Spillover
of Political Populism into Public Health: Populist Attitudes and
Perceptions on Democratic and Public Health Institutions and Policies in
the United States Constitutional Political Economy, 2026 DataverseAbstract (click to expand)Combating communicable diseases produces public goods, and public health
institutions play a crucial role when transaction costs are high. Over
the past decade, populism has surged globally, shaping public attitudes
toward these institutions. Rooted in anti-elitism and a Manichean
worldview that divides society into “the good people” and “the corrupt
elites,” populists often target individuals while undermining the
institutions that constrain them. Using data from the 2020 American
National Election Studies, this study examines how individual-level
populist attitudes are associated with views of democratic institutions,
public health institutions and officials, and COVID-19 related policies
in the United States during the COVID-19 pandemic. The findings show
that populist attitudes are associated with opposition to democratic
institutions for checks and balances, support for unconstrained
executive authority, lower trust in both the federal government and
people in general, negative perceptions of public health organizations
and officials, including the CDC, the WHO, Dr. Fauci, and scientists,
and opposition to expert-recommended pandemic policies, even when for
anti-intellectualism, party identification, liberal–conservative
ideology, education, race, sex, and income. By contrast, although more
educated respondents also express skepticism toward the federal
government and negative perceptions of the CDC and the WHO, they tend to
show higher levels of interpersonal trust and stronger support for
democratic institutions as well as checks and balances. This contrast
suggests that skepticism toward government does not necessarily entail
targeting individual actors or opposing the universalistic institutional
constraints that sustain democratic governance.
Finite
pool of worry and emotions in climate change tweets during
COVID-19 (with Oleg Smirnov and Ignacio Urbina) Journal of Environmental Psychology, 2025Abstract (click to expand)Whether the COVID-19 pandemic diverted public attention away from the
issue of climate change is a topic that divided scholars in recent
years. Two competing theories have emerged: the ‘finite pool of worry’,
which asserts that concerns over the pandemic have overshadowed those
for climate change, and the ‘finite pool of attention’, which argues
that although attention to climate change has waned, worry has remained
steady or even intensified – in line with affect generalization theory.
Survey research appears to support the latter hypothesis more strongly.
In this study, we investigate this theoretical discourse and revisit
these conclusions by conducting an emotional content analysis on a novel
dataset of nearly 24 million Twitter posts related to climate change
from 2018 to 2022. Employing three lexicons—LIWC, NRC Lex, and VADER—we
find that climate change tweets exhibit a decline in expressions of
fear, anxiety, and other negative emotions concurrent with COVID-19
surges. Our daily-level analysis incorporates controls such as media
coverage of climate change, the occurrence of climate-related disasters
like hurricanes and wildfires, and the impact of major political events,
including the 2020 presidential election. The negative association
between COVID-19 severity and climate change worry was strongest in
2020, weakening progressively in 2021 and 2022.
Deservingness heuristics drive redistributive
choices, but weights on recipient effort vary (with Reuben Kline) Humanities and Social Sciences Communications, 2025 oTree Program CodeDataverseNoBeC TalkAbstract (click to expand)Building on the deservingness heuristic–evaluating recipients based on
need and effort–from evolutionary psychology, this study integrates it
with the conditional altruism model from political science and economics
to understand individual similarities and differences in redistributive
preferences. We found that, when a recipient’s effort is known, most
participants’ choices could be explained by the need and effort effect
derived from the model. Furthermore, participants mainly fall into three
categories: highly responsive to effort, less responsive, and
self-interested. These categories reflect how much weight individuals
place on recipient effort in the utility function. However, when effort
information is indirect and partial, income becomes the primary factor,
even if effort can be partially inferred.
Psychological reactance to vaccine mandates on
Twitter: a study of sentiments in the United States Journal of Public Health Policy, 2025 Machine Learning CodeFine-Tuned Model on HuggingfaceDataverseAbstract (click to expand)This study examines the relationship between vaccine mandates and public
sentiment toward vaccines and health officials on Twitter. I analyzed
6.6 million vaccine-related tweets from July 2021 to February 2022 in
the United States. Leveraging a large language model, BERT, I identified
tweets discussing vaccine mandates even when lacking explicit keywords.
Compared to non-mandate tweets, those mentioning mandates exhibit
greater negativity, anger, and freedom-related language. Furthermore,
increased state-level discussion of mandates correlates with rising
levels of negativity and anger toward both vaccines and public health
officials. Finally, greater disparity in vaccination progress across
counties within a state is associated with increased anger in tweets
directed toward both.
COVID-19,
climate change, and the finite pool of worry in 2019-2021 Twitter
discussions (with Oleg Smirnov) Proceedings of the National Academy of Sciences (PNAS),
2022 DataverseAbstract (click to expand)Climate change mitigation has been one of the world’s most salient
issues for the past three decades. However, global policy attention has
been partially diverted to address the COVID-19 pandemic for the past
two years. Here, we explore the impact of the pandemic on the frequency
and content of climate change discussions on Twitter for the period of
2019 – 2021. Consistent with the “finite pool of worry” hypothesis, both
at the annual level and on a daily basis, a larger number of COVID-19
cases and deaths is associated with a smaller number of “climate change”
tweets. Climate change discussion on Twitter decreased despite (a) a
larger Twitter daily active usage in 2020 and 2021, (b) greater coverage
of climate change in traditional media in 2021, (c) a larger number of
North Atlantic ocean hurricanes, and (d) a larger wildland fires area in
the United States in 2020 and 2021. Further evidence supporting the
finite pool of worry is the significant relationship between daily
COVID-19 cases/deaths on the one hand and the public sentiment and
emotional content of “climate change” tweets on the other. In
particular, increasing COVID-19 numbers decrease negative sentiment in
climate change tweets and the emotions related to worry and anxiety,
such as fear and anger.
Working Papers in Preparation (Pre-prints
available on request)
Concentrated but Misaligned: Trends in Voter Agreement on
Party Ideology in the United States, 1972-2020 (Under
Review) PreprintAbstract (click to expand)While Americans increasingly recognize differences between the major
parties, it remains unclear whether they agree on each party’s
ideological image. Using cumulative ANES data from 1972 to 2020, this
study finds that partisan bias in perceptions of party ideology was
moderate before 2008 but became dominant thereafter. First, voter
agreement on party ideology peaked in 2000, declined sharply by 2008,
and rose again in subsequent years. Second, consistent with research on
asymmetric ideological clarity, partisans in both camps agreed more on
the Republican Party’s ideology than on the Democratic Party’s from 1978
to 2004. Since 2008, however, partisans have shown greater agreement on
the opposing party’s ideology than on their own. Third, although the
partisan gap in perceived extremity has persisted, it remained moderate
before 2000 but has grown steadily since 2008—doubling by 2020. These
findings suggest that partisan bias has intensified, increasingly
outweighing shared perceptions across party lines.
Tweets Attacking the U.S. Public Health Officials During the
COVID-19 Pandemic, October 2019 to December 2022 (Under
Review) Abstract (click to expand)During the COVID-19 pandemic, online hostility toward U.S. public health
officials surged. This study analyzed English-language tweets (X)
referencing the Centers for Disease Control and Prevention (CDC) and
Dr. Anthony Fauci between October 2019 and December 2022. Political
DEBATE (DeBERTa Algorithm for Textual Entailment), a natural language
inference classifier fine-tuned to detect hate speech, was used to
identify attacks and advocacy of violence. Aggressive content increased
following the national emergency declaration. Approximately 30% of
tweets attacked public health officials, and 1% advocated violence.
Anger and disgust were expressed in 80% of attacking tweets, while
non-attacking tweets displayed a more diverse range of emotions. Using
zero-inflated negative binomial models, I found that tweets attacking
the CDC and those advocating violence against Fauci were more popular
and widely shared, while tweets attacking Fauci were more viral but
received fewer likes. These findings suggest that online aggression
toward public health officials was widespread during the pandemic, with
important implications for public health institutions. While policies
should remain open to debate and critique, personal attacks risk
deterring professionals from public service and weakening public health
institutions.
Are People Averse to Mandating Costly Cooperative Behavior
When Voluntary Cooperation Is Available? (Under Review) Abstract (click to expand)Many societal challenges, such as climate change and epidemics, require
collective mitigation efforts. Whether coercive policies should be used
to mandate cooperative behaviors remains a longstanding debate. While
research empirically shows that individuals value their own autonomy,
few studies examine whether they also have a procedural preference for
others’ freedom. This study investigates whether individuals discount
coercive measures. To test this, I designed an incentivized experiment
in which players exerted real labor effort to cooperate in a prisoner’s
dilemma. One player could choose to implement either a voluntary
rule—where they played with a voluntary cooperator—or a mandatory
rule—where both players were required to cooperate. Both rules ensured
the same outcome distribution. If individuals intrinsically valued
others’ freedom beyond instrumental values, they would choose voluntary
cooperation. The results do not indicate a general aversion to coercing
others , even when both rules produce the same outcome. However,
participants who selected voluntary cooperation and those who opted for
mandatory cooperation differed in their views on the government’s role
when freedom and social welfare are at odds, their trust in government,
and their support for COVID-19 mask mandates. These findings suggest
heterogeneity in preferences for others’ freedom.
The Effect of Extreme Income Inequality and Meritocratic
Systems on Redistributive Preferences (Under Review) (with Daniella P. Alva) Abstract (click to expand)As income differences become more disparate, researchers need to study
behavior and preferences under extreme inequality. We use an
incentivized experiment to test for differences in redistributive
behavior on two distinct samples: one online sample (N = 797) and one
laboratory sample (N = 488). In our experiment, players are given an
initial distribution of tokens and can propose redistribution. We
manipulate the level of inequality experienced as well as the source of
inequality. For some of our players, the distribution of tokens is
performance-based (merit). For the rest, tokens are randomly-generated
(luck). We find that people care about merit even at extreme
inequality—those who experience merit-based inequality prefer equality
less than those who experience luck-based inequality. Most players that
experience any inequality redistribute to create equal outcomes, but
those exposed to extreme inequality want more equality overall. In other
words, people who experience extreme inequality, even when merit-based,
still want more equality than those who experience non-extreme types of
inequality.
Belief Updating and the Strength and Stability of
Norms (with Cristina Bicchieri) Abstract (click to expand)New norms often spread through social learning, as individuals update
their beliefs about others’ behaviors through sequential interactions.
Consequently, learning mechanisms—specifically, how beliefs are
updated—play a key role in norm strength and stability. This study uses
agent-based modeling in dyadic coordination games to compare three
stylized learning modes: (1) perfect Bayesian updating, (2) limited
memory, and (3) confirmation bias. We find that norms are more
persistent when maintained by confirmation-biased agents, especially
when they form the majority. Minority groups influenced by confirmation
bias are also more likely to shift prevailing norms. Even when
expectations are inaccurate, confirmation bias anchors behavior and
sustains norms. When both groups rely on confirmation-biased learning,
norm convergence slows or fails entirely. These findings suggest that
persistent misperceptions and norm divergence may result from cognitive
biases, highlighting the need to account for learning dynamics when
designing policies to change behaviors shaped by social norms.
Work in Progress
Similarity-Based Topic Segmentation for Political
Corpora (with Michael Burnham) Abstract (click to expand)This study demonstrates how unsupervised topic segmentation can improve
preprocessing of long, unstructured political speech transcripts—such as
campaign rallies and State of the Union addresses—for downstream NLP
tasks like topic modeling. Although automatic speech recognition (ASR)
has made such transcripts widely available, their unstructured,
multi-topic nature challenges standard methods, including LDA and
transformer-based models. I apply similarity-based topic segmentation
that detects topic shifts via semantic change in sentence-transformer
embeddings. Using two corpora—Biden’s 2024 State of the Union address
and campaign rally transcripts from the 2024 U.S. presidential
election—I show that segmentation yields more coherent and interpretable
LDA topics than either uniform slicing or using full documents. These
findings offer practical guidance for political scientists working with
spoken political texts, demonstrating how segmentation can enhance the
clarity of computational analyses.
A Case for Applying NLP Techniques for Characterizing
Beliefs on Twitter: An Analysis on American Attitudes Towards the
Police (with Gilvir Gill, Srivardhan Jangili, Veronica Oelerich, Rosa M.
Bermejo, Amie Paige, and Tori Peña) Abstract (click to expand)
Plant-Rich Diets, Policies, and Conditional
Cooperation (with Daniella
P. Alva & Shawn Kim)Abstract (click to expand)Approximately one-third of global greenhouse gas emissions are
attributed to food systems, and adopting a plant-rich diet is an
effective way to mitigate climate change by individuals. Nevertheless,
the entire population will benefit from this individual’s efforts to
adopt a plant-rich diet, resulting in an incentive for individuals to
free-ride. Despite the fact that adopting a plant-rich diet is effective
in reducing greenhouse gas emissions, it remains unpopular with
Americans. In the current proposal, we design a vignette experiment to
examine how others’ behavior and perceptions of reciprocity influence
Americans’ willingness to adopt plant-rich diets and support policies
that promote plant-rich diets, such as federal diet guidelines, meat
taxes, the proposed PLANT Act, and school meals. Our pilot study found
that the dynamic trend treatment and the reciprocity cue influenced
several policy attitudes promoting a plant-rich diet.
Sanctions and their audiences: the analysis of US and
Russian popular responses to anti-Russian sanctions (with Guzel Garifullina)Abstract (click to expand)Economic sanctions have become a major instrument of international
politics. However, the effectiveness of economic sanctions depends on
both public opinions in the targeted country and the country that
imposes sanctions. We explored whether certain emotions—anxiety and
anger—were associated with specific attitudes towards the sanctions
against Russia in 2022. Existing literature in psychology and political
science has demonstrated that anxiety and anger provoke different risk
perceptions and risk attitudes and, in turn, change political attitudes,
such as support for a war. Therefore, we proposed that the fear of the
consequences of the economic sanctions, such as inflation and shortage,
resulted in negative attitudes toward the sanctions among Americans and
negative attitudes toward the invasion among Russians. In contrast, the
American citizens angered by the invasion were more likely to support
the sanctions, and the Russian citizens angered by the sanctions were
more likely to support the invasion. Using the data from Twitter and
VKontakte, we applied emotion analysis and automated text classification
to measure the emotions and attitudes toward the sanctions in Americans
and Russians. In the preliminary analysis, we found that the
pro-sanction tweets were angrier than the anti-sanction tweets. In
contrast, the anti-sanction tweets were more anxious than the
pro-sanction tweets.
By Methods
Machine Learning/Natural Language Processing
Peer-Reviewed Publications
Finite
pool of worry and emotions in climate change tweets during
COVID-19 (with Oleg Smirnov & Ignacio Urbina) Journal of Environmental Psychology, 2025Abstract (click to expand)Whether the COVID-19 pandemic diverted public attention away from the
issue of climate change is a topic that divided scholars in recent
years. Two competing theories have emerged: the ‘finite pool of worry’,
which asserts that concerns over the pandemic have overshadowed those
for climate change, and the ‘finite pool of attention’, which argues
that although attention to climate change has waned, worry has remained
steady or even intensified – in line with affect generalization theory.
Survey research appears to support the latter hypothesis more strongly.
In this study, we investigate this theoretical discourse and revisit
these conclusions by conducting an emotional content analysis on a novel
dataset of nearly 24 million Twitter posts related to climate change
from 2018 to 2022. Employing three lexicons—LIWC, NRC Lex, and VADER—we
find that climate change tweets exhibit a decline in expressions of
fear, anxiety, and other negative emotions concurrent with COVID-19
surges. Our daily-level analysis incorporates controls such as media
coverage of climate change, the occurrence of climate-related disasters
like hurricanes and wildfires, and the impact of major political events,
including the 2020 presidential election. The negative association
between COVID-19 severity and climate change worry was strongest in
2020, weakening progressively in 2021 and 2022.
Psychological reactance to vaccine mandates on
Twitter: a study of sentiments in the United States Journal of Public Health Policy, 2025 Machine Learning CodeFine-Tuned Model on HuggingfaceDataverseAbstract (click to expand)This study examines the relationship between vaccine mandates and public
sentiment toward vaccines and health officials on Twitter. I analyzed
6.6 million vaccine-related tweets from July 2021 to February 2022 in
the United States. Leveraging a large language model, BERT, I identified
tweets discussing vaccine mandates even when lacking explicit keywords.
Compared to non-mandate tweets, those mentioning mandates exhibit
greater negativity, anger, and freedom-related language. Furthermore,
increased state-level discussion of mandates correlates with rising
levels of negativity and anger toward both vaccines and public health
officials. Finally, greater disparity in vaccination progress across
counties within a state is associated with increased anger in tweets
directed toward both.
COVID-19,
climate change, and the finite pool of worry in 2019-2021 Twitter
discussions (with Oleg Smirnov) Proceedings of the National Academy of Sciences (PNAS),
2022 DataverseAbstract (click to expand)Climate change mitigation has been one of the world’s most salient
issues for the past three decades. However, global policy attention has
been partially diverted to address the COVID-19 pandemic for the past
two years. Here, we explore the impact of the pandemic on the frequency
and content of climate change discussions on Twitter for the period of
2019 – 2021. Consistent with the “finite pool of worry” hypothesis, both
at the annual level and on a daily basis, a larger number of COVID-19
cases and deaths is associated with a smaller number of “climate change”
tweets. Climate change discussion on Twitter decreased despite (a) a
larger Twitter daily active usage in 2020 and 2021, (b) greater coverage
of climate change in traditional media in 2021, (c) a larger number of
North Atlantic ocean hurricanes, and (d) a larger wildland fires area in
the United States in 2020 and 2021. Further evidence supporting the
finite pool of worry is the significant relationship between daily
COVID-19 cases/deaths on the one hand and the public sentiment and
emotional content of “climate change” tweets on the other. In
particular, increasing COVID-19 numbers decrease negative sentiment in
climate change tweets and the emotions related to worry and anxiety,
such as fear and anger.
Working
Papers in Preparation (Pre-prints available on request)
Tweets Attacking the U.S. Public Health Officials During the
COVID-19 Pandemic, October 2019 to December 2022 Abstract (click to expand)During the COVID-19 pandemic, online hostility toward U.S. public health
officials surged. This study analyzed English-language tweets (X)
referencing the Centers for Disease Control and Prevention (CDC) and
Dr. Anthony Fauci between October 2019 and December 2022. Political
DEBATE (DeBERTa Algorithm for Textual Entailment), a natural language
inference classifier fine-tuned to detect hate speech, was used to
identify attacks and advocacy of violence. Aggressive content increased
following the national emergency declaration. Approximately 30% of
tweets attacked public health officials, and 1% advocated violence.
Anger and disgust were expressed in 80% of attacking tweets, while
non-attacking tweets displayed a more diverse range of emotions. Using
zero-inflated negative binomial models, I found that tweets attacking
the CDC and those advocating violence against Fauci were more popular
and widely shared, while tweets attacking Fauci were more viral but
received fewer likes. These findings suggest that online aggression
toward public health officials was widespread during the pandemic, with
important implications for public health institutions. While policies
should remain open to debate and critique, personal attacks risk
deterring professionals from public service and weakening public health
institutions.
Work in Progress
Similarity-Based Topic Segmentation for Political
Corpora (with Michael Burnham)Abstract (click to expand)This study demonstrates how unsupervised topic segmentation can improve
preprocessing of long, unstructured political speech transcripts—such as
campaign rallies and State of the Union addresses—for downstream NLP
tasks like topic modeling. Although automatic speech recognition (ASR)
has made such transcripts widely available, their unstructured,
multi-topic nature challenges standard methods, including LDA and
transformer-based models. I apply similarity-based topic segmentation
that detects topic shifts via semantic change in sentence-transformer
embeddings. Using two corpora—Biden’s 2024 State of the Union address
and campaign rally transcripts from the 2024 U.S. presidential
election—I show that segmentation yields more coherent and interpretable
LDA topics than either uniform slicing or using full documents. These
findings offer practical guidance for political scientists working with
spoken political texts, demonstrating how segmentation can enhance the
clarity of computational analyses.
Sanctions and their audiences: the analysis of US and
Russian popular responses to anti-Russian sanctions (with Guzel Garifullina)Abstract (click to expand)Economic sanctions have become a major instrument of international
politics. However, the effectiveness of economic sanctions depends on
both public opinions in the targeted country and the country that
imposes sanctions. We explored whether certain emotions—anxiety and
anger—were associated with specific attitudes towards the sanctions
against Russia in 2022. Existing literature in psychology and political
science has demonstrated that anxiety and anger provoke different risk
perceptions and risk attitudes and, in turn, change political attitudes,
such as support for a war. Therefore, we proposed that the fear of the
consequences of the economic sanctions, such as inflation and shortage,
resulted in negative attitudes toward the sanctions among Americans and
negative attitudes toward the invasion among Russians. In contrast, the
American citizens angered by the invasion were more likely to support
the sanctions, and the Russian citizens angered by the sanctions were
more likely to support the invasion. Using the data from Twitter and
VKontakte, we applied emotion analysis and automated text classification
to measure the emotions and attitudes toward the sanctions in Americans
and Russians. In the preliminary analysis, we found that the
pro-sanction tweets were angrier than the anti-sanction tweets. In
contrast, the anti-sanction tweets were more anxious than the
pro-sanction tweets.
Experiments
Peer-Reviewed Publications
Deservingness heuristics drive redistributive
choices, but weights on recipient effort vary (with Reuben Kline) Humanities and Social Sciences Communications, 2025 oTree Program CodeDataverseNoBeC TalkAbstract (click to expand)Building on the deservingness heuristic–evaluating recipients based on
need and effort–from evolutionary psychology, this study integrates it
with the conditional altruism model from political science and economics
to understand individual similarities and differences in redistributive
preferences. We found that, when a recipient’s effort is known, most
participants’ choices could be explained by the need and effort effect
derived from the model. Furthermore, participants mainly fall into three
categories: highly responsive to effort, less responsive, and
self-interested. These categories reflect how much weight individuals
place on recipient effort in the utility function. However, when effort
information is indirect and partial, income becomes the primary factor,
even if effort can be partially inferred.
Working
Papers in Preparation (Pre-prints available on request)
Are People Averse to Mandating Costly Cooperative Behavior
When Voluntary Cooperation Is Available? (Under Review) Abstract (click to expand)Many societal challenges, such as climate change and epidemics, require
collective mitigation efforts. Whether coercive policies should be used
to mandate cooperative behaviors remains a longstanding debate. While
research empirically shows that individuals value their own autonomy,
few studies examine whether they also have a procedural preference for
others’ freedom. This study investigates whether individuals discount
coercive measures. To test this, I designed an incentivized experiment
in which players exerted real labor effort to cooperate in a prisoner’s
dilemma. One player could choose to implement either a voluntary
rule—where they played with a voluntary cooperator—or a mandatory
rule—where both players were required to cooperate. Both rules ensured
the same outcome distribution. If individuals intrinsically valued
others’ freedom beyond instrumental values, they would choose voluntary
cooperation. The results do not indicate a general aversion to coercing
others , even when both rules produce the same outcome. However,
participants who selected voluntary cooperation and those who opted for
mandatory cooperation differed in their views on the government’s role
when freedom and social welfare are at odds, their trust in government,
and their support for COVID-19 mask mandates. These findings suggest
heterogeneity in preferences for others’ freedom.
The Effect of Extreme Income Inequality and Meritocratic
Systems on Redistributive Preferences (Under Review) (with Daniella P. Alva)Abstract (click to expand)As income differences become more disparate, researchers need to study
behavior and preferences under extreme inequality. We use an
incentivized experiment to test for differences in redistributive
behavior on two distinct samples: one online sample (N = 797) and one
laboratory sample (N = 488). In our experiment, players are given an
initial distribution of tokens and can propose redistribution. We
manipulate the level of inequality experienced as well as the source of
inequality. For some of our players, the distribution of tokens is
performance-based (merit). For the rest, tokens are randomly-generated
(luck). We find that people care about merit even at extreme
inequality—those who experience merit-based inequality prefer equality
less than those who experience luck-based inequality. Most players that
experience any inequality redistribute to create equal outcomes, but
those exposed to extreme inequality want more equality overall. In other
words, people who experience extreme inequality, even when merit-based,
still want more equality than those who experience non-extreme types of
inequality.
Work in Progress
Plant-Rich Diets, Policies, and Conditional
Cooperation (with Daniella P. Alva & Shawn Kim)Abstract (click to expand)Approximately one-third of global greenhouse gas emissions are
attributed to food systems, and adopting a plant-rich diet is an
effective way to mitigate climate change by individuals. Nevertheless,
the entire population will benefit from this individual’s efforts to
adopt a plant-rich diet, resulting in an incentive for individuals to
free-ride. Despite the fact that adopting a plant-rich diet is effective
in reducing greenhouse gas emissions, it remains unpopular with
Americans. In the current proposal, we design a vignette experiment to
examine how others’ behavior and perceptions of reciprocity influence
Americans’ willingness to adopt plant-rich diets and support policies
that promote plant-rich diets, such as federal diet guidelines, meat
taxes, the proposed PLANT Act, and school meals. Our pilot study found
that the dynamic trend treatment and the reciprocity cue influenced
several policy attitudes promoting a plant-rich diet.
Survey Data
Peer-Reviewed Publications
Spillover
of Political Populism into Public Health: Populist Attitudes and
Perceptions on Democratic and Public Health Institutions and Policