing weather
and climate statements (𝑀 𝑝𝑟𝑒 = 6.36, SD = 1.87) (𝑀 𝑝𝑜𝑠𝑡 = 6.70, SD = 1.73), and in defining
climate (𝑀 𝑝𝑟𝑒 = 1.91, SD = 1.40) (𝑀 𝑝𝑜𝑠𝑡 = 3.42, SD = 1.61). I next conducted three repeated
measures ANOVAs to understand the influence of thematic frames on understanding the
individual dimensions of climate. I first analyzed changes in understanding the ‘where’ of
climate change, specifically the spatial and temporal relationship. Results indicated a main effect
of time (p = .002), but no interaction (p = .709), indicating all individuals increased their
understanding of how time and space relate to define climate. For the ‘what’ dimension of
climate, results indicated no effect of time (p = .584), or interaction (p = .225) between time and
thematic frame. However, for the final dimension of climate ‘when’ results indicated a
significant main effect of time F(1, 368) = 467.29, p < .001, partial η-squared = .559 such that
all individuals increased their understanding of the time needed to define climate from pre (M =
0.41, SD = 0.88) to post (M = 1.77, SD = 1.11). Additionally, this effect was qualified by a
significant time by thematic frame interaction F(2, 368) = 4.79, p = .009, partial η-squared =
.025. Simple effects tests with a with a Bonferroni correction indicated that individuals possessed
a significantly better temporal understanding of climate at post-test after learning from either a
72
polar (M = 1.96, SD = 1.07) or weather themed ER (M = 1.87, SD = 1.11) compared to a
scientific communication frame with no ER (M = 1.46, SD = 1.08).
Changes in Climate Attitude
To explore the influence of thematic frame on climate change attitudes (RQ1b) I
conducted the same 2-Time x 3-Thematic frame repeated-measures mixed ANOVA. Results
indicated an effect of time F(1, 368) = 16.04, p < .001, partial η-squared = .042 such that all
individuals regardless of message theme shifted their attitudes from more unsure (M = 3.68, SD
= 0.82) at pre-test toward being less unsure that climate change is happening (M = 3.76, SD =
0.81) at post-test. However, results indicated no time by thematic frame interaction (p = .754).
Changes in Two Types of Climate Policy Preferences
To assess the extent to which ERs thematic frame influences policy preferences
(governmental policies and tax policies), I conducted two 2-Time x 3-Thematic frame repeated-
measures ANOVAs. Results indicated no effect of time (p = .254), or time by thematic frame
interaction (p = .522) on governmental policies. However, for tax policy preferences results
indicated a significant effect of time F(1, 364) = 7.77, p = .006, partial η-squared = .021
wherein all individuals increased their support for tax policies from pre (M = 3.02, SD = 0.81) to
post (M = 3.07, SD = 0.85). No other effects reached statistical significance (p values > .360).
Impact of Alignment on Knowledge, Attitude, and Policy Preferences
To answer my second research question, To what extent does an external
representation’s cognitive alignment influence knowledge (categorization, definitions), attitudes,
and policy preferences? I conducted eight two-way mixed design ANOVAs, five for knowledge,
one for attitude, and two others to assess impacts on climate policy preferences.

73

Changes in Knowledge of the Distinctions Between Climate and Weather
First, I conducted a 2-Time x 2-Alignment (aligned, misaligned) repeated-measures
mixed ANOVA with time as a within-subjects factor and alignment as a between-subjects factor.
For categorization accuracy, results indicated a main effect of time F(1, 255) = 14.36, p < .001,
partial η-squared = .053 such that everyone increased their correct responses from pre-test (M =
6.37, SD = 1.90) to post-test (M = 6.72, SD = 1.69), but no interaction (p = .226). A similar
pattern emerged for climate definitions, specifically a main effect of time F(1, 255) = 213.46, p
< .001, partial η-squared = .456 such that everyone increased the accuracy of their climate
definition from pre-test (M = 1.91, SD = 1.38) to post-test (M = 3.52, SD = 1.62).
I next investigated the impact of the ERs alignment on three dimensions of climate
definitions specifically the ‘what’, ‘where’ and ‘when’ of climate. Overall, I documented main
effects of time for the ‘where’ (p = .033) and ‘when’ (p < .001) dimensions of climate meaning
individuals increased their understanding of the relationship between time and space in
determining climate and specifying the exact time range. Additionally, no effect of time was
documented for the ‘what’ dimension (p = .322), and no interactions were documented for any of
the climate dimensions.
Changes in Climate Attitude
To address the influence of ERs alignment on attitude I conducted a 2-Time x 2-
Alignment repeated-measures ANOVA. Results indicated a main effect of time F(1, 255) = 9.48,
p = .002, partial η-squared = .036, but no time by alignment interaction (p = .317). Again, all
individuals regardless of ER alignment shifted their attitudes from less sure pre-intervention (M
74
= 3.66, SD = 0.87) toward more sure that climate change is happening post-intervention (M =
3.74, SD = 0.87).
Changes in Two Types of Climate Policy Preferences
Finally, to address the potential impact of alignment on climate policy preferences
(RQ2c), I conducted two repeated-measures ANOVAs. For both policy preference scales results
indicated no effects of time (p values > .107), or time by alignment interaction (p = .323),
although the tax policy preferences interaction approached significance (p = .083).
Impact of Thematic Frame and Alignment on Knowledge, Attitude, and Policy Preferences
To address my third research question, To what extent does an external representation’s
thematic frame and its cognitive alignment impact individuals’ knowledge, attitudes, and policy
preferences? I conducted eight three-way mixed design ANOVAs, five for knowledge, one for
attitude, and two to assess impacts on climate policy preferences.
Changes in Knowledge of the Distinctions Between Climate and Weather
First, I conducted a 2-Time x 2-Thematic frame x 2-Alignment repeated-measures
ANOVA with time as a within-subjects factor. For categorization accuracy (DWCM) results
indicated a significant effect of time F(1, 253) = 14.66, p < .001, partial η-squared = .055 such
that regardless of ER everyone increased their categorization accuracy from pre-test (M = 6.37,
SD = 1.90) to post-test (M = 6.72, SD = 1.69). Finally, results indicated no significant two-way
(p values > .247), or three-way interactions (p values > .270).
I conducted the same analysis using definitions of climate (WCA). A similar pattern to
categorization accuracy emerged, such that regardless of ER all individuals F(1, 253) = 214.64, p
< .001, partial η-squared = .459 significantly increased the quality of their overall definition for
climate from pre-test (M = 1.91, SD = 1.38) to post-test (M = 3.51, SD = 1.62). Next, I analyzed
75
climate definitions along three dimensions, spatial (where, what) and temporal (when). Here,
results indicated a main effect of time (p = .031) for the ‘where’ dimension such that everyone
increased their understanding of the temporal and spatial relationship in defining climate. For the
second spatial dimension of climate ‘what’, results indicated no main effect of time (p = .521),
two-way (p values > .084), or three-way interactions (p = .521).
Finally, I analyzed the temporal dimension of climate ‘when’, to assess the impact of
thematic frame and alignment. Results indicated a significant main effect of time (p < .001), such
that again regardless of ER everyone increased their understanding of the time that determines
climate from pre (M = 0.43, SD = 0.90) to post (M = 1.91, SD = 1.09). However, this main effect
was qualified by a significant three-way interaction between time, thematic frame, and alignment
F(1, 253) = 4.39, p = .037, partial η-squared = .017.
Simple effects tests indicated significant main effects of alignment for the polar ERs at
pre-test F(1, 253) = 6.13, p = .014 and for the weather ERs at post-test F(1, 253) = 5.06, p =
.025, partial η-squared = .026. At pre-test those in the aligned polar ER condition (M = 0.23, SD
= 0.83) had significantly lower scores on the temporal dimension of climate compared to those in
the misaligned polar ER (M = 0.63, SD = 0.97). However, at post-test those who learned from
the misaligned weather ER (M = 2.10, SD = 0.95) had a significantly more detailed
understanding of the temporal dimension of climate compared to the aligned weather ER (M =
1.67, SD = 1.20).
Changes in Climate Attitudes
To address the impact of an ERs characteristics on attitude I conducted a 2-Time x 2-
Thematic frame x 2-Alignment repeated-measures ANOVA. Results indicated a main effect of
time F(1, 253) = 9.32, p = .003, partial η-squared = .036 such that all individuals regardless of
76
ER shifted their attitude from unsure (M = 3.66, SD = 0.87) pre-intervention toward agreeing (M
= 3.74, SD = 0.87) post-intervention that climate change is happening. Additionally, results
indicated no interactions of time with alignment (p = .316), thematic frame (p = .477), or their
combination (p = .717).
Changes in Two Types of Climate Policy Preferences
Next, to assess impacts of an ER on policy preference change (RQ3c), I conducted two
three-way repeated-measures ANOVAs, one on general policy preferences, and another on tax
policy preferences. For general policy preferences, results indicated no effect of time (p = .101),
or two-way interactions (p values > .347), although I documented a significant three-way
interaction F(1, 253) = 3.91, p = .049, partial η-squared = .015. Simple effects tests on the three-
way interaction indicated a significant effect of time for the cognitively misaligned weather ER
F(1, 253) = 6.33, p = .013, partial η-squared = .024. Here, only individuals who saw the
misaligned weather ER significantly increased their support for climate policies from pre-
intervention (M = 3.03, SD = 0.59) to post-intervention (M = 3.13, SD = 0.65) (see Figure 3).
For tax policy preferences, a different pattern emerged. Results indicated non-significant
effects of time (p = .096), time by alignment (p = .091), and time by thematic frame (p = .680).
However, similar to general policy preferences, results indicated a significant three-way
interaction between time, thematic frame, and alignment F(1, 250) = 5.33, p = .022, partial η-
squared = .021. Simple effects tests (see Figure 4) indicated a significant effect of time but only
for the cognitively misaligned weather ER F(1, 250) = 6.67, p = .010, partial η-squared = .026,
such that individuals who saw the misaligned weather ER significantly increased their support
for climate tax policies from pre (M = 3.04, SD = 0.77) to post (M = 3.16, SD = 0.81). No other
comparisons reached statistical significance (all p values > .177).
77
Figure 3
Interaction of Thematic Frame, Alignment, and Time on Climate Policy Preferences

Figure 4
Interaction of Thematic Frame, Alignment, and Time on Climate Tax Policy Preferences

1
2
3
4
5
Aligned polar Misaligned polar Aligned weather Misaligned weather
Self-reported climate policy support
Interaction of Thematic Frame, Alignment, and Time on General
Climate Policy Preferences
Pre-intervention Post-intervention
1
2
3
4
5
Aligned polar Misaligned polar Aligned weather Misaligned weather
Self-reported tax climate policy support
Interaction of Thematic Frame, Alignment, and Time on Tax
Climate Policy Preferences
Pre-intervention Post-intervention
p = .010
p = .013
78
Impact of Thematic Frame and Alignment on Self-reported Emotions
To address my fourth research question, Do an external representation’s thematic frame
and alignment impact the intensity of self-reported emotions, the number of emotion objects
selected for a given emotion, the number of emotion objects identified across emotions, and
cognitive construals along affective and concrete dimensions? I conducted a series of three-way
mixed design ANOVAs, to assess impacts on the indices of emotion (see Table 14 for means and
standard deviations of all emotion variables). Here, I leveraged the reading time data to create
three groups that corresponded to time spent reading the message. I am interested in epistemic
emotions, or those that occur when thinking about information, however, since many individuals
did not read the content for more than a few seconds, I explored emotional reactions relative to
time spent reading.
Table 14
Descriptive Statistics for all Emotion Variables Overall and by Condition

Full
Sample
(n = 257)
Aligned
Polar ER
(n = 64)
Misaligned
Polar ER
(n = 60)
Aligned
Weather ER
(n = 72)
Misaligned
Weather ER
(n = 61)
Mean SD Mean SD Mean SD Mean SD Mean SD
Anger 2.36 1.41 2.30 1.41 2.30 1.47 2.32 1.38 2.54 1.39
Sources 1.41 1.81 1.16 1.56 1.45 1.93 1.49 1.96 1.54 1.76
Anxious 2.37 1.32 2.50 1.18 2.25 1.43 2.22 1.26 2.51 1.41
Sources 1.82 1.95 2.20 2.03 1.55 1.88 1.61 1.92 1.95 1.94
Curious 2.84 1.29 2.77 1.11 2.88 1.37 2.92 1.37 2.77 1.31
Sources 1.58 1.48 1.78 1.57 1.52 1.62 1.56 1.39 1.46 1.35
Enjoyment 1.73 1.09 1.61 1.02 1.82 1.17 1.81 1.15 1.69 1.04
Sources 0.60 1.07 0.61 1.18 0.60 1.03 0.63 1.05 0.57 1.04
Frustrated 2.49 1.41 2.41 1.32 2.67 1.56 2.46 1.43 2.46 1.32
Sources 1.45 1.72 1.42 1.66 1.52 1.87 1.39 1.74 1.48 1.66
Hopeful 2.32 1.20 2.27 1.04 2.43 1.20 2.29 1.31 2.28 1.25
Sources 1.21 1.40 1.20 1.25 1.57 1.72 1.07 1.39 1.02 1.15
79

