Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Climate change communication: challenges and insights on misinformation, new technology, and social media outreach
(USC Thesis Other)
Climate change communication: challenges and insights on misinformation, new technology, and social media outreach
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Copyright 2023 Lauren Lutzke
Climate change communication: Challenges and insights on misinformation,
new technology, and social media outreach
by
Lauren Lutzke
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PSYCHOLOGY)
May 2023
ii
Acknowledgements
First and foremost, I would like to express my sincerest thanks to my advisor, Dr. Joe
Árvai. His mentorship perfectly balanced offering invaluable guidance and direction, while also
challenging me to learn how to defend my own ideas. Further, his unending support and
enthusiasm about our projects has made me feel like I matter as a whole person instead of as only
a cog in a research machine. I hope to extend the same type of mentorship to those I work with in
the future. I am forever grateful!
I would also like to express my gratitude to the rest of my dissertation committee, Drs.
Wändi Bruine de Bruin, Caitlin Drummond Otten, Norbert Schwarz, and Gale Sinatra, for all of
their support and for inspiring me with their research. To Caitlin, I am grateful to have had your
guidance and encouragement since before I became a PhD student. Having you by my side for
my first conference presentation in particular, helped get me to where I am today.
I am also deeply grateful to Dr. Paul Slovic, who welcomed me to visit Decision
Research as a master’s student, and to Dr. Michael Siegrist for welcoming me into the Consumer
Behavior Group at ETH Zürich as I finished writing my dissertation. Both trips have shaped who
I am professionally and personally.
To the past and present Judgement and Decision Making Lab members, I feel incredibly
lucky to call you all friends, and I am grateful for the many memories we all share. And to Alex
Segrè Cohen, I do not know where to start—thank you for everything.
I would also like to thank Cory Miller, my high school psychology teacher, Dr. Shelly
Schreier, my Introduction to Psychology professor, and Dr. Ed O’Brien, who hired me as an
undergraduate research assistant, all for helping to instill my love for psychology.
To the rest of my friends and family, thank you—I could not have done this without you!
iii
Table of Contents
Acknowledgements..........................................................................................................................ii
List of Tables..................................................................................................................................iv
List of Figures..................................................................................................................................v
Abstract...........................................................................................................................................vi
Introduction......................................................................................................................................1
Chapter I: Priming critical thinking: Simple interventions limit the influence of fake news about
climate change on Facebook................................................................................................5
Chapter II: Consumer acceptance of products from carbon capture and utilization......................22
Chapter III: All atwitter about climate change: Do polite and informative Twitter debates
influence support for climate policy?................................................................................49
Concluding Remarks......................................................................................................................78
References......................................................................................................................................82
Appendices.....................................................................................................................................92
Appendix A: Chapter I Tables and Figures.......................................................................92
Appendix B: Chapter II Tables and Figures......................................................................97
Appendix C: Chapter II Supplementary Materials..........................................................101
Appendix D: Chapter III Tables and Figures...................................................................111
Appendix E: Chapter III Supplementary Materials.........................................................117
iv
List of Tables
Appendix A, Table 1. Sample characteristics……………………………………………………93
Appendix A, Table 2. Mean ratings of perceived trustworthiness, likelihood of “liking,”, and
likelihood of “sharing” across post type (fake news and real news)…………………….94
Appendix A, Table 3. Regression analyses for climate change doubters and believers on
perceived trustworthiness of, likelihood of “liking”, and likelihood of “sharing” posts
based on fake news.……………………………………………………………………...95
Appendix A, Table 4. Regression analyses for climate change doubters and believers on
perceived trustworthiness of, likelihood of “liking”, and likelihood of “sharing” posts
based on real news……………………………………………………………………….96
Appendix B, Table 1. Sample Characteristics…………………………………………………...97
Appendix B, Table 2. Regression analysis for willingness to consume or use a CCU-based
product…………………………………………………………………………………...98
Appendix D, Table 1. Distribution of Participants by Condition………………………………111
Appendix D, Table 2. Summary of Main Findings…………………………………………….112
v
List of Figures
Appendix A, Figure 1. The fake news and real news posts used in this experiment…………….92
Appendix B, Figure 1. Distribution of participant responses regarding willingness to consume or
use a CCU-based product, perception of climate change mitigation benefits, and
perception of health risks………………………………………………………………...99
Appendix B, Figure 2. ANOVA results and Tukey’s post-hoc tests comparing the willingness to
consume or use different product types from DAC and point source capture………….100
Appendix D, Figure 1. Hypothetical Twitter conversation debating a carbon tax (positive tone
and high informativeness)………………………………………………………………113
Appendix D, Figure 2. Hypothetical Twitter conversation debating a carbon tax (negative
tone and low informativeness)………………………………………………………….114
Appendix D, Figure 3. Responses by tone, informativeness, and climate policy………………115
Appendix D, Figure 4. Responses by tone, informativeness, and political party (republican vs.
democrat), collapsed across climate policy…………………………………………….116
vi
Abstract
Although many scientific and policy solutions for addressing climate change have been
developed or proposed, major barriers to their enactment are related to public perception and
political opposition to climate action. With this, implementing effective communication
strategies for conveying relevant climate information to the public, along with addressing
barriers to effective communication (e.g., misinformation on social media), is of vital
importance. Therefore, through detailing the findings of three online survey experiments, this
dissertation aims to highlight challenges and insights for effective climate communication in
three separate contexts. The first chapter focuses on misinformation (i.e., fake news) about
climate change on Facebook and tests simple interventions for their effectiveness in limiting the
influence of fake climate news among Facebook users. The next chapter shifts to analyzing
public opinion of a new technology called carbon capture and utilization (CCU) which removes
CO2 from ambient air or from emissions sources (e.g., a factory or power plant) and uses the CO2
to create consumer goods. The third chapter turns to examining communication about climate
policies on Twitter. This study manipulated whether politicians were polite versus impolite, and
informative versus uninformative, in their tweets (as they debate the merits of a climate policy).
We then measured participants’ judgements of the politicians and their messaging. The findings
of these three studies have implications for advancing climate goals in a politicized environment.
1
Introduction
There are many scientific and policy solutions for addressing climate change that have
been developed. For example, technological advances in renewable energy (solar, wind, battery
storage, etc.) and proposed policy initiatives, such as a price on carbon, have great potential to
reduce greenhouse gas emissions, thereby limiting the negative consequences that correspond
with a warmer planet (IPCC, 2022). However, while these scientific and policy solutions already
exist, major barriers to their enactment are related to public perception and political opposition to
climate action. And although a majority of those in the United States believe climate change is
occurring and is human caused
1
, these beliefs alone have not been enough to spur climate
initiatives at the necessary scale.
Thus, in addition to possessing this basic understanding of climate science, more is
needed from the public to advance climate agendas. For one, people must choose to prioritize
climate action, even if that involves tradeoffs (e.g., higher costs or a rejection of NIMBYism, the
idea that people are accepting of an initiative until they are faced with it themselves “in their own
backyard”)(Cole et al., 2022; Dear, 1992). And, greater bipartisanship is needed to overcome
certain political barriers (e.g., the 60 votes needed to avoid a filibuster in the U.S. Senate), which
remains challenging given the state of political polarization surrounding issues related to climate
change.
With this, implementing effective communication strategies for conveying relevant
climate information to the public (and to business leaders, government officials, etc.), along with
addressing barriers to effective communication with the public (e.g., misinformation on social
media, which fuels political polarization), is of vital importance. Therefore, this dissertation aims
1
https://climatecommunication.yale.edu/visualizations-data/ycom-us/
2
to highlight challenges and insights for effective climate communication in three separate
contexts.
The first chapter of this dissertation focuses on misinformation (i.e., fake news) about
climate change on Facebook. With one analysis finding that climate misinformation receives up
to approximately 1.3 million views per day on the social media platform, there is a need to limit
its reach (Stop Funding Heat, 2021). Meta (Facebook’s parent company) has taken several steps
to limit the influence of climate misinformation. For example, the company has partnered with
fact-checkers to flag posts and has created a Climate Science Center landing page where
Facebook users can learn about climate change. However, daily encounters with fake climate
news far outnumber user visits to the Climate Science Center (one hundred thousand daily
visits)
2
. Ultimately, without removing climate misinformation from the platform, as Meta does
for COVID-19 and election misinformation, these efforts fall short and additional measures to
combat fake news are necessary. Therefore, the study presented in chapter I tests simple
interventions for their effectiveness in limiting the influence of fake climate news.
At the time this research began in 2018, the onus was on Facebook users themselves to
decide to educate themselves on media literacy. If they did in fact choose to do this, they could
search through the Facebook Help Center where they would find a list of tips for spotting fake
news online. While it seemed unlikely that the average Facebook user would take these steps,
especially unprompted, we questioned whether exposure to guidelines for evaluating news online
would actually help people recognize misinformation. Through an online survey experiment, we
explored how exposure to media literacy guidelines influenced how people perceived and
engaged with both “fake” and legitimate news about climate change on Facebook. Our
2
https://www.cnn.com/2021/11/07/tech/facebook-climate-change-misinformation/index.html
3
motivation behind this study being that, if the interventions were effective, Meta could find
creative ways to place these guidelines in front of people while using the platform—thus helping
to address one barrier to effective climate communication.
This dissertation’s second chapter shifts to analyzing public opinion of a new technology
called carbon capture and utilization (CCU). This technology can remove CO2 from ambient air
or from emissions sources (e.g., a factory or power plant) and use the CO2 to create consumer
goods, such as carbonated beverages, plastics, synthetic fuel, and more (Koytsoumpa et al.,
2018). CCU products have the potential to benefit the climate because using captured carbon as a
raw material in manufacturing 1) takes that CO2 out of the atmosphere and 2) helps to displace
whatever untapped resource would have been used instead (e.g., reducing the need to drill for
more oil to make plastics) (Quadrelli et al., 2015). In addition, the profits from these products
can be used to support the development and expansion of carbon removal technology more
generally, which includes technology used to store large amounts of captured carbon
underground indefinitely. This process removes carbon from the atmosphere at a scale greater
than is possible solely through the creation of new products (Global CCS Institute, 2019).
However, without broad public acceptance, there may not be a reliable end-market for
CCU-based products, hindering the economic viability and potential mitigative benefits of CCU.
In order to glean public attitudes related to the technology and its resulting consumer goods, we
conducted a study of adults in U.S. where we measured their willingness to use CCU-based
products, along with other variables we expected may be associated with such acceptance (such
as perceptions of benefit and risk). We collected data on these additional variables in hopes of
uncovering useful insights for effective communication about CCU.
4
Finally, the third chapter here turns to examining communication about climate policies
on Twitter, a social media platform where many in the U.S. report turning to in order to keep up
with the news (Pew Research Center, 2021). Indeed, many politicians are active on Twitter to
reach this group, and the median follower count among members of Congress is approximately
40,000 (Pew Research Center, 2020). With politicians using this platform to communicate about
their ideas and with their constituents, it is important to consider the communication styles they
employ, and how differing styles may influence how readers perceive politicians and their
tweets. Both civility and informativeness, as represented in politicians’ tweets, can vary widely.
And civility in particular, despite the professional nature of politicians’ roles, is often absent
from tweets (notably in many of former President Trump’s messages).
With this, we wanted to examine the effectiveness of polite versus impolite, and
uninformative versus informative tweets in the context of conveying information about climate
policies. We therefore created hypothetical Twitter conversations between a Democratic and
Republican Congress member and manipulated whether politicians were civil versus uncivil, and
uninformative versus informative, as they debate the merits of a climate policy. We then
measured participants’ judgements of the politicians, perceived learning about the policy,
evaluations of the arguments presented both for and against the climate policy, and ultimately
support for the climate policy itself. Especially, in light of the recent acquisition of Twitter by
Elon Musk, which may alter norms surrounding the use of civil language on the platform with
fewer restrictions on hateful speech, findings related to civility may be especially pertinent to
communicating effectively through tweets.
Although these three chapters span three unique contexts, insights from each are in some
ways related, and this will be discussed in the conclusion of the dissertation.
5
Chapter I
Priming critical thinking: Simple interventions limit the influence of fake news about
climate change on Facebook
3
Since the period leading to and following the American election cycle of 2016, media
outlets have warned that people in the U.S. are being exposed to “fake news”. Fake news in this
context refers to fabricated information intended to mislead consumers, which mimics the
appearance of legitimate reporting (Lazer et al., 2018). A wide range of strategies can be utilized
by fraudsters to create fake news; they include misrepresenting data or recommendations,
presenting fabricated information, or sharing so-called insights or recommendations about a
subject from people who lack the qualifications for offering them (Björnberg et al., 2017).
Though fake news has persisted for decades (Beiler & Kiesler, 2018), its reach and potentially
deleterious influence has been exacerbated by its prevalence on a wide range of social media
platforms, and by its purported role in influencing voters during the 2016 federal election
(Allcott & Gentzkow, 2017; Grinberg et al., 2019; Guess et al., 2018). Recent research, for
example, estimates that the average American adult viewed between one and three fake news
stories on social media in the month leading to the 2016 election (Allcott & Gentzkow, 2017).
And a separate study of Twitter users estimated that fake news accounted for approximately 6%
of total news consumption on the site (Grinberg et al., 2019).
In light of concerns about fake news, researchers and media providers have been
searching for ways to limit its spread and influence (Google News Initiative; Mosseri, 2017;
Roozenbeek & van der Linden, 2019). Many websites now offer advice about how to detect fake
3
Chapter I is adapted from Lutzke, L., Drummond, C., Slovic, P., & Árvai, J. (2019). Priming critical thinking:
Simple interventions limit the influence of fake news about climate change on Facebook. Global environmental
change, 58, 101964. https://doi.org/10.1016/j.gloenvcha.2019.101964
6
news, or to evaluate the credibility of information online (Facebook Help Center; International
Federation of Library Associations and Institutions, 2019; Kiely & Robertson, 2016; Smith; van
der Linden, 2017). Facebook, in its online help center, offers advice for spotting fake news,
including “be skeptical of headlines” and “investigate the source”; the public-facing Psychology
Today offers similar advice. Despite the increasingly widespread availability of this kind of
advice, its effects on consumers of information have not been thoroughly examined. Thus, our
study sought to understand if exposure to these kinds of guidelines for evaluating the credibility
of news could make people less likely to trust, engage with, and share fake news on social media.
The challenges associated with combatting fake news on social media are manifold. We
know, for example, that social media sites are popular sources of news; 47% of Americans report
that they use social media to check the news “sometimes” or “often” with Facebook being the
most popular platform for this purpose (Shearer & Gottfried, 2017). We also know that social
media providers typically do not police the accuracy or the sources of content posted to their
platforms; and, because of how social media functions, information—be it true or false—can be
shared or promoted quickly, easily, and repeatedly. Social media’s capability for speedy
dissemination poses an especially acute challenge regarding false information because it tends to
be shared or promoted more often than accurate information (Vosoughi et al., 2018).
Although fake news can cover any topic, our research focused on false information—in
the form of posts from the Facebook newsfeed—about climate change. We did so because of the
importance of climate change to global environmental, social, and economic affairs. Fake news
about climate change typically states that climate change is not occurring, that it is not caused by
humans, and that it does not pose a threat to humans and the environment (Farrell et al., 2019),
thus perpetuating common and dangerous misconceptions.
7
To evaluate the effectiveness of guidelines in helping to limit the influence and spread of
fake climate news, we asked participants to rate a post’s trustworthiness after reviewing it. To
mimic the choices Facebook users encounter on the platform, participants were also asked how
likely they would be to “like” (which is a reflection of users’ interest in content) and “share”
(which leads to the proliferation of content among Facebook users) the post. Thus,
trustworthiness, liking, and sharing represented the dependent variables in our research.
Prior research has offered complementary theories for why individuals may be
susceptible to fake news. First, the messages imparted by fake news may align with deeply held
political beliefs which, in turn, triggers identity protective cognition. People tend to be motivated
to protect their beliefs from evidence to the contrary and may, therefore, align themselves with
information that confirms what they already believe to be true or right (Kunda, 1990; Nir, 2011).
For example, prior research suggests when people are the recipients of fake news that is in line
with their preexisting beliefs or values, they will be less motivated to engage in critical reflection
about its accuracy (Allcott & Gentzkow, 2017; Taber & Lodge, 2006). Secondly, recent research
has suggested that a general lack of critical thinking—which may be independent of partisan
motivation—is responsible for an individual’s susceptibility to fake news. For example,
controlling for political ideology, Pennycook and Rand (2018) found that individuals who scored
highly on an assessment of analytical reasoning ability were better able to distinguish between
fake and real news headlines.
In light of these perspectives, we were generally skeptical about the ability of mere
exposure to guidelines to inoculate consumers against the effects of fake news; in our view at the
outset, guidelines would not be powerful enough to overcome the partisan tug of motivated
reasoning or the absence of critical thinking that may be common to consuming false information
8
while scrolling through the Facebook News Feed while in cognitive autopilot. Thus, we
hypothesized that people who simply read guidelines for spotting fake news immediately before
being exposed to inaccurate Facebook posts would be no less likely to trust, like, or share them
when compared to a control group that did not receive the guidelines.
We did, however, speculate that encouraging people to more deeply process guidelines
could prove powerful enough to subsequently influence their willingness to trust, like, and share
fake news about climate change. Therefore, we tested a second intervention—which we labeled
Enhanced Guidelines—where participants first received a series of guidelines and then were
asked to rate the importance of each one in terms of helping to determine the credibility of news
received on Facebook. In the same way that attribute weighting tools in research on decision
support help people to make more internally consistent choices (Bessette et al., 2019; Gregory et
al., 2016), we hypothesized that taking the time to rate individual guidelines would help people
to consider them more deeply; this, in turn, would lead people to trust, like, and share fake news
less when compared to a control group.
It is worth noting that a potential challenge associated with attempting to limit the
influence and spread of fake news is a spillover effect whereby interventions aimed at false
information would also limit the influence and spread of accurate information. Thus, in addition
to studying the effect of guidelines and enhanced guidelines on fake news, we also tested them
on real news about climate change.
Methods
Design
Our study adopted a 3 ⨉ 2 experimental design involving two fake news interventions
(Guidelines and Enhanced Guidelines) and a control (no intervention), and two types of news
9
about climate change (fake and real). Participants were randomly assigned to just one of the six
possible experimental variations, and their progression through the experimental design followed
the same sequence of tasks.
Conditions
Participants in the control condition were informed that they would view a Facebook post
about climate change, and then be asked to answer questions about what they saw.
In the Guidelines condition, participants were informed that they would view a Facebook
post about climate change. Next, they were asked to consider a series of four questions (i.e., the
guidelines) that would help them to evaluate the credibility of news online. The questions were:
(1) Do I recognize the news organization that posted the story?; (2) Does the information in the
post seem believable?; (3) Is the post written in a style that I expect from a professional news
organization?; and (4) Is the post politically motivated? These guidelines reflected common
recommendations for identifying fake news (Facebook Help Center; International Federation of
Library Associations and Institutions, 2019; Kiely & Robertson, 2016; Smith; van der Linden,
2017).
In the Enhanced Guidelines condition, participants were also informed that they would
view a Facebook post about climate change, and they were also asked to consider the same four
questions from the Guidelines condition. But, participants in this condition were also asked to
rate the importance of each guideline (on a 1 – 10 scale from not at all important to very
important) in terms of its ability to help them evaluate the credibility of news online.
Fake vs. Real News
10
In total, our study contained six Facebook news posts about climate change; three
contained fake news and the other three contained real news (Figure 1). Participants were
randomly assigned to view one of the six posts.
The three fake news posts were drawn from websites of three different hyper-partisan
media outlets: Breitbart, InfoWars, and Natural News. Each of these outlets is known for
peddling in conspiracy theories and disinformation, and content from is heavily biased in favor
of an ultra-conservative political ideology (Marwick & Lewis 2017). We utilized the search
function on these three websites to identify posts containing false information about climate
change. We confirmed that the Breitbart (Now 400 scientific papers in 2017 say “global
warming” is a myth) and Natural News (NASA confirms sea levels have been falling across the
planet for two years…media silent) contained falsehoods by cross-checking them with an
independent fact-checking database (snopes.com). The post from InfoWars (Al Gore insists
global warming causes global cooling) was confirmed as false by tracing the sources and
information it cited; specifically, we located the blog post shared by Mr. Gore (from The Climate
Reality Project) which explained that climate change may in some cases lead to colder weather,
but not a cooler climate overall.
The three real news posts were drawn from NASA, USA Today, and Scientific
American; these articles were selected because they reflect mainstream climate science, because
they were published by reputable media outlets, and because they mirrored the themes in the
opposing fake news posts: climate change is not a myth, sea levels are rising globally, and
climate change may also lead to cold weather.
Facebook posts are accompanied by tags—which vary by post—including the number of
“likes”, the number of “shares”, and a range of user-selected emojis that convey their emotional
11
reactions to them. To control for their influence, we standardized the number of likes, number of
shares, and types of emojis displayed for all six fake and real news posts (Figure 1). We also
standardized the post date. Aside from these edits, all of the text and the images in the post were
as they appeared in the original stories from the three fake news and three real news outlets.
Measures
After reviewing their randomly assigned post, participants were asked to rate its
trustworthiness on a 10-point scale from 1 (not at all trustworthy) to 10 (very trustworthy).
Participants also rated each post in terms of perceived accuracy on a 10-point scale from 1 (not at
all accurate) to 10 (very accurate). Judged trustworthiness and accuracy were combined to create
a single item index variable for trust (Cronbach’s a = 0.93). Participants were also asked to
indicate their likelihood of “liking” or “sharing” their assigned post; these responses were
collected on 11-point bi-polar scales from -5 (definitely not) to 0 (neutral) to +5 (definitely yes).
These scales were converted to continuous (1-11) scales for analysis.
In terms of covariates, participants were asked if they recognized the source of the
Facebook post, choosing between bipolar (yes or no) response options. We hypothesized that
climate doubters, who tend to be more politically conservative (Hornsey et al., 2016), would be
more likely to trust fake news from sources recognized for having a conservative political
ideology.
Next, we measured participants’ domain-specific knowledge about climate change using
an 11-item
4
scale used in previous research by us and others (Shi et al., 2015; Shi et al., 2016;
Tobler et al., 2012). This scale included three subscales representing three different forms of
knowledge: knowledge about the physics underlying climate change, knowledge about the
4
The twelfth item typically included in the scale (“CO 2 is harmful to plants.”) was unintentionally omitted from the physical
knowledge subscale.
12
reasons that climate change is happening, and knowledge about the different natural hazards and
environmental effects of climate change. We hypothesized that participants who were more
knowledgeable would be more discerning consumers—i.e., less likely to trust, like, and share—
of fake news.
Because of increasing public concern and regulator scrutiny regarding Facebook’s role in
the spread of disinformation as well as the company’s handling of consumer data and privacy
issues, we also asked participants to rate their current attitudes toward the platform; responses
were collected on an 11-point bi-polar scale from -5 (strong negative feelings) to 0 (neutral) to
+5 (strong positive feelings). We speculated that a more positive attitude toward Facebook would
lead people to view all posts as more trustworthy regardless of whether they contained real or
fake news.
Finally, we collected demographic information from participants regarding their gender,
age, education level, and political orientation (measured on a 5-point continuous scale from very
conservative to moderate to very liberal).
Participants
Data collection took place in September 2018 using an online Qualtrics panel. The
instrument was sent to adults over the age of 18 in the United States. Quota sampling was used to
balance gender and belief in anthropogenic climate change; 50% of recruited participants
believed climate change is human-caused (labeled “believers”), and 50% did not or were unsure
(labeled “doubters”).
Initially 4,212 participants responded to our instrument, of which 370 were removed
because they did not complete the study. An additional 1,015 participants were removed from the
sample because they failed an attention check (a multiple choice question which instructed
13
participants to select a particular answer; n = 998 deleted) or because they selected the same
response for every question in a 12-item scale (which was not part of this study; n = 17 deleted).
Other participants were removed because they completed the experiment in less than half the
median time (n = 69 deleted) or because they provided gibberish responses to a series of open-
ended questions (which were not part of this study; n = 8 deleted). This left us with a final
sample of 2,750 participants
5
(Table 1).
Results
With respect to the relationship between our dependent variables, trust in Facebook posts
was relatively strongly correlated with the intention to like it (Cronbach’s a = 0.67) and to share
it (Cronbach’s a = 0.59); the correlation between the intention to like and share a post was even
higher (Cronbach’s a = 0.82).
Table 2 presents mean ratings for our dependent variables by condition for both post
types: fake and real news. We collapsed the three individual fake news (Figure 1: A, B, and C)
and real news (Figure 1: C, D, and E) items for each post type.
A two-way ANOVA detected a significant main effect of post type (F1, 2744 = 710.8, p <
0.001), and a significant interaction between post type and condition (F2, 2744 = 4.53, p = 0.011)
for trust. Here, the effect of condition on trust depended upon whether participants received real
or fake news. We also detected a significant main effect of post type (F1, 2744 = 275.4, p < 0.001)
on “liking”; for this variable, the interaction between post type and condition approached, but
ultimately was not significant (F2, 2744 = 2.7, p < 0.066). For “sharing”, the ANOVA detected a
significant main effect of post type (F1, 2744 = 171.9, p < 0.001), a significant main effect of
condition (F2, 2744 = 3.1, p = 0.047), and a significant interaction between post type and condition
5
Data will be made available in a data repository.
14
(F2, 2744 = 4.53, p = 0.011); thus, participants’ likelihood of sharing a Facebook post was
influenced by both condition and post type. Overall, participants reported significantly lower
ratings for trust, liking, and sharing when confronted with posts based on fake news as compared
to real news, and, being exposed to Guidelines or Enhanced Guidelines had a downward effect
on trust and “sharing”.
Next, we used multiple linear regressions to more thoroughly study the effect of
condition on the dependent measures when controlling for our covariates. We conducted 12
regressions, predicting each of our three dependent measures separately for doubters and
believers who saw fake or real news.
Fake News
Linear regression analyses for the posts based on fake news (Table 3) indicated that
climate change doubters exposed to the Guidelines condition were less likely to trust (η
2
p =
0.011) and like (η
2
p = 0.009) these posts. Doubters in Enhanced Guidelines condition were less
likely to like (η
2
p = 0.012) and share (η
2
p = 0.006) fake news when compared to doubters in the
control condition. Climate change believers exposed to the Guidelines condition were, by
contrast, less likely to share (η
2
p = 0.009) posts based on fake news, while exposure to the
Enhanced Guidelines condition led believers to be less likely to trust (η
2
p = 0.008) and share
(η
2
p = 0.019) fake climate news.
When controlling for the other covariates, participants’ ability to recognize the sources of
posts based on fake news also influenced their responses to the dependent measures (Table 3).
Specifically, participants who were climate change doubters and who recognized either Breitbart
or Natural News as sources of a fake news post were more likely to trust, like, and share the
15
posts. Climate change believers were more likely to trust, like, and share a fake news post if they
recognized Natural News as the source.
Beyond recognizing the source, and when controlling for other covariates, higher levels
of domain-specific knowledge about climate change (Table 3) led believers to report lower levels
of trust, and a lower likelihood of liking and sharing fake news. Higher levels of domain-specific
knowledge had no significant effect on trusting, liking, and sharing amongst climate change
doubters.
For both doubters and believers, the more positive a participant’s attitude toward
Facebook (Table 3), the more likely they were to trust, like, and share a post based on fake news.
And, in terms of political orientation, believers who self-identified as being more conservative
reported higher levels of trust in, and were more likely to like and share, fake news posts (Table
3).
Real News
Linear regression analyses for the posts based on real news (Table 4) revealed that, when
controlling for other covariates, believers of climate change who were exposed to the Guidelines
condition were more likely to trust (η
2
p = 0.017) real climate news. Exposure to the Guidelines
or Enhanced Guidelines conditions had no effect on doubters of climate change.
Participants’ ability to recognize the sources of posts once again influenced their
responses to the dependent measures (Table 4). For both doubters and believers, recognizing the
source of a post based on real news was associated with higher levels of trust, and a greater
likelihood of liking and sharing the post almost half of the time.
When controlling for other covariates, higher levels of domain-specific knowledge about
climate change (Table 4) led doubters to report greater trust in, and a higher likelihood of liking
16
and sharing real news. For believers, higher scores on the scale measuring domain-specific
knowledge led to higher levels of trust in posts based on real climate news.
As was the case with fake news posts, the more positive a doubter’s or believer’s attitude
toward Facebook (Table 4), the more likely they were to trust, like, and share posts based on real
news. And, in terms of political orientation, doubters and believers who self-reported higher
levels of alignment with a conservative ideology were less likely to trust and like real climate
news.
Discussion
This study tested the effect of reading or interacting with guidelines for evaluating the
credibility of Facebook news posts on individuals’ likelihood to trust, like, and share fake and
real news about climate change. It is noteworthy that we detected relatively high correlations
between these three dependent variables. These relationships make sense in that trusting content
on social media would be positively associated with liking it and sharing it; the even higher
correlations between liking and sharing also make sense in that both are a form of online
expression, and in most cases, represent approval of, or praise for, content.
With respect to our independent variable, participants in the Guidelines condition simply
read suggestions for detecting fake news while those in the Enhanced Guidelines condition read
the same guidelines but rated each one in terms of its importance for detecting fake news. A
control group was not exposed to guidelines of any sort. Participants were then asked to evaluate
either a fake or real Facebook news post about climate change. We hypothesized that simply
reading guidelines would not be a powerful enough intervention to influence a person’s
likelihood to trust, like, and share news online. However, we did anticipate that the additional
rating task in the Enhanced Guidelines condition would help participants think more carefully
17
about guidelines and, in turn, lead them to trust, like, and share fake news to a lesser extent
relative to a control. We also predicted that these interventions would not negatively impact real
climate news.
In line with our hypotheses, participants who saw enhanced guidelines were significantly
less likely to trust, like, or share fake climate news. Contrary to our hypothesis, participants who
only read simple guidelines were also less likely to trust, like, or share fake news. Both
conditions had consistently small effect sizes for each dependent variable. Importantly, these
interventions did not lower a participant’s trust, like, and share ratings for real climate news.
Prior research has shown that people with higher analytical thinking abilities are better
able to recognize fake news. Pennycook and Rand (2018), for example, show that individuals—
independent of political ideology—who score highly on a modified version of the Cognitive
Reflection Test (CRT; see Frederick, 2005) as well as a non-numeric version of the CRT
(Thomson & Oppenheimer, 2016) were better able to distinguish between fake and real news
headlines. Furthermore, Bronstein et al. (2019) also used the same two sets of CRT questions to
measure the association between analytical thinking and assessment of fake news, and found that
higher analytical reasoning scores were positively correlated with the ability to discern between
real and fake news.
Our results seem to support these findings in that critical thinking may indeed play an
important role in the evaluation of fake news. However, while prior research studied critical
thinking ability as a covariate, our research is novel for treating it as a treatment effect. In other
words, our research did not focus on preexisting ability, but instead relied upon simple
interventions to prime critical thinking on the part of participants. The effectiveness of these
primes is further supported by the observation that, when we control for education level and
18
domain-specific knowledge about climate change, participants exposed to the Guidelines or
Enhanced Guidelines treatment were still less likely to trust, like, and share fake news about
climate change (Table 3).
In addition to critical thinking, our results suggest that motivated reasoning also
contributes to a person’s evaluation of fake news. We know that, when confronted with
information that is inconsistent with deeply held beliefs or ideological viewpoints, people are
often motivated to reject it in favor of information that is more closely aligned with their
preexisting beliefs (Taber and Lodge 2006). Along these lines, our results show that the more
politically conservative a participant was, the more likely they were to trust fake climate news
and mistrust real climate news. But, despite the powerful effect of motivated reasoning, our
interventions led doubters of climate change to trust, like, and share fake climate news to a lesser
degree.
Independent of our interventions, our covariates point to other factors (knowledge of
climate change and attitudes toward Facebook) that also influenced a person’s likelihood to trust,
like, and share fake climate news. For example, it is noteworthy that climate change believers
with higher levels of domain-specific knowledge about climate change were even less likely to
trust, like, and share fake climate news when compared to climate change believers with lower
levels of domain-specific knowledge. And, climate change doubters with greater domain-specific
knowledge were more likely to trust, like, and share real climate news (compared to other
climate doubters). Although there are many recent examples where objective facts related to
politicized issues are discounted or ignored (Beck, 2017)—even amongst those that do not
believe in, or who are unsure about, anthropogenic climate change—a high level understanding
of relevant facts can be influential even when these facts conflict with a person’s prior beliefs.
19
These results point to the critical importance of continuing to educate the public about climate
change.
In addition, a positive disposition toward Facebook—regardless of whether a participant
was a doubter of or believer in climate change, or whether they were exposed to real or fake
news—had a significant, positive, and consistent effect on trusting, liking, and sharing content.
This observation highlights an important challenge when it comes to preventing the spread of
fake news on social media; that is, users who are positively disposed toward a social media
platform may not be as critical of information they encounter on the site. Since Facebook (as
well as other social media platforms like Twitter) does not moderate posts for accuracy, critical
evaluation of posts is essential for consumers wishing to make more accurate judgments about
the credibility of news online
6
. Thus, the mechanism behind the association between a positive
disposition toward Facebook and a potential reduction in critical thinking warrants additional
research. For now, these results suggest that simple—and, possibly, more elaborate—
interventions will likely be less effective for users who are extremely fond of their chosen social
media platform.
Limitations
Our results, while statistically significant, were associated with small effect sizes. For
example, we observed a partial eta-squared (ηp²) that was between 0.006 to 0.019 for each effect
of our interventions on a participant’s likelihood to trust, like, or share climate news on
Facebook. However, the fact that Facebook sees over 1.5 billion active users per day, coupled
with the small political margins that seem to be increasingly pervasive in American politics
(Smidt, 2017)—e.g., the approximately 107,000 votes required in 2016 to tip the electoral
6
At the same time, an even more effective step would be the enactment of legislation that requires social media
platforms that double as news sites to factcheck and moderate posts for accuracy.
20
college in favor of the sitting U.S. president accounted for only 0.0008% of the total number of
ballots cast—we would argue that these small effects are still practically meaningful. For similar
reasons, interventions that influence only a small fraction of social media users may nevertheless
be important and impactful given the exponential rate at which information can be shared on
social media. Since false information can quickly go “viral”, preventing even a small number of
people from sharing fake news has the potential to prevent an exponentially larger number of
others from seeing and sharing the same content (Dow et al., 2013).
The composition of our sample may also be a limitation of our study in that it may not
accurately reflect those who are most likely to spread fake news. For example, prior research
(Guess et al., 2019) found that adults over the age of 65 are more likely to share fake news when
compared to younger internet users. Although our research included 313 participants over the
age of 65, we did not detect an association between age and trusting, liking, or sharing fake
news. Future research on interventions designed to stem the spread of fake news may benefit
from oversampling older adults.
Other limitations of our research are related to our study’s design. Specifically, the
window of time between exposure to our interventions and exposure to fake (and real) news was
short, and, our intervention was only tested on one post per participant. Thus, we should not
assume that priming critical thinking once would be effective in light of repeated exposure to
fake news. We believe that our understanding of interventions aimed at preventing the
proliferation and influence of fake news would be enhanced by future research that adopts a
longitudinal design, offers more than one opportunity for critical thinking, and presents
participants with multiple news stimuli.
21
Likewise, our study utilized short fake news posts, which we selected to mimic the rapid-
fire nature of viewing the Facebook News Feed. Future research should focus on the
effectiveness of interventions for helping people to detect disinformation in longer and more
detailed fake news stories. And, since fake news is a problem that transcends the topic of climate
change, future research should also focus on the generalizability of interventions that target fake
news across a wide range of socially, economically, and environmentally important subjects.
Conclusion
This study highlights the potential of simple interventions that prime critical thinking and
slow the spread of fake news about climate change on social media platforms. However, several
challenges—e.g., motivated reasoning, a strongly conservative political ideology, and low levels
of domain-specific knowledge about climate change—continue to stand in the way of
interventions designed to address the problem. Thus, a multiplex of approaches (rolled out in
collaboration with social media providers) will be necessary to effectively combat the problems
posed to society by fake news; chief among them are efforts to both improve the critical thinking
abilities of people who rely on social media for their news and educate people about climate
change.
22
Chapter II
Consumer acceptance of products from carbon capture and utilization
7
Deploying carbon capture technology has been highlighted as one element of a climate
mitigation strategy that aims to limit global warming to 2°C. Specifically, carbon capture and
utilization (CCU) has the capability to contribute to carbon dioxide removal by capturing CO2
directly from ambient air or from point sources (e.g., a smokestack), and then utilizing that CO2
(or the carbon from it) to create new materials, including consumer goods (Koytsoumpa et al.,
2018; Markewitz et al., 2012). This process has the potential to sequester carbon in long-lasting
materials such as polycarbonates and concrete (Baena-Moreno et al., 2019; CO2 Sciences and
The Global CO2 Initiative, 2016; Jang et al., 2016). And, CCU has the additional benefit of
displacing the extraction of virgin carbon-based resources, in that whenever captured carbon is
utilized, other natural resources (such as hydrocarbons) that would typically be used to make the
same products are required to a lesser extent (Quadrelli et al., 2015).
However, an important limitation of the technology lies in its narrow capacity to
sequester carbon in products at a meaningfully large scale; in order to meaningfully address the
risks associated with climate change, the amount of CO2 that would need to be removed from the
atmosphere far exceeds the demand for products that could be derived from captured carbon
(Mac Dowell et al., 2017; Quadrelli et al., 2015). Thus, a major benefit of CCU comes from its
potential to financially support carbon capture and storage, which is the process of capturing and
then storing carbon dioxide underground in deep saline formations or depleted oil reservoirs.
Specifically, creating products from what would otherwise be a CO2 waste stream may add value
to the carbon capture and storage (CCS) process (Peters et al., 2011). This connection is
7
Chapter II is adapted from Lutzke, L., & Árvai, J. (2021). Consumer acceptance of products from carbon capture
and utilization. Climatic Change, 166(1), 1-20. https://doi.org/10.1007/s10584-021-03110-3
23
significant because CCS has the potential to sequester far more CO2 than CCU (Global CCS
Institute, 2019). That said, it is important to emphasize that CCS and CCU are distinct, in that
once carbon dioxide has been captured, storage versus utilization are alternative paths.
Therefore, it cannot be assumed that CCU will certainly benefit CCS through the relationship
described, as a CCU project may operate without any ties to storing CO2 underground.
Both CCU and CCS have received their share of criticisms. Large energy requirements,
high operating costs, and the potential to distract from the development of renewables, have been
highlighted as concerns for both technologies (Nisbet, 2019). However, despite these concerns,
the huge excess of CO2 already in the atmosphere points to the need to address historical
emissions (IPCC, 2018). And while the prevention of future emissions (e.g., through the
transition to renewables) is more than essential, carbon capture technology will also likely play a
part in attempting to limit global temperatures.
With this need in mind, researchers (e.g., at the Global CO2 Initiative at the University of
Michigan and the Carbon Capture, Utilization, and Storage Center at MIT) and companies (e.g.,
Climeworks in Switzerland, Carbon Engineering in Canada, and Global Thermostat in the United
States) are working to advance CCU technology (and related CCU-based products) such that it
can be deployed on a global scale. However, as work on overcoming the current technical
barriers to CCU continues, it is equally important to also consider potential social barriers to the
technology (Jones, 2015), chief among them a potential lack of consumer acceptance of products
made from captured carbon. Without broad consumer acceptance, there may not be a reliable
end-market for CCU-based products, hindering the economic viability—and the mitigative
benefits from a climate change perspective—of CCU.
Current research on CCU acceptance
24
Researchers have begun to survey public opinions regarding CCU, and studies conducted
in Germany, the UK, and China have generally shown low initial awareness of CCU among
participants, but then an openness toward the technology after an introduction (K. Arning et al.,
2019; Katrin Arning et al., 2019; Jones et al., 2017; Li et al., 2017; Linzenich et al., 2019; Perdan
et al., 2017). In addition, themes have emerged related to potential factors which either bolstered
or hindered acceptance of CCU. Most commonly, these studies found that participants, on
average, expect CCU will likely have a beneficial effect on the climate (K. Arning et al., 2019;
Katrin Arning et al., 2019; Li et al., 2017; Perdan et al., 2017). Further, perceiving climate
benefits of CCU was indeed found to predict acceptance of the technology generally (K. Arning
et al., 2019; Katrin Arning et al., 2019).
At the same time, several studies have revealed that risks were perceived in relation to
CCU by some participants (K. Arning et al., 2019; Katrin Arning et al., 2019; Jones et al., 2017;
Li et al., 2017; Linzenich et al., 2019; Perdan et al., 2017) which, in turn, have been found to be
negatively associated with CCU acceptance (K. Arning et al., 2019). Risk perceptions ranged
from general concerns that CCU posed unspecified environmental or personal health risks
(Perdan et al., 2017), to specific worries that developing CCU will delay investments in other
sustainable technologies (K. Arning et al., 2019). Regarding the later concern, multiple studies
found that at least some participants believed CCU may interfere with other sustainability and
climate mitigation initiatives (K. Arning et al., 2019; Katrin Arning et al., 2019; Jones et al.,
2017; Perdan et al., 2017). However, the relationship between this belief and acceptance of CCU
is unclear. For example, in qualitative interviews conducted by Jones et al. (2017), some
participants suggested that CCU may enable the continued use of fossil fuels, and expressed this
as an undesirable aspect of CCU, which would seemingly weaken support for the technology.
25
Yet, Katrin Arning et al. (2019) measured perceptions of similar sustainability risks (e.g., “CCU
only delays the problem of increasing emissions”) and found that agreement with such
statements was not predictive of general risk perceptions or acceptance of CCU.
Current research on perceptions of CCU-based products
Despite a raft of prior research on the acceptance of CCU generally, relatively little is
known about consumers’ opinions regarding the products that can be derived from CCU.
Qualitative data from some of the research cited above do, however, offer some preliminary
insights (Jones et al., 2017). For example, when respondents were provided with a diagram
highlighting a wide range of end-uses for captured carbon—e.g., plastics, liquid fuels,
fertilizers—some participants noted that products would differ in their ability to store carbon
(i.e., some products are more permanent than others), and that plastic as an end-use may seem
contrary to environmental goals. However, respondents also acknowledged the beneficial role
CCU could play in reducing consumption of raw materials and encouraging a circular economy.
From the perspective of a consumer, some participants noted that a higher cost for these products
could be acceptable in return for a more eco-friendly purchase. But, others questioned the safety
of the products, wondering if unknown risks could be associated with a “waste material”
touching the skin (e.g., through CCU-based plastic in garments).
Interviews conducted by van Heek et al. (2017a) provide additional detail as to the health
risks consumers may perceive. While respondents reported a generally positive outlook on CCU,
they also voiced concern about health impacts which they believed could be brought on by close
interaction with a CCU-based product. Concerns about allergies, difficulty breathing, and the
related risk of suffocation, were among those highlighted by respondents; the perceived
mechanism behind these worries being the belief that CO2 may leak from CCU-based products.
26
This belief was incorrect as the products in question were solid objects, rather than liquid
products (i.e., carbonated beverages) where CO2 would, in fact, escape.
Taken together, these (Jones et al., 2017; van Heek et al., 2017a) qualitative studies seem
to point to similarities between the risk and benefit perceptions that emerge when participants
consider CCU-based products specifically, as the ones that emerged when participants
considered CCU technology on the whole.
Aside from Jones et al. (2017) and van Heek et al. (2017a), current research on the
subject of CCU-based products mainly consists of studies which sought to gauge perceptions of a
CCU-based polyurethane foam mattress (made with carbon captured from industrial point
sources, not from carbon captured from ambient air) among German consumers (Arning et al.,
2018; van Heek et al., 2017b). Unsurprisingly, factors similar to those found to predict general
acceptance of CCU were also found to predict acceptance of mattresses. For example, an online
survey conducted by van Heek et al. (2017b) identified perceived health risks of using CCU-
based products (e.g., concern contact with products will irritate the skin) and environmental
benefits (e.g., capacity to conserve fossil resources) as traits which influenced satisfaction with
the CCU-based mattress.
A separate online survey by Arning et al. (2018) identified participant characteristics and
beliefs which influenced acceptance of a CCU-based mattress. Specifically, three groups of
consumers were identified within their sample, the first being participants that were cautiously
open to using a CCU-based mattress. These participants were more climate focused—reporting
higher levels of engagement in environmentally conscious behavior (e.g., using public transit)
and acknowledging the potential climate benefits of CCU. Alternatively, a smaller group rejected
the technology as they did not see climate benefits in CCU, and instead indicated greater concern
27
for sustainability risks (e.g., distracting from renewables). Finally, the group most accepting of
mattresses had a more technical professional background and may have been attracted to the
innovativeness of the technology. This would be consistent with prior work which found that
many participants positively rated CCU technology as innovative (Linzenich et al., 2019).
Overall, in line with findings regarding positive public opinions of CCU, these studies found
participants to be generally open to CCU-based products, with additional evidence for the
acceptance of a CCU-based mattress, as many participants indicated an intention or willingness
to buy and use the product (Arning et al., 2018; van Heek et al., 2017b).
Gaps in current knowledge of CCU-based product acceptance
Although these prior studies are important for advancing our understanding of
perceptions surrounding CCU, questions remain. Specifically, with the majority of relevant
research having been conducted in Europe, how might consumers in the U.S. respond to CCU-
based products? While Europe indeed represents an important market for CCU-based products,
the level of concern about climate change across the region is elevated and driven by different
cultural variables when compared to the United States (Lorenzoni & Pidgeon, 2006; Shi et al.,
2016). This greater concern about climate change, paired with the fact that capturing carbon is a
process tied to the mitigation of climate change, suggests that acceptance as measured in a
European context may not be a reliable proxy for consumer perceptions in the U.S.
A second question relates to consumer perceptions of additional CCU-based products,
alongside the foam mattress which has been studied in depth. Due to the mattress’s relatively
high level of technological feasibility in an area of materials science that is just beginning to
emerge, the reasoning for initially focusing on perceptions of a CCU-based foam mattress is
clear. However, given the wide range of products that could be manufactured from captured
28
carbon, measuring perceptions of additional consumer goods in depth would be of additional
value. Specifically, the ability to make comparisons between products could serve as a guide in
prioritizing the manufacturing of some CCU-based products over others, as well as highlight
products which may need stronger marketing or communication strategies to garner consumer
acceptance.
Third, it is not known how the method of carbon capture may influence consumer
perceptions of products made from CCU. Currently, CO2 is most frequently captured from point
sources like industrial smokestacks (Olajire, 2010), yet, the technology to capture CO2 from
ambient air via Direct Air Capture (DAC) is also becoming commercially available. Indeed,
DAC may become more prevalent as one of the strategies for mitigating the risks associated with
climate change (Breyer et al., 2019). Despite the projected role of DAC, other studies on public
or consumer perceptions of CCU-based products have not explicitly examined differences based
on capture method, and most have provided participants with introductions only to carbon
captured from point sources.
Aim of present study
With this as backdrop, the intent of this study was to expand upon our existing
knowledge by measuring acceptance levels for a wide range of products manufactured from
carbon captured through either Direct Air Capture, or from point sources, among U.S.
consumers. Specifically, we asked participants in this study if they would be willing to use—or,
in one case, consume—one of four CCU-based products, each created through one of two
capture methods.
The four products presented to participants included carbonated beverages, plastic food
storage containers, furniture made with foam or plastic, and shatterproof glass (i.e., glass that is
29
sandwiched between thin layers of polycarbonate plastic). These products were chosen to
represent an array of CCU-based products likely to become commercially available soon. They
were also selected to account for another characteristic that might affect product acceptance, that
being, the degree to which a consumer must physically interact with the product in order to use
it. These varying degrees were represented through products meant to be consumed (carbonated
beverages), products meant to hold items to be consumed (plastic food storage containers),
products one must touch to use (furniture made with foam or plastic), and products that are
seldom the subject of physical contact (shatterproof glass). With participants in multiple
subsequent studies displaying hesitancy toward physical contact with CCU-based products
(Arning et al., 2018; Jones et al., 2017; van Heek et al., 2017a, 2017b), we hypothesized that, in
this study, the more interaction a type of product requires in order to be used, the less accepting
consumers will be of that product. Thus, we anticipated shatterproof glass would be most
accepted, followed by furniture made with foam and plastic, then plastic food storage containers,
and finally carbonated beverages.
In addition to product type, there is evidence that the origin of the captured carbon might
influence consumer perceptions. For example, in a study by Duetschke et al. (2014), participants
viewed CCS less favorably when carbon was captured from a coal fired power plant as opposed
to from a biomass or industrial facility, possibly because of negative consumer perceptions of
coal as an energy source. Therefore, in this current study, we hypothesized that capture method
would influence acceptance, in that consumers would be more willing to consume or use
products created with carbon captured directly from ambient air (via DAC) versus carbon
captured from a point source (i.e., carbon captured from factories and power plants). We
30
hypothesized further that this would especially be the case for products that people consume, and
for products that are used to store food and beverages.
Apart from exploring characteristics inherent to CCU-based products (i.e., product type
and capture method), another aim of this research is to continue to examine how a variety of
consumer characteristics and perceptions may impact a person’s willingness to consume or use a
CCU-based product.
Perceptions related to risks and benefits have been highlighted as especially important in
the context of acceptance of CCU technology and its resulting consumer goods. As discussed,
multiple studies have found that perceiving climate benefits increases acceptance, and perceiving
health risks decreases acceptance (K. Arning et al., 2019; Katrin Arning et al., 2019; van Heek et
al., 2017b). We expect to find the same pattern here, in that perceiving greater climate change
mitigation benefits of CCU-based products will increase product acceptance, while believing that
consuming or using a CCU-based product poses a risk to personal health will decrease product
acceptance.
In addition to perceived health risks, participants in other studies have perceived
sustainability risks (e.g., that CCU will enable the continued use of fossil fuels) to be a concern
as well. These types of risks are also often referred to as a moral hazard, or, the idea that
technological “solutions” to climate change may be perceived as insurance, thus emboldening
people to continue with risky environmental practices (Campbell-Arvai et al., 2017). Because
perceiving sustainability risks has been associated with less acceptance of CCU-based products
for at least some participants in past work (Arning et al., 2018), we hypothesized that a high level
of concern about moral hazard as it pertains to CCU would dampen consumer acceptance
regarding CCU-based products.
31
Often related to risk and benefit perceptions is affect, a concept which consists of the
intuitive emotional responses people experience when exposed to stimuli (Finucane et al., 2000).
Here, measuring affect will likely offer additional insight in capturing the emotion behind
participants’ beliefs related to health risks and climate benefits of CCU-based products.
Furthermore, several studies demonstrate that affect may influence consumer decision-making,
in that people who experience positive emotional responses (e.g., optimism, happiness, etc.) to a
product are more likely to purchase or use it; for negative emotional responses, the opposite is
true (Etale et al., 2018; Segrè Cohen et al., 2020; Siegrist et al., 2007). Thus, we expected that
participants who felt mostly afraid or worried in response to CCU-based products would be less
willing to consume or use them.
Along with consumer perceptions of CCU-based products, individual characteristics of
participants—e.g., their knowledge, beliefs, and past behaviors—likely will also influence a
person’s willingness to consume or use new products. For example, both the tendency to partake
in environmentally conscious behavior and a proclivity toward technology have been associated
with greater acceptance of CCU (Arning et al., 2018). In this study, we anticipated that those
who tend to perform environmentally-friendly actions, and those who exhibit technological
optimism (characterized by a high level of trust in the potential for technology to address the
world’s problems, as well as limited concern regarding risks associated with new technologies
(Achterberg, 2014)) will be more accepting of products derived from captured carbon.
Likewise, a participant’s domain-specific knowledge about climate change may also impact
acceptance of CCU-based products. For example, a high level of understanding about the risks of
climate change may make the need for mitigation seem more urgent. In addition, possessing
knowledge specifically related to the properties of the gas CO2 may contribute to a better
32
understanding of the CCU process, as well as the characteristics of the resulting products (e.g.,
that CO2 will not “leak” out of solid objects). Therefore, we hypothesized that participants with
higher levels of domain-specific knowledge regarding both climate change and CO2 would be
more willing to consume or use CCU-based products.
Finally, we also deemed a participant’s political ideology to be a likely predictor of CCU-
based product acceptance. Because the development of carbon capture technology is motivated
by environmental concerns, combined with the fact that climate change remains a politically
polarizing issue in the U.S. (Chinn et al., 2020), we hypothesized that more politically
conservative consumers would be less accepting of CCU-based products.
Methods
This research was determined exempt by the Health Sciences and Behavioral Sciences
Institutional Review Board (protocol number HUM00167873) at the University of Michigan.
Design
This study adopted a 2 ⨉ 4 experimental design where participants were provided with
information about one of two carbon capture methods (DAC or point source capture), and one of
four types of CCU-based products (carbonated beverages, plastic food storage containers,
furniture made with foam or plastic, or shatterproof glass). Participants were assigned to one of
these eight possible experimental treatments.
After providing their informed consent, participants were first asked about their prior
knowledge of carbon capture and utilization technology. Responses were provided on a 3-point
scale: No; Yes, but I don’t fully understand what it is; and Yes, and I know what it is.
Next, participants read an introductory vignette explaining how CCU uses technology to
create products using captured carbon. A total of eight vignettes were developed. Four of them,
33
provided to approximately half of the participants (see section 2.2, below), included an
introductory section that outlined the DAC process; the other four outlined the process of CO2
capture at an emissions site (i.e., point source capture). The remainder of each vignette outlined
how the captured carbon was used in the manufacture of one of the four product types noted
above. To assist with comprehension of the CCU process, each vignette was accompanied by a
simple schematic that visually represented the main steps involved in capturing CO2 and then
using the captured carbon in the manufacture of their assigned product. (See the Supplemental
Materials section for each introductory vignette and schematic).
Participants then responded to a question aimed at the study’s main dependent variable:
willingness to consume or use. For those in the carbonated beverage condition, participants were
asked to respond to the following statement: I would personally consume carbonated beverages
made using CCU technology in my daily life. Responses were recorded on a Likert scale
(1=definitely no, 4=not sure, and 7=definitely yes). Using the same response scale, participants
in the other three product conditions responded to a similar statement: I would personally use
[plastic food storage containers; furniture made with foam or plastic; shatterproof glass] made
using CCU technology in my home.
In addition to our dependent variable, we collected data across a series of covariates (see
Table S1 for all covariate survey items). First, we asked participants if they thought consuming
or using a CCU-based product would pose a risk to their health. Second, we asked participants if
they believed making products using CCU technology would help to slow or stop the negative
effects of climate change. Responses were collected on Likert scales for both covariates
(1=definitely no, 4=not sure, and 7=definitely yes). For the later question, an eighth response
34
option allowed participants to opt-out of answering by indicating that they do not believe in
climate change.
Next, participants responded to a series of questions aimed at their affective responses to
CCU. Specifically, we asked participants about the degree to which CCU technology made them
feel enthusiastic, afraid, interested, or worried. Responses were recorded on Likert scales (1=not
at all, 4=somewhat, and 7=very). The four measures were combined to create an index, with
“enthusiastic” and “interested” reverse coded (Cronbach’s α = 0.77). We labeled this index
“negative affect”.
Participants then indicated their level of concern about possible moral hazard when
deploying CCU through their responses to three statements that were adapted from prior research
(L’Orange Seigo et al., 2014). (See Table S1 for survey items.) Responses were recorded on
Likert scales (1=completely disagree, 4=unsure, and 7=completely agree). Internal reliability
was lower than anticipated (Cronbach’s α = 0.39), however this scale was still included in our
analyses as these (and similar) questions have been used by others (K. Arning et al., 2019; Katrin
Arning et al., 2019; Arning et al., 2018; L’Orange Seigo et al., 2014). These questions were not
shown to participants who indicated that they do not believe in climate change.
Each participant also responded to a series of questions aimed at ascertaining their
specific knowledge about climate change and, separately, their specific knowledge about CO2;
both scales were adapted from prior research by one of this paper’s authors (see L’Orange Seigo
et al., 2014). In each, participants responded to a set of true or false statements. With respect to
participants’ knowledge about CO2, participants responded to seven statements, three of which
were added to the scale by us (see Table S1 for survey items).
35
Based on a validated scale developed by Arning et al. (2018), we then measured
participants’ behavioral tendencies regarding environmentally conscious behavior via 3-items
(Cronbach’s α = 0.68). Participants responded to items on Likert scales (1=never, 4=sometimes,
and 7=always). We also used a 4-item scale adapted from Achterberg (2014) to gauge
participants’ technological optimism (Cronbach’s α = 0.65). Responses were again recorded on
Likert scales (1=completely disagree, 4=unsure, and 7=completely agree).
Finally, we collected demographic information from each participant including their
political ideology, level of education, gender, and age.
Participants
Data collection took place in August 2019 using an online U.S.-based panel from Lucid
®
(which is the largest panel provider in the United States). Participants in this study were adults
over the age of 18. Quota sampling was used to obtain a sample that was approximately
nationally representative for political ideology, education level, gender, and age. A total of 3,887
participants initially responded to our survey, of which 74 did not provide consent. An additional
26 participants did not live in the United States and were removed from the sample. Other
participants were removed because they completed the experiment in less than half the median
time (2m:59s; n = 163 deleted) or because they failed a multiple-choice attention check which
required them to correctly recall the name of the technology (carbon capture and utilization) that
they had read about on the preceding page (n = 1,224 deleted)
8
. Although the number of
participants that failed the attention check is large, it is not uncommon among studies that sample
from the Lucid
®
panel (Aronow et al., 2020). This left us with a final sample of 2,400
participants. The demographic makeup of participants by treatment group was equivalent in that
8
Incorrect response options to the multiple-choice attention check included: Carbon Extraction and Purification;
Raw Material Conversion; Atmospheric Correction and Stabilization; or None of the above.
36
one-way ANOVAs found no significant differences between participants in the eight conditions
for gender (F7, 2392 = 0.28, p = 0.96), age (F7, 2392 = 0.51, p = 0.83), education (F7, 2392 = 1.42, p =
0.19), and political ideology (F7, 2392 = 0.72, p = 0.65)(Table 1). Data have been deposited in
Open Science Framework (https://osf.io/y597w/).
Results and discussion
Knowledge of CCU
The vast majority of participants were not familiar with CCU upon the start of this study.
A total of 1,591 participants (approximately 66% of our sample) did not know what CCU was,
while 715 participants (approximately 30% of our sample) had heard of the technology but did
not fully understand it. Only 94 participants (approximately 4% of our sample) had heard of
CCU and self-reported that they understood it. For those familiar with CCU, average willingness
to consume or use a CCU-based product was 5.7 (SD = 1.4) on a 7-point Likert scale. Averages
were 5.3 (SD = 1.4) and 4.9 (SD = 1.4) for those somewhat familiar and those unfamiliar,
respectively and an exploratory one-way ANOVA was conducted. A significant difference
between groups (F2, 2397 = 30.32, p < 0.001) was detected and a post-hoc Tukey HSD found that
those who were familiar with (p < 0.001), or were somewhat familiar with (p < 0.001), CCU,
were more accepting of CCU-based products than those who were unfamiliar with CCU.
Findings related to knowledge of CCU are consistent with a host of other studies which also
found awareness of the technology to be low (K. Arning et al., 2019; Katrin Arning et al., 2019;
Jones et al., 2017; Li et al., 2017; Linzenich et al., 2019; Perdan et al., 2017). Further, our
exploratory results are consistent with Perdan et al. (2017) which found that those with more
awareness of CCU technology were more likely to favor development of CCU in the UK.
3.2 Acceptance of CCU-based products
37
After reading an introduction to CCU, a majority of participants indicated an openness to
consume or use CCU-based products. Collapsing across all conditions (i.e., capture method and
product type), we found that approximately 69% of participants (n = 1,645) indicated that they
were likely to consume or use a CCU-based product; these participants gave responses that
ranged from “probably yes” to “definitely yes” (Figure 1). Similarly, approximately 69% of
participants (n = 1,659) believed that manufacturing products using CCU technology would
likely help to mitigate climate change; these respondents selected “probably yes”, “yes”, or
“definitely yes”. Personal health risks of using CCU-based products were only perceived by
approximately 13% of participants (n = 320, providing responses that ranged from “probably
yes” to “definitely yes”).
Thus, despite the political and cultural differences between the United States and Europe
(where research on CCU-based products has been conducted), these results are congruent with
prior work in that this study also finds consumers to be generally open to CCU-based products,
with many acknowledging CCU’s potential climate benefits, and some indicating concerns about
personal health risks (Arning et al., 2018; Jones et al., 2017; van Heek et al., 2017a, 2017b).
Comparisons of CCU-based products
Next, a two-way ANOVA was conducted to explore the impacts of both product type
(four-level factor) and capture method (two-level factor) on willingness to consume or use.
Because capture method could impact willingness to consume or use differently depending on
the type of product (e.g., point source capture could decrease acceptance for carbonated
beverages but increase acceptance for shatterproof glass), an interaction term was included. The
model did in fact find a significant interaction between product type and capture method (F3, 2392
= 3.59, p = 0.013), meaning, to some extent, both product type and capture method had an effect
38
on willingness to consume or use. In addition, the interaction reveals that for at least one type of
product, acceptance depended on whether the carbon was captured via DAC or from a point
source. To uncover the differences between the CCU-based products that drove the significant
interaction, a one-way ANOVA with eight levels (consisting of four product types by two
capture methods), was conducted. The ANOVA indicated at least one difference between groups
(F7, 2392 = 16.62, p < 0.001) in terms of willingness to consume or use, and a post-hoc Tukey
HSD analysis revealed where differences were present (Figure 2).
Contrary to our hypothesis, regardless of capture method, no significant differences were
found between plastic food storage containers, furniture made with foam or plastic, and
shatterproof glass. However, in line with our hypothesis, capture method did affect willingness to
consume carbonated beverages, in that participants were significantly less likely to indicate a
willingness to consume if the beverage contained carbon captured from point sources versus
DAC. In addition, participants were generally less accepting of carbonated beverages compared
to other products. Specifically, participants were less accepting of carbonated beverages from
point source capture compared to all other product combinations (i.e., product type ⨉ capture
method). And, participants were less accepting of carbonated beverages from DAC compared to
plastic food storage containers and shatterproof glass (but not compared to furniture made with
foam or plastic, where no difference was found).
We anticipated that the degree to which a product requires interaction in order to be used
would play a larger role in influencing willingness to consume or use a CCU-based product. In
prior studies, participants raised concerns about using CCU-based products, often driven by
worries that the products could irritate or harm the skin (Jones et al., 2017; van Heek et al.,
2017a, 2017b). Here, we did not find evidence of that concern, as no significant differences in
39
acceptance were found between shatterproof glass (which would not commonly come in contact
with skin) and furniture or plastic containers (which would often come in contact with skin).
Instead, whether or not a product is intended for consumption seemed to be the main factor
driving differences in acceptance between products. This finding shares similarities with
consumer perceptions of recycled water and genetically modified food, where products meant to
be consumed are less accepted due to unsupported concerns regarding contamination or the
naturalness of the product (Nemeroff & Rozin, 2018).
This extra sensitivity toward products meant to be consumed may help to explain why
capture method influenced acceptance for carbonated beverages, but not other products in this
study. In support of this idea, prior research that examined public perceptions of DAC (in
relation to CO2 storage, not utilization) noted that participants were less concerned about the
chemical processes involved in DAC than researchers expected (Cox et al., 2020). Thus, we
speculate that if participants in our study were also not very concerned about the chemical
processes involved in DAC, the influence of capture method on acceptance of carbonated
beverages could be due to concerns about contamination being more felt by participants in our
point source conditions (if they, for example, associate contamination with the industrial
processes that occur at factories and power plants).
Predictors of product acceptance
Next, we conducted a regression analysis to highlight additional variables that may
impact willingness to consume or use a CCU-based product, including negative affect, perceived
risk to health, perceived benefit to climate change mitigation, environmentally conscious
behavior, concern of moral hazard, technological optimism, knowledge of climate change,
knowledge of carbon dioxide, political ideology, level of education, gender, and age. This
40
analysis collapses the data across all product types and both capture methods. A total of 52
participants that indicated that they did not believe in climate change were excluded from this
analysis because they did not answer questions related to moral hazard or the potential climate
change mitigation benefits of CCU. Variables were included in the regression model after no
multicollinearity was found. The strongest associations between predictor variables were
between negative affect and perceived risk to health (r = 0.53, p < 0.001), and negative affect
and perceived benefit to climate (r = -0.53, p < 0.001) (see Table S2 for correlation matrix, p-
values were adjusted for multiple comparisons).
With willingness to consume or use as the main dependent variable (Table 2) the
regression revealed that as participants’ negative affective responses related to CCU increased,
their reported willingness to consume or use decreased significantly (p < 0.001). (Increased
perception of personal health risks related to CCU exhibited the same pattern; p < 0.001.)
Conversely, higher perceived benefit regarding CCU’s potential to mitigate the negative
consequences of climate change led participants to be more willing to consume or use CCU-
based products (p < 0.001). Similarly, the more participants displayed environmentally conscious
behaviors in other domains, the more willing they were to consume or use CCU-based products
(p < 0.01). The same was true of possessing a high level of technological optimism; the more a
participant trusted technology, the more willing they were to consume or use CCU-based
products (p < 0.01). These findings were all in line with our hypotheses.
Unexpectedly, when controlling for the other covariates, participants with higher levels
of climate change knowledge were less likely to consume or use CCU-based products (p < 0.05).
Also conflicting with our hypotheses, neither participants’ perception of moral hazard, nor their
41
domain-specific knowledge of CO2, were significant predictors of willingness to consume or use
a CCU-based product. The same was true of political ideology.
Neither education level, gender, nor age were predictors of willingness to consume or
use.
The results from the regression model (Table 2) largely mirror results found when
examining the uncontrolled relationships between acceptance of CCU-based products and the
other variables (Table S2). However, several variables did not show a consistent pattern. For
example, without controlling for other variables, domain-specific knowledge of both climate
change (r = 0.21, p < 0.001) and carbon dioxide (r = 0.18, p < 0.001) were found to have
positive relationships with the dependent variable (contrary to findings in the regression). In
addition, perception of moral hazard was found to have a slight negative relationship with
acceptance of a CCU-based product (r = -0.11, p < 0.001), as was gender (women were less
accepting; r = -0.07, p < 0.05).
Overall, we see several similarities between our findings here and those found in prior
studies on acceptance of CCU-based products. Specifically, perceiving CCU-based products to
be potentially risky to personal health predicted less consumer acceptance, while perceiving the
products to have environmental benefits predicted more acceptance. Personal characteristics such
as engaging in environmentally conscious behaviors and possessing a liking of technology have
also been identified as likely predictors of acceptance of CCU-based products (Arning et al.,
2018; van Heek et al., 2017b). Our findings regarding moral hazard (i.e., sustainability risks)
relate to prior research as well. While we found concern of moral hazard to have a slight negative
association with acceptance without controlling for other variables, the variable became
insignificant in our regression model. Likewise, in other studies, there has been evidence both for
42
and against the connection between perceived sustainability risks and acceptance of CCU and
CCU-based products (Katrin Arning et al., 2019; Arning et al., 2018; Jones et al., 2017).
Practical implications and communication strategies
With the majority of participants indicating an openness to consume or use a CCU-based
product, there is initial encouragement that consumers in the United States could potentially be a
reliable end-market for products derived from CCU. This openness toward products, paired with
the finding that product type and capture method did little to influence that acceptance, suggest
that consumers may be open to a wide range of CCU-based products, regardless of how the
carbon was sourced. Products meant to be consumed, however, may be the exception, as
participants were generally less accepting of carbonated beverages compared to other non-edible
products. Thus, advertising a product as CCU-based may be less beneficial for edible products
(particularly when this carbon is captured from industrial emissions) as compared to products not
meant for consumption.
Relevant to advertising, demographics, such as age, gender, education, and political
ideology were not found to be significant predictors of product acceptance in our model. This
indicates that CCU-based products may have the potential to appeal to a large audience, not
limited by characteristics inherent to some groups. When other variables were not controlled for,
it is of note that women were associated with slightly less acceptance of CCU-based products (r
= -0.07, p < 0.05). Women were also associated with perceiving the products to be riskier to
personal health (r = 0.07, p < 0.05). Thus, it is possible that gender may not be a significant
predictor of acceptance in our regression because perceiving more risk is what drives women to
be less accepting.
43
Considering the political tension in the United States surrounding most environmental
goals, it was surprising to us that political ideology did not predict acceptance of CCU-based
products. Due to the unexpected nature of this finding, we conducted an additional exploratory
test. Although this variable was treated as continuous in the linear regression, it was treated as
seven levels in a one-way ANOVA to observe the impact of political ideology without
controlling for other variables (Table S3). This model confirmed that there were no significant
differences in product acceptance between even the most liberal and most conservative groups
(F6, 2393 = 1.31, p = 0.249).
This finding suggests that CCU may be one aspect of a broad climate mitigation strategy
that could benefit from bipartisan support. This is in line with the bipartisanship we have already
seen surrounding carbon capture in the political sphere, as evidenced by the 45Q tax credit (for
capturing and storing carbon underground) along with other measures taken to support R&D
funding for DAC
9
. Yet, it is important to note that participants in this study had limited
information about CCU, and with exposure to more politicized messaging about the technology
in the future, support for CCU could easily differ by party. That said, participants here were
made aware of CCU’s intended climate benefits in the survey, and conservative participants still
indicated an openness to the products. Further, the more conservative a participant was, the less
they perceived CCU to offer climate benefits (r = -0.12, p < 0.001), which could suggest there
are additional factors unrelated to the climate driving acceptance among this group. Moving
forward, researchers should aim to test if this study’s finding, that willingness to consume or use
a CCU-based product is independent of political ideology, is also indicative of bipartisan support
9
https://www.thirdway.org/memo/getting-serious-about-direct-air-capture
44
for policy measures related to CCU (as well as if said support, of policy or products, remains
stable over time).
Despite the positive reception to CCU discussed thus far, because only 4% of our sample
indicated they understood the technology prior to the study, communicating to the public
effectively about CCU and CCU-based products will be especially important. While only 13% of
participants thought that consuming or using a CCU-based product would likely be risky to their
personal health, there were still 36% of participants who were “unsure”, thus, clear messaging
about risk will be one element of a strong communication strategy. Low knowledge about the
characteristics of carbon dioxide—e.g., not knowing whether or not CO2 is toxic—may be one
factor that drives this risk perception. Associations between this variable and others support this
idea, as outside of our regression model, less knowledge about CO2 was associated with both
higher perceived risk (r = -0.22, p < 0.001) and a lower willingness to consume or use a CCU-
based product (r = 0.18, p < 0.001). Because perceived personal health risk had a significant
effect (η
2
= 0.05) on willingness to consume or use, and thus could be a barrier to consumer
acceptance, more research is needed to test if specific messaging surrounding the risk of CCU
and CCU-based products, along with educating consumers about the properties of CO2, could
address a consumer’s potential hesitations.
On the other hand, participants who indicated greater technological optimism were more
accepting of CCU-based products (r = 0.29, p < 0.001) and less likely to perceive health risks
associated with CCU (r = -0.16, p < 0.001). Technological optimism was also associated with
less negative affect (r = -0.33, p < 0.001). In line with prior work, this suggests that some people
may positively view CCU-based products as innovative and less risky.
45
The perceived benefit of manufacturing CCU-based products, in terms of stemming the
negative effects of climate change, was one other factor that predicted acceptance, and may also
encourage positive perceptions of the technology. The influence of benefit perception, combined
with the positive influence of a participant’s inclination toward environmentally conscious
behavior on willingness to consume or use, suggests that a large portion of consumers may be
receptive to climate-friendly messaging surrounding CCU-based products in the future.
Furthermore, negative affect, which was the strongest predictor of willingness to consume or use
in our model (η
2
= 0.14), was also negatively associated with perceived benefit (r = -0.53, p <
0.001), meaning as perceived climate benefits of using CCU-based products increased, negative
emotions surrounding CCU decreased. This suggests that promoting CCU as a climate-friendly
technology may resonate with potential consumers by increasing positive feelings associated
with CCU.
Finally, understanding how the domain-specific knowledge one possesses about climate
change influences perceptions of CCU-based products may play an important role in
communicating with consumers about the technology. Without controlling for other variables,
increased knowledge of climate change is associated with greater willingness to consume or use
a CCU-based product (r = 0.21, p < 0.001). Thus, climate-friendly messaging (as mentioned
above) may also be effective among those with more knowledge of climate change. But, because
greater climate knowledge decreased acceptance within our regression model, that indicated that
there must be at least some participants who had higher domain-specific knowledge about
climate change but also reported lower acceptance of CCU-based products. In a prior study,
some participants rejected CCU products likely due to their concerns about the technology
hindering environmental progress (Arning et al., 2018). Therefore, one plausible explanation for
46
the connection between greater climate knowledge and less acceptance in our model is that some
participants with higher climate change knowledge were also concerned about moral hazard,
diminishing acceptance of the products. Regardless of if this is the case, as awareness of CCU
increases, it will be important to convey to the public that carbon capture technology is limited in
its potential to mitigate climate change, and that any celebration of this technology should take
place alongside continued support for other climate mitigation strategies. Clear communication
on this front may alleviate concerns some have about the connection between CCU and moral
hazard. Further, more research should be undertaken to better understand the relationship
between concerns of moral hazard and perceptions of CCU.
Limitations
Typical of most online consumer survey research, our main dependent variable
(willingness to consume or use) was based on responses to a realistic, but ultimately hypothetical
scenario in that it did not measure actual consumer behavior (i.e., actual use of CCU-based
products). Future research should offer participants opportunities to interact with or actually
purchase these products in person. In addition, future research should measure willingness-to-
pay, as CCU-based products may be more expensive than their conventional counterparts.
Further, in this study, the type of information participants received about CCU likely
influenced their responses throughout the survey. For example, it is not surprising that 69% of
participants saw environmental benefits in the technology when our explanation of CCU
mentioned an environmental benefit. Thus, it is important to note that consumers may respond
differently if they received more information on the advantages and disadvantages of the
technology. Along the same line, providing different information about the capture process could
also influence responses. While we asked participants to read broadly about carbon captured
47
from “factories and power plants”, reading more detailed information about the type of capture
method (e.g., carbon captured from a coal power plant or waste incinerator), may garner different
reactions to products. Notably, the Cronbach’s alpha for the scale we used to measure moral
hazard was low; this may have also been a result of the information we presented to participants,
as they may have required more nuanced information about CCU in order to evaluate the more
complex statements included in the scale. Future research should continue to examine how
exposure to different introductions to the technology (in terms of type and depth of information
provided) influences public perceptions.
Also of note was that this study’s sample included only 52 participants—from a total of
2,400—that indicated that they do not believe in climate change; this accounts for 2.17% of our
sample. This number is inconsistent with 2020 data from the Yale Program on Climate Change
which found 12% of adults in the U.S. do not believe climate change is occurring
10
. Related, a
relatively large number of participants were excluded from our survey for failing our attention
check. Although participants had read about the correct answer (carbon capture and utilization)
on the page immediately preceding the attention check, it is possible that it was easier to select
the appropriate answer if one was already familiar with terminology (e.g. carbon dioxide)
typically used when discussing climate change. Taken together, this may mean that our sample
skews in the direction of people who may better understand, or are more open to, climate
science. However, even for those who agree that climate change exists, a person’s political
ideology, rather than their scientific knowledge, has been shown to better predict opinions on
polarizing topics (Drummond & Fischhoff, 2017). Therefore, because our sample was
10
https://climatecommunication.yale.edu/visualizations-data/ycom-us/
48
approximately representative for political ideology, opposition to CCU-based products due to the
technology’s connection to climate change should have still been accounted for in our study.
In addition, although we surveyed participants on their familiarity with CCU, their
responses were subjective. Thus, it is possible that participants who indicated an understanding
of CCU may not actually fully comprehend the technology. This posed a limitation because we
could not control for a participant’s familiarity with CCU in our analysis, and for some
participants, we cannot attribute their impressions of CCU to the supporting information
provided within the experiment. Regardless, in line with past research which demonstrates that
familiarity increases positive perceptions (Bornstein, 1989; Park & Stoel, 2005), it is a hopeful
finding that those who indicated familiarity with CCU were also more accepting of CCU-based
products.
Conclusion
In the end, this study offers initial evidence that consumers may be open to consuming or
using products manufactured using carbon derived from carbon capture and utilization. In
addition, insights regarding the drivers of and barriers to acceptance can inform messaging to
accurately and effectively communicate about CCU-based products to consumers. Messaging
which conveys the environmental benefits and technological innovativeness of CCU, or
advertises the products as an environmentally conscious alternative, may resonate with some.
Encouragingly, political ideology was not a barrier to acceptance, which further indicates the
potential for CCU-based products to be accepted widely, and perhaps, for CCU generally to
receive less political opposition as one aspect of a mitigation strategy that can contribute to the
fight against climate change.
49
Chapter III
All atwitter about climate change: Do polite and informative Twitter debates influence
support for climate policy?
11
Though a majority of Americans now believe in the reality and gravity of climate change,
what to do about it remains a controversial and polarizing topic. According to recent polling
data
12
, approximately 72% of Americans believe that climate change is happening, and 57%
believe it is caused by humans. However, support for policies aimed at addressing climate
change diverges sharply according to political ideology. For example, 87% of Democrats believe
that CO2 should be regulated as a pollutant, and 85% believe that strict greenhouse gas emissions
limits should be imposed upon coal-fired power plants; support for these policies falls to 60%
and 50%, respectively, among conservative Republicans
13
.
Given this partisan divide, there is a pressing need to communicate effectively about
climate policy options so that people may move forward with impactful solutions. An
increasingly prominent platform for informing and attempting to persuade people about a wide
range of issues—including climate change—is social media (Gil de Zúñiga et al., 2012; Nielsen
& Schrøder, 2014). This is especially the case given social media sites are popular sources of
news; a majority of Americans report that they routinely rely on social media for the news (Pew
Research Center, 2021). And among social media platforms, Twitter stands out as one of the
most popular among news-focused users; according to recent data from the Pew Research Center
(2021), 59% of adult Twitter users get information about current events from the platform.
11
Chapter III is adapted from a manuscript co-authored with Sanghamitra Sen, Caitlin Drummond Otten, and Joseph
Árvai.
12
https://climatecommunication.yale.edu/visualizations-data/ycom-us/
13
https://climatecommunication.yale.edu/visualizations-data/americans-climate-views/
50
Twitter’s popularity (among both posters and information seekers) may be due, in part, to
the fact that it limits posts to 280 characters; thus, posts are easy to craft, brief, and easy to read
and understand. Users of Twitter may also post web-links to information that can be consulted if
one desires more information than can be shared in 280 characters. In addition, Twitter allows
users to “converse” with one another in exchanges where multiple people can comment in
response to an original tweet. It is perhaps for these reasons that Twitter has become popular
among high-profile influencers such as politicians.
Indeed, the vast majority of federal politicians have official Twitter accounts
14
and many
of these politicians post regularly. These tweets have a large audience, too. For example, on the
extreme end of the spectrum, the current U.S. President Joseph R. Biden has over 25 million
followers, and his posts typically receive thousands of engagements (e.g., likes and retweets,
which further the reach of the post). When it comes to members of Congress, the median
follower count is almost 37,000, as of 2020 (Pew Research Center, 2020). Therefore, with
information being conveyed by politicians on Twitter, it is important to consider its influence,
and how differing communication styles represented through tweets may affect readers’
perceptions.
Communication Styles
A communication style is the characteristic way an individual (or spokesperson for a
group or position) transmits verbal, paraverbal, or non-verbal signals in a social interaction
(Yuan et al., 2019). How people interact with others (e.g., with civility vs. incivility), as well as
the amount or type of information they provide, are elements of communication styles. With a
great number of politicians on Twitter, it is unsurprising that the communications styles
14
https://pressgallery.house.gov/member-data/members-official-twitter-handles; https://twitter.com/i/lists/88345660;
and https://twitter.com/i/lists/63915645
51
represented in political tweets can vary widely. For example, despite the professional nature of a
politician’s role, tweets that come from politicians sometimes do not reflect this professionalism
when it comes to the civility and informativeness of their messaging.
Civility, in particular, can be lacking. On social media platforms like Twitter, politicians
often aggressively criticize or denounce people or ideas that they don’t find favorable
(Theocharis et al., 2020; Zompetti, 2019). In fact, using artificial intelligence to evaluate over a
million tweets by U.S. politicians, Frimer et al. (2022) found that incivility (i.e., rudeness or
disrespect) represented in politicians’ tweets has increased 23% from 2009 to 2019 (Frimer et al.,
2022). And, in a study which examined uncivil tweets related to climate change, almost 90% of
these tweets mentioned politics (Anderson & Huntington, 2017). With this, the authors suggest
politicized topics may bring about the uncivil tones.
This trend is not isolated to a single political ideology; Democrats and Republicans alike
have come under scrutiny for aggressive and uncivil tweets directed at others. For example,
Democrat Neera Tanden’s 2021 nomination to the post of Director of the White House Office of
Management and Budget was withdrawn for her Twitter posts that claimed that vampires have
more heart than Senator Ted Cruz, and that likened Senator Mitch McConnell to the literary
villain Voldemort. Likewise, former U.S. President Donald J. Trump, a Republican, was well-
known for aggressive and uncivil tweets; for example, upset with TIME Magazine’s decision to
name climate activist Greta Thunberg as its Person of the Year, he posted “So ridiculous. Greta
must work on her Anger Management problem, then go to a good old fashioned movie with a
friend! Chill Greta, Chill!” And concerns about civility on Twitter remain top-of-mind for many
users and observers of the platform now that Elon Musk has taken over as its owner; on the day
Mr. Musk took over ownership of the platform, he said that he wished for it to be a “more
52
freewheeling place for all kinds of commentary”, and that he planned to restore the accounts of
users previously banned for uncivil conduct and for spreading falsehoods
15
.
Further, in spite of their brevity, the level of informativeness that can be conveyed in
280-character tweets may vary widely. For example, upset with information that encouraged
energy conservation, Senator Ted Cruz once tweeted
16
“California is now unable to perform even
basic functions of civilization, like having reliable electricity. Biden/Harris/AOC want to make
CA’s failed energy policy the standard nationwide. Hope you don’t like air conditioning!” No
detail was provided about why Senator Cruz believes an energy policy to be failure, or how this
failure might relate to “basic functions of civilization”. Senator Elizabeth Warren similarly
posted
17
“Climate change is the existential threat of our time. By officially declaring it a national
emergency, the Biden-Harris administration can unlock more tools and resources to tackle this
crisis head on.” It’s not clear from this tweet why climate change is an existential threat, nor is it
clear what tools would be unlocked by declaring it a national emergency.
These posts stand in contrast to more informative tweets such as one from the United
Nations Framework Convention on Climate Change about policy changes needed to achieve net-
zero emissions; along with an infographic, they posted
18
“@IEA’s comprehensive roadmap for
the global energy sector to reach net-zero greenhouse gas emissions by 2050 covers all energy
sectors, including transport. By 2030, 60% of all new car sales world-wide must be electric.” We
are able to understand from this tweet a timeframe for achieving net-zero emissions (2050) and
what kinds of changes would be required to get there (60% of new vehicle sales must be
electric).
15
https://www.nytimes.com/2022/10/27/technology/elon-musk-twitter-deal-complete.html
16
https://twitter.com/tedcruz/status/1296134869320380419
17
https://twitter.com/ewarren/status/1355560084714221569
18
https://twitter.com/UNFCCC/status/1395010174126022660
53
The Present Research
Testing the influence of tone and informativeness
When it comes to communicating about climate change on Twitter, differing
communication styles such as those exemplified above may lead to different outcomes when it
comes to peoples’ perceptions. Referring to climate communication generally, renowned climate
scientist, Dr. Katharine Hayhoe, was quoted in the New York Times
19
saying “If you begin a
conversation with, ‘You’re an idiot,’ that’s the end of the conversation, too.” Our study
questioned if the incivility commonly found in politicians’ tweets could be a detriment to
advancing constructive communication about climate change on the platform, in that those who
do not already agree with the statements being conveyed may be less open to hearing others out
if they use uncivil language.
Yet, despite Dr. Hayhoe’s insight, there is conflicting evidence about how people
perceive or respond to a communicator’s positive (i.e., civil, polite, respectful) versus negative
(i.e., uncivil, aggressive, rude) tone. For example, communicating in a negative tone with
forceful language and attacks of a personal nature may be appealing to some (assuming one’s
own group is not the target) because it conveys a strong, no-nonsense attitude on the part of the
communicator, and because doing so may increase outside interest or attention in whatever is
being shared (Lau et al., 2007; Metzger et al., 2003; Mutz & Reeves, 2005; Yuan et al., 2019).
Likewise, an aggressive communication style may be effective in that it may damage trust in
whoever or whatever the communicator is attacking (Mutz & Reeves, 2005). Further, posts on
Twitter with language containing political out-group animosity (e.g., messaging that attacks an
19
https://www.nytimes.com/2016/10/11/science/katharine-hayhoe-climate-change-science.html
54
opposing political party) have been found to generate significant engagement on the platform
(Rathje et al., 2021).
Alternatively, others may find a positive tone may be more appealing. For example, an
aggressive or negative tone has been shown to result in boomerang effects whereby trust in,
likeability, or credibility, of the communicator, may also decline (Lau et al., 2007; Thorson et al.,
2010; Yuan et al., 2018). These boomerang effects may be the result of expectancy violations
(Burgoon & Hale, 1988); that is, communication styles that violate people’s expectations about
how interactions ought to unfold (e.g., scientists or politicians adopting a rude or aggressive
communication style) may lead to communicators being viewed with greater skepticism and
shared information being perceived as low quality. Communicators using a more positive tone
may in turn be viewed as more credible. For example, Fiske and Dupree (2014) suggest that
when it comes to science communicators (e.g. researchers), being perceived as an expert is not
enough for effective communication with the public. Scientists must also be perceived as warm
and trustworthy to be deemed credible communicators (as their motives may otherwise be
questioned).
Therefore, because evidence suggests that people could find either a positive or negative
tone appealing (in terms of evaluating both communicators and their messages), we conducted an
experiment to better understand the role tone may play on social media. In this study, we created
hypothetical conversations between two politicians on Twitter (a Democrat and a Republican)
where we manipulated the tone they used as they debated the merits of a policy meant to address
climate change.
We explored whether respectful communication between politicians would improve
participants’ perceptions of said politicians (in terms of their trustworthiness, competence, and
55
likeability). Further, we questioned if a positive tone would cause participants to be more open to
the points made in the climate policy debate, as indicated by their purported learning about the
policy, their evaluations of the arguments presented both for and against the policy, and
ultimately by their indicated support for the policy. We also recognized that the influence of tone
may be dependent upon the influence of other variables, such as participants’ political
orientation, which we describe later on.
There is a great deal of literature supporting the idea that people form perceptions about
people, places, and things based on salient cues that are available to them around the time that
judgments are rendered (Árvai & Gregory, 2021; Lichtenstein & Slovic, 2006). Extended to
communications, the manner in which information is shared (e.g., be it in a positive or negative
tone) sends influential signals to receivers about the content of communications (e.g., is the
information credible?) and the communicators themselves (e.g., are they strong, decisive, or
trustworthy?). Yet, from a normative perspective of judgement and decision-making, a positive
tone should have no bearing on participants’ policy support or perceptions of learning and
argument strength, as altering tone does not change the fundamentals of the information and
arguments being communicated.
On the other hand, the informative or technical content of communications about
controversial subjects such as climate change should (Árvai, 2014) be particularly important in
messages and deliberations that are intended to inform judgments (e.g., about policies or
behaviors). The information that a communicator can offer vis-à-vis the technical or mechanistic
details about a problem or policy (i.e., how a policy works to address a problem) may serve two
important purposes in communication. First, technical or mechanistic details may help others
(e.g., participants in a dialogue, or observers of it) to learn about a problem or proposed solution
56
(Sturgis & Allum, 2004) so that they may construct a more reasoned judgment of their own
(Árvai & Gregory, 2021; Campbell-Arvai et al., 2018). And second, the presence or absence of
technical or mechanistic details from the communicator should send important signals about their
expertise (Lesgold, 1983).
Against this theoretical background, we also manipulated the level of informativeness
that was represented in the Twitter climate policy conversations created for this experiment. We
expected that when participants were exposed to more informative Twitter conversations (i.e.,
with more relevant information being provided about the climate policy), they would evaluate
the arguments presented to them as stronger and report that they learned more. Further, we
predicted that higher informativeness will lead to evaluating politicians in a more positive light,
as the ability to communicate knowledgably about important issues should reflect positively on a
politician’s competence.
When it comes to policy support, in cases where both the Democratic and Republican
politicians incorporate substantive arguments to back up their views, each side may benefit
equally from improved informativeness, giving neither party an advantage that would sway a
participant’s opinion. Yet, we also suspected that the influence of informativeness may depend
on interactions with other variables, detailed below.
Finally, to the best of our knowledge, prior research has not examined whether tone and
informativeness may interact to predict judgments of speakers discussing contested policies.
While relatively unexplored, these interactions may be plausible given the findings described
above. For example, participants may respond to the informativeness of conversations only when
those are delivered in a respectful tone, dismissing all disrespectful communications. To test the
possibility that the effects of tone on our dependent measures, mentioned above, may depend on
57
informativeness, and vice versa, we examined exploratory interaction effects between tone and
informativeness for each dependent measure.
Testing for potential interactions with political party affiliation
In addition to testing for interactions between tone and informativeness, we also
examined potential interactions between informativeness, tone, and participants’ political
orientation, to determine if our manipulations impact Democrats versus Republicans differently.
As noted above, particularly when the tone of the conversation is positive and does not promote
partisan animosity, it is also possible that exposure to a detailed argument from the opposing
party could prompt participants to give more consideration to alternative viewpoints. This greater
consideration for the other side may lead some participants to report more moderate standpoints
on the policy. If this were the case, positive and informative conversations could influence policy
support differently depending on participants’ political orientation (e.g., increasing support more
amongst Republicans than Democrats).
At the same time, given that support for climate policy is more typical among Democrats,
there may be a ceiling effect in that Democrats largely indicate high support for the policy, while
Republican responses could be more readily influenced by the tenor of the debate (i.e., the tone
or level of informativeness). (We did not expect to see a floor effect among Republicans
opposing the policy because of the many Republicans who do in fact see climate change as a
concern
20
).
Interactions between our independent variables and political orientation may be relevant
for our other dependent variables as well. For example, participants may judge politicians that
share their political orientation as favorable regardless of the tone used, while evaluating the
20
https://news.gallup.com/poll/394955/republicans-environmental-worry-varies-age.aspx
58
opposing party’s representative as more favorably only when a positive tone is conveyed. Or, if
participants disagree with the stances of the opposing party, they may rate argument strength as
low regardless of the politician’s level of informativeness. We explored these possibilities
through our analyses.
Methods
Experimental Design
Our experiment had a 2 (climate policy) ´ 2 (tone) ´ 2 (informativeness) between-
subjects design. All participants read a hypothetical Twitter conversation between a Republican
and a Democratic member of Congress, debating the merits of a climate policy. The conversation
was created to look identical to what one would see on the platform, except that participants
were told that they would only be able to see the party affiliations of the Congresspeople, and
that all comments, likes, and retweets (which are typically displayed on tweets) were removed.
In all versions of the conversation, the Democratic member was in support of and the Republican
member was opposed to the climate policy. All versions of the Twitter conversation were similar
in length (8 tweets in total) and tweets were similar in length across all versions. Two example
Twitter conversations are located in Figures 1 and 2; all versions are located in the Supplemental
Materials.
Climate Policy Condition
Participants were randomly assigned to read a Twitter conversation about either a carbon
tax or CAFE standards. Under a federal carbon tax, the government would set a price that those
who emit greenhouse gases (through the burning of fossil fuels) would have to pay for each ton
of emissions that they release into the atmosphere. With stricter Corporate Average Fuel
Economy (or CAFE) standards, manufacturers would be required to make new vehicles more
59
fuel efficient, so to reduce the carbon emissions from cars and trucks. These climate policies
were selected because they represent policies currently being debated by federal and state
politicians. For example, see the Energy Innovation and Carbon Dividend Act (H.R. 2307) that
was been referred to the Congressional Subcommittee on Energy in 2021
21
, as well as the 2021
National Highway Traffic Safety Administration’s withdrawal of a Trump administration rule
that would have prevented states from setting their own greenhouse gas emissions standards for
vehicles
22
. Before they read the conversation, participants were shown a brief informative written
description about the policy, to ensure that all participants had a baseline understanding of the
policy itself, before reading a conversation about its relative merits and drawbacks. Each
conversation was written to contain specific arguments for or against the carbon tax or CAFE
standards drawn from the policy debates surrounding these policies
23
.
Tone Condition
Participants were randomly assigned to read a conversation with either a positive or a
negative tone. In the positive tone conversation, the Republican and Democrat were polite to
each other, using phrases like, “I respect your position” and “I hear you”. In contrast, in the
negative tone condition, the Republican and Democrat were impolite to each other, using phrases
like, “Your concerns are stupid” and “You’re being ridiculous!” The Twitter conversations were
written so that the sentences containing positive or negative tone, like those quoted above,
occurred at the beginning of each tweet, and tweets were otherwise identical.
21
https://www.congress.gov/bill/117th-congress/house-bill/2307
22
https://www.nhtsa.gov/press-releases/cafe-preemption-final-rule
23
https://www.thebalancemoney.com/carbon-tax-definition-how-it-works-4158043;
https://www.heritage.org/energy-economics/commentary/why-the-carbon-tax-would-backfire-america;
https://energyathaas.wordpress.com/2018/04/08/does-cafe-work/; https://www.heritage.org/government-
regulation/report/fuel-economy-standards-are-costly-mistake; https://www.eisneramper.com/fuel-economy-cafe-
standards-adi-0317/; and https://rollcall.com/2016/09/22/house-gop-seeks-to-ease-or-end-vehicle-emissions-
standards/
60
Informativeness Condition
Participants were randomly assigned to read a conversation with either low and high
informativeness. We manipulated informativeness by having the Twitter debaters either give a
reason behind their assertion or follow their assertion with uninformative commentary. We chose
this over other methods of conveying informativeness on Twitter (e.g., providing hyperlinked
articles) because we wanted our hypothetical Twitter debaters to have a back-and-forth
conversation and because we did not want participants to click links (thereby adding additional
variables to our design). Thus, in the low informativeness conversation, the Republican and
Democrat provided uninformative and circuitous reasons for their viewpoints, for example
explaining opposition to a carbon tax by indicating that, “I think having extra money in their
purses and wallets is important to pretty much everybody in this country”. In the high
informativeness condition, participants provided more substantive and mechanistic reasons for
their viewpoints, for example explaining opposition to a carbon tax by indicating that, “A carbon
tax will mean that energy prices will rise, and that will make things more difficult for businesses
and families.” The Twitter conversations were written so that after the positive/negative tone
sentences described above, the next sentence justified the position with an argument of either
low or high informativeness.
Dependent Measures
Policy Support: Participants were asked, “To what extent would you support a carbon tax
[CAFE standards] in the United States?” on a 7-point scale from 1 = strongly oppose to 7 =
strongly support.
Perceived Strength of Argument for the Policy Proponent (Democrat) and Opponent
(Republican): Participants were asked about the extent to which they thought that the
61
representative in favor of the carbon tax [CAFE standards] provided strong arguments to support
their position, on a 7-point scale from 1 = strongly disagree to 7 = strongly agree. Participants
also responded to this question in relation to the arguments presented by the representative
opposed to the climate policy.
Learning: Participants were asked, “How much do you feel like you learned about the
arguments for and against CAFE standards [a carbon tax]?” using a 7-point scale from 1= none
at all to 7 = a great deal.
Trust in Policy Proponent (Democrat) and Opponent (Republican): Participants were
asked to rate the representative in favor of the carbon tax [CAFE standards] on three dimensions,
trustworthiness, likability, and competence, on a 5-point scale from 1= not at all to 5 =
extremely. Participants evaluated the representative opposed to the climate policy on these same
dimensions. An average of these responses was taken for the two representatives (Cronbach’s
alpha for representative in favor: 0.9; Cronbach’s alpha for representative in opposition: 0.91).
Covariates and Demographics
Participants reported their political party (with response options of: Democrat,
Republican, Independent, No party/not interested in politics, or Prefer not to say) as well as their
political ideology, on a 5-point scale from very conservative to very liberal, including a not sure
option (participants who selected that option were treated as missing data for analyses involving
political ideology). Participants reported their gender, age, and education and other covariate
measures not analyzed here.
Participants and Participant Exclusions
62
The University of Michigan Health Sciences and Behavioral Sciences Institutional
Review Board approved this study as exempt from ongoing review. APA guidelines were
followed and all participants gave their informed consent.
We recruited American adult participants, using Lucid, to partake in our online survey.
We set a target N of 200 participants per condition for each of the 8 conditions in our 2 ´ 2 ´ 2
between-subjects experiment, for a target total N of 1,600 participants. To ensure participants
were familiar with Twitter and conversations on the platform, we included a screener question at
the beginning of the survey to restrict our sample to those who reported having a Twitter
account. Thus, we asked Lucid to recruit a sample of 1,600 U.S. participants that have a Twitter
account, balancing for political orientation (approximately an even split between Democrats,
Republicans, and Independents). Due to an error, Lucid over-recruited Democrats resulting in
595 excess Democrats. We excluded these excess Democrats from our analyses in the main text,
in line with our original recruitment strategy. In the Supplemental Material, we report the below
analyses including these excess Democrats, as a robustness check. Results are very similar.
Excluding the 595 excess Democrats, our final sample consisted of 1,831 participants. Within
this final sample, 43% reported their gender as “man”, and the mean age of participants was 34.5
(SD = 13.9). One percent reported having less than a high school education; 18% had graduated
high school; 22% had completed some college; 10% had a two-year degree; 27% held a four-
year degree; and 22% had a post-graduate degree. Thirty-six percent described themselves as
Democrats, 25% as Independents and 31% as Republicans; mean political liberalism, on a 5-
point scale from very conservative to very liberal, was 3.0 (SD = 1.3).
A total of 2,451 additional participants were excluded from the final sample (or excluded
from taking the survey) because they did not provide their informed consent (N = 125), they
63
indicated they did not have a twitter account (N = 1,982), or they failed an attention check (N =
344). There were two attention checks included in this study. Before participants were randomly
assigned to condition, they were asked to answer an instructed-choice attention check question.
Those who answered the attention check question incorrectly (169 participants) were screened
out at that point. Our second attention check occurred after participants completed the
experimental task of reading a Twitter conversation: participants were asked to answer a
multiple-choice question to correctly identify the topic of the conversation they read; the other
two unrelated options were “movies” and “sports”. Of those who read about a carbon tax, 85
participants (8.3% of those assigned to the carbon tax condition) failed this attention check; of
those who read about CAFE standards, 90 participants (9.1% of those assigned to the CAFE
standards condition) failed this attention check.
Manipulation Check
Before conducting the experiment reported here, we conducted a manipulation check to
assess respondent’s assessment of tone and informativeness within each climate policy. The
manipulation check had a 3 ´ 2 ´ 2 between-subjects design in which participants read a debate
about one of three climate policies: carbon taxes, CAFE standards, and a moratorium on burning
coal to generate electricity. We recruited 494 participants using Lucid, the same recruiter as for
the main study. The survey asked participants to read one Twitter conversation and then rate how
polite (measuring tone) and how detailed (measuring informativeness) they perceived the
conversation to be.
Using t-tests, we found that participants rated the negative tone version as less polite than
the positive tone for all three climate policy conditions (CAFE: t(125) = -12.70, p < .001, Carbon
Tax: t(129) = -9.60, p < .001, Coal Ban: t(136) = -12.52, p < .001). We found that participants
64
rated the high informativeness version as more detailed than the low informativeness version in
both the CAFE and carbon tax treatments, but not for the coal ban treatment (CAFE: t(151) =
2.43, p = .016, Carbon Tax: t(151) = 2.10, p = 0.037, Coal Ban: t(169) = -0.32, p = .753).
Because the coal moratorium condition failed the manipulation check for informativeness, it was
excluded from this study. There was also an error in uploading one of the four iterations of the
coal ban Twitter conversation to the manipulation check survey (half of one conversation was
essentially missing), preventing us from fully evaluating the differences between tone and
informativeness for that treatment.
Full results from the manipulation check, including means and standard deviations for
tone and informativeness by climate policy treatment, along with the coal ban Twitter
conversations, can be found in the Supplemental Materials.
Data Availability Statement
Code, data, and materials are available at:
https://osf.io/bwsrj/?view_only=d15146e2c2214a33874e0459538e946c.
Results
Assignment to Condition
The distribution of participants was approximately even among treatments (Table 1). T-
tests showed that participants in the carbon tax and CAFE standards conditions did not differ by
gender, age, education, or political ideology (all p’s > 0.36); participants who saw the
conversation with a negative vs. a positive tone also did not differ by gender, age, education, or
political ideology (all p’s > 0.49); participants who saw the conversation with a high vs. low
informativeness also did not differ by gender, age, education, or political ideology (all p’s >
0.12).
65
Results by Condition
Descriptive Results: Figure 3 depicts means and standard errors by condition for each of
our dependent measures, described below.
Analysis Strategy: For each of our main dependent measures, we conducted a three-way
ANOVA predicting the dependent measure as a function of tone, informativeness, climate
policy, and their interactions. Our intention in creating multiple climate policy conditions was to
determine if the potential influence of tone and informativeness on our dependent variables
would be consistent across policies. We tested this by examining whether we observed
statistically significant two- and three-way interactions between tone, informativeness, and
climate policy. If we did not, we reran the model with climate policy as a main effect only, and
examined the interaction between tone and informativeness. We planned to use two-way
ANOVAs and t-tests where needed to further examine statistically significant interactions. A
summary of the main findings from this section is represented in Table 2.
Policy Support: We did not observe a statistically significant three-way interaction of
tone, informativeness, and policy, F(1, 1823) = 0.5, p = 0.48, partial η
2
< 0.001, nor were any of
the two-way interactions statistically significant (all p’s > 0.35). We reran the model including
climate policy as a main effect only. There was no main effect of tone (F(1, 1826) = 0.03, p =
0.9, partial η
2
< 0.001) nor informativeness (F(1, 1826) = 0.01, p = 0.9, partial η
2
< 0.001) and no
interaction between tone and informativeness (F(1, 1826) = 0.79, p = 0.38, partial η
2
< 0.001).
We did observe a significant main effect of climate policy (F(1, 1826) = 30, p < 0.001, partial η
2
= 0.016). Participants preferred CAFE standards to a carbon tax (Figure 3).
Learning: We did not observe a statistically significant three-way interaction of tone,
informativeness, and policy, F(1, 1823) = 0.01, p = 0.95, partial η
2
< 0.001, nor were any of the
66
two-way interactions statistically significant (all p’s > 0.30). We reran the model including
climate policy as a main effect only. There was a main effect of tone (F(1, 1826) = 10.9, p =
0.001, partial η
2
= 0.006), a main effect of informativeness (F(1, 1826) = 18.9, p < 0.001, partial
η
2
= 0.01), and no interaction between tone and informativeness (F(1, 1826) = 1.1, p = 0.3,
partial η
2
= 0.001). We did not observe a significant main effect of climate policy (F(1, 1826) =
1.3, p = 0.25, partial η
2
= 0.001). Participants felt that they learned more when in the positive vs.
negative tone condition, and when in the high informativeness vs. low informativeness condition
(Figure 3).
Perceived Strength of Argument (Policy Proponent – Democrat): We did not observe a
statistically significant three-way interaction of tone, informativeness, and policy, F(1, 1823) =
2.9, p = 0.09, partial η
2
= 0.002, nor were any of the two-way interactions statistically significant
(all p’s > 0.11). We then ran an ANOVA which included climate policy as a main effect only and
examined the interaction between tone and informativeness. We observed a main effect of tone
(F(1, 1826) = 5.8, p = 0.016, partial η
2
= 0.003), a main effect of informativeness (F(1, 1826) =
5.7, p = 0.018, partial η
2
= 0.003), and no interaction between tone and informativeness (F(1,
1826) = 1.6, p = 0.2, partial η
2
= 0.001), but a significant main effect of climate policy (F(1,
1826) = 5.9, p = 0.015, partial η
2
= 0.003). As depicted in Figure 3, participants viewed the
proponent’s arguments as stronger in the positive tone condition compared to the negative tone
condition. Participants viewed the proponent’s arguments as stronger in the high informativeness
condition compared to the low informativeness condition. Arguments were perceived as stronger
in the CAFE standards condition compared to the carbon tax condition.
Perceived Strength of Argument (Policy Opponent – Republican): We observed a
statistically significant interaction between tone, informativeness, and climate policy (F(1, 1823)
67
= 4.9, p = 0.03, partial η
2
= 0.003). To further probe this interaction, we next ran separate two-
way ANOVAS for those in the carbon tax and those in the CAFE standards conditions. For those
participants who saw the carbon tax conversation, we observed a main effect of tone (F(1, 933) =
12.9, uncorrected p < 0.001, partial η
2
= 0.014), no main effect of informativeness (F(1, 933) =
2.5, uncorrected p = 0.12, partial η
2
= 0.003) and no interaction between tone and
informativeness (F(1, 933) = 2.5, uncorrected p = 0.12, partial η
2
= 0.003). Participants who saw
the carbon tax conversation viewed the opponent’s arguments as stronger in the positive tone
condition (Figure 3). For those participants who saw the CAFE standards conversation, we
observed a marginal main effect of tone (F(1, 890) = 3.8, uncorrected p = 0.051, partial η
2
=
0.004), a marginal main effect of informativeness (F(1, 890) = 3.2, uncorrected p = 0.076, partial
η
2
= 0.004) and no interaction between tone and informativeness (F(1, 890) = 2.4, uncorrected p
= 0.12, partial η
2
= 0.003). Participants who saw the CAFE standards conversation viewed the
opponent’s arguments as similarly strong regardless of tone and informativeness.
Trust in Policy Proponent (Democrat): We did not observe a statistically significant
three-way interaction of tone, informativeness, and policy, F(1, 1823) = 0.02, p = 0.9, partial η
2
<
0.001, nor were any of the two-way interactions statistically significant (all p’s > 0.76). We then
ran an ANOVA including climate policy as a main effect only and examined the interaction
between tone and informativeness. We observed a main effect of tone (F(1, 1826) = 18.3, p <
0.001, partial η
2
= 0.01), no main effect of informativeness (F(1, 1826) = 2.6, p = 0.11, partial η
2
= 0.001), no interaction between tone and informativeness (F(1, 1826) = 0.006, p = 0.94, partial
η
2
< 0.001), and no main effect of climate policy (F(1, 1826) = 1.8, p = 0.19, partial η
2
= 0.001).
Self-reported trust in the policy proponent was greater in the positive tone condition (Figure 3).
68
Trust in Policy Opponent (Republican): We did not observe a statistically significant
three-way interaction of tone, informativeness, and policy, F(1, 1823) = 0.46, p = 0.5, partial η
2
<
0.001, nor were any of the two-way interactions statistically significant (all p’s > 0.20). We ran
an ANOVA which included climate policy as a main effect only and examined the interaction
between tone and informativeness. We observed a main effect of tone (F(1, 1826) = 41.6, p <
0.001, partial η
2
= 0.022), but no main effect of informativeness (F(1, 1826) = 2.5, p = 0.11,
partial η
2
= 0.001), no interaction between tone and informativeness (F(1, 1826) = 1.1, p = 0.29,
partial η
2
= 0.001), and no main effect of climate policy (F(1, 1826) = 0.25, p = 0.61, partial η
2
<
0.001). Self-reported trust in the policy opponent was greater in the positive tone condition
(Figure 3).
Effects of Condition by Political Party
In exploratory analyses, we next asked whether Democrats and Republicans responded
differently to our experimental manipulations of tone and informativeness when debating climate
policy. For these analyses, we created a binary variable indicating whether a respondent self-
identified as a Republican (n = 571) vs. a Democrat (n = 661), excluding Independents and those
who did not indicate a political party preference (n = 599). Figure 4 depicts means and standard
errors by tone, informativeness, and party (Republican vs. Democrat), collapsed across climate
policy condition, for each of our dependent measures. We adopted a similar analysis strategy as
used for our main results: we ran three-way ANOVAS predicting each dependent measure as a
function of tone, informativeness, and self-reported political party (Republican vs. Democrat)
and their interactions. Because the main results reported above were largely consistent across
climate policy, we included climate policy as a main effect only in these analyses.
69
For each of the dependent variables, we did not observe any statistically significant two-
way or three-way interactions. This suggests that the effects of condition on policy support,
perceptions of argument strength, perceptions of learning, and trustworthiness of the
representatives did not differ by political party. Main effects suggested that Republicans reported
less policy support than Democrats. Democrats rated the climate policy proponent’s arguments
as stronger and reported greater trust in the climate policy proponent; similarly, Republicans
rated the climate policy opponent’s arguments as stronger and reported greater trust in the
climate policy opponent. Democrats reported learning more from reading the Twitter
conversation than Republicans. Figure 4 illustrates these findings. Full details of these analyses
are reported in the Supplemental Materials.
Discussion
With important information about climate change policy being conveyed on Twitter, this
study set out to understand how the communication styles represented in a Twitter conversation
about a climate policy influenced readers’ perceptions. We recruited U.S. adults with a Twitter
account to read hypothetical Twitter exchanges (between a Republican and a Democrat serving
in the United States Congress) about a proposed federal carbon tax and more stringent Corporate
Average Fuel Economy (CAFE) standards. These exchanges varied by the climate policy being
debated (carbon tax vs. CAFE standards), the tone of the conversation (positive/polite vs.
negative/impolite), and the informativeness of the policymakers’ arguments (more informative
vs. less informative). We tested how altering tone and informativeness would impact
participants’ 1) judgements of the politicians posing the arguments; 2) perceived learning about
the policy; 3) evaluations of the arguments presented both for and against the climate policy; and
ultimately 4) support for the climate policy.
70
Influence of tone on judgements of politicians, learning, and argument strength
With respect to the potential influence of tone, prior work provided evidence that a
negative or uncivil tone could lead to either improved or worsened perceptions of
communicators and their messages. Here we found that a when comparing the impacts of
politicians on Twitter using a negative versus positive tone in their debate, it was a positive tone
that often led to more favorable perceptions from participants.
For one, participants indicated that both politicians (the Democrat and the Republican)
were viewed more favorably— on a scale that averaged perceptions of trustworthiness, likability,
and competence—when communicating in a more polite and positive tone. This supports
research which suggests that uncivil tones may backfire and reduce trust in or likeability of the
speaker (Lau et al., 2007; Thorson et al., 2010; Yuan et al., 2018). Importantly, our analyses
found no differences in the impact of tone by participants’ party identity, in that a positive tone
improved evaluations of politicians even when participants were evaluating an opposing party
member. This is not to say there were no differences between the parties in our sample;
Democrats still preferred both climate policies more so than Republicans, and each party rated
their own representative more favorably. But, both Democratic and Republican participants
viewed both Democratic and Republican representatives more positively when conversations
were polite.
From a normative perspective of judgement and decision-making, tone should have
nothing to do with evaluating the content of an argument or the amount one has learned (Slovic,
1995; Slovic et al., 1977). Yet, there is a great deal of evidence that non-rational cues influence
the judgements and decisions people make (Gilovich et al., 2002). Our research builds on this
evidence, as tone did in fact influence perceived learning and argument strength.
71
When evaluating the strength of the Democrat’s argument in favor of the climate policy,
a positive tone led participants to report that they perceived the policy proponent’s argument to
be stronger, as compared to the perceptions of participants in the negative tone condition.
However, for the opponent of the climate policy, the influence of tone depended upon whether
participants were reading about a carbon tax or CAFE standards—positive tone increased
argument strength in the carbon tax condition, but was only marginally significant in improving
perceptions of strength in the CAFE standards condition. Overall though, the pattern shows that a
polite and positive tone increased people’s perceptions that the arguments being presented to
them were strong. Further, perceived learning was also influenced by tone, in that a positive tone
increased how much participants felt they learned about the climate policy. Again, results
regarding the effect of tone on learning and argument strength did not differ depending on
participants’ party identity.
With perceptions of greater learning and stronger arguments, it is possible that a positive
tone, as compared to negative and polarizing language, made people feel more open to
considering the information being conveyed. It is also possible that tone was influential in
improving perceptions of arguments and learning because a positive tone increased favorable
opinions of the representative, and thus the content of their tweets (Nisbett & Wilson, 1977).
And, work by Fiske and Dupree (2014) supports the idea that warmth conveyed by science
communicators matters when it comes to the public’s judgement of a scientist’s credibility.
Influence of informativeness on judgements of politicians, learning, and argument strength
Normatively, a higher level of informativeness represented in tweets is a factor which should
influence our dependent variables. We therefore predicted that when politicians convey more
relevant information in their tweets about the policy being debated that it would improve
72
perceptions of the politicians posing the arguments, as well as increase perceived argument
strength and purported learning.
When it comes to judgements of the politicians, we instead found the level of
informativeness represented in the tweets had no significant bearing on how likeable,
trustworthy, and competent politicians come off as to participants. Considering the influence
tone had here, it is perhaps that when Twitter users are on social media their attention when it
comes to making evaluative judgements is drawn elsewhere (e.g., toward a politician’s
demeanor), rather than a politicians’ grasps of the policies they support. While prior literature
suggests that trustworthiness is comprised of warmth and competence (e.g. Fiske & Dupree,
2014), which roughly map on to our manipulations of tone and informativeness, our results may
suggest that warmth may be more easily manipulated on Twitter, via tone, and easier for readers
to evaluate.
With respect to learning, participants did report learning more in the high informativeness
condition, as expected. However, higher informativeness improved perceptions of the
politicians’ arguments in only some cases. For the Democratic representative in favor of a carbon
tax or more stringent CAFE standards, a more informative argument led participants to perceive
the arguments to be stronger, also as expected. However, the opponent of the climate policies did
not see the same benefit. A higher level of informativeness in the carbon tax condition did not
improve perceptions of the Republican’s arguments when it comes to strength. In the CAFE
standards condition, higher informativeness had a marginally significant positive relationship
with perceptions of argument strength. With even many Republicans understanding that climate
change is human-caused
24
, it is possible that the arguments opposing the climate policy were not
24
https://climatecommunication.yale.edu/visualizations-data/americans-climate-views/
73
viewed as stronger because they ultimately fail to offer solutions for an issue that many members
of the representative’s own party view as legitimate.
Again, it is worth noting that because our analyses did not uncover any significant
interactions between tone, informativeness, and political orientation, this means the findings
above hold for both Democratic and Republican participants. For example, both Democratic and
Republican participants perceived the Democrat’s argument as stronger when they provided
more well-reasoned arguments. And, the level of informativeness included in the tweets had no
influence on either group’s judgements of the politicians.
Influence of tone and informativeness on policy support
Despite influences on argument strength, learning, and perceptions of politicians, our
independent variables stopped short of shifting participants’ support for the climate policies—
neither tone nor informativeness influenced support for either a carbon tax or more stringent
CAFE standards. This may be because in all conditions, communication styles were consistent
between the representatives: in other words, both parties were polite or both were impolite, and
both parties were either informative or uninformative. Therefore, each side may have benefited
equally from improved civility and greater informativeness, giving neither an advantage that
would sway a reader’s opinion.
At the same time, if participants perceived both representatives to be presenting strong
arguments, we questioned if participants’ stances on the policy would be moderated. For
example, participants may have been compelled to moderate their views of the climate policy
after becoming more aware of or open to the other side’s reasoning (due to exposure to a high
level of informativeness or a positive tone). While we did find participants to be more open to
arguments of the other side when political animosity was absent from tweets (as evidenced by
74
rating arguments as stronger in the positive tone condition), policy support did not increase
among Republicans, or decrease among Democrats.
Thus, even though social media platforms such as Twitter are arenas in which users may
attempt to persuade others (Gil de Zúñiga et al., 2012; Nielsen & Schrøder, 2014), it is plausible
that users of these platforms gravitate to them—and to certain spaces within a social media
landscape—so as to confirm or affirm their previously held viewpoints. Indeed, several recent
studies (e.g., Modgil et al. (2021), Workman (2018)) suggest that users of these platforms may
be prone to confirmation bias (Nickerson, 1998) and, thereby, cherry-pick information that
supports the opinions that they brought to the platform vs. the opinions contained in the
information they are being exposed to.
Implications
Because we did not compare the exposure to a Twitter conversation about climate policy
to exposure to a conversation completely unrelated to climate policy, we cannot make any
statements about the influence of encountering climate information on Twitter versus not
encountering it at all. However, since greater informativeness in particular did not increase
policy support (at least when matched with an opposing argument) it may point to the minimal
influence that conveying more information about climate change on Twitter may have when it is
encountered in small doses. Consistent with work by Drummond et al. (2020) that found a small
exposure to online misinformation on climate change has little effect on beliefs and attitudes, our
findings here may point to the minimal influence of reading a single conversation on policy
support. More research should be done to examine the cumulative impact of exposure to
information about climate change on social media.
75
Even though participants’ policy opinions may not be moved by exposure to more
informative conversations, a higher level of informativeness (and a more positive tone) did
increase perceived learning. Thus, participants still were aware that they encountered new
information, which could, with repeated exposure, indicate that Twitter can play a role in making
its users more politically informed. Relatedly, it may be somewhat concerning to see tone play
an outsized role in determining people’s perceptions of learning and argument strength. This
could, for example, mean that if politicians were to communicate respectfully, people that
witness those conversations may leave with a false sense of feeling informed.
It is also worth reiterating that our findings related to our main independent variables
(tone and informativeness) did not differ by participants’ political orientations. Overall,
arguments were viewed as stronger when the dialogue was informative and polite regardless of
whether the reader’s political orientation aligned with that of the representative’s. And,
respectful politicians were viewed more favorably (as compared to rude or aggressive
politicians) by participants from both sides of the political aisle. We view this finding as
somewhat uplifting—despite high levels of reported polarization in U.S. society (Pew Research
Center, 2022), this supports the idea that animosity is still not how many members of the public
would prefer to see our representatives engage with one another.
Ultimately, while policy support was not affected by our manipulations, the perceptions
people have of arguments for climate policy (Druckman et al., 2013), and the perceptions people
have of political representatives (Harring & Jagers, 2013), are likely important factors for
garnering policy support over time. Therefore, the takeaway for those communicating about
climate change may still be to favor polite and informative communication styles on social
media.
76
Limitations and future research
Although the Twitter conversations we created for this study were inspired by the types
of communication we have seen on the platform, the tweets were not pulled directly from
Twitter, and thus there is a limitation to how generalizable our results are in terms of how readers
might interact with real Twitter conversations. Relatedly, participants were not actually using the
Twitter platform, thus their behavior may differ from choices they make when truly using the
site. For example, people may have read our Twitter conversation more carefully (in order to
pass attention checks, ensure payment, etc.) than if they were to quickly encounter it in their
Twitter feed. In addition, because many statistical tests were conducted in our analysis (due to
measuring a variety of independent and dependent variables) it is important to interpret our
results keeping in mind that p-values were not adjusted for multiple comparisons (although
oftentimes significant p-values were less than 0.001).
As discussed above, we did not test the impact of reading a climate policy dialogue
compared to a neutral control dialogue; rather, we compared the influence of differing types of
climate policy conversations to each other. Comparing to a neutral conversation could offer
insight as to if reading tweets about climate change can influence policy support as compared to
reading nothing about climate change on Twitter at all. Future research should explore this
potential impact.
Relatedly, we did not create conversations where the tone and level of informativeness
varied between representatives (i.e., one using polite communication while the other uses
aggressive or rude communication). Exploring this further could uncover useful insights about
climate policy communication on Twitter—for example, perhaps taking the time to provide
77
detailed explanations about climate policy is even more worthwhile (in terms of swaying
opinions) when a policy opponent is cursory with their language.
Finally, we cannot generalize the findings of our study beyond the realm of politicians on
Twitter. For example, the effects of tone and informativeness that influenced participants
perceptions in this study may have been a function of people’s expectations about how
policymakers can and should communicate. When it comes to other groups of people
communicating about politics (e.g., policy advocates, lobbyists, everyday citizens), we might not
see the same results. Remaining in the domain of communication among politicians, though, the
positive role of polite and informative communication found here may indicate that studying
these stylistic choices when communicating about climate change outside of the Twitter space
(e.g., in a political debate) may be worthwhile as well.
Conclusion
Addressing climate change and advancing climate policy will require cooperation between
groups of people that have differing or opposing interests and values. Yet, cooperative
conversations usually rely on some amount of openness towards others, which is often not
extended across the political aisle, especially when communication is taking place on social
media. This study, though, may offer some incentive for politicians to consider changing their
ways. When debating the merits of a climate policy on Twitter in a respectful manor, politicians
were viewed as more trustworthy, likeable, and competent by participants. In addition,
politicians’ arguments were also viewed as stronger. Offering more relevant information about
the policy sometimes had this effect as well. Thus, despite the trolls, bullying, and incivility that
can be characteristic of communication online, participants here seemed to prefer when elected
officials debate in a more collegial communication style. Politicians on Twitter take note.
78
Concluding Remarks
The research covered in this dissertation explores three different contexts—
misinformation interventions on Facebook, perceptions of a new climate technology, and
conveying information about climate policies on Twitter—all related to the pursuit of effective
communication about climate change. Through becoming familiar with these topics, and through
examining these studies’ findings, in my view, several significant themes have emerged. The
first being that there is no one “right” answer to our questions and no one right way to move
forward, as many varied approaches will be required to tackle the problems we face. In climate
circles, it seems much energy is spent discussing what people view as dichotomies. The current
quintessential example of this being whether we should focus the bulk of our climate efforts on
personal behavior change or policy/systemic change, even though ultimately most agree that
both are necessary. In my own research, I have also found “both” to often be the correct
response.
In the context of misinformation online, some question whether it is more political bias or
a lack of critical thinking that drives social media users to share fake news (Pennycook & Rand,
2018). With participants in our study being significantly less likely to trust, like, and share fake
climate news after exposure to a critical thinking intervention, our results suggest critical
thinking plays an important role. At the same time, our results also point to the challenge
political polarization poses to climate communication and climate progress. In our sample, there
was in fact a portion of conservative participants that believe in human caused climate change
yet chose to like and share fake climate news anyways (likely because it aligned with their
79
political priorities).
25
With this, both are likely contributors to the spread of misinformation
online, so the question should not be if critical thinking interventions like these should be
implemented, instead, we should say, this and what else?
When it comes to carbon capture and utilization (CCU), some oppose the technology
because they believe carbon removal is a distraction from the more important or more
established methods for limiting climate change. For example, some are concerned that carbon
removal will be prioritized over the development of renewables (Nisbet, 2019). However, again,
I do not believe this should be an either/or scenario—legitimate promotors of the technology
emphasize that the expansion of carbon removal is only one component that will be necessary for
reaching net-zero carbon emissions. And even though transitioning to renewables is of the
utmost importance, there are still reasons for working to expand carbon removal alongside. One
being that, even after society reaches net-zero emissions, the concentration of CO2 in the air will
still be too high
26
. Ideally, when we reach that net-zero goal, the technology will have already
been developed at a scale such that it is ready to remove a meaningful amount of CO2 from the
atmosphere. Therefore, rather than debating the merits of where to focus our efforts, considering
our study’s finding of bipartisan acceptance for CCU-based products, we may want to instead
explore the carbon removal space as a rare opportunity for bipartisan collaboration.
Another theme that has stood out to me is that there are plenty of reasons for both hope
and hopelessness when it comes to the future of addressing climate challenges. The results from
our misinformation study, for example, can beget both hopeful and discouraging interpretations.
On one hand, it could be seen as positive that improved media literacy (i.e., understanding
25
Likewise, considering our study on Twitter communication styles, both informativeness and civility mattered in
influencing participants’ judgements, but ultimately, these judgements were still also influenced by one’s political
orientation.
26
https://www.carbonbrief.org/explainer-will-global-warming-stop-as-soon-as-net-zero-emissions-are-reached/
80
certain cues to look for when evaluating news online) does make a significant difference when it
comes to reducing trust in and engagement with misinformation. On the other, approximately a
third of our sample indicated at least a slight openness to potentially sharing fake climate news, a
figure which, in and of itself, is daunting.
For carbon capture and utilization, we see encouraging evidence for bipartisan support in
our study. Yet with conservatives often opposed to climate initiatives, and voices from some
progressives expressing resistance to carbon removal technology, the support we found could
become diminished from either or both sides of the aisle.
And for effective communication about climate change on Twitter, I find it uplifting that
a positive tone improved evaluations of politicians (in terms of their likeability, trustworthiness,
and competence), even when participants were evaluating an opposing party member. Yet, it is
less uplifting that participants were not sensitive to the quality of information conveyed in
tweets, in that politicians that were more informative were not viewed as more competent (or
trustworthy or likeable). In addition, with Twitter’s new CEO Elon Musk allowing hate speech
on the site
27
, I do not feel hopeful about a reduction in incivility on the platform. Although, on
November 27
th
, 2022, Musk did tweet, “Just a note to encourage people of different political or
other views to engage in civil debate on Twitter. Worst case, the other side has a slightly better
understanding of your views,”
28
which ironically (considering the platform’s new speech policy)
aligns with what I view to be a key takeaway of our study.
Finally and unsurprisingly, political polarization, which is already known to pose
challenges for climate progress, has been at the crux of many of the issues that have been
27
https://www.brookings.edu/blog/how-we-rise/2022/11/23/why-is-elon-musks-twitter-takeover-increasing-hate-
speech/
28
https://twitter.com/elonmusk/status/1596942103937921025?cxt=HHwWgsCtgb6YvaksAAAA
81
discussed. Indeed, the less hopeful conclusions covered here are in many ways related to political
polarization. And considering the current state of a vastly divided nation, at times identifying
ways to move forward can feel like an insurmountable challenge. But despite the frustration and
despair some experience in response to this, expressing these emotions through disdain for the
“other” side may serve to exacerbate the division that is the problem. With this, the research that
has been presented in these chapters had led me to believe that effective communication about
climate change should, when possible, aim to highlight common ground, or at the very least, not
feed into the polarization that already exists. In the end, I am convinced the best way to do this is
to communicate with empathy and without judgement.
82
References
Achterberg, P. (2014, 2014/05/01). Knowing hydrogen and loving it too? Information provision,
cultural predispositions, and support for hydrogen technology among the Dutch. Public
Understanding of Science, 23(4), 445-453. https://doi.org/10.1177/0963662512453117
Allcott, H., & Gentzkow, M. (2017). Social media and fake news in the 2016 election. Journal of
Economic Perspectives, 31(2), 211-236. https://doi.org/10.1257/jep.31.2.211
Anderson, A. A., & Huntington, H. E. (2017, 2017/10/01). Social Media, Science, and Attack
Discourse: How Twitter Discussions of Climate Change Use Sarcasm and Incivility.
Science Communication, 39(5), 598-620. https://doi.org/10.1177/1075547017735113
Arning, K., Offermann-van Heek, J., Linzenich, A., Kaetelhoen, A., Sternberg, A., Bardow, A.,
& Ziefle, M. (2019). Same or different? Insights on public perception and acceptance of
carbon capture and storage or utilization in Germany. Energy Policy, 125, 235-249.
https://doi.org/10.1016/j.enpol.2018.10.039
Arning, K., Offermann-van Heek, J., Sternberg, A., Bardow, A., & Ziefle, M. (2019, 5/2019).
Risk-benefit perceptions and public acceptance of Carbon Capture and Utilization.
Environmental Innovation and Societal Transitions, 35, 292-308.
https://doi.org/10.1016/j.eist.2019.05.003
Arning, K., van Heek, J., & Ziefle, M. (2018, 1/7/2018
7/2018). Acceptance profiles for a carbon-derived foam mattress. Exploring and segmenting
consumer perceptions of a carbon capture and utilization product. Journal of Cleaner
Production, 188, 171-184. https://doi.org/10.1016/j.jclepro.2018.03.256
Aronow, P. M., Kalla, J., Orr, L., & Ternovski, J. (2020). Evidence of Rising Rates of
Inattentiveness on Lucid in 2020. Investment Weekly News, Preprint, 402.
Árvai, J. (2014). The end of risk communication as we know it. Journal of Risk Research, 17,
1245-1249. https://doi.org/10.1080/13669877.2014.919519
Árvai, J., & Gregory, R. (2021). Beyond choice architecture: A building code for structuring
climate risk management decisions. Behavioural Public Policy, 5(4), 556-575.
https://doi.org/10.1017/bpp.2020.37
Baena-Moreno, F. M., Rodríguez-Galán, M., Vega, F., Alonso-Fariñas, B., Vilches Arenas, L.
F., & Navarrete, B. (2019, 18/6/2019
6/2019). Carbon capture and utilization technologies: a literature review and recent advances.
Energy Sources, Part A: Recovery, Utilization, and Environmental Effects, 41(12), 1403-
1433. https://doi.org/10.1080/15567036.2018.1548518
83
Beck, J. (2017). This Article Won’t Change Your Mind.
https://www.theatlantic.com/science/archive/2017/03/this-article-wont-change-your-
mind/519093/
Beiler, M., & Kiesler, J. (2018). “Lügenpresse! Lying press!” Is the Press Lying? In K. Otto &
A. Köhler (Eds.), Trust in Media and Journalism: Empirical Perspectives on Ethics,
Norms, Impacts and Populism in Europe (pp. 155-179). Springer Fachmedien
Wiesbaden. https://doi.org/10.1007/978-3-658-20765-6_9
Bessette, D. L., Wilson, R. S., & Arvai, J. L. (2019, 2019). Do people disagree with themselves?
Exploring the internal consistency of complex, unfamiliar, and risky decisions. Journal of
Risk Research, 1-13. https://doi.org/10.1080/13669877.2019.1569107
Björnberg, K. E., Karlsson, M., Gilek, M., & Hansson, S. O. (2017, 20/11/2017 11/2017).
Climate and environmental science denial: A review of the scientific literature published
in 1990–2015. Journal of Cleaner Production, 167, 229-241.
https://doi.org/10.1016/j.jclepro.2017.08.066
Bornstein, R. F. (1989, 9/1989
1989). Exposure and Affect. Psychological Bulletin, 106(2), 265-289.
https://doi.org/10.1037/0033-2909.106.2.265
Breyer, C., Fasihi, M., Bajamundi, C., & Creutzig, F. (2019, 2019/09/18/). Direct Air Capture of
CO2: A Key Technology for Ambitious Climate Change Mitigation. Joule, 3(9), 2053-
2057. https://doi.org/https://doi.org/10.1016/j.joule.2019.08.010
Bronstein, M. V., Pennycook, G., Bear, A., Rand, D. G., & Cannon, T. D. (2019, 3/2019 2019).
Belief in fake news is associated with delusionality, dogmatism, religious
fundamentalism, and reduced analytic thinking. Journal of Applied Research in Memory
and Cognition, 8(1), 108-117. https://doi.org/10.1016/j.jarmac.2018.09.005
Burgoon, J. K., & Hale, J. L. (1988, 1988/03/01). Nonverbal expectancy violations: Model
elaboration and application to immediacy behaviors. Communication Monographs, 55(1),
58-79. https://doi.org/10.1080/03637758809376158
Campbell-Arvai, V., Bessette, D., Wilson, R., & Arvai, J. (2018). Decision-making about the
environment. In T. Marsden (Ed.), The Sage Handbook of Nature (pp. 487-511). Sage.
Campbell-Arvai, V., Hart, P. S., Raimi, K. T., & Wolske, K. S. (2017, 2017/08/01). The
influence of learning about carbon dioxide removal (CDR) on support for mitigation
policies. Climatic change, 143(3), 321-336. https://doi.org/10.1007/s10584-017-2005-1
Chinn, S., Hart, P. S., & Soroka, S. (2020, 2/2020
2020). Politicization and Polarization in Climate Change News Content, 1985-2017. Science
Communication, 42(1), 112-129. https://doi.org/10.1177/1075547019900290
84
CO2 Sciences and The Global CO2 Initiative. (2016). A Roadmap for the Implementation of
Carbon Utilization Technologies (Executive Summary).
https://assets.ctfassets.net/xg0gv1arhdr3/5VPLtRFY3YAIasum6oYkaU/48b0f48e32d6f4
68d71cd80dbd451a3a/CBPI_Roadmap_Executive_Summary_Nov_2016_web.pdf
Cole, J. C., Ehret, P. J., Sherman, D. K., & Van Boven, L. (2022, 2022/07/18). Social norms
explain prioritization of climate policy. Climatic change, 173(1), 10.
https://doi.org/10.1007/s10584-022-03396-x
Cox, E., Spence, E., & Pidgeon, N. (2020, 2020/08/01). Public perceptions of carbon dioxide
removal in the United States and the United Kingdom. Nature Climate Change, 10(8),
744-749. https://doi.org/10.1038/s41558-020-0823-z
Dear, M. (1992). Understanding and overcoming the NIMBY syndrome. Journal of the
American planning association, 58(3), 288-300.
https://doi.org/https://doi.org/10.1080/01944369208975808
Dow, A. P., Adamic, L. A., & Friggeri, A. (2013). The Anatomy of Large Facebook Cascades.
Proceedings of the Seventh International AAAI Conference on Weblogs and Social
Media,
Druckman, J. N., Peterson, E., & Slothuus, R. (2013). How Elite Partisan Polarization Affects
Public Opinion Formation. The American political science review, 107(1), 57-79.
https://doi.org/10.1017/S0003055412000500
Drummond, C., & Fischhoff, B. (2017). Individuals with greater science literacy and education
have more polarized beliefs on controversial science topics. Proceedings of the National
Academy of Sciences, 114(36), 9587. https://doi.org/10.1073/pnas.1704882114
Drummond, C., Siegrist, M., & Árvai, J. (2020). Limited effects of exposure to fake news about
climate change. Environmental Research Communications, 2(8), 081003.
https://doi.org/10.1088/2515-7620/abae77
Duetschke, E., Schumann, D., Pietzner, K., Wohlfarth, K., & Höller, S. (2014, 2014). Does it
Make a Difference to the Public Where CO2 Comes from and Where it is Stored? Energy
Procedia, 63, 6999-7010. https://doi.org/10.1016/j.egypro.2014.11.733
Etale, A., Jobin, M., & Siegrist, M. (2018, 1/2/2018
2/2018). Tap versus bottled water consumption: The influence of social norms, affect and image
on consumer choice. Appetite, 121, 138-146. https://doi.org/10.1016/j.appet.2017.11.090
Facebook Help Center. Tips to Spot False News.
https://www.facebook.com/help/188118808357379
85
Farrell, J., McConnell, K., & Brulle, R. (2019, 2019/03/01). Evidence-based strategies to combat
scientific misinformation. Nature Climate Change, 9(3), 191-195.
https://doi.org/10.1038/s41558-018-0368-6
Finucane, M. L., Alhakami, A., Slovic, P., & Johnson, S. M. (2000). The affect heuristic in
judgments of risks and benefits. Journal of behavioral decision making, 13(1), 1-17.
https://doi.org/10.1002/(SICI)1099-0771(200001/03)13:1<1::AID-BDM333>3.0.CO;2-S
Fiske, S. T., & Dupree, C. (2014). Gaining trust as well as respect in communicating to
motivated audiences about science topics. Proceedings of the National Academy of
Sciences - PNAS, 111(Supplement 4), 13593-13597.
https://doi.org/10.1073/pnas.1317505111
Frederick, S. (2005, 2005). Cognitive reflection and decision making. The Journal of Economic
Perspectives, 19(4), 25-42. https://doi.org/10.1257/089533005775196732
Frimer, J. A., Aujla, H., Feinberg, M., Skitka, L. J., Aquino, K., Eichstaedt, J. C., & Willer, R.
(2022). Incivility Is Rising Among American Politicians on Twitter. Social psychological
& personality science. https://doi.org/10.1177/19485506221083811
Gil de Zúñiga, H., Jung, N., & Valenzuela, S. (2012). Social media use for news and individuals'
social capital, civic engagement and political participation. Journal of computer-mediated
communication, 17(3), 319-336. https://doi.org/10.1111/j.1083-6101.2012.01574.x
Gilovich, T., Griffin, D., & Kahneman, D. (2002). Heuristics and biases: The psychology of
intuitive judgment. Cambridge university press.
Global CCS Institute. (2019). The Global Status of CCS: 2019.
Google News Initiative. Building a stronger future for journalism.
https://newsinitiative.withgoogle.com
Gregory, R., Satterfield, T., & Hasell, A. (2016, 19/1/2016 1/2016). Using decision pathway
surveys to inform climate engineering policy choices. Proceedings of the National
Academy of Sciences of the United States of America, 113(3), 560-565.
https://doi.org/10.1073/pnas.1508896113
Grinberg, N., Joseph, K., Friedland, L., Swire-Thompson, B., & Lazer, D. (2019). Fake news on
Twitter during the 2016 U.S. presidential election. Science, 363(6425), 374-378.
https://doi.org/10.1126/science.aau2706
Guess, A., Nagler, J., & Tucker, J. (2019). Less than you think: Prevalence and predictors of fake
news dissemination on Facebook. Science Advances, 5(1), eaau4586.
https://doi.org/10.1126/sciadv.aau4586
86
Guess, A., Nyhan, B., & Reifler, J. (2018). Selective exposure to misinformation: Evidence from
the consumption of fake news during the 2016 U.S. presidential campaign [Working
Paper]. http://www.dartmouth.edu/~nyhan/fake-news-2016.pdf
Harring, N., & Jagers, S. C. (2013). Should we trust in values? Explaining public support for
pro-environmental taxes. Sustainability, 5(1), 210-227.
https://doi.org/10.3390/su5010210
Hornsey, M. J., Harris, E. A., Bain, P. G., & Fielding, K. S. (2016, 25/5/2016 5/2016). Meta-
analyses of the determinants and outcomes of belief in climate change. Nature Climate
Change, 6(6), 622-626. https://doi.org/10.1038/nclimate2943
International Federation of Library Associations and Institutions. (2019). How to Spot Fake
News. https://www.ifla.org/publications/node/11174
IPCC. (2018). Summary for Policymakers: Global Warming of 1.5°C. An IPCC Special Report
on the impacts of global warming of 1.5°C above pre-industrial levels and related global
greenhouse gas emission pathways, in the context of strengthening the global response to
the threat of climate change, sustainable development, and efforts to eradicate poverty.
IPCC. (2022). Climate Change 2022: Mitigation of Climate Change. Contribution of Working
Group III to the Sixth Assessment Report of the Intergovernmental Panel on Climate
Change.
Jang, J. G., Kim, G. M., Kim, H. J., & Lee, H. K. (2016, 30/11/2016
11/2016). Review on recent advances in CO2 utilization and sequestration technologies in
cement-based materials. Construction and Building Materials, 127, 762-773.
https://doi.org/10.1016/j.conbuildmat.2016.10.017
Jones, C. R. (2015). Chapter 15 - Understanding and Assessing Public Perceptions of Carbon
Dioxide Utilisation (CDU) Technologies. In P. Styring, E. A. Quadrelli, & K. Armstrong
(Eds.), Carbon Dioxide Utilisation (pp. 273-283). Elsevier.
https://doi.org/https://doi.org/10.1016/B978-0-444-62746-9.00015-3
Jones, C. R., Olfe-Kräutlein, B., & Kaklamanou, D. (2017). Lay perceptions of Carbon Dioxide
Utilisation technologies in the United Kingdom and Germany: An exploratory qualitative
interview study. Energy research & social science, 34, 283-293.
https://doi.org/10.1016/j.erss.2017.09.011
Kiely, E., & Robertson, L. (2016). How to Spot Fake News.
https://www.factcheck.org/2016/11/how-to-spot-fake-news/
Koytsoumpa, E. I., Bergins, C., & Kakaras, E. (2018, 2/2018). The CO2 economy: Review of
CO2 capture and reuse technologies. The Journal of Supercritical Fluids, 132, 3-16.
https://doi.org/10.1016/j.supflu.2017.07.029
87
Kunda, Z. (1990, 1990). The case for motivated reasoning. Psychological Bulletin, 108(3), 480-
498. https://doi.org/10.1037//0033-2909.108.3.480
L’Orange Seigo, S., Arvai, J., Dohle, S., & Siegrist, M. (2014, 6//). Predictors of risk and benefit
perception of carbon capture and storage (CCS) in regions with different stages of
deployment. International Journal of Greenhouse Gas Control, 25(0), 23-32.
https://doi.org/http://dx.doi.org/10.1016/j.ijggc.2014.03.007
Lau, R. R., Sigelman, L., & Rovner, I. B. (2007, 2007/11/01). The effects of negative political
campaigns: A meta-analytic reassessment. The Journal of politics, 69(4), 1176-1209.
https://doi.org/10.1111/j.1468-2508.2007.00618.x
Lazer, D. M. J., Baum, M. A., Benkler, Y., Berinsky, A. J., Greenhill, K. M., Menczer, F.,
Metzger, M. J., Nyhan, B., Pennycook, G., Rothschild, D., Schudson, M., Sloman, S. A.,
Sunstein, C. R., Thorson, E. A., Watts, D. J., & Zittrain, J. L. (2018). The science of fake
news. Science, 359(6380), 1094. https://doi.org/10.1126/science.aao2998
Lesgold, A. M. (1983). Acquiring expertise.
Li, Q., Liu, G., Cai, B., Leamon, G., Liu, L.-C., Chen, Z.-A., & Li, X. (2017, 10/2017). Public
awareness of the environmental impact and management of carbon dioxide capture,
utilization and storage technology: the views of educated people in China. Clean
Technologies and Environmental Policy, 19(8), 2041-2056.
https://doi.org/10.1007/s10098-017-1387-0
Lichtenstein, S., & Slovic, P. (2006). The Construction of Preference. Cambridge University
Press. https://doi.org/10.1017/CBO9780511618031
Linzenich, A., Arning, K., Offermann-van Heek, J., & Ziefle, M. (2019). Uncovering attitudes
towards carbon capture storage and utilization technologies in Germany: Insights into
affective-cognitive evaluations of benefits and risks. Energy research & social science,
48, 205-218. https://doi.org/10.1016/j.erss.2018.09.017
Lorenzoni, I., & Pidgeon, N. F. (2006). Public views on climate change: European and USA
perspectives. Climatic Change, 77(1-2), 73-95.
Mac Dowell, N., Fennell, P. S., Shah, N., & Maitland, G. C. (2017). The role of CO2 capture and
utilization in mitigating climate change. Nature climate change, 7(4), 243-249.
https://doi.org/10.1038/nclimate3231
Markewitz, P., Kuckshinrichs, W., Leitner, W., Linssen, J., Zapp, P., Bongartz, R., Schreiber, A.,
& Müller, T. E. (2012, 2012). Worldwide innovations in the development of carbon
capture technologies and the utilization of CO 2. Energy and Environmental Science,
5(6), 7281-7305. https://doi.org/10.1039/c2ee03403d
88
Metzger, M. J., Flanagin, A. J., Eyal, K., Lemus, D. R., & McCann, R. M. (2003, 2003/01/01).
Credibility for the 21st Century: Integrating Perspectives on Source, Message, and Media
Credibility in the Contemporary Media Environment. Annals of the International
Communication Association, 27(1), 293-335.
https://doi.org/10.1080/23808985.2003.11679029
Modgil, S., Singh, R. K., Gupta, S., & Dennehy, D. (2021, 2021/11/20). A Confirmation Bias
View on Social Media Induced Polarisation During Covid-19. Information Systems
Frontiers. https://doi.org/10.1007/s10796-021-10222-9
Mosseri, A. (2017). Working to Stop Misinformation and False News.
https://www.facebook.com/facebookmedia/blog/working-to-stop-misinformation-and-
false-news
Mutz, D. C., & Reeves, B. (2005). The new videomalaise: Effects of televised incivility on
political trust. American Political Science Review, 99(1), 1-15.
https://doi.org/10.1017/S0003055405051452
Nemeroff, C., & Rozin, P. (2018). Back in Touch with Contagion: Some Essential Issues.
Journal of the Association for Consumer Research, 3(4), 612-624.
https://doi.org/10.1086/699971
Nickerson, R. S. (1998, 1998/06/01). Confirmation Bias: A Ubiquitous Phenomenon in Many
Guises. Review of General Psychology, 2(2), 175-220. https://doi.org/10.1037/1089-
2680.2.2.175
Nielsen, R. K., & Schrøder, K. C. (2014, 2014/10/02). The relative importance of social media
for accessing, finding, and engaging with news. Digital Journalism, 2(4), 472-489.
https://doi.org/10.1080/21670811.2013.872420
Nir, L. (2011). Motivated reasoning and public opinion perception. Public Opinion Quarterly,
75(3), 504-532. https://doi.org/10.1093/poq/nfq076 %J Public Opinion Quarterly
Nisbet, M. C. (2019). The Carbon Removal Debate: Asking Critical Questions About Climate
Change Futures.
Nisbett, R. E., & Wilson, T. D. (1977). The halo effect: Evidence for unconscious alteration of
judgments. Journal of personality and social psychology, 35(4), 250-256.
https://doi.org/10.1037//0022-3514.35.4.250
Olajire, A. A. (2010, 2010/06/01/). CO2 capture and separation technologies for end-of-pipe
applications – A review. Energy, 35(6), 2610-2628.
https://doi.org/https://doi.org/10.1016/j.energy.2010.02.030
Park, J., & Stoel, L. (2005, 1/2/2005
89
2/2005). Effect of brand familiarity, experience and information on online apparel purchase.
International Journal of Retail & Distribution Management, 33(2), 148-160.
https://doi.org/10.1108/09590550510581476
Pennycook, G., & Rand, D. G. (2018, 2018/06/20/). Lazy, not biased: Susceptibility to partisan
fake news is better explained by lack of reasoning than by motivated reasoning.
Cognition, 188, 39-50. https://doi.org/https://doi.org/10.1016/j.cognition.2018.06.011
Perdan, S., Jones, C. R., & Azapagic, A. (2017, 4/2017). Public awareness and acceptance of
carbon capture and utilisation in the UK. Sustainable Production and Consumption, 10,
74-84. https://doi.org/10.1016/j.spc.2017.01.001
Peters, M., Köhler, B., Kuckshinrichs, W., Leitner, W., Markewitz, P., & Müller, T. E. (2011,
9/19/2011
). Chemical Technologies for Exploiting and Recycling Carbon Dioxide into the Value Chain.
ChemSusChem, 4(9), 1216-1240. https://doi.org/10.1002/cssc.201000447
Pew Research Center. (2020). Congress Soars to New Heights on Social Media.
https://www.pewresearch.org/internet/2020/07/16/1-the-congressional-social-media-
landscape/
Pew Research Center. (2021). News Use Across Social Media Platforms in 2020.
https://www.pewresearch.org/journalism/2021/01/12/news-use-across-social-media-
platforms-in-2020/
Pew Research Center. (2022). As Partisan Hostility Grows, Signs of Frustration With the Two-
Party System. https://www.pewresearch.org/politics/2022/08/09/as-partisan-hostility-
grows-signs-of-frustration-with-the-two-party-system/
Quadrelli, E. A., Armstrong, K., & Styring, P. (2015). Chapter 16 - Potential CO2 Utilisation
Contributions to a More Carbon-Sober Future: A 2050 Vision. In P. Styring, E. A.
Quadrelli, & K. Armstrong (Eds.), Carbon Dioxide Utilisation (pp. 285-302). Elsevier.
https://doi.org/https://doi.org/10.1016/B978-0-444-62746-9.00016-5
Rathje, S., Van Bavel, J. J., & van der Linden, S. (2021, 2021/06/29). Out-group animosity
drives engagement on social media. Proceedings of the National Academy of Sciences,
118(26). https://doi.org/10.1073/pnas.2024292118
Roozenbeek, J., & van der Linden, S. (2019, 1/12/2019 12/2019). Fake news game confers
psychological resistance against online misinformation. Palgrave Communications, 5(1),
1-10. https://doi.org/10.1057/s41599-019-0279-9
Segrè Cohen, A., Love, N. G., Nace, K. K., & Árvai, J. (2020, 21/4/2020
4/2020). Consumers' Acceptance of Agricultural Fertilizers Derived from Diverted and Recycled
Human Urine. Environmental science & technology, 54(8), 5297-5305.
https://doi.org/10.1021/acs.est.0c00576
90
Shearer, E., & Gottfried, J. (2017). News Use Across Social Media Platforms 2017 (Policy File,
Issue. https://www.journalism.org/2017/09/07/news-use-across-social-media-platforms-
2017/
Shi, J., Visschers, V. H. M., & Siegrist, M. (2015, 12/2015 2015). Public Perception of Climate
Change: The Importance of Knowledge and Cultural Worldviews. Risk Analysis, 35(12),
2183-2201. https://doi.org/10.1111/risa.12406
Shi, J., Visschers, V. H. M., Siegrist, M., & Arvai, J. (2016). Knowledge as a driver of public
perceptions about climate change reassessed. Nature Climate Change, 6(8), 759-762.
https://doi.org/10.1038/nclimate2997
Siegrist, M., Cousin, M.-E., Kastenholz, H., & Wiek, A. (2007, 2007). Public acceptance of
nanotechnology foods and food packaging: The influence of affect and trust. Appetite,
49(2), 459-466. https://doi.org/10.1016/j.appet.2007.03.002
Slovic, P. (1995). The construction of preference. American psychologist, 50(5), 364.
https://doi.org/10.1037/0003-066X.50.5.364
Slovic, P., Fischhoff, B., & Lichtenstein, S. (1977). Behavioral decision theory. Annual review of
psychology, 28(1), 1-39. https://doi.org/10.1146/annurev.ps.28.020177.000245
Smidt, C. D. (2017, 1/4/2017
4/2017). Polarization and the decline of the American floating voter. American Journal of
Political Science, 61(2), 365-381. https://doi.org/10.1111/ajps.12218
Smith, S. The Quick Guide to Spotting Fake News. https://www.freedomforuminstitute.org/first-
amendment-center/primers/fake-news-primer/
Stop Funding Heat. (2021). InDenial – Facebook’s Growing Friendship With Climate
Misinformation. https://stopfundingheat.info/facebook-in-denial/
Sturgis, P., & Allum, N. (2004, 2004/01/01). Science in society: Re-evaluating the deficit model
of public attitudes. Public Understanding of Science, 13(1), 55-74.
https://doi.org/10.1177/0963662504042690
Taber, C. S., & Lodge, M. (2006, 7/2006 2006). Motivated skepticism in the evaluation of
political beliefs. American Journal of Political Science, 50(3), 755-769.
https://doi.org/10.1111/j.1540-5907.2006.00214.x
Theocharis, Y., Barberá, P., Fazekas, Z., & Popa, S. A. (2020, 2020/04/01). The dynamics of
political incivility on twitter. SAGE Open, 10(2), 2158244020919447.
https://doi.org/10.1177/2158244020919447
91
Thomson, K. S., & Oppenheimer, D. M. (2016, 1/1/2016 1/2016). Investigating an alternate form
of the cognitive reflection test. Judgment and Decision Making, 11(1), 99-113.
Thorson, K., Vraga, E., & Ekdale, B. (2010, 2010/06/03). Credibility in context: How uncivil
online commentary affects news credibility. Mass Communication and Society, 13(3),
289-313. https://doi.org/10.1080/15205430903225571
Tobler, C., Visschers, V. H., & Siegrist, M. (2012). Consumers’ knowledge about climate
change. Climatic change, 114(2), 189-209.
van der Linden, S. (2017). How to Spot Fake News.
https://www.psychologytoday.com/us/blog/socially-relevant/201711/how-spot-fake-news
van Heek, J., Arning, K., & Ziefle, M. (2017a, 2017/07/01/). Differences between Laypersons
and Experts in Perceptions and Acceptance of CO2-utilization for Plastics Production.
Energy Procedia, 114, 7212-7223.
https://doi.org/https://doi.org/10.1016/j.egypro.2017.03.1829
van Heek, J., Arning, K., & Ziefle, M. (2017b, 6/2017
2017). Reduce, reuse, recycle: Acceptance of CO2-utilization for plastic products. Energy
Policy, 105, 53-66. https://doi.org/10.1016/j.enpol.2017.02.016
Vosoughi, S., Roy, D., & Aral, S. (2018). The spread of true and false news online. Science,
359(6380), 1146. https://doi.org/10.1126/science.aap9559
Workman, M. (2018). An empirical study of social media exchanges about a controversial topic:
Confirmation bias and participant characteristics. The Journal of Social Media in Society,
7(1), 381-400.
Yuan, S., Besley, J. C., & Lou, C. (2018, 2018/04/03). Does being a jerk work? Examining the
effect of aggressive risk communication in the context of science blogs. Journal of Risk
Research, 21(4), 502-520. https://doi.org/10.1080/13669877.2016.1223159
Yuan, S., Ma, W., & Besley, J. C. (2019, 2019/06/01). Should scientists talk about GMOs
nicely? exploring the effects of communication styles, source expertise, and preexisting
attitude. Science Communication, 41(3), 267-290.
https://doi.org/10.1177/1075547019837623
Zompetti, J. P. (2019). Rhetorical incivility in the Twittersphere: A comparative thematic
analysis of Clinton and Trump’s tweets during and after the 2016 presidential election.
Journal of Contemporary Rhetoric, 9, 29-54.
92
Appendix A: Chapter I Tables and Figures
Figure 1. The fake news (A, B, and C) and real (D, E, and F) news posts used in this experiment.
93
Table 1. Sample Characteristics
n Age
Percent
Women
Percent
College
*
Percent
Skeptic
**
Fake News 1,397 44.40 (sd = 15.26) 50% 43% 50%
Breitbart 462 43.29 (sd = 14.72) 49% 41% 49%
Control 154 43.41 (sd = 15.12) 53% 35% 51%
Guidelines 159 42.57 (sd = 14.33) 45% 44% 47%
Enhanced 149 43.95 (sd = 14.77) 48% 45% 49%
Info Wars 458 45.32 (sd = 15.87) 50% 45% 49%
Control 165 45.22 (sd = 15.87) 51% 49% 52%
Guidelines 151 44.66 (sd = 15.82) 50% 41% 50%
Enhanced 142 46.13 (sd = 15.99) 49% 44% 44%
Natural News 477 44.58 (sd = 15.16) 51% 44% 51%
Control 158 44.34 (sd = 15.06) 53% 40% 52%
Guidelines 154 43.85 (sd = 15.38) 49% 48% 50%
Enhanced 165 45.48 (sd = 15.10) 50% 43% 50%
Real News 1,353 43.81 (sd = 14.93) 50% 45% 50%
NASA 452 43.75 (sd = 15.27) 48% 47% 48%
Control 151 43.47 (sd = 15.48) 50% 49% 48%
Guidelines 146 44.51 (sd = 15.50) 50% 42% 49%
Enhanced 155 43.30 (sd = 14.90) 44% 49% 48%
Scientific American 437 44.11 (sd = 14.66) 51% 45% 52%
Control 145 44.37 (sd = 14.64) 50% 46% 50%
Guidelines 143 43.66 (sd = 14.62) 51% 40% 52%
Enhanced 149 44.30 (sd = 14.81) 52% 48% 53%
USA Today 464 43.58 (sd = 14.86) 52% 44% 51%
Control 161 41.16 (sd = 14.64) 54% 39% 50%
Guidelines 164 44.71 (sd = 14.49) 54% 46% 51%
Enhanced 139 45.04 (sd = 15.31) 47% 47% 53%
Total Sample 2,750 44.11 (sd = 15.10) 50% 44% 50%
*
Reflects percentage of participants who completed a degree in higher education (associates, bachelor, or graduate degree).
**
Reflects percentage of participants who did not believe in anthropogenic climate change, or who were unsure.
€
x
94
Table 2. Mean ratings of perceived trustworthiness, likelihood of “liking,”, and likelihood of “sharing”
across post type (fake news and real news).
TRUST
LIKE
SHARE
Fake News Mean SD Mean SD Mean SD
All Conditions Collapsed 4.08 2.53 4.23 3.21 4.10 3.21
Control 4.28 2.55 4.55 3.25 4.51 3.32
Guidelines 3.93 2.60 4.12 3.27 4.00 3.24
Enhanced Guidelines 4.03 2.44 4.00 3.09 3.80 3.03
Real news
All Conditions Collapsed 6.56 2.34 6.24 3.16 5.69 3.19
Control 6.38 2.34 6.17 3.23 5.65 3.20
Guidelines 6.69 2.40 6.32 3.17 5.78 3.15
Enhanced Guidelines 6.60 2.27 6.22 3.09 5.65 3.21
95
Table 3. Regression analyses for climate change doubters and believers on perceived trustworthiness of, likelihood of liking, and likelihood of
sharing posts based on fake news.
96
Table 4. Regression analyses for climate change doubters and believers on perceived trustworthiness of, likelihood of liking, and likelihood of
sharing posts based on real news.
97
Appendix B: Chapter II Tables and Figures
Table 1. Sample Characteristics
n Mean Age
Percent
Women
Percent
College*
Percent
Liberal**
Percent
Independent
Percent
Conservative***
Direct Air Capture 1,176 42.97 (sd = 15.04) 51% 49% 39% 31% 31%
Carbonated beverage 296 44.21 (sd = 14.83) 50% 44% 40% 27% 33%
Plastic container 290 41.97 (sd = 15.33) 50% 50% 40% 32% 28%
Furniture 292 42.75 (sd = 14.49) 52% 53% 37% 32% 31%
Shatterproof glass 298 42.91 (sd = 15.49) 51% 51% 37% 32% 31%
Point source capture 1,224 42.74 (sd = 15.33) 54% 46% 38% 33% 28%
Carbonated beverage 310 42.84 (sd = 16.07) 53% 44% 38% 30% 33%
Plastic container 284 42.54 (sd = 15.49) 54% 49% 41% 33% 26%
Furniture 315 42.63 (sd = 14.74) 53% 46% 41% 33% 26%
Shatterproof glass 315 42.93 (sd = 15.08) 54% 44% 34% 37% 29%
Total Sample 2,400 42.85 (sd = 15.19) 52% 48% 38% 32% 30%
*Reflects percentage of participants who completed a degree in higher education (associates, bachelor, or graduate degree).
**Includes: somewhat liberal, liberal, and very liberal
***Includes: somewhat conservative, conservative, and very conservative
98
Table 2. Regression analysis for willingness to consume or use a CCU-based product
†
Willingness to Consume or Use
β 95% CI η
2
Intercept 5.26*** 4.82; 5.70 -
Negative affect -0.46*** -0.51; -0.42 0.139
Risk to health -0.22*** -0.26; -0.18 0.053
Benefit to climate 0.22*** 0.18; 0.26 0.050
Environmentally conscious behavior 0.05** 0.02; 0.09 0.003
Moral hazard 0.04 -0.01; 0.08 0.001
Technological optimism 0.07** 0.02; 0.12 0.004
Knowledge of climate change -0.04* -0.06; -0.01 0.002
Knowledge of carbon dioxide 0.00 -0.03; 0.03 0.000
Political ideology
††
0.03 0.00; 0.05 0.002
Education -0.00 -0.03; 0.03 0.000
Gender (0 = man) -0.05 -0.14; 0.03 0.000
Age -0.00 0.00; 0.00 0.000
R
2
0.47
F(df1, df2) 175(12, 2335)
* p < 0.05, ** p < 0.01, *** p < 0.001.
†
Excludes 52 participants that indicated they do not believe in climate change. Mean
willingness to consume or use a CCU-based product was 4.2 (SD = 1.7) for these participants
††
Political ideology: continuous variable from very liberal to very conservative
99
Figure 1. Distribution of participant responses regarding willingness to consume or use a CCU-based product,
perception of climate change mitigation benefits, and perception of health risks.
100
Figure 2. ANOVA results and Tukey’s post-hoc tests comparing the willingness to consume or use different product
types from DAC and point source capture. Error bars represent standard errors. Key to post-hoc comparisons:
a
DAC-CB ≠ DAC-PC (p < 0.001); DAC-SG (p < 0.001); PSC-CB (p < 0.01); PSC-PC (p < 0.05); PSC-SG (p <
0.001)
b
PSC-CB ≠ PSC-PC (p < 0.001); PSC-F (p < 0.001); PSC-SG (p < 0.001); DAC-PC (p < 0.001); DAC-F (p <
0.001); DAC-SG (p < 0.001)
101
Appendix C: Chapter II Supplementary Materials
Table S1. Descriptive statistics for survey items.
102
Table S1. Descriptive statistics for survey items continued.
103
Table S2. Pearsons correlation matrix for variables measured, with probability values adjusted for multiple comparisons.
* p < 0.05, ** p < 0.01, *** p < 0.001..
104
Table S3. ANOVA results comparing willingness to consume or use a CCU-based product
between political ideologies.
Willingness to
consume or use
SE
Very liberal 5.23 0.09
Liberal 5.04 0.09
Somewhat liberal 5.01 0.07
Independent 4.97 0.05
Somewhat conservative 4.97 0.08
Conservative 5.00 0.08
Very conservative 4.91
0.14
F 1.13
p 0.249
η
2
0.003
€
x
105
Participant Stimuli
CCU PRIMER – DIRECT AIR CAPTURE
Every day, large amounts of carbon dioxide (CO 2) are generated by the combustion of fossil
fuels – like coal, oil, and natural gas – at factories and power plants. The rise in the amount of
CO 2 in the Earth’s atmosphere is one of the most important contributing factors to climate
change.
A technology called Direct Air Capture is available to reduce the amount of CO 2 already present
in the atmosphere. Using machines that are several stories tall, this technology relies on fans to
draw air into the machine, where it undergoes a chemical process to remove the CO 2.
After the CO 2 has been captured, it is purified and then taken to a production facility where it
can be used as a raw material to make a variety of products to be used by people. For example:
• Captured CO 2 can be used directly as the gas that puts the “fizz” in drinks like sodas, seltzer
water, sparkling wine, and beer; and
• The carbon in captured CO 2 can be removed and used as a raw material for creating
polycarbonates, otherwise known as plastic.
When CO 2 is captured and then used as a raw material in another product, the process is called:
“Carbon Capture and Utilization”, or “CCU” for short
Carbon Capture and Utilization (CCU) is growing in popularity because it can reduce the
amount of CO 2 in the atmosphere, thereby reducing the risk of climate change.
On the next page, we tell you more about specific products that could be created from Carbon
Capture and Utilization (CCU). Then, we will ask you to answer a few questions about them.
106
CCU PRIMER – POINT SOURCE CAPTURE
Every day, large amounts of carbon dioxide (CO 2) are generated by the combustion of fossil
fuels – like coal, oil, and natural gas – at factories and power plants. The rise in the amount of
CO 2 in the Earth’s atmosphere is one of the most important contributing factors to climate
change.
Technologies are available to reduce the amount of CO 2 entering the atmosphere from
factories and power plants that burn fossil fuels. Specifically, it is possible to capture CO 2 inside
of a smokestack before the CO 2 would enter the atmosphere.
After the CO 2 has been captured, it is purified and then taken to a production facility where it
can be used as a raw material to make a variety of products to be used by people. For example:
• Captured CO 2 can be used directly as the gas that puts the “fizz” in drinks like sodas, seltzer
water, sparkling wine, and beer; and
• The carbon in captured CO 2 can be removed and used as a raw material for creating
polycarbonates, otherwise known as plastic.
When CO 2 is captured and then used as a raw material in another product, the process is called:
“Carbon Capture and Utilization”, or “CCU” for short
Carbon Capture and Utilization (CCU) is growing in popularity because it can reduce the
amount of CO 2 in the atmosphere, thereby reducing the risk of climate change.
On the next page, we tell you more about specific products that could be created from Carbon
Capture and Utilization (CCU). Then, we will ask you to answer a few questions about them.
107
PRODUCT DESCRIPTIONS
A: DIRECT CONSUMPTION – DAC
Once purified CO 2 has been recovered at a Direct Air Capture facility, the gas can be used
directly by injecting it into liquids to make “fizzy”, carbonated beverages – such as sodas (e.g.,
cola), seltzer water, sparkling wine, and beer – for people to consume.
A: DIRECT CONSUMPTION – POINT SOURCE CAPTURE
After CO 2 has been captured from a smokestack at a factory or power plant, the gas can be
purified and used directly by injecting it into liquids to make “fizzy”, carbonated beverages –
such as sodas (e.g., cola), seltzer water, sparkling wine, and beer – for people to consume.
108
B: CONTACT (NEAR) – DAC
Once purified CO 2 has been recovered at a Direct Air Capture facility, the carbon can be
removed from the CO 2 to make plastic. This plastic can then be used to make a variety of
products for daily use by people like reusable water bottles and containers for storing food.
B: CONTACT (NEAR) – POINT SOURCE CAPTURE
After CO 2 has been captured from a smokestack at a factory or power plant, the gas can be
purified, and the carbon can be removed from the CO 2 to make plastic. This plastic can then be
used to make a variety of products for daily use by people like reusable water bottles and
containers for storing food.
109
C: CONTACT (FAR) – DAC
Once purified CO 2 has been recovered at a Direct Air Capture facility, the carbon can be
removed from the CO 2 to make plastic and foam. This plastic and foam can then be used to
make a variety of products for daily use by people, like chairs to sit in, and the foam used in
sofa cushions.
C: CONTACT (FAR) – POINT SOURCE CAPTURE
After CO 2 has been captured from a smokestack at a factory or power plant, the gas can be
purified, and the carbon can be removed from the CO 2 to make plastic and foam. This plastic
and foam can then be used to make a variety of products for daily use by people, like chairs to
sit in, and the foam used in sofa cushions.
110
D: CONTACT (MINIMAL) – DAC
Once purified CO 2 has been recovered at a Direct Air Capture facility, the carbon can be
removed from the CO 2 to make plastic. This plastic can then be used to make shatterproof glass
used in windows and doors for houses and buildings.
D: CONTACT (MINIMAL) – POINT SOURCE CAPTURE
After CO 2 has been captured from a smokestack at a factory or power plant, the gas can be
purified, and the carbon can be removed from the CO 2 to make plastic. This plastic can then be
used to make shatterproof glass used in windows and doors for houses and buildings.
111
Appendix D: Chapter III Tables and Figures
Table 1. Distribution of Participants by Condition
Scenario Informativeness
Carbon Tax Low High
Tone Negative 243 233
Positive 229 232
Informativeness
CAFE Standards Low High
Tone Negative 231 216
Positive 225 222
112
Table 2. Summary of Main Findings
113
Figure 1. Hypothetical Twitter Conversation Debating a Carbon Tax (Positive Tone and High Informativeness)
Continues to top right.
114
Figure 2. Hypothetical Twitter Conversation Debating a Carbon Tax (Negative Tone and Low Informativeness)
Continues to top right.
115
Figure 3. Responses by Tone, Informativeness, and Climate Policy
Note: Total N = 1831. Error bars represent plus or minus 1 standard error.
116
Figure 4. Responses by Tone, Informativeness, and Political Party (Republican vs. Democrat),
Collapsed Across Climate Policy
Note: Total N = 1232. Error bars represent plus or minus 1 standard error.
117
Appendix E: Chapter III Supplementary Materials
Results from Manipulation Check
Table S1. Tone ratings by condition
Tone Rating
CAFE Mean Standard Dev.
Negative tone 2.98 1.78
Positive tone 5.97 1.08
Carbon Tax
Negative tone 3.41 1.78
Positive tone 5.77 1.21
Coal Ban
Negative tone 3.29 1.77
Positive tone 6.01 1.03
A significant difference was found in negative versus positive tone for all three Twitter conversation
topics (CAFE: t(125) = -12.70, p < .001, Carbon Tax: t(129) = -9.60, p < .001, Coal Ban: t(136) = -12.52,
p < .001).
Table S2. Informativeness ratings by condition
Informativeness Rating
CAFE Mean Standard Dev.
High info. 5.32 1.15
Low info. 4.82 1.44
Carbon Tax
High info. 5.16 1.04
Low info. 4.78 1.22
Coal Ban
High info. 5.12 1.36
Low info. 5.18 1.11
A significant difference was found in low versus high explanatory depth for CAFE and Carbon Tax
treatments, but not for the Coal Ban treatment (CAFE: t(151) = 2.43, p = .016, Carbon Tax: t(151) = 2.10,
p = 0.037, Coal Ban: t(169) = -0.32, p = .753).
118
Supplemental Results from Main Study: Effects of Condition by Political Party
We conducted exploratory analyses assessing whether Democrats and Republicans responded differently
to our experimental manipulations of tone and informativeness. We created a binary variable indicating
whether a respondent self-identified as a Republican (n = 571) vs. a Democrat (n = 661), excluding
Independents and those who did not indicate a political party preference (n = 599). For each dependent
measure, we ran a three-way ANOVA predicting the dependent measure as a function of tone,
informativeness, and self-reported political party (Republican vs. Democrat) and their interactions.
Because the main results reported above were largely consistent across climate policy, we included
climate policy as a main effect in the ANOVAs.
Policy Support: We ran an ANOVA examining the three-way interaction between tone, informativeness,
and self-reported political party (Republican vs. Democrat), with climate policy included as a main effect.
We did not observe any statistically significant two-way or three-way interactions (all p’s > 0.27),
suggesting that the effects of condition on policy support did not differ by political party. We observed a
significant main effect of political party (F(1, 1223) = 107, p < 0.001, partial η
2
= 0.08). Republicans
reported less policy support than Democrats (Figure 4). As found in the full sample, we again observed a
main effect of climate policy (F(1, 1223) = 17.5, p < 0.001, partial η
2
= 0.013), but no main effect of
informativeness (F(1, 1223) = 0.9, p = 0.34, partial η
2
= 0.001) or tone (F(1, 1223) = 0.8, p = 0.37, partial
η
2
= 0.001).
Learning: We ran an ANOVA examining the three-way interaction between tone, informativeness, and
self-reported political party (Republican vs. Democrat), with climate policy included as a main effect. We
did not observe any statistically significant two-way or three-way interactions (all p’s > 0.21), suggesting
that the effects of condition on learning did not differ by political party. We observed a significant main
effect of political party (F(1, 1223) = 4, p = 0.046, partial η
2
= 0.003). Democrats reported learning more
than Republicans (Figure 4). As in the full sample, we again observed a main effect of tone (F(1, 1223) =
11.2, p = 0.001, partial η
2
= 0.009), a main effect of informativeness (F(1, 1223) = 17, p < 0.001, partial η
2
= 0.013) and no effect of climate policy (F(1, 1223) = 0.69, p = 0.41, partial η
2
= 0.001).
Perceived Strength of Argument (Policy Proponent): We did not observe any statistically significant
two-way or three-way interactions (all p’s > 0.06), suggesting that the effects of condition on extremity of
policy support did not differ by political party. We did observe significant main effects of political party
(F(1, 1223) = 20.6, p < 0.001, partial η
2
= 0.016): Democrats rated the policy proponent’s arguments as
stronger (Figure 4). As found above, in the main analyses, informativeness (F(1, 1223) = 8.1, p = 0.004,
partial η
2
= 0.006), tone (F(1, 1223) = 4.3, p = 0.038, partial η
2
= 0.003), and climate policy (F(1, 1223) =
4.5, p = 0.034, partial η
2
=0.004) were associated with perceived argument strength.
Perceived Strength of Argument (Policy Opponent): We ran an ANOVA examining the three-way
interaction between tone, informativeness, and self-reported political party (Republican vs. Democrat),
with climate policy included as a main effect. We did not observe any statistically significant two-way or
three-way interactions (all p’s > 0.34), suggesting that the effects of condition on policy support did not
differ by political party. We observed a significant main effect of political party (F(1, 1223) = 28.5, p <
0.001, partial η
2
= 0.023). Republicans perceived the policy opponent to have stronger arguments (Figure
4). We observed a significant main effect of tone (F(1, 1223) = 8.1, p = 0.004, partial η
2
= 0.006), but no
119
main effect of informativeness (F(1, 1223) = 2.5, p = 0.11, partial η
2
= 0.002), or climate policy (as found
above, F(1, 1223) = 0.06, p = 0.8, partial η
2
< 0.001).
Trust in Policy Proponent: We ran an ANOVA examining the three-way interaction between tone,
informativeness, and self-reported political party (Republican vs. Democrat), with climate policy included
as a main effect. We did not observe any statistically significant two-way or three-way interactions (all
p’s > 0.23), suggesting that the effects of condition on policy support did not differ by political party. We
observed a significant main effect of political party (F(1, 1223) = 29, p < 0.001, partial η
2
= 0.023);
Republicans reported less trust in the policy proponent than Democrats (Figure 4). As found above in the
full sample, we again observed a main effect of tone (F(1, 1223) = 11.5, p = 0.001, partial η
2
= 0.009), but
no main effect of informativeness (F(1, 1223) = 3.24, p = 0.07, partial η
2
= 0.003) or climate policy (F(1,
1223) = 1.1, p = 0.29, partial η
2
= 0.001).
Trust in Policy Opponent: We ran an ANOVA examining the three-way interaction between tone,
informativeness, and self-reported political party (Republican vs. Democrat), with climate policy included
as a main effect. We did not observe any statistically significant two-way or three-way interactions (all
p’s > 0.18), suggesting that the effects of condition on policy support did not differ by political party. We
observed a significant main effect of political party (F(1, 1223) = 28, p < 0.001, partial η
2
= 0.022);
Republicans reported more trust in the policy opponent than Democrats (Figure 4). As we found in the
full sample, we again observed a main effect of tone (F(1, 1223) = 22.3, p < 0.001, partial η
2
= 0.017), but
no main effect of informativeness (F(1, 1223) = 1.6, p = 0.21, partial η
2
= 0.001) or climate policy (F(1,
1223) = 0.43, p = 0.5, partial η
2
< 0.001).
120
Supplemental Analyses from Main Study: Analyses Including Extra Democrats Who Were
Accidentally Recruited
Participants
We reran the main analyses, including 595 respondents who self-identified as Democrats and were
recruited in error, above and beyond the quota for Democrats, due to an error on the part of the survey
panel provider. We reran these analyses as a robustness check.
A small number of these extra Democrats were screened out due to our final participant exclusion
criterion, which occurred after participants read the Twitter conversation and asked participants to
identify the topic of that conversation. Of those who read about a carbon tax in this expanded sample (N =
1305), 114 participants (8.7% of those assigned to the carbon tax conversation) failed this attention check;
of those who read about CAFE standards (N = 1296), 123 participants (9.5% of those assigned to the
CAFE standards conversation) failed this attention check.
In the final expanded sample of 2364 participants, 42% reported their gender as “man”, and the mean age
of participants was 34.8 (SD = 14.5). One percent reported having less than a high school education; 18%
had graduated high school; 22% had completed some college; 10% had a two-year degree; 27% held a
four-year degree; and 22% had a post-graduate degree. Fifty-one percent described themselves as
Democrats, 20% as Independents and 24% as Republicans; mean political liberalism, on a 5-point scale
from very conservative to very liberal, was 3.1 (SD = 1.3).
Results
We adopted the same analytical strategy as in the main analyses. Overall, results were very similar to
those reported in the main analyses. We report the full results for this expanded sample below for
completeness.
Policy Support: We did not observe a statistically significant three-way interaction of tone,
informativeness, and policy, F(1, 2356) = 0.3, p = 0.6, partial η
2
< 0.001, nor were any of the two-way
interactions statistically significant (all p’s > 0.4). We reran the model including climate policy as a main
effect only. There was no main effect of tone (F(1, 2359) = 0.1, p = 0.7, partial η
2
< 0.001) nor
informativeness (F(1, 2359) = 0.7, p = 0.4, partial η
2
< 0.001) and no interaction between tone and
informativeness (F(1, 2359) = 0.7, p = 0.4, partial η
2
< 0.001). We did observe a significant main effect of
climate policy (F(1, 2359) = 34.9, p < 0.001, partial η
2
= 0.015). Participants preferred CAFE standards to
a carbon tax.
Learning: We did not observe a statistically significant three-way interaction of tone, informativeness,
and policy, F(1, 2356) = 0.01, p = 0.9, partial η
2
< 0.001, nor were any of the two-way interactions
statistically significant (all p’s > 0.4). We reran the model including climate policy as a main effect only.
There was a main effect of tone (F(1, 2359) = 15.8, p < 0.001, partial η
2
= 0.007), a main effect of
informativeness (F(1, 2359) = 18.7, p < 0.001, partial η
2
= 0.008), and no interaction between tone and
informativeness (F(1, 2359) = 0.6, p = 0.4, partial η
2
< 0.001). We did not observe a significant main
effect of climate policy (F(1, 2359) = 2, p = 0.16, partial η
2
= 0.001). Participants felt that they learned
more when in the positive vs. negative tone condition, and when in the high informativeness vs. low
informativeness condition.
121
Perceived Strength of Argument (Policy Proponent): We did not observe a statistically significant three-
way interaction of tone, informativeness, and policy, F(1, 2356) = 1.9, p = 0.17, partial η
2
= 0.001, nor
were any of the two-way interactions statistically significant (all p’s > 0.11). We then ran an ANOVA
which included climate policy as a main effect only and examined the interaction between tone and
informativeness. We observed a main effect of tone (F(1, 2359) = 4.8, p = 0.03, partial η
2
= 0.002), a main
effect of informativeness (F(1, 2359) = 8.5, p = 0.004, partial η
2
= 0.004), and no interaction between tone
and informativeness (F(1, 2359) = 2.6, p = 0.11, partial η
2
= 0.001), but a significant main effect of
climate policy (F(1, 2359) = 6.0, p = 0.014, partial η
2
= 0.003). Participants viewed the proponent’s
arguments as stronger in the positive (vs. negative) condition, the high (vs. low) informativeness
condition, and the CAFE standards (vs. carbon tax) condition.
Perceived Strength of Argument (Policy Opponent): We observed a statistically significant interaction
between tone, informativeness, and climate policy (F(1, 2356) = 3.9, p = 0.05, partial η
2
= 0.002). To
further probe this interaction, we next ran separate two-way ANOVAS for those in the carbon tax and
those in the CAFE standards conditions. For those participants who saw the carbon tax conversation, we
observed a main effect of tone (F(1, 1187) = 18, uncorrected p < 0.001, partial η
2
= 0.015), a main effect
of informativeness (F(1, 1187) = 3.8, uncorrected p = 0.05, partial η
2
=0.003) and no interaction between
tone and informativeness (F(1, 1187) = 2.8, uncorrected p = 0.09, partial η
2
=0.002). Participants who saw
the carbon tax conversation viewed the opponent’s arguments as stronger in the positive tone condition
and high informativeness condition. For those participants who saw the CAFE standards conversation, we
observed a main effect of tone (F(1, 1169) = 8.9, uncorrected p = 0.003, partial η
2
= 0.008), no main effect
of informativeness (F(1, 1169) = 0.7, uncorrected p = 0.4, partial η
2
= 0.001) and no interaction between
tone and informativeness (F(1, 1169) = 1.2, uncorrected p = 0.3, partial η
2
= 0.001). Participants who saw
the CAFE standards conversation viewed the opponent’s arguments as stronger in the positive tone
condition.
Trust in Policy Proponent: We did not observe a statistically significant three-way interaction of tone,
informativeness, and policy, F(1, 2356) = 0.03, p = 0.9, partial η
2
< 0.001, nor were any of the two-way
interactions statistically significant (all p’s > 0.4). We then ran an ANOVA including climate policy as a
main effect only, and examined the interaction between tone and informativeness. We observed a main
effect of tone (F(1, 2359) = 20.2, p < 0.001, partial η
2
= 0.008), a main effect of informativeness (F(1,
2359) = 5.1, p = 0.02, partial η
2
= 0.002), no interaction between tone and informativeness (F(1, 2359) =
0.3, p = 0.6, partial η
2
< 0.001), and no main effect of climate policy (F(1, 2359) = 2.3, p = 0.13, partial η
2
= 0.001). Self-reported trust in the policy proponent was greater in the positive tone condition and the
high informativeness condition.
Trust in Policy Opponent: We did not observe a statistically significant three-way interaction of tone,
informativeness, and policy, F(1, 2356) = 0.5, p = 0.5, partial η
2
< 0.001, nor were any of the two-way
interactions statistically significant (all p’s > 0.07). We ran an ANOVA which included climate policy as
a main effect only and examined the interaction between tone and informativeness. We observed a main
effect of tone (F(1, 2359) = 57.5, p < 0.001, partial η
2
=0.024), but no main effect of informativeness (F(1,
2359) = 3.2, p = 0.07, partial η
2
= 0.001), no interaction between tone and informativeness (F(1, 2359) =
1.8, p = 0.18, partial η
2
= 0.001), and no main effect of climate policy (F(1, 2359) = 0.1, p = 0.7, partial η
2
< 0.001). Self-reported trust in the policy opponent was greater in the positive tone condition.
Abstract (if available)
Abstract
Although many scientific and policy solutions for addressing climate change have been developed or proposed, major barriers to their enactment are related to public perception and political opposition to climate action. With this, implementing effective communication strategies for conveying relevant climate information to the public, along with addressing barriers to effective communication (e.g., misinformation on social media), is of vital importance. Therefore, through detailing the findings of three online survey experiments, this dissertation aims to highlight challenges and insights for effective climate communication in three separate contexts. The first chapter focuses on misinformation (i.e., fake news) about climate change on Facebook and tests simple interventions for their effectiveness in limiting the influence of fake climate news among Facebook users. The next chapter shifts to analyzing public opinion of a new technology called carbon capture and utilization (CCU) which removes CO2 from ambient air or from emissions sources (e.g., a factory or power plant) and uses the CO2 to create consumer goods. The third chapter turns to examining communication about climate policies on Twitter. This study manipulated whether politicians were polite versus impolite, and informative versus uninformative, in their tweets (as they debate the merits of a climate policy). We then measured participants’ judgements of the politicians and their messaging. The findings of these three studies have implications for advancing climate goals in a politicized environment.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Individual differences, science communication, and critical thinking for emergent risks
PDF
How misinformation exploits moral values and framing: insights from social media platforms and behavioral experiments
PDF
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
Metacognitive experiences in judgments of truth and risk
PDF
Hard to argue against coherence: a metacognitive approach to correction of misinformation
PDF
Can I make the time or is time running out? That depends in part on how I think about difficulty
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
Towards combating coordinated manipulation to online public opinions on social media
PDF
Facilitating learning and conceptual change through an intervention on text graphic processing
PDF
What do you notice? What do you wonder? A mixed-methods investigation into community science data talks
PDF
Perceived social consensus: a metacognitive perspective
PDF
Thinking Of Trump and Weinstein: the impact of prominent cases of sexual harassment on perceptions of sexual harassment across countries
PDF
Improving information diversity on social media
PDF
Climate change, air pollution and health: how policy can induce adaptation
PDF
The interplay between social connection and substance use
PDF
School staff perceptions of school climate: a mixed-methods multistudy examination of staff school climate at the state, regional, and school levels
PDF
Beyond active and passive social media use: habit mechanisms are behind frequent posting and scrolling on Twitter/X
PDF
Rebels with a cause: Youth, social movements, and media
PDF
Rhetorical strategies in contemporary responses to science and modernity: legitimizing religion in human origins and climate change controversies
PDF
Essays in climate change adaptation: role of market power in incentivizing adaptation behavior
Asset Metadata
Creator
Lutzke, Lauren
(author)
Core Title
Climate change communication: challenges and insights on misinformation, new technology, and social media outreach
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Psychology
Degree Conferral Date
2023-05
Publication Date
01/11/2025
Defense Date
12/12/2022
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
carbon capture,climate change,Communication,misinformation,OAI-PMH Harvest,social media
Format
theses
(aat)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Árvai, Joseph (
committee chair
), Bruine de Bruin, Wändi (
committee member
), Drummond Otten, Caitlin (
committee member
), Schwarz, Norbert (
committee member
), Sinatra, Gale (
committee member
)
Creator Email
llutzke@umich.edu,lutzke@usc.edu
Unique identifier
UC112711066
Identifier
etd-LutzkeLaur-11407.pdf (filename)
Legacy Identifier
etd-LutzkeLaur-11407
Document Type
Dissertation
Format
theses (aat)
Rights
Lutzke, Lauren
Internet Media Type
application/pdf
Type
texts
Source
20230111-usctheses-batch-1000
(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
Tags
carbon capture
climate change
misinformation
social media