Self-reported Intensity of Epistemic and Epistemically Related Emotions
To assess the influences of thematic frame and alignment on self-reported emotional
intensity I conducted a 2-Thematic frame x 2-Alignment x 3-Reading group MANOVA. Here, I
created a grouping variable that allowed me to analyze emotional reactions not only as a function
of the thematic frame and its alignment but also relative to message exposure. The groups
corresponded to reading for less than 33 seconds, reading between 33 and 124 seconds, and more
than 124 seconds.
Results indicated a significant multivariate effect of reading group Pillai’s Trace F (22,
472) = 1.89, p = .009, and a thematic frame by alignment by reading group interaction F (22,
Hopeless 2.08 1.25 2.14 1.30 2.05 1.35 1.97 1.20 2.16 1.20
Sources 1.31 1.78 1.50 1.89 1.20 1.78 1.07 1.64 1.51 1.82
Interest 3.25 1.23 3.17 1.27 3.38 1.28 3.25 1.18 3.20 1.21
Sources 2.10 1.78 1.94 1.70 2.33 1.87 2.21 1.81 1.92 1.76
Surprise 1.69 1.03 1.64 0.86 1.63 1.04 1.79 1.19 1.67 0.98
Sources 0.67 1.06 0.83 1.28 0.65 1.18 0.56 0.82 0.64 0.93
Bored 1.53 1.01 1.58 1.05 1.53 0.99 1.49 0.98 1.52 1.03
Sources 0.42 0.89 0.41 0.79 0.47 1.00 0.42 0.98 0.41 0.80
Confused 1.52 0.94 1.50 0.89 1.55 1.03 1.56 0.93 1.46 0.94
Sources 0.42 0.82 0.45 0.83 0.47 1.02 0.44 0.77 0.33 0.63
Emotion Objects
Text 3.08 1.71 3.14 1.66 2.76 1.75 3.25 1.76 3.14 1.67
People 2.75 1.71 3.00 1.77 3.17 1.99 2.43 1.42 2.55 1.68
Animals 2.86 1.73 3.19 1.78 3.23 1.91 2.60 1.47 2.52 1.72
Nature 2.83 1.68 2.92 1.53 2.82 1.99 2.87 1.55 2.70 1.70
Politicians 2.66 1.51 2.97 1.66 3.00 1.65 2.35 1.50 2.30 1.09
CC is
happening
3.13 1.81 3.20 1.82 3.14 2.20 2.84 1.42 3.42 1.84

Construal
Affect 0.91 1.38
Concrete 0.49 0.50
80
472) = 1.88, p = .009. These significant multivariate effects were followed up with a series of
one-way ANOVAs to assess their impact on each emotion variable. For the reading groups,
results indicated significantly more intense anger (M = 2.67, SD = 1.50) and confusion (M =
1.87, SD = 1.23) among those who read for less than 33 seconds compared to either reading for
between 33 and 124 seconds (M = 2.20, SD = 1.39; M = 1.33, SD = 0.77) or more than 124
seconds (M = 2.24, SD = 1.30; M = 1.37, SD = 0.68). Additionally, those who read for less than
33 seconds indicated significantly (p = .012) less interest (M = 2.93, SD = 1.31) compared to
those who read for more than 2 minutes (M = 3.47, SD = 1.21) and marginally less interest than
those who read for between 33 seconds and 2 minutes (M = 3.33, SD = 1.11), although not
statistically significant (p = .09).
The majority of the epistemic and epistemically related emotions (i.e., hopeful, hopeless,
bored, confused, curious, enjoyment, frustration) were not influenced by the ERs thematic frame,
cognitive alignment or their interaction.
For anxiety however, results indicated no effects of alignment or theme (p values > .730),
but a significant two-way interaction between reading group and alignment F(2, 245) = 4.42, p =
.013 that was qualified by a significant three-way interaction F(2, 245) = 3.12, p = .046, partial
η-squared = .025. Simple effects tests revealed a significant effect of reading group for the
misaligned polar ERs F(2, 245) = 6.52 p = .002, η-squared = .051 with self-reported anxiety
being significantly more intense among those who read the ER for less than 33 seconds (M =
3.17, SD = 0.30) compared to those who read for between 33 and 124 seconds (M = 1.91, SD =
0.28) or more than 124 seconds (M = 1.80, SD = 0.28). Simple effects tests also indicated
significant effects of alignment for the polar themed ERs within the reading groups less than 33
seconds F(1, 245) = 4.51 p = .035, partial η-squared = .018 and between 33- and 124-seconds
81
F(2, 245) = 5.57 p = .019, partial η-squared = .022. Specifically, those who viewed the
misaligned polar ER (M = 3.17, SD = 1.43) compared to the aligned polar ER (M = 2.30, SD =
1.15) for less than 33 seconds reported significantly more intense (p = .035) anxiety.
Additionally, those who viewed the aligned polar ER (M = 3.85, SD = 1.42) for between 33 and
124 seconds reported greater anxiety than those who viewed the aligned weather ER (M = 1.96,
SD = 1.15) for the same amount of time.
For interest, results indicated a non-significant reading group by alignment interaction
F(2, 245) = 2.77, p = .065. While not significant, simple effects tests indicated that when read for
less than 33 seconds interest was more intense from a brief exposure to a misaligned ER (M =
3.29, SD = 1.24) compared to an aligned ER (M = 3.21, SD = 1.22). Additionally, interest in the
aligned ERs was more intense when read for either between 33 and 124 seconds, or greater than
124 seconds compared to less than 33 seconds.
Finally, I repeated the previous analysis focusing on self-reported surprise. Results
indicated a significant three-way interaction F(2, 245) = 5.42, p = .005, partial η-squared = .042.
Simple effects tests indicated several relevant effects. First, I documented a significant effect of
thematic frame on aligned ERs read between 33 seconds and 124 seconds F(1, 245) = 6.67, p =
.01. Here, individuals were more surprised when viewing the aligned weather ER (M = 2.00, SD
= 1.45) compared to the aligned polar ER (M = 1.20, SD = 0.41). Following this pattern, I
documented an effect of reading groups for the misaligned weather ER F(2, 245) = 3.94, p =
.021. Specifically, those who read for 33 seconds or less indicated more intense surprise (M =
2.10, SD = 1.17) than those who read for between 33 and 124 seconds (M = 1.23, SD= 0.61).
Finally, simple effects tests indicated an effect of cognitive alignment among readers who
viewed a weather framed ER for between 33- and 124-seconds F(1, 245) = 6.54, p = .011.
82
Specifically, those who viewed the aligned weather ER indicated significantly more surprise (M
= 2.0, SD = 1.45) compared to the misaligned weather ER (M = 1.23, SD = .61).
Number of Emotion Objects Identified for a Given Emotion
Of interest was not only the intensity of self-reported emotions, but to identify their
object focus (RQ4b). I conducted a series of two-way ANOVAs to assess if thematic frame and
alignment influenced the number of emotion objects (6 possible) selected when indicating one’s
emotion. For all emotions other than frustration and anxiety (i.e., anger, bored, curious,
confused, enjoyment, hopeless, interest, hopeful, and surprise) results indicated no effects of
thematic frame, alignment, or their interaction. However, for frustration, results indicated no
effect of thematic frame or a theme by alignment interaction but a main effect of alignment F(1,
253) = 480.58, p = .029. Those who viewed a misaligned ER identified more emotion objects as
frustrating (M = 1.50, SD = 1.76) compared to an aligned ER (M = 1.40, SD = 1.70). I next
analyzed the sources of emotion objects for frustrated to see if the considerations activated by the
experimental conditions differed. Results indicated no effect of alignment for any of the emotion
objects (i.e., ideas in the text, people, animals, nature impacted by climate change, the fact that
climate change is happening, or politicians) indicating that overall learning from a misaligned
ER activated more considerations than an aligned ER.
For anxiety I documented no main effects of alignment or thematic frame, but their
interaction F(1, 253) = 4.17, p = .042. Simple effects tests revealed an insightful but non-
significant effect of thematic frame for the aligned ERs (p = .077) such that descriptively more
anxious emotion objects were selected when viewing the aligned polar compared to weather ER.
An additional non-significant (p = .063) effect of alignment was documented for the polar
themed ERs where more objects were selected after viewing the aligned compared to misaligned
83
polar ER. I analyzed the specific emotion objects for anxiety documenting no impact on the
frequencies of ideas in the text, animals impacted, nature impacted, and the fact that climate
change is happening. However, results indicated an interaction between thematic frame and
alignment on ‘people impacted by climate change’ F(1, 253) = 1.20, p = .008. Simple effects
tests indicated a main effect of theme for the aligned ERs F(1, 253) = 5.52, p = .020 such that the
aligned polar ER resulted in more focus on the people impacted by climate change (M = 0.47,
SD = 0.50) compared to the aligned weather ER (M = 0.28, SD = 0.45). Simple effects tests also
indicated a main effect of alignment for the polar ERs, F(1, 253) = 5.65, p = .018 where ‘people
impacted by climate change’ was selected more for the aligned polar ER (M = 0.47, SD = 0.50)
compared to the misaligned polar ER (M = 0.27, SD = 0.45). Finally, for the emotion object
‘politicians’ results indicated a main effect of thematic frame F(1, 253) = 427.56, p = .031 where
those who saw a polar themed (M = 0.16, SD = 0 .37) ER identified politicians significantly
more than those who viewed a weather ERs (M = 0.11, SD = 0.32).
Number of Emotion Objects Identified Across Emotions
In addition to intensity, and number of emotion objects associated with each individual
emotion, research question 4c addresses how an ERs thematic frame and alignment impact the
number of times the same emotion objects (i.e., ‘ideas in the text’ or ‘politicians’) were identified
across individual emotions. I conducted six (one for each emotion object) 2-Alignment (aligned,
misaligned) x 2-Thematic frame (polar, weather) ANOVAs. Three of the emotion objects, 1)
ideas in the text, 2) fact that climate change is happening, and 3) nature impacted by climate
change were not influenced by alignment (p values > .312), thematic frame (p values > .296), or
their interaction (p values > .223). However, the other three emotion objects 1) people impacted
by climate change, 2) animals impacted by climate change, and 3) politicians were all influenced
84
by thematic frame. Individuals identified ‘people impacted by climate change’ significantly more
F(1, 169) = 5.24, p = .023, partial η-squared = .030, when viewing a polar (M = 3.08, SD =
1.86) compared to a weather themed message (M = 2.48, SD = 1.54). This pattern repeated for
‘animals impacted by climate change’ with results indicating that ‘animals impacted’ was
selected significantly more F(1, 150) = 5.45, p = .021 after viewing a polar (M = 3.21, SD =
1.83) compared to a weather themed message (M = 2.56, SD = 1.59). The final emotion object,
‘politicians’, followed the same pattern of results F(1, 117) = 5.91, p = .017 where politicians
were identified more after viewing a polar (M = 2.98, SD = 1.65) compared to a weather themed
message (M =2.32, SD = 1.29).
Cognitive Construal of the Task
To address the final aspect of the fourth research question, Do an external
representation’s thematic frame and alignment impact the cognitive construal of the learning
event along dimensions of affect (positive, negative, mixed) and explicit connection with the
material (concrete construal). For affect, a preliminary analysis was conducted on the coded data,
and indicated a non-significant effect of thematic frame (p = .075) where the weather frame
resulted in descriptively more negative construals of the situation with no effect of alignment (p
= .161) or their interaction (p = .823).
I conducted a second 2-Thematic Frame (polar, weather) x 2-Alignment (aligned,
misaligned) ANOVA to understand their effect on an individual’s concrete construal, here
defined, as making an explicit reference or connection to the climate change communication.
Results indicated no effect of alignment or theme (p values > .576), and a non-significant
interaction (p = .064), although simple effects tests indicated an effect of thematic frame for the
misaligned ERs (p = .021) where concrete construals (i.e., construals that explicitly referred to
85
the text) occurred descriptively more in the misaligned polar compared to misaligned weather
ER, with no other effects approaching significance.

86
Chapter 5

Discussion

Prior research indicates the importance and ubiquity of framed
5
climate communication
messages specifically their use of external representations (ERs) and associated themes, (Rebich-
(Rebich-Hespanha et al., 2015) relevance, and affect (Lehman et al., 2019; Leiserowitz, 2006).
These message characteristics have documented influences on knowledge (Danielson et al.,
2016; Mason et al., 2017; Nussbaum et al., 2017), attitude (Isberner et al., 2013; Ranney &
Clark, 2016), and policy preferences (Feldman & Hart, 2018; Lorenzoni et al., 2006;
Leiserowitz, 2006). However, debate exists regarding the role of accuracy (i.e., alignment) in
depicting climate and how this alignment influences affect.
Carney and Levin (2002) state ERs can be decorational, organizational, or
transformational, but this only considers the cognitive dimension of ERs, specifically relevance
and alignment. However, relevance is a ‘cold’ variable, whereas Schneider et al. (2018)
highlighted the importance of an ERs cognitive relevance and affective charge (i.e., affective
relevance). This distinction is necessary to discuss the recommendation of Lewandowsky and
Whitmarsh (2018) to seek accurate representations and legitimate affective triggers, as study
results question both of these recommendations, although for different reasons.
Here, I invoked a framework of emotional thought (Immordino-Yang & Damasio, 2007) to
better understand how a climate communication paired with an ER depicting different impacts of
climate change (polar, weather) influenced knowledge, attitude, emotions, and preferences for
how to address climate change. Given my interest in change, I selected the CRKM to account for

5
Chong, (1993) “The concept of framing assumes that some representations of an issue are more persuasive than
others, so that attitudes and opinions will be swayed in predictable ways if attention can be concentrated on those
representations.” (p. 151).
87
the role of learner and message characteristics, and the ITPC to provide a cognitive explanation
for how an ERs thematic frame, and alignment in depicting climate change influenced
understanding of climate change and choices to mitigate it. This study attempted to
systematically manipulate two climate communication thematic frames, polar impacts and
weather impacts, varied on cognitive alignment to the concept climate and to assess their impact
on climate knowledge, attitudes, policy preferences, and self-reported emotions. In this chapter, I
discuss the study results and their relevance to the Cognitive Reconstruction of Knowledge
Model (CRKM) and the Integrative Theory of Text and Picture Comprehension (ITPC).
Specifically, I review the four central research questions before discussing study limitations,
implications, and future directions.
Thematic Frames Matter
My first research question explored how a climate message’s thematic frame influenced
knowledge, attitude, and policy preferences for the topic climate change. Here, results indicated
that an educated group of predominately white US men and women across the political spectrum
can improve their understanding of climate change and ability to define its key elements. Here,
even brief exposure to informational materials resulted in significant increases in their ability to
categorize statements as weather or climate, ability to define climate overall, attitude toward
climate change, and support for policies to mitigate climate change.
However, the ERs thematic frame did differentially impact specific dimensions of ones’
climate definition. Analyses revealed that viewing a polar or weather themed ER, namely any
message with an external representation, resulted in a significantly better understanding of the
temporal ‘when’ dimension of climate compared to a science communication frame. There was
also a main effect of time for the spatial ‘where’ dimension of climate, meaning that all
88
individuals increased their awareness of the relationship between time and space. This main
effect contextualizes the non-significant effect of time for the temporal/spatial ‘what’ dimension
of defining climate. Here, viewing any message with an ER facilitated understanding of this
difficult concept in line with the recommendations of Lombardi and Sinatra (2010), Nussbaum et
al. (2017), and Lewandowsky and Whitmarsh (2018).
This general pattern of results (i.e., increased accurate knowledge and increased positive
attitude) provides support for Sinatra and Seyranian’s (2016) knowledge-attitude link model
observed by Heddy et al. (2016), Jacobson et al. (2021) and Thacker et al. (2020) and tentative
support for knowledges’ impact on climate change behaviors, specifically policy preferences
(van Valkengoed & Steg, 2019). It also parallels findings by Ranney and Clark (2016) who
documented that increased scientific knowledge of climate change was associated with what they
termed justified climate acceptance and provides disconfirming evidence for Kahan (2013b)
stasis theory, which argues that scientific information cannot be used to resolve climate change
conflicts.
This finding provides strong support to models of text and graphic comprehension, and
the multimedia principle (Mayer, 2005). Here, a refutation text with any type of ER increased the
conceptual change process, compared to the expository infographic or a refutation text alone.
Importantly, all individuals increased their understanding of climate change and their attitude
toward it, but only when the concept is decomposed could I determine how the ERs facilitated
conceptual change, specifically, by depicting the concept in a way that allowed for meaning to be
made beyond that of the science communication messages.

89
Cognitive Alignment May Not Matter Alone
My second research question sought to investigate the role of accuracy or inaccuracy in
depicting the concept climate, and how this alignment influenced conceptual knowledge,
attitude, and policy preferences. Results indicated that alignment, or representational accuracy
was not a driving force alone in communicating the distinctions between weather and climate.
Here, I documented that regardless of the messages’ accuracy in representing the concept
climate, all individuals increased their categorization accuracy, ability to define climate, and
attitude toward climate change, but not policy preferences.
Again, this pattern of increased knowledge and positive attitude supports Sinatra and
Seyranian’s (2016) specification of knowledge-attitude links and the continued role of informing
the general public about climate change. It also provides support for the belief that scientific
information has a place in public discourse, and that persuasion is not the only approach forward.
Instead, as Ranney and Clark (2016) documented, even brief interventions to facilitate
conceptual knowledge of climate can influence attitude and policy preferences.
The findings for cognitive alignment deviated from my hypotheses as I based them on the
ITPCs focus on analogical structure mapping and literal similarity. While analogical structure
mapping can account for some of the benefits of external representations, study results suggest a
broader lens is needed to identify the full range of relationships present between a text and ER. A
broader framework, one that includes analogy is Relational Reasoning (Dumas et al., 2013) and
specifies four relationships (analogy, anomaly, antinomy, and antithesis). Danielson and Sinatra
(2017) bridge text and graphic comprehension to relational reasoning and provide succinct
definitions of the relations pertinent for misaligned ERs, “…anomalies are unexpected deviations
from known patterns, antinomies present non-examples or paradoxes, and antitheses present
90
opposing arguments” (p. 55). Mason et al. (2017) provided support for this view documenting
equally beneficial learning outcomes using either a standard (i.e., analogical relation) or
refutational (i.e., antithetical) ER as both facilitated conceptual change. Their study also
illuminates the role of prior knowledge in detecting conflict from a graphic. Here, when students
were not given instructions on how to detect conflicts they detected more when viewing the
aligned graphic (15%) compared to the misaligned graphic (5.2%), however after receiving
instruction conflict detection was higher for the misaligned graphic (17.4%) compared to the
aligned graphic (13%). This finding highlights the importance of prior knowledge not only for
the content, but of strategies to learn from graphics, diagrams and other external representations.
Prior content knowledge determines what relationships are seen, specifically knowledge of
structural similarities between sources of information (Dumas et al., 2013; Kendeou et al., 2017).
Kendeou et al. (2017) argue that relational reasoning relies on the comparison of information to
identify what is similar or dissimilar, and that this only occurs if the individual has sufficient
prior knowledge (Gentner & Holyoak, 1997; Holyoak & Thagard, 1997). Sufficient is a loaded
term as it ignores inaccurate knowledge, beliefs, or misconceptions as the basis for identification
of similarities and dissimilarities. This issue is present in the text and graphic comprehension
literature and is termed an expertise reversal effect (Kalyuga & Renkl, 2010; Schnotz, 2010) for
it describes benefits of ERs for low knowledge individuals but detriments for high knowledge
individuals. Here, individuals were knowledgeable about some aspects of climate and weather
but demonstrated both improvement of their current understanding and acquisition of key
distinctions.
Above, I offered a theoretical explanation to account for the benefit of misaligned external
representations on climate change understanding. I also highlighted the role of prior knowledge
91
in relational reasoning, and now connect it to a measurement issue. First, the scoring rubric
published by Nussbaum et al. (2017) only captures accurate scientific understanding, and treats
inaccurate responses (i.e., misconceptions) the same as no response. This methodological tension
relates to scientific communicators often failing to account for the starting point of the general
publics’ knowledge of climate change. Climate change is a socio-scientific issue (Sadler, 2004)
but climate research often under attends to the context in which the public form their views of
the issue. While all individuals increased their accurate understanding, it is likely that synthetic
conceptions (i.e., attempts to assimilate prior beliefs with new scientific explanations) emerged
(Vosniadou, 2009; Vosniadou & Mason, 2012) and would benefit from additional analyses to
better understand the roles of attitude and political identity in the conceptual change process.
Thematic Frame and Cognitive Alignment Interact But Not In the Way I Predicted
My third research question explored the interactive effects of thematic frame and
cognitive alignment on conceptual understanding, attitude, and policy preferences. Here, I
documented an overall increase in categorization accuracy and definition quality (3a), and
attitude (3c) consistent with Sinatra and Seyranian’s (2016) and Ranney and Clark’s (2016) work
on knowledge attitude links for socio-scientific issues (see Table 15). Results support the view
that the general public can be informed about climate change, and that communicators should not
be overly concerned about inducing backfire effects (Jacobson et al., 2021; Nyhan & Reifler,
2010) or inducing reactance with climate messages (van der Linden et al., 2021).

92
Table 15
Hypothesized and Obtained Results for Research Questions 3(a), 3(b), and 3(c)

(3a) Conceptual Change (3b) Attitude Change
(3c) Policy Preference
Change
Alignment
Hypothesis Result Hypothesis Result Hypothesis Result
Align
>
Misalign
Misaligned
Weather
>
Aligned
Weather

*Temporal
(when)
Align
>
Misalign
n.s.
Align
>
Misalign
Misaligned
Weather ER

General: pre
< post
Tax: pre <
post

The results obtained across the first three research questions provide support for the
importance of studying (Leiserowitz, 2006; Lewandowsky & Whitmarsh, 2018; Rebich-
Hesphana et al., 2015) and employing (Nussbaum et al., 2017; Lombardi & Sinatra, 2012)
external representations in climate change communication. Results regarding research question 1
indicated it may be advisable to include a weather or polar themed ER regardless of alignment,
as conceptual change was greatest among those who viewed a message with any thematic
external representation. Key here is that I was attempting to inform individuals of factual
information, not to persuade their attitudes and emotions. I had strong evidence to predict that
increased accurate knowledge would accord with increased positive attitude (Sinatra &
Seyranian, 2016; Clark & Ranney, 2016) and that changes in knowledge may shift policy
preferences. Results of research question 2 provided insight into the possible functions served by
an ER, specifically that individuals here increased their understanding of climate equally well
when learning from a representation aligned with climate (i.e., what it is) or one misaligned with
climate (i.e., what it is not). However, results regarding research question 3 inform us that only
by considering the interaction between a thematic frame and its cognitive alignment with the
93
concept, can we see differences in what the ERs differentially communicate. Here, results
indicated that the temporal understanding of climate was best supported by the misaligned
weather ER (i.e., extreme weather impact), and that it was significantly different from the
aligned weather ER. Thematic frame and alignment also interacted to influence policy
preferences such that only those who viewed the misaligned weather ER increased their support
for mitigation strategies from pre-test to post-test. This link between increased conceptual
knowledge and policy preferences is supported by Ranney and Clark’s (2016) finding that brief
interventions on knowledge can impact policy preferences, and that knowledge is related to
policy preferences (van Valkengoed & Steg, 2019).
Research question 3 provided partial support for my hypotheses, specifically that the
misaligned weather ER would produce policy preference shifts compared to the misaligned polar
ER, however, the significant difference was for the misaligned weather ER over time, as it was
the only message that induced policy preference changes. I predicted this effect for affective
reasons (i.e., personal risk perception, negative affect, experience with natural disasters, and
place attachment) however, the misaligned weather ER best facilitated conceptual change for
understanding the temporal aspect of climate. Instead the misaligned weather ER may function
similar to the mechanism of justified climate acceptance described by Ranney and Clark (2016).
Finally, the influence of thematic frame from RQ1 and the null findings of alignment from RQ2
challenge my view on the roles of alignment and misalignment in climate communication
(Lewandowsky & Whitmarsh, 2018).
What I’ve termed misaligned ERs mean inaccurate representations of the scientific concept
climate along spatial and temporal dimensions. Here, I documented no main effects of alignment,
which indicate that learning occurred regardless of alignment, and requires an alternative
94
explanation. The ITPC draws on analogical structure mapping, which is only one of four types of
relational reasoning, along with antinomy, anomaly, and antithesis (Dumas et al., 2013).
Danielson and Sinatra (2017) wrote a theoretical integration piece combining relational
reasoning with theories of text and graphic comprehension and have conducted primary studies
assessing the role of refutational graphics (i.e., misaligned ERs) (Mason et al., 2017, Danielson
et al., 2016). These studies, while nascent are advancing theoretical considerations of text and
graphic comprehension and require consideration of new variables (emotions) and methods
(think aloud studies).
Another possible explanation for the benefit of the misaligned weather ER relates to the
connectedness (i.e., semantic relations), and affective charge (i.e., positive or negative
association) between the text and ER (Schneider et al., 2018). All ERs used in this study were
normed for relevance, arousal, and valence (Lehman et al., 2019) and provided evidence for
connectedness and affective charge. However, relevance ratings in the norming study related to
climate change, whereas I assessed connectedness to the concept climate. Additionally, it is not
only the connection between the ER and the concept, but also between the ER and the text.
These ERs were not normed relative to their connectedness with the refutational text, and given
they are an uncommon format, connectedness may provide a possible explanation. For instance,
the misaligned weather ER could be perceived as relevant to climate change, but at an
experiential compared to conceptual level. However, when this same image was paired with a
text designed to restructure ones’ understanding of climate, its connectedness may either increase
or decrease depending on the characteristics and engagement of the individual. Refutation texts
function by alerting the individual to conflict between their current understanding and the to-be-
learned information. It is possible then that a similar type of learning could occur for a
95
misaligned representation of content. If detected, a conflict between ones’ prior knowledge, the
text, and/or the ER would influence later processing. This however requires the learner to detect
and then resolve conflicts between and among these representations.
The second aspect, affective charge, may also provide insight into the beneficial effect of
the misaligned weather ER. Affect, specifically negative affect (i.e., anger, fear, anxiety, and
hope) are known to influence climate adaptation (van Volkengoed & Steg, 2019) and policy
preferences (Born, 2019; Lehman et al., 2019; Leviston et al., 2014). However, positive affect
(i.e., surprise, hope, interest) facilitate conceptual change (Broughton et al., 2011; Muis et al.,
2015). Data revealed that those who viewed the messages for less than 33 seconds reported more
intense anger and confusion and less intense interest. The most illustrative example of the
interaction between connectedness and affective charge relates to self-reported surprise. Here,
individuals who read for between 33 and 124 seconds indicated more surprise when viewing the
aligned weather ER compared to the misaligned weather ER. Again, while speculative, this
finding illuminates the role of surprise in the conceptual change process. I defined the alignment
of the ERs a priori based on the correct conceptualization of climate, however, given the ubiquity
of this misconception it is quite possible that learners experienced the ERs reversed, where the
‘misaligned’ ER aligned with their current understanding and these sources conflicted with the
text. This alternative explanation runs counter to Lewandowsky and Whitmarsh’s (2018) focus
on avoiding misrepresentation and locating legitimate affective triggers for climate
communication; for this study documented benefits of misrepresentation when communicating
the accurate science about climate change.

96
Thematic Frame and Cognitive Alignment Influence Some Aspects of Emotion
My fourth research question sought to understand the influence of an ERs thematic frame
and its alignment on indices of emotion. I documented interactive effects of thematic frame,
alignment, and reading time on self-reported anxiety and surprise (4a). Many of the effects
related to differences in emotional intensity based on reading time. For instance, among those
who viewed the polar ERs for less than 33 seconds, anxiety was significantly greater when
viewing the polar bears (misaligned) compared to the sea ice (aligned) ER. Also, when viewed
for between 33 and 124 seconds, the aligned polar ER resulted in more intense anxiety compared
to the aligned weather ER. Research question 4b, examined the sum and type of emotion objects
selected for a given emotion, but did not yield significant results. However, 4c shed light on the
emotion objects selected as sources of emotions. Results indicated that the emotion objects
‘politicians’, ‘people impacted by climate change’ and ‘animals impacted by climate change’
were identified significantly more after viewing a polar compared to weather themed ER. This
finding provides evidence that even when a text is the exact same, the external representation’s
thematic frame directs attention, focus, and emotions either toward or away from key
considerations. A similar pattern emerged for self-reported surprise except individuals were more
surprised when viewing the aligned weather ER compared to the aligned polar ER. These
findings typify the impact of framing on socio-scientific issues, specifically that even when the
ERs are communicating the same accurate temporal information, the thematic content
differentially influenced the intensity of two emotions, one epistemic (surprise), and one topic or
social (anxiety). Additionally, results provide evidence of what considerations are brought to
attention via framing manipulations and provide a powerful tool for crafting personalized
messages that employ frames in line with individual preferences. Finally, regarding construals of
97
the task results indicated a non-significant effect of thematic frame on affect (p =.075) where
construals were more negative after seeing a weather message, and a non-significant interaction
on concreteness (p =.075) where construals connected to the text more after viewing a
misaligned polar compared to misaligned weather ER.
Overall, the findings about emotional intensity provide evidence that external
representations can influence emotions and object focus, however, predictions deviated from
results. Here, I aimed to assess the influence of ERs on emotions, specifically epistemic
emotions or those experienced when learning or resolving conflict. Surprise, anxiety, and interest
are strongly associated with conceptual change however, here individuals had relatively high
prior knowledge that may have depressed surprise and interest ratings, although surprise and
interest were associated with reading times. Surprise was also driven by reading time and
thematic frame but only for the aligned ERs indicating that thematic frame may be more
important than alignment.
Another possible explanation for the null findings around emotions relates to the affective
connection between the content and representation. Carney and Levin (2002) describe
decorational, and transformational ERs among others but either of these ERs may increase
engagement or situational interest (Schiefele, 2009; Thomas & Kirby, 2020) that enables
students to engage in relational reasoning and recognize the ER serving as anomaly, antinomy, or
antithesis (Danielson et al., 2016; Mason et al., 2017).
In this study, results indicated that even brief exposure to a climate communication
message can increase accurate knowledge, ability to define climate, and attitude, but not
typically policy preferences. These findings are positive given the sample of individuals spanned
the political spectrum and included overt climate deniers who read the message for anywhere
98
between one-second and several minutes. This study also provided evidence for the importance
of external representations and thematic frames in climate communication (RQ1), and the need
to reconsider the kinds of relationships that can be drawn between a text and an ER (RQ2).
However, when the interaction between thematic frame and alignment is considered the
affordances of ERs became clear. Here, the misaligned weather ER best facilitated awareness of
the temporal dimension of climate change and resulted in significantly greater support for tax
and government climate policies. Also, while the thematic weather frame increased conceptual
knowledge, the polar theme resulted in selection of different emotion objects compared to the
weather theme and provides a fruitful avenue for future research on conceptual, attitude, and
policy preference change.
Limitations
Like all studies, this current investigation has limitations. First, this study investigated a
relatively small (N = 371) homogeneous (i.e., 76% white) sample of older (Mean age = 41)
educated (41% had a bachelor’s degree) individuals. These sample characteristics decrease the
generalizability of results as they related primarily to educated, older, white adults, although this
too is an important portion of the public (Leiserowitz et al., 2021). Additionally, the sample also
limited the role of epistemic emotions in the study, specifically surprise, as the they had
generally accurate knowledge. A final sample-based limitation is the number of participants
excluded from the final sample. Supplementary analyses provided a rationale for these cleaning
decisions; however, it must be acknowledged these choices influenced the sample characteristics.
Second, several limitations are present in my experimental materials. Here, I only
manipulated two thematic frames while many others are present (Rebich-Hesphana et al., 2015)
and need to be investigated (Feldman & Hart, 2018). Additionally, I only used two ERs for each
99
thematic frame (aligned, misaligned) although more ERs could be selected to represent a specific
theme. Finally, the experimental text was approximately 700 words, and participants spent less
time reading than predicted, indicating the intervention may have lacked fidelity, however briefer
interventions have produced conceptual change (Ranney & Clark, 2016).
Finally, this study was limited regarding measurement. The absence of a delayed post-test
limits the interpretation of the learning, attitude, and policy preference findings as some concepts
take repeated instruction (Lombardi et al., 2013; Vosniadou, 2008) while others can be impacted
briefly and persist (Munnich & Ranney, 2019). Also, the reliability of the categorization task was
low (< .6), although it mirrors findings that forced response items may not be sensitive to
conceptual change for this topic (Nussbaum et al., 2017). This study also relied on self-report
data which belies the richness of the in vivo experience. Inclusion of in-person methods and
think-aloud procedures may illuminate the specific roles of the theme and alignment variables
(Kendeou et al., 2019).
Support and Implications for CRKM and ITPC
The CRKMs focus on learner characteristics was supported by the study results,
specifically the relationship of accurate science knowledge and increased positive attitude toward
the science. This provides additional support for the inclusion of knowledge-attitude links in
socio-scientific communication, and the potential to craft specific communications based on their
specific knowledge/attitude link (i.e., high-pro, low-pro, high-con, and low-con). The CRKMs
attention to message characteristics received partial support, given the model does not identify a
cognitive process to explain the utility of an ER on conceptual change, but does acknowledge an
ERs impact on attitude. I continue my discussion of the implications for the CRKMs
100
consideration of message characteristics below, as my suggestion is the incorporation of the
ITPC.
The ITPC specifies two processes of text and graphic comprehension, analogical
structure-mapping and thematic selection. However, results suggest that analogical structure-
mapping is only one of four ways to reason relationally. Danielson and Sinatra (2017) wrote a
review piece arguing for the importance of three other types of relationships specifically for text
and graphic comprehension. First, anomaly which are events, data, or occurrences that are
unusual or do not fit expectations (Chinn & Brewer, 1993) (e.g., Michael Mann’s hockey stick
graph paired with a text). Second, antinomy “allows the thinker to understand what something is
by ascertaining what it is not” (Dumas et al., 2013, p. 395). An example of antimony being a
dual y-axis line graph (with different scales) created by Americans United For Life juxtaposing
two unrelated concepts like cancer screenings and abortions to misrepresent a relationship
(Danielson & Sinatra, 2017). Third, antithesis describes situations when two representations are
in oppositional relation to one another (Kreezer & Dallenbach, 1929), an example being a side-
by-side comparison of two diagrams representing models of seasonal change that that explicitly
identify the accurate and inaccurate understanding. I argue results from this study are better
explained by a relational reasoning approach compared to an analogical structure-mapping
approach only. I hesitate to specify the exact type of relationships derived by learners between
the refutational text and misaligned weather ER although I can see how the ER alone could
facilitate awareness of anomaly, and antimony, and how the text and ER together could yield
antithetical reasoning.
Regarding thematic selection, this study provides support for its importance in text and
graphic comprehension as evidenced by increased learning gains when viewing a message with
101
an ER compared to not. The topic of this investigation was climate change, a controversial and
politicized issue, and not photosynthesis (Södervik et al., 2014) or Newton’s Laws (Hynd &
Alvermann, 1985). For this reason it is difficult to exactly state how a specific topic defines the
role of an external representation, for socio-scientific issues are much more closely tied to
identity (Heddy et al., 2013; Trevors et al., 2016) and are therefore subject to increased top-down
influences (Kuklinski et al., 2000). Socio-scientific topics fuse the importance of knowledge and
attitudes, dependent on the topic, and make conceptual change more difficult when they are in
conflict (Jacobson et al., 2021). Therefore, at this point topic can only be productively described
as scientific or socio-scientific, for this informs the communicator that attitudes, knowledge,
identity and emotions are likely to be activated. For instance, Born (2018) documented that polar
bear imagery creates personal concern for some, but produces an individualized and localized
account of climate for others that masks climate changes’ wider impacts. Research question 4c
also provided evidence that thematic frames, regardless of alignment, activate different
considerations and self-reported emotions. Thematic framing drove the majority of study
findings, however, the significant interactions between thematic frame and alignment indicates
that both comprehension processes are involved in ways we do not yet fully understand.
Implications for Instruction
The communication approach used in this study (i.e., define thematic frame and assess
alignment with key concepts) can be adapted and applied to diverse scientific concepts, both
controversial and not. Additionally, media literacy, specifically scientific media literacy is an
emerging necessity for individuals across the political spectrum and ages (Lewandowsky et al.,
2020). Recent interest in inoculation strategies (Lewandowsky & Van Der Linden, 2021),
epistemic interventions (Thacker, 2020) and text and graphic interventions (Mason et al., 2017
102
(Danielson, 2017) indicate a desire for increased scientific media literacy, and its communication
to educators. These various intervention strategies all rely on instruction and the specification of
tasks which do influence how text and graphics are processed when learning. For instance,
McCrudden, Magliano, and Schraw (2010) found that different instructions for a text resulted in
significant differences in reading times and strategy use and provides direct support for the utility
of graphical literacy instruction (Danielson, 2017; Mason et al., 2017). Schnotz and Preuß (1997)
found that learning tasks were facilitated or inhibited by the specific representation paired with a
task, for instance planning a circumnavigation route or solving for time differences.
News and media coverage are not centered on informing, but entertaining. Therefore, it
may be useful to instruct students and teachers to critically analyze science messages for
information, persuasion, and leading frames, as they do influence decision making. Ironically,
students and educators tend to focus on one source of information, usually the text, at the
expense of the ER, and may provide the start point for text and graphic literacy interventions
(Cromley et al., 2010; Mason, 2013). Teaching students to reason relationally would allow them
to analyze ERs both in school and in ‘the wild’, which should support the skills in both formal
and informal learning spaces.
Another key finding in this study relates to measuring knowledge for controversial topics,
Recently, McCarthy and McNamara (2021) proposed a new framework for assessing knowledge
specifically commitment/confidence in a belief. Commitment can be conceptualized two ways,
one regarding confidence in the belief’s accuracy, the other relates to political commitment. This
is directly relevant to climate communication for certain types of knowledge are subject to
partisan influence (Hamilton, 2020). The study also speaks to the measurement of emotions
specifically the value in capturing intensity and identifying emotion objects. Here, the
103
implication for instruction relates to exploring the thoughts, emotions, considerations, and
societal connections individuals leverage when learning about controversial topics. This
recommendation relates to facilitating meaningful, personally relevant, and value driven
associations with science (Immordino-Yang & Gotlieb, 2020).
Future Directions
Future analyses, and studies are necessary to address mediation or explain how (if at all)
knowledge, attitude, and emotions together impact decision-making for climate-change.
Additionally, future studies should explore if misalignment is beneficial for specific topics only,
or as a general approach but must consider the specific prior knowledge and mental images the
individuals bring with them. For instance, a motivated reasoning (Kunda, 1990) approach may
prove fruitful as it will foreground how someone’s identity may yield a focus on directional or
accuracy-based reasoning goals. There is also a possibility to leverage ones’ mental image and
have the individual assess similarities and differences with the external representation in a think
aloud or draw aloud protocol. However, these methods are demanding of participants and would
therefore benefit from including a measure of need for cognition (Petty and Cacioppo, 1982) as it
is related to conceptual change (Dole & Sinatra, 1998). Overall, this study serves as proof of
concept for the consideration thematic framing and cognitive alignment in climate
communication but would benefit from scrutiny by alternative theoretical perspectives such as
Construal Level Theory (Trope & Liberman, 2010) or Cognitive Load Theory (Sweller, 2011).
Conclusions
Findings from this study contribute to our growing understanding of how to best facilitate
an accurate understanding of the concepts climate and weather, and how these conceptual
changes relate to attitude and policy preferences. Overall I documented support for the
104
importance of visual message characteristics in conceptual change. This finding suggests the
CRKM’s (Dole & Sinatra, 1998) consideration of message characteristics could be extended by
incorporating the ITPCs processes of thematic selection and analog structure mapping and the
ITPC by including relational reasoning (Schnotz, 2005; 2014). Here, I documented that thematic
external representations facilitated conceptual change for a difficult concept, but that
consideration of alignment alone did not. However, the interaction between thematic frame and
alignment provided insight regarding what the ERs differentially communicated (time, and
emotion objects) and potentially how they influenced policy preferences. Overall, this study
provides evidence for the importance of climate communication messages in informing the
general public about climate, provides two methods to assess their role (thematic selection and
cognitive alignment) and assessed the impacts of informational science communication on
knowledge, attitudes, and climate policy preferences.

105
References

Ballew, M. T., Pearson, A. R., Goldberg, M. H., Rosenthal, S. A., & Leiserowitz, A. (2020).
Does socioeconomic status moderate the political divide on climate change? The roles of
education, income, and individualism. Global Environmental Change, 60.
https://doi.org/10.1016/j.gloenvcha.2019.102024

Ballew, M. T., Rosenthal, S. A., Goldberg, M. H., Gustafson, A., Kotcher, J. E., Maibach, E. W.,
& Leiserowitz, A. (2020). Beliefs about others' global warming beliefs: The role of party
affiliation and opinion deviance. Journal of Environmental Psychology.
https://doi.org/10.1016/j.jenvp.2020.101466

Born, D. (2019). Bearing Witness? Polar Bears as Icons for Climate Change Communication in
National Geographic. Environmental Communication, 13(5), 649-663.
https://doi.org/10.1080/17524032.2018.1435557

Broughton, S. H., Sinatra, G. M., & Nussbaum, E. M. (2011). “Pluto Has Been a Planet My
Whole Life!” Emotions, Attitudes, and Conceptual Change in Elementary Students’
Learning about Pluto’s Reclassification. Research in Science Education, 43(2), 529-550.
https://doi.org/10.1007/s11165-011-9274-x

Brun, G., & Kuenzle, D. (2008). Introduction: A New Role for Emotions in Epistemology. In G.
Brun, U. Doguoglu, & D. Kuenzle (Eds.), Epistemology and Emotions. Ashgate.

Cacioppo, J. T., & Petty, R. E. (1982). The need for cognition. Journal of Personality and Social
Psychology, 42(1), 116.

Capstick, S., Whitmarsh, L., Poortinga, W., Pidgeon, N., & Upham, P. (2015). International
trends in public perceptions of climate change over the past quarter century. Wiley
Interdisciplinary Reviews: Climate Change, 6(1), 35-61. https://doi.org/10.1002/wcc.321

Carney, R. N., & Levin, H. M. (2002). Pictorial Illustrations Still Improve Students ’ Learning
From Text. Educational Psychology Review, 14(1), 5-26.

Chan, M.-p. S., Jones, C. R., Hall Jamieson, K., & Albarracin, D. (2017). Debunking: A Meta-
Analysis of the Psychological Efficacy of Messages Countering Misinformation.
Psychological Science, 28(11), 1531-1546. https://doi.org/10.1177/0956797617714579

Chapman, D. A., Corner, A., Webster, R., & Markowitz, E. M. (2016). Climate visuals: A mixed
methods investigation of public perceptions of climate images in three countries. Global
Environmental Change, 41, 172-182. https://doi.org/10.1016/j.gloenvcha.2016.10.003

Chi, M. T. H. (2008). Three Types of Conceptual Change Belief Revision, Mental Model
Transformation, and Categorical Shift. In S. Vosniadou (Ed.), Handbook of research on
conceptual change (pp. 61-82). Erlbaum.

106
Chinn, C. A., & Brewer, W. F. (1993). The role of anomalous data in knowledge acquisition: A
theoretical framework and implications for science instruction. Review of Educational
Research, 63(1), 1-49.

Chong, D. (1993). How People Think, Reason, and Feel About Rights and Liberties. American
Journal of Political Science, 37, 867-899.

Chong, D. (2000). Rational lives: Norms and values in politics and society. University of
Chicago Press.

Cook, J., & Lewandowsky, S. (2011). The Debunking Handbook. University of Queensland.
http://sks.to/debunk

Cromley, J. G., Snyder-Hogan, L. E., & Luciw-Dubas, U. A. (2010). Reading comprehension of
scientific text: A domain-specific test of the direct and inferential mediation model of
reading comprehension. Journal of Educational Psychology, 102(3), 687-700.
https://doi.org/10.1037/a0019452

Damasio, A., & Carvalho, G. B. (2013). The nature of feelings: evolutionary and neurobiological
origins. Nature Reviews Neuroscience, 14(2), 143-152.

Damasio, A. R. (2004). Emotions and Feelings. In Feelings and Emotions (pp. 49-57).
https://doi.org/10.1017/cbo9780511806582.004

Danielson, R. W. (2017). Facilitating learning and conceptual change through an intervention
on text graphic processing University of Southern California].

Danielson, R. W., & Sinatra, G. M. (2017). A Relational Reasoning Approach to Text-Graphic
Processing. Educational Psychology Review, 29, 55-72. https://doi.org/10.1007/s10648-
016-9374-2

Danielson, R. W., Sinatra, G. M., & Kendeou, P. (2016). Augmenting the Refutation Text Effect
with Analogies and Graphics. Discourse Processes, 53(5-6), 392-414.
https://doi.org/10.1080/0163853x.2016.1166334

Dole, J. A., & Sinatra, G. M. (1998). Reconceptualizing Change in the Cognitive Construction of
Knowledge. Educational Psychologist, 33(2/3), 109-128.

Duchi, L., Lombardi, D., Paas, F., & Loyens, S. M. (2020). How a growth mindset can change
the climate: The power of implicit beliefs in influencing people's view and action.
Journal of Environmental Psychology, 70, 101461.

Dumas, D., Alexander, P. A., & Grossnickle, E. M. (2013). Relational Reasoning and Its
Manifestations in the Educational Context: a Systematic Review of the Literature.
Educational Psychology Review, 25(3), 391-427. https://doi.org/10.1007/s 1 0648-0 1 3-
9224-4
107

Feldman, L., & Hart, P. S. (2018, Mar). Is There Any Hope? How Climate Change News
Imagery and Text Influence Audience Emotions and Support for Climate Mitigation
Policies. Risk Anal, 38(3), 585-602. https://doi.org/10.1111/risa.12868

Gaines, B. J., Kuklinski, J. H., Quirk, P. J., Peyton, B., & Verkuilen, J. (2007). Same Facts,
Different Interpretations- Partisan Motivation and Opinion on Iraq. Journal of Politics,
69(4), 957–974.

Gamson, W. A., & Modigliani, A. (1989). Media discourse and public opinion on nuclear power:
A constructionist approach. American Journal of Sociology, 95(1), 1-37.

Gentner, D. (1983). Structure-Mapping: A Theoretical Framework for Analogy*. Cognitive
Science, 7, 155-170.

Gentner, D. (1989). Analogical learning. Similarity and analogical reasoning.

Gentner, D., & Holyoak, K. J. (1997). Reasoning and learning by analogy: Introduction.
American Psychologist, 52(1), 32.

Gill, M. G., Trevors, G. J., Greene, J. A., & Algina, J. (2020). Don’t Take It Personally? The
Role of Personal Relevance in Conceptual Change. The Journal of Experimental
Education, 1-22. https://doi.org/10.1080/00220973.2020.1754152

Gowda, M. V. R., Fox, J. C., & Magelky, R. D. (1997). Students’ Understanding of Climate
Change: Insights for Scientists and Educators. Educational Affairs, 78(10).

Guzzetti, B. J., Snyder, T. E., Glass, G. V., & Gamas, W. S. (1993). Promoting Conceptual
Change in Science: A Comparative Meta-Analysis of Instructional Interventions from
Reading Education and Science Education. Reading Research Quarterly, 28(2), 116-159.
https://doi.org/10.2307 / 747886

Hamilton, L. C. (2020). Surveys Find" Two Kinds" of Public Knowledge about Polar Regions.
Witness the Arctic.

Harp, S. F., & Mayer, R. E. (1997). The role of interest in learning from scientific text and
illustrations- On the distinction between emotional interest and cognitive interest. Journal
of Educational Psychology, 89(1), 92-102.

Hart, P. S., & Nisbet, E. C. (2012). Boomerang Effects in Science Communication.
Communication Research, 39(6), 701-723. https://doi.org/10.1177/0093650211416646

Harvey, J. A., van den Berg, D., Ellers, J., Kampen, R., Crowther, T. W., Roessingh, P.,
Verheggen, B., Nuijten, R. J. M., Post, E., Lewandowsky, S., Stirling, I., Balgopal, M.,
Amstrup, S. C., & Mann, M. E. (2018, Apr 1). Internet Blogs, Polar Bears, and Climate-
108
Change Denial by Proxy. Bioscience, 68(4), 281-287.
https://doi.org/10.1093/biosci/bix133

Heddy, B. C., Danielson, R. W., Sinatra, G. M., & Graham, J. (2016). Modifying Knowledge,
Emotions, and Attitudes Regarding Genetically Modified Foods. The Journal of
Experimental Education, 85(3), 513-533.
https://doi.org/10.1080/00220973.2016.1260523

Hegarty, M. (2011). The Cognitive Science of Visual-Spatial Displays: Implications for Design.
Topics in Cognitive Science, 3(3), 446-474. https://doi.org/10.1111/j.1756-
8765.2011.01150.x

Holyoak, K. J., & Thagard, P. (1997). The analogical mind. American Psychologist, 52(1), 35.

Hynd, C. R., & Alvermann, D. E. (1985). The Role of Refutation Text in Overcoming Difficulty
with Science Concepts. College Reading and Learning Assistance Technical Report 85-
08.
http://libproxy.usc.edu/login?url=https://search.proquest.com/docview/63283633?accoun
tid=14749

Immordino-Yang, M. H. (2010). Toward a Microdevelopmental, Interdisciplinary Approach to
Social Emotion. Emotion Review, 2(3), 217-220.
https://doi.org/10.1177/1754073910361985

Immordino-Yang, M. H., Christodoulou, J. A., & Singh, V. (2012). Rest Is Not Idleness:
Implications of the Brain's Default Mode for Human Development and Education.
Perspect Psychol Sci, 7(4), 352-364. https://doi.org/10.1177/1745691612447308

Immordino-Yang, M. H., & Damasio, A. (2007). We feel, therefore we learn: The relevance of
affective and social neuroscience to education. Mind, Brain, and Education, 1(1), 3-10.
https://doi.org/10.1111/j.1751-228X.2007.00004.x

Immordino-Yang, M. H., McColl, A., Damasio, H., & Damasio, A. (2009). Neural correlates of
admiration and compassion. Proceedings of the National Academy of Science, 106(19),
8021-8026. https://doi.org/10.1073/pnas.0810363106

Isberner, M.-B., Richter, T., Maier, J., Knuth-Herzig, K., Horz, H., & Schnotz, W. (2013).
Comprehending conflicting science-related texts: graphs as plausibility cues.
Instructional Science, 41(5), 849-872. https://doi.org/10.1007/s11251-012-9261-2

Jacobson, N. G., Thacker, I., & Sinatra, G. M. (2021). Here’s Hoping It’s not Just Text Structure:
The Role of Emotions in Knowledge Revision and the Backfire Effect. Discourse
Processes, 1-23. https://doi.org/10.1080/0163853x.2021.1925059

Johnson-Laird, P. N. (1980). Mental Models in Cognitive Science. Cognitive Science, 4, 71-115.

109
Kahan, D. M. (2013). Ideology, motivated reasoning, and cognitive reflection. Judgment and
Decision Making, 8(4), 407-424.

Kalyuga, S., & Renkl, A. (2010). Expertise reversal effect and its instructional implications :
introduction to the special issue. 38(3), 209-215. https://doi.org/10.1007/sl

Kaplan, J. T., Gimbel, S. I., & Harris, S. (2016). Neural correlates of maintaining one’s political
beliefs in the face of counterevidence. Scientific Reports, 6, 1-11.
https://doi.org/10.1038/srep39589

Kendeou, P., Butterfuss, R., Kim, J., & Van Boekel, M. (2019, Jan). Knowledge revision through
the lenses of the three-pronged approach. Mem Cognit, 47(1), 33-46.
https://doi.org/10.3758/s13421-018-0848-y

Kendeou, P., Butterfuss, R., Van Boekel, M., & O’Brien, E. J. (2017). Integrating Relational
Reasoning and Knowledge Revision During Reading. Educational Psychology Review,
29(1), 27-39. https://doi.org/10.1007/s10648-016-9381-3

Kendeou, P., & O’Brien, E. J. (2014). The Knowledge Revision Components (KReC)
Framework: Processes and Mechanisms. In D. N. Rapp & J. L. Braasch (Eds.),
Processing inaccurate information: Theoretical and applied perspectives from cognitive
science and the educational sciences (pp. 353-375). MIT Press.
https://doi.org/doi.org/10.7551/mitpress/9737.003.0022

Kreezer, G., & Dallenbach, K. M. (1929). Learning the relation of opposition. The American
Journal of Psychology, 41(3), 432-441.

Kuklinski, J. H., Quirk, P. J., Jerit, J., Schwieder, D., & Rich, R. F. (2000). Misinformation and
the currency of democratic citizenship. The Journal of Politics, 62(3), 790-816.

Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulletin, 108(3), 480.

Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (2008). International affective picture system
(IAPS): affective ratings of pictures and instruction manual.

Lehman, B., Thompson, J., Davis, S., & Carlson, J. M. (2019). Affective Images of Climate
Change. Front Psychol, 10, 960. https://doi.org/10.3389/fpsyg.2019.00960

Leiserowitz, A. (2006). Climate Change Risk Perception and Policy Preferences: The Role of
Affect, Imagery, and Values. Climatic Change, 77(1-2), 45-72.
https://doi.org/10.1007/s10584-006-9059-9

Leiserowitz, A., Maibach, E., Rosenthal, S., Kotcher, J., Bergquist, P., Ballew, M., Goldberg,
M., Gustafson, A., & Wang, X. (2020). Climate Change in the American Mind: April
2020. Yale University and George Mason University.

110
Leiserowitz, A., Maibach, E. W., & Roser-Renouf, C. (2009). Climate Change in the American
Mind: Americans' Climate Change Beliefs, Attitudes, Policy Preferences, and Actions

Leiserowitz, A., Roser-Renouf, C., Marlon, J., & Maibach, E. (2021). Global Warming’s Six
Americas: a review and recommendations for climate change communication. Current
Opinion in Behavioral Sciences, 42, 97-103.

Leviston, Z., Price, J., & Bishop, B. (2014). Imagining climate change: The role of implicit
associations and affective psychological distancing in climate change responses.
European Journal of Social Psychology, 44(5), 441-454.
https://doi.org/10.1002/ejsp.2050

Lewandowsky, S., & Whitmarsh, L. (2018). Climate communication for biologists: When a
picture can tell a thousand words. PLOS Biology, 16(10), 1-6.
https://doi.org/10.1371/journal.pbio.2006004

Linnenbrink-Garcia, L., & Patall, E. A. (2015). Motivation. In L. Corno & E. Anderman (Eds.),
Handbook of Educational Psychology (3 ed., pp. 91–103).

Linnenbrink-Garcia, L., Patall, E. A., & Pekrun, R. (2016). Adaptive Motivation and Emotion in
Education. Policy Insights from the Behavioral and Brain Sciences, 3(2), 228-236.
https://doi.org/10.1177/2372732216644450

Lombardi, D. (2019). Thinking scientifically in a changing world.

Lombardi, D., & Danielson, R. W. (2021). On Learning and Teaching for Conceptual Change. In
A. M. O’Donnell, J. Reeve, & N. Barnes (Eds.), Oxford Handbook.

Lombardi, D., Seyranian, V., & Sinatra, G. M. (2014). Source Effects and Plausibility Judgments
When Reading About Climate Change. Discourse Processes, 51(1-2), 75-92.
https://doi.org/10.1080/0163853x.2013.855049

Lombardi, D., & Sinatra, G. M. (2012). College Students’ Perceptions About the Plausibility of
Human-Induced Climate Change. Research in Science Education, 42(2), 201-217.
https://doi.org/10.1007/s11165-010-9196-z

Lorenzoni, I., Leiserowitz, A., De Franca Doria, M., Poortinga, W., & Pidgeon, N. F. (2006).
Cross‐National Comparisons of Image Associations with “Global Warming” and
“Climate Change” Among Laypeople in the United States of America and Great Britain1.
Journal of Risk Research, 9(3), 265-281. https://doi.org/10.1080/13669870600613658

Maier, M., Rothmund, T., Retzback, A., Lukas, O., & Besley, J. C. (2014). Informal Learning
Through Science Media Usage. Educational Psychologist, 49(2), 86-103.
https://doi.org/10.1080/00461520.2014.916215

111
Marlon, J. R., Bloodhart, B., Ballew, M. T., Rolfe-Redding, J., Roser-Renouf, C., Leiserowitz,
A., & Maibach, E. (2019). How Hope and Doubt Affect Climate Change Mobilization.
Frontiers in Communication, 4. https://doi.org/10.3389/fcomm.2019.00020

Mason, L. (2013). An eye-tracking study of learning from science text with concrete and abstract
illustrations.pdf>. Journal of Experimental Education, 81(3), 356-384.

Mason, L., Baldi, R., Di Ronco, S., Scrimin, S., Danielson, R. W., & Sinatra, G. M. (2017).
Textual and graphical refutations: Effects on conceptual change learning. Contemporary
Educational Psychology, 49, 275-288. https://doi.org/10.1016/j.cedpsych.2017.03.007

Mason, L., & Gava, M. (2007). Effects of epistemological beliefs and learning text structure on
conceptual change. In S. Vosniadou, A. Baltas, & X. Vamvakoussi (Eds.), Reframing the
conceptual change approach in learning and instruction (pp. 165–196). Elsevier.

Mayer, R. E. (2005). Cognitive theory of multimedia learning. In R. E. Mayer (Ed.), Cambridge
Handbook of Multimedia Learning (Vol. 1, pp. 31-48). Cambridge University Press.

McCarthy, K. S., & McNamara, D. S. (2021). The Multidimensional Knowledge in Text
Comprehension framework. Educational Psychologist, 1-19.
https://doi.org/10.1080/00461520.2021.1872379

Muis, K. R., Chevrier, M., & Singh, C. A. (2018). The Role of Epistemic Emotions in Personal
Epistemology and Self-Regulated Learning. Educational Psychologist, 53(3), 165-184.
https://doi.org/10.1080/00461520.2017.1421465

Muis, K. R., Etoubashi, N., & Denton, C. A. (2020). The catcher in the lie: The role of emotions
and epistemic judgments in changing students’ misconceptions and attitudes in a post-
truth era. Contemporary Educational Psychology, 62.
https://doi.org/10.1016/j.cedpsych.2020.101898

Muis, K. R., Pekrun, R., Sinatra, G. M., Azevedo, R., Trevors, G. J., Meier, E., & Heddy, B. C.
(2015). The curious case of climate change: Testing a theoretical model of epistemic
beliefs, epistemic emotions, and complex learning. Learning and Instruction, 39, 168-
183. https://doi.org/10.1016/j.learninstruc.2015.06.003

Munnich, E. L., & Ranney, M. A. (2019, Jan). Learning From Surprise: Harnessing a
Metacognitive Surprise Signal to Build and Adapt Belief Networks. Topics in Cognitive
Science, 11(1), 164-177. https://doi.org/10.1111/tops.12397

Nussbaum, M. E., Cordova, J. R., & Rehmat, A. P. (2017). Refutation Texts for Effective
Climate Change Education. Journal of Geoscience Education, 65(1), 23-34.

Nyhan, B., & Reifler, J. (2010). When Corrections Fail: The Persistence of Political
Misperceptions. Political Behavior, 32(2), 303-330. https://doi.org/10.1007/s11109-010-
9112-2
112

Nyhan, B., & Reifler, J. (2019). The roles of information deficits and identity threat in the
prevalence of misperceptions. Journal of Elections, Public Opinion and Parties, 29(2),
222-244. https://doi.org/10.1080/17457289.2018.1465061

O’Neill, S. (2017). Engaging with Climate Change Imagery. In Oxford Research Encyclopedia
of Climate Science. https://doi.org/10.1093/acrefore/9780190228620.013.371

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

O’Neill, S. J., Boykoff, M., Niemeyer, S., & Day, S. A. (2013). On the use of imagery for
climate change engagement. Global Environmental Change, 23(2), 413-421.
https://doi.org/10.1016/j.gloenvcha.2012.11.006

O’Neill, S. J., & Smith, N. (2014). Climate change and visual imagery. Wiley Interdisciplinary
Reviews: Climate Change, 5(1), 73-87. https://doi.org/10.1002/wcc.249

Papadimitriou, V. (2004). Prospective primary teachers’ understanding of climate change,
greenhouse effect, and ozone layer depletion. Journal of Science Education and
Technology, 13(2), 299–307.

Pekrun, R. (2006). The Control-Value Theory of Achievement Emotions: Assumptions,
Corollaries, and Implications for Educational Research and Practice. Educational
Psychology Review, 18(4), 315-341. https://doi.org/10.1007/s10648-006-9029-9

Pekrun, R., & Meier, E. (2011). Epistemic emotion scales (EES) [Unpublished Manuscript].
University of Munich.

Pekrun, R., & Perry, R. P. (2014). Control-value theory of achievement emotions. In R. Pekrun
& L. Linnenbrink-Garcia (Eds.), International Handbook of Emotions in Education (pp.
120-141). Routedge.

Pekrun, R., & Stephens, E. J. (2012). Academic emotions. In K. R. Harris, S. Graham, T. Urdan,
J. M. Royer, & M. Zeidner (Eds.), APA educational psychology handbook (Vol. 2, pp. 3-
31). American Psychological Association. https://doi.org/10.1037/13274-001

Pekrun, R., Vogl, E., Muis, K. R., & Sinatra, G. M. (2017). Measuring emotions during
epistemic activities: the Epistemically-Related Emotion Scales. Cognition and Emotion,
31(6), 1268-1276. https://doi.org/10.1080/02699931.2016.1204989

Pfirman, S., Hamilton, L., Turrin, M., Narveson, C., & Lloyd, C. A. (2021). Polar knowledge of
US students as indicated by an online Kahoot! quiz game. Journal of Geoscience
Education, 1-26. https://doi.org/10.1080/10899995.2021.1877526

113
Pintrich, P. R., Marx, R. W., & Boyle, R. A. (1993). Beyond Cold Conceptual Change: The Role
of Motivational Beliefs and Classroom Contextual Factors in the Process of Conceptual
Change. Review of Educational Research, 63(2), 167-199.
https://doi.org/10.3102/00346543063002167

Ranney, M. A., & Clark, D. (2016). Climate change conceptual change: Scientific information
can transform attitudes. Topics in Cognitive Science, 8(1), 49-75.
https://escholarship.org/content/qt5rv6s50g/qt5rv6s50g.pdf?t=o5xbij

Read, D., Bostrom, A., Morgan, M. G., Fischhoff, B., & Smuts, T. (1994). What do people know
about global climate change? 2. Survey studies of educated laypeople. Risk analysis : an
official publication of the Society for Risk Analysis, 14(6), 971-982.

Rebich-Hespanha, S., Rice, R. E., Montello, D. R., Retzloff, S., Tien, S., & Hespanha, J. P.
(2015). Image Themes and Frames in US Print News Stories about Climate Change.
Environmental Communication, 9(4), 491-519.
https://doi.org/10.1080/17524032.2014.983534

Resnick, I., Davatzes, A., Newcombe, N. S., & Shipley, T. F. (2017, Mar 01). Using Relational
Reasoning to Learn about Scientific Phenomena at Unfamiliar Scales. Educational
Psychology Review, 29(1), 11-25.

Reynolds, T. W., Bostrom, A., Read, D., & Morgan, M. G. (2010). Now what do people know
about global climate change? Survey studies of educated laypeople. Risk Analysis,
30(10), 1520-1538. https://doi.org/10.1111/j.1539-6924.2010.01448.x

Roser-Renouf, C., Myers, T., & Maibach, E. (2020). Informal Science Education on the NASA
Climate Change Website: Addressing the Challenge of Educating Diverse Audiences.
https://doi.org/10.13021/g112-4686

Sadler, T. D. (2004). Informal reasoning regarding socioscientific issues: A critical review of
research. Journal of Research in Science Teaching, 41(5), 513-536.
https://doi.org/10.1002/tea.20009

Schneider, S., Dyrna, J., Meier, L., Beege, M., & Rey, G. D. (2018). How affective charge and
text–picture connectedness moderate the impact of decorative pictures on multimedia
learning. Journal of Educational Psychology, 110(2), 233-249.
https://doi.org/10.1037/edu0000209

Schnotz, W. (2002). Towards an integrated view of learning from text and visual displays.
Educational Psychology Review, 14(1), 101-120.

Schnotz, W. (2005). An integrated model of text and picture comprehension. In R. E. Mayer
(Ed.), The Cambridge Handbook of Multimedia Learning. Cambridge University Press.

114
Schnotz, W. (2010). Reanalyzing the expertise reversal effect. Instructional Science, 38(3), 315-
323. https://doi.org/10.1007/sl

Schnotz, W. (2014). Integrated Model of Text and Picture Comprehension. In R. E. Mayer (Ed.),
The Cambridge Handbook of Multimedia Learning (Vol. 2). Cambridge University Press.

Schnotz, W., & Preuß, A. (1997). Task-dependent construction of mental models as a basis for
conceptual change. European Journal of Psychology of Education, 12(2), 185-211.

Sinatra, G. M. (2005). The "Warming Trend" in Conceptual Change Research: The Legacy of
Paul R. Pintrich. Educational Psychologist, 40(2), 107-115.
https://doi.org/10.1207/s15326985ep4002_5

Sinatra, G. M., Broughton, S. H., & Lombardi, D. (2014). Emotions in Science Education In R.
Pekrun & L. Linnenbrink-Garcia (Eds.), Handbook of Emotions and Education.

Sinatra, G. M., Kardash, C. M., Taasoobshirazi, G., & Lombardi, D. (2012). Promoting attitude
change and expressed willingness to take action toward climate change in college
students. Instructional Science, 40(1), 1-17. https://doi.org/10.1007/s11251-011-9166-5

Södervik, I., Mikkilä-Erdmann, M., & Vilppu, H. (2014, Aug 01). Promoting the Understanding
of Photosynthesis among Elementary School Student Teachers through Text Design.
Journal of Science Teacher Education, 25(5), 581-600.
http://libproxy.usc.edu/login?url=https://usc.illiad.oclc.org/illiad/illiad.dll/OpenURL??ctx
_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-
8&rfr_id=info:sid/ProQ%3Aeric&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=ar
ticle&rft.jtitle=Journal+of+Science+Teacher+Education&rft.atitle=Promoting+the+Unde
rstanding+of+Photosynthesis+among+Elementary+School+Student+Teachers+through+
Text+Design&rft.au=S%C3%B6dervik%2C+Ilona%3BMikkil%C3%A4-
Erdmann%2C+Mirjamaija%3BVilppu%2C+Henna&rft.aulast=S%C3%B6dervik&rft.auf
irst=Ilona&rft.date=2014-08-
01&rft.volume=25&rft.issue=5&rft.spage=581&rft.title=Journal+of+Science+Teacher+E
ducation&rft.issn=1046560X

Stenseth, T., Bråten, I., & Strømsø, H. I. (2016). Investigating interest and knowledge as
predictors of students' attitudes towards socio-scientific issues. Learning and Individual
Differences, 47, 274-280. https://doi.org/10.1016/j.lindif.2016.02.005

Sweller, J. (2011). Cognitive load theory. In Psychology of Learning and Motivation (pp. 37-76).
Academic Press.

Tabachnick, B. G., & Fidell, L. S. (2013). Using multivariate statistics (6 ed.). Pearson.

Thacker, I., Sinatra, G. M., Muis, K. R., Danielson, R. W., Pekrun, R., Winne, P. H., & Chevrier,
M. (2020). Using persuasive refutation texts to prompt attitudinal and conceptual change.
115
Journal of Educational Psychology, 112(6), 1085-1099.
https://doi.org/http://dx.doi.org/10.1037/edu0000434

Thomas, C. L., & Kirby, L. A. J. (2020). Situational interest helps correct misconceptions: An
investigation of conceptual change in university students. Instructional Science, 48(3),
223-241. https://doi.org/10.1007/s11251-020-09509-2

Tippett, C. D. (2010). Refutation Text in Science Education: A Review of Two Decades of
Research. International Journal of Science and Mathematics Education, 8(6), 951-970.
https://doi.org/10.1007/s10763-010-9203-x

Trevors, G. J., Bohn-Gettler, C., & Kendeou, P. (2021). The Effects of Experimentally Induced
Emotions on Revising Common Vaccine Misconceptions. Quarterly Journal of
Experimental Psychology.

Trevors, G. J., & Kendeou, P. (2019). The Effects of Positive and Negative Emotions on
Knowledge Revision.

Trevors, G. J., Muis, K. R., Pekrun, R., Sinatra, G. M., & Winne, P. H. (2016). Identity and
epistemic emotions during knowledge revision: A potential account for the backfire
effect. Discourse Processes, 53(5-6), 339-370.

Trope, Y., & Liberman, N. (2010). Construal level theory. Handbook of theories of social
psychology, 1, 118-134.

Van Boekel, M., Lassonde, K. A., O’Brien, E. J., & Kendeou, P. (2017). Source credibility and
the processing of refutation texts. Memory & Cognition, 45(1), 168-181.

van der Linden, S., Cook, J., Maertens, R., Goldberg, M., Kerr, J., Lewandowsky, S., Maibach,
E., & Leiserowitz, A. (2021). Do Climate Change Consensus Messages Cause
Reactance? A Comment on Chinn and Hart (2021).

van Valkengoed, A. M., & Steg, L. (2019). Meta-analyses of factors motivating climate change
adaptation behaviour. Nature Climate Change, 9(2), 158-163.
https://doi.org/10.1038/s41558-018-0371-y

Vosniadou, S. (2009). The framework theory approach to the problem of conceptual change. In
International handbook of research on conceptual change (pp. 31-62). Routledge.

Vosniadou, S., & Mason, L. (2012). Conceptual change induced by instruction: A complex
interplay of multiple factors. In K. R. Harris, S. Graham, T. Urdan, S. Graham, J. M.
Royer, & M. Zeidner (Eds.), APA Educational Psychology Handbook: Individual
differences and cultural and contextual factors. (Vol. 2, pp. 221-246). American
Psychological Association.

116
Yang, X. F., Pavarini, G., Schnall, S., & Immordino-Yang, M. H. (2018). Looking up to virtue:
averting gaze facilitates moral construals via posteromedial activations. Social Cognitive
Affective Neuroscience, 13(11), 1131-1139. https://doi.org/10.1093/scan/nsy081

117
Appendix A
Rationale for Data Cleaning
0 issues 1 issue 2 issues 3 issues 4 issues 5 issues 6 issues
N 373 63 30 41 44 64 120
Mean duration in
minutes (SD)
112.5
(578.48)
123.30
(790.41)
215.70
(631.34)
135.41
(690.03)
117.32
(465.57)
248.03
(987.80)
214.84
(927.44)
Min, Max

3.72, 5958.42

4.97, 6296.7

6.40,
2858.28

8.38,
4445.22

7.10,
2678.05

5.73,
5978.90

4.15,
7220.20

Median 21.33 21.68 20.28 24.05 17.36 15.11 16.35
Mean reading time in
minutes (SD)
1.78
1
(1.80) 1.07 (2.16) 0.53 (0.85) 0.61 (0.94) 0.46 (1.07) 0.27 (0.49) 0.25 (0.82)
Min, Max 0.01, 13.75 0.02, 16.50 0.02, 3.53 0.02, 3.94 0.01, 6.11 0.01, 2.93 0.01, 6.73
Median 1.30 0.36 0.20 0.18 0.11 0.09 0.07
DWCM mean change
(SD)
+0.34
2
(1.59) -0.11 (2.29) -0.03 (1.90) -0.22 (1.54) -0.50 (2.19) 0.00 (1.79) -0.03 (1.85)
Attitude mean change
(SD)
+0.08
3
(0.39) +0.01 (0.40) -0.05 (0.44) -0.02 (0.34)
-0.004
(0.41)
-0.06 (0.37) +0.04 (0.35)
General policy
preference mean
change (SD)
+0.02 (0.31) +0.09 (0.34) +0.06 (0.52) -0.03 (0.45) +0.01 (0.54) +0.07 (0.38) +0.06 (0.41)
118
Tax policy preference
mean change (SD)
+0.05 (0.37) +0.10 (0.45) +0.09 (0.52) -0.01 (0.55) +0.03 (0.51) -0.02 (0.52) +0.11 (0.64)
Mean Anger (SD) 2.32
4
(1.38) 2.63 (1.37) 2.50 (1.46) 2.73 (1.48) 3.18 (1.44) 3.02 (1.32) 3.43 (1.28)
Mean Anxiety (SD) 2.33
5
(1.29) 2.65 (1.27) 3.33 (1.37) 2.61 (1.24) 3.25 (1.30) 3.13 (1.22) 3.48 (1.24)
Mean Boredom (SD)

1.53
6
(0.99) 1.94 (1.16) 2.60 (1.50) 2.51 (1.34) 3.00 (1.36) 3.08 (1.15) 3.22 (1.28)
Mean Confused (SD)

1.51
7
(0.92) 2.08 (1.34) 2.43 (1.38) 2.12 (1.31) 3.11 (1.48) 2.98 (1.37) 3.41 (1.31)
Mean Curious (SD)

2.79
8
(1.25) 2.81 (1.38) 2.93 (1.31) 3.12 (1.27) 3.41 (1.34) 3.19 (1.21) 3.52 (1.27)
Mean Enjoyment (SD)

1.73
9
(1.11) 2.65 (1.48) 3.20 (1.56) 3.02 (1.31) 3.45 (1.21) 3.42 (1.23) 3.49 (1.21)
Mean Frustration (SD)

2.47
10
(1.37) 2.59 (1.33) 2.73 (1.48) 2.39 (1.26) 3.09 (1.31) 3.03 (1.15) 3.43 (1.20)
Mean Hopeful (SD)

2.37
11
(1.20) 3.02 (1.44) 3.10 (1.35) 3.49 (1.10) 3.75 (0.99) 3.59 (1.07) 3.48 (1.14)
Mean Hopeless (SD)

2.06
12
(1.20) 2.54 (1.33) 2.57 (1.50) 2.59 (1.26) 3.07 (1.52) 3.02 (1.28) 3.44 (1.32)
Mean Interest (SD)

3.26
13
(1.22) 3.25 (1.37) 3.40 (1.38) 3.59 (1.16) 3.64 (1.10) 3.64 (1.10) 3.61 (1.15)
Mean Surprise (SD)

1.70
14
(1.05) 2.70 (1.52) 2.90 (1.40) 2.95 (1.36) 3.43 (1.23) 3.47 (1.10) 3.53 (1.20)
Knowledge-Attitude
correlation (p-value)
+0.07
(p = .164)
+0.06
(p = .660)
+0.11
(p = .573)
+0.09
(p = .592)
+0.17
(p = .274)
-0.17
(p = .187)
0.09
(p = .324)
119
1. Indicates a significant effect of WCA Issues on time spend reading the experimental condition (F(6, 728) = 24.42, p < .001. All
post-hoc comparisons were significantly different from the 0 issues group and was confirmed with a significant deviation contrast
comparing the reading time of 0 issues to the mean of all other groups.
2. Indicates a significant effect of WCA Issue on change in categorization accuracy (DWCM) (F(6, 728) = 2.54, p = .019. Deviation
contrasts indicated change in accuracy was significantly different for 0 issues compared to the mean of all other issue totals.
3. Indicates a significant deviation contrast where the 0 Issues group differed significantly (p = .001) from the average change of all
other Issues groups, although the main effect was not significant ( p = .188).
4. Indicates a significant effect of WCA Issue on self-reported anger (F(6, 728) = 11.96, p < .001, and is supported by a significant
deviation contrast (p < .001) indicating less anger for the 0 issues group compared to the mean of all other groups.
5. Indicates a significant effect of WCA Issue on self-reported anger (F(6, 728) = 16.40, p < .001, and is supported by a significant
deviation contrast (p < .001) indicating less anxiety for the 0 issues group compared to the mean of all other groups.
6. Indicates a significant effect of WCA Issue on self-reported boredom (F(6, 728) = 49.11, p < .001, and is supported by a significant
deviation contrast (p < .001) indicating less anxiety for the 0 issues group compared to the mean of all other groups.
7. Indicates a significant effect of WCA Issue on self-reported confusion (F(6, 728) = 54.09, p < .001, and is supported by a
significant deviation contrast (p < .001) indicating less confusion for the 0 issues group compared to the mean of all other groups.
8. Indicates a significant effect of WCA Issue on self-reported curiosity (F(6, 728) = 6.23, p < .001, and is supported by a significant
deviation contrast (p < .001) indicating less curiosity for the 0 issues group compared to the mean of all other groups.
9. Indicates a significant effect of WCA Issue on self-reported enjoyment (F(6, 728) = 52.33, p < .001, and is supported by a
significant deviation contrast (p < .001) indicating less enjoyment for the 0 issues group compared to the mean of all other groups.
10. Indicates a significant effect of WCA Issue on self-reported frustration (F(6, 728) = 9.76, p < .001, and is supported by a
significant deviation contrast (p < .001) indicating less frustration for the 0 issues group compared to the mean of all other groups.
11. Indicates a significant effect of WCA Issue on self-reported hopefulness (F(6, 728) = 26.23, p < .001, and is supported by a
significant deviation contrast (p < .001) indicating less hopefulness for the 0 issues group compared to the mean of all other groups.
12. Indicates a significant effect of WCA Issue on self-reported hopelessness (F(6, 728) = 21.65, p < .001, and is supported by a
significant deviation contrast (p < .001) indicating less hopelessness for the 0 issues group compared to the mean of all other groups.
13. Indicates a significant effect of WCA Issue on self-reported interest (F(6, 728) = 2.43, p = .025, and is supported by a significant
deviation contrast (p = .006) indicating less interest for the 0 issues group compared to the mean of all other groups.
14. Indicates a significant effect of WCA Issue on self-reported surprise (F(6, 728) = 58.28, p < .001, and is supported by a
significant deviation contrast (p < .001) indicating less surprise for the 0 issues group compared to the mean of all other groups.
I next analyzed the 373 participants who had 0 issues with their WCA open questions again to assess the impact of reading time on
performance on the DWCM at post-test. However, first I removed two participants for failure of attention checks. I conducted a one-
way ANCOVA on post-test DWCM using pre-test scores as the covariate and a grouping variable to assess outcomes for those who
read for less than 20 seconds (N = 64 or more than 20 seconds (N = 307) . Results indicated a significant effect of reading time F(1,
120
368) = 12.35, p < .001, partial eta squared = .032. Those who read for more than 20 seconds demonstrated significantly higher post-
test DWCM (𝑀 𝑎𝑑𝑗 = 6.82, SE = .077) compared to those who read for less than 20 seconds (𝑀 𝑎𝑑𝑗 = 6.16, SE = .17)
121
Appendix B
Distinctions Between Weather and Climate Measure
(DWCM; Lombardi & Sinatra, 2012)
Read the following statements. Decide if each statement best fits into the category of weather or
climate and select the appropriate option.
Response format: Forced choice (i.e., Climate Weather). Correct answers are presented
alongside measure items.
1. There was a heat wave last summer. Weather
2. It rarely snows in Southern Alabama. Climate
3. It is colder than normal outside. Weather
4. By mid-May, it is usually warm enough to go to the beach. Climate
5. The monsoon rains will probably begin in June. Climate
6. Clouds cover about 40% of the sky. Weather
7. For the last 4 years, the least rainfall occurred during October. Weather
8. There was considerable fogginess on the drive from Los Angeles to Santa Barbara.
Weather
9. The average annual temperature in Reno is 51°F. Climate
10. Miami’s record low temperature is 30°F. Climate
11. We are predicting that temperatures will be greater than normal this autumn.
Weather
12. Over the last 7 years, a drought has caused lake levels to drop about 13 feet.
Weather
13. Tree rings reveal that the region received greater rainfall 700 years ago.
Climate

122
Appendix C
Weather and Climate Assessment
(WCA; Nussbaum et al., 2017)

Read and respond to the following questions to the best of your ability.

Response format: Open-Ended Answer Questions. Scoring rubric presented below assessment
items.

Assessment items:

1. What is climate?
2. What is weather?
4. What evidence indicates that climate change is taking place?
5. What time period is considered to determine climate change?
6. What impact do large bodies of water have on an area’s climate?

Scoring Rubric for the Weather vs. Climate Refutation Text Constructed-Response Questions

Question 1: What is climate? (Choose one of the following)
Spatial (where) dimension
• 0 – no spatial description of climate
• 0.5 – particular/given area OR geographical location (earth, planet)
• 1.0 – acknowledges multiple ‘where’ (i.e., seasonal, regional, global) and temporal
interaction (in a certain area and season)
Spatial (what) dimension
• 0 – no spatial description of climate
• 0.5 – environment/environmental conditions, conditions
• 1.0 – weather/temperature (rain, fire, hear, cold), atmosphere
• 1.5 – temperature & humidity (tropical, arid)
Temporal (when) dimension
• 0 – no temporal description of climate
• 0.5 – time but no qualification (change of year, over period of time)
• 1.0 – time with some qualification (over many years, over an extended period of time)
• 2 – time qualified with long-term (long-term, over a long period of time)
• 3 – time described as 30-years

Question 2: What is weather? (Choose one of the following)
Spatial dimension
• 0 – no spatial description of weather
• 0.5 – spatial description of weather that mentions: heat, wind, rain, snow, or temperature
• 1.0 – spatial description of weather that explicitly states atmosphere
Temporal dimension
• 0 – no temporal description of weather
123
• 1.0 – temporal description of weather including: day-to-day changes, short-term
conditions)

Question 4: What evidence indicates that climate change is taking place? (Choose one of the
following):
• 0 – does not mention the warming of the oceans or large bodies of water
• 1.0 – mentions the warming of the oceans or large bodies of water

Question 5: What time period is considered to determine climate change? (Choose one of the
following):
• 0 – no temporal description of climate
• 0.5 – temporal description of climate as some period of time but without a description or
elaboration of how much time “period of time” OR “several years.”
• 1.0 – temporal description of climate that describes "shorter trends/duration/intervals of
at least several years (e.g., 5) OR more than 100 years
• 2.0 – temporal description of climate that describes "long-term trends/duration/interval
ranging from several decades to 50 years BUT NOT more than 100 years
• 3.0 – temporal description of climate explicitly states duration/period/interval as 30 years

Question 6: What impact do large bodies of water have on an area’s climate? (Choose all that
apply):
• 0 – none of the below
• 0.5 – states that bodies of water can make climate warmer, OR can make climate colder,
OR affects temperature
• 0.5 – mentions/describes water (e.g., lots of moisture or rain, OR changes in
wetness/dryness or humidity)
• 1.0 – states that large bodies of water store heat. Thus, large bodies of water heat up and
cool down more slowly than nearby land masses
• 1.0 – states that bodies of water prevent dramatic fluctuations of temperatures between
day and night
• 1.0 – states that bodies of water keep an area's climate within a moderate range

124
Appendix D
Attitudes Toward Global Warming Measure
(Sinatra, Kardash, Taasoobshirazi, & Lombardi, 2012)

Rate the degree to which you agree with the following statements.

Response format: 5-point Likert scale.

1 2 3 4 5
Strongly agree Agree Unsure Disagree Strongly disagree
Measure items:
1. Human activity has been the driving force behind the warming trend over the last 50
years.
2. The release of CO
2
(carbon dioxide) from human activity (such as smoke stacks and car
emissions) has played a central role in raising the average surface temperature of the
earth.
3. The Greenland ice cap is melting faster than had previously been thought.
4. The speed with which the melting ice caps may raise sea levels is uncertain.
5. The likelihood that emissions are the main cause of the observed warming trend of the
last 50 years is between 90 and 99%.
6. An increase in CO
2
(carbon dioxide) is directly related to an increase in global
temperature.

125
Appendix E
Climate Policy Preferences
(CPP; Leiserowitz, 2006)

Read the following policies and indicate how much you support or oppose each policy. If you do
not support or oppose a policy, indicate "Don't know".

Response format: 5-point Likert scale with a “Don’t know” option

1 2 3 4 5
Strongly oppose Oppose Support Strongly support Don’t know

Items in bold are components of the governmental tax policy preferences, with non-bolded items
corresponding to tax policy preferences

1. Regulate carbon dioxide (the primary greenhouse gas) as a pollutant.
2. Sign an international treaty that requires the United States to cut its emissions of carbon
dioxide 90% by the year 2050.
3. Require electric utilities to produce at least 20% of their electricity from wind, solar, or
other renewable energy sources, even if it cost the average household an extra $100 a
year.
4. Establish a special fund to help make buildings more energy efficient and teach
Americans how to reduce their energy use. This would add a $2.50 surcharge to the
average household’s monthly electric bill.
5. Create a new national market that allows companies to buy and sell the right to emit the
greenhouse gases said to cause global warming. The federal government would set a
national cap on emissions. Each company would then purchase the right to emit a portion
of this total amount. If a company then emitted more than its portion, it would have to
buy more emission rights from other companies or pay large fines.
6. Increase taxes on gasoline by 25 cents per gallon and return the revenues to taxpayers by
reducing the federal income tax.
7. Expand offshore drilling for oil and natural gas off the U.S. coast
6
.
8. Drill for oil in the Arctic National Wildlife Refuge.
9. Build more nuclear power plants.
10. Require automakers to increase the fuel efficiency of cars, trucks, and SUVS, to 45mpg,
even if it means a new vehicle will cost up to $1,000 more to buy.
11. Provide tax rebates for people who purchase energy-efficient vehicles or solar panels.
12. Provide a government subsidy to replace old water heaters, air conditioners, light bulbs,
and insulation. This subsidy would cost the average household $5 a month in higher
taxes. Those who took advantage of the program would save money on their utility bills.
13. Fund more research into renewable energy sources, such as solar and wind power.

6
Items 7 and 8 are reverse scored.
126
Appendix F
Cognitive Construal Task
(Immordino-Yang et al., 2009; Yang et al., 2018)

How does this information about climate change make you feel?

Construal
Dimension
Code Goal Value Value
Label
Code Example
Affect:
Sensitive to
responses
with emotion
words, or
statements of
feeling.
Does the
response
provide a
description
of
emotional
feelings?
0 neutral Neutral was applied to
codes that stated, they
didn’t feel any
different, or didn’t feel
about the topic or
responses that referred
to some aspect of
learning, feeling
informed or
understanding.
‘I already knew
this’,
‘informed’, ‘it
doesn’t make
me feel any type
of way’
1 positive Positive codes were
applied to statements
where a positive
emotion or feeling
word was indicated.
empowered’,
‘inspired’,
‘happy’, ‘very
good’,
2 negative Negative codes were
applied to responses
that described a
negative emotion or
feeling was indicated
‘concerned, ‘it
makes me feel
anxious”, “sad”,
“afraid”,
“scared”,
“irritated”
“depressed”,
“worried”.
3 neutral,
positive
Neutral and positive
applied to responses
that referred to learning
but were paired with a
positive emotion not
focused on the
information directly.
‘I know more. I
feel more
hopeful’ OR
‘informed, and
hopeful we can
do something’
4 neutral,
negative
Neutral and negative
applied to responses
that referred to
learning, but also
described a negative
emotion or feeling not
focused on the
information.
‘I don’t feel any
way… maybe
annoyed
because I don’t
believe it is
caused by man’.
127
5 positive,
negative
Positive and negative
codes applied to
responses such as that
referred to positive and
negative emotions or
feelings not focused on
the information.
‘concerned…but
hopeful our
actions can slow
it’, ‘amazed at
just how long
term. I feel
afraid’
Construal
Dimension
Code Value Value
label
Code Guide Example
Concreteness:
Captures
construals
based on
contrast with
an external
influence.
Does the
response
provide a
statement(s)
that
describes a
comparison
or contrast
between
themselves
and the
source of
the feeling
(i.e., the
text)?
0 absent A response was offered
but didn’t meet the
criteria.
‘that I was
almost right’, ‘a
little uneasy
about the
future’, ‘in line
with what I
already
thought’,
‘empowered’
1 present A response was offered
that did meet the
criteria.
‘it makes me
feel vindicated’,
‘it makes me
feel justified’,
‘it makes me
worry and gives
me anxiety’
Construal
Dimension
Code Value Value
Label
Code Guide Example
Abstractness:
Captures
construals
based on
general
beliefs,
values, and
attitudes.
Does the
response
provide a
statement(s)
that are
based on
general
beliefs,
values, and
attitudes
0 absent A response was offered
but didn’t meet the
criteria.
‘that I was
almost right’, ‘a
little uneasy
about the
future’, ‘in line
with what I
already
thought’,
‘empowered’
1 present A response was offered
that did meet the
criteria.
it makes me feel
vindicated’, ‘it
makes me feel
justified’, ‘it
makes me worry
and gives me
anxiety’

128
Appendix G
Epistemically Related Emotions Scale
(EES; Pekrun et al., 2017)

We are interested in the emotions you experienced when thinking about the scientific view on
the weather and climate distinction. For each of the following emotions we may ask one or two
questions.

First, you will be asked to indicate the intensity of your emotional response by selecting the
option that best describes your experience.

Response format: 5-point Likert intensity scale.

1 2 3 4 5
Not at all Very little Moderate Strong Very strong

1. Angry
2. Anxious
3. Bored
4. Confused
5. Curious
6. Enjoyment
7. Frustrated
8. Hopeless
7

9. Hopeful
10. Interested
11. Surprised

Second, you may also be asked to select the source of that emotion.

Please indicate the source of your anger. (select all that apply)

1. Ideas in the text
2. People impacted by climate change
3. Animals impacted by climate change
4. Nature impacted by climate change
5. Fact that climate change is happening
6. Politicians

7
Emotion added to the EES based on findings from Heddy et al. (2016)
129
Appendix H
Political Ideology and Partisanship Measures
(Nyhan & Reifler, 2010)

Select the option that best describes your political ideology.

1. Other: _______
2. Very liberal
3. Liberal
4. Lean liberal
5. Centrist
6. Lean conservative
7. Conservative
8. Very conservative
9. Prefer not to state

Select the response that best describes you.

1. Other: _______
2. Strong Democrat
3. Democrat
4. Lean Democrat
5. Independent
6. Lean Republican
7. Republican
8. Strong Republican
9. Prefer not to state

130
Appendix I
Demographics Questionnaire and Attention Checks
1. What is your current age? (in years)
2. Please select you gender below (female, male, I self-identify as, prefer not to state)
3. What is your ethnicity? (select all that apply) (African American/Black, American
Indian/Alaska Native, Asian American/Asian, Native Hawaiian/Pacific Islander, Mexican
American/ Chicano, Puerto Rican, Other Latino, White/Caucasian, Other:_____, Prefer
not to state)
4. What is your highest level of formal education? (some high school, high school
diploma/GED, some college, associate degree, bachelor’s degree, master’s degree,
doctoral degree, prefer not to state)
1. Drag the slider to the value 72.
2. Select the third option in the list below. (first, second, third, fourth)
3. Select the option ‘blue’ from the list below. (bule, bleu, Blue, blue)

131
Appendix J
Experimental Control Text
(National Oceanic and Atmospheric Administration Weather and Climate Infographic)

132
Appendix K
Experimental Refutation Text
(Nussbaum et al., 2017)

Abstract (if available)

Abstract Communicating scientifically accurate information about climate change is difficult, partly due to characteristics of the general public (inaccurate knowledge, negative attitudes, negative emotions) and partly due to characteristics of climate change messages (inaccurate, disengaging). The purpose of this dissertation study was to investigate the impact of a climate change message paired with a thematically framed external representation (i.e., polar, weather) that depicted climate change impacts accurately (aligned) or inaccurately (misaligned) on climate change understanding, attitudes, policy preferences, and emotions. Results indicated an online sample of adults (N = 371) increased their understanding and attitude toward climate change. However, thematically framed ERs resulted in greater climate understanding compared to control (specifically the temporal aspect of climate change) and results suggested a beneficial role of ERs overall. These effects were qualified by thematic frame and alignment interactions on policy preferences where support for tax and general climate policies increased only for the misaligned weather ER. Finally, for emotions, results indicated that a polar compared to weather frame resulted in more focus on the nature, people, and animals impacted by climate change. Together results provide support for the importance of external representations in climate communication, for these choices about theme and alignment can differentially influence object focus, understanding of climate, and climate policy preferences, although everyone learned.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Facilitating learning and conceptual change through an intervention on text graphic processing

PDF

The influence of numerical estimation skills and epistemic cognition in plausibility judgments and conceptual change

PDF

Beyond greenhouse gases and towards urban-scale climate mitigation: understanding the roles of black carbon aerosols and the urban heat island effect as local to regional radiative forcing agents

PDF

The role of the timing of school changes and school quality in the impact of student mobility: evidence from Clark County, Nevada

PDF

Understanding the critical role of parents in improving the identification and support of twice exceptional learners in the K–12 system

PDF

Participation in higher education diversity, equity, and inclusion work: a relational intersectionality of organizations analysis

Asset Metadata

Creator Jacobson, Neil G. (author)

Core Title What does climate change look like to you? The role of internal and external representations in facilitating conceptual change about the weather and climate distinction

School Rossier School of Education

Degree Doctor of Philosophy

Degree Program Urban Education Policy

Degree Conferral Date 2021-12

Publication Date 12/17/2021

Defense Date 11/17/2021

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag climate change,conceptual change,external representations,graphics,OAI-PMH Harvest,refutation text

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Sinatra, Gale M. (committee chair), Chong, Dennis (committee member), Immordino-Yang, Mary Helen (committee member)

Creator Email njacobson@mail.csuchico.edu,njacobson5@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC18652764

Unique identifier UC18652764

Legacy Identifier etd-JacobsonNe-10307

Document Type Dissertation

Format application/pdf (imt)

Rights Jacobson, Neil G.

Type texts

Source 20211221-wayne-usctheses-batch-905-nissen (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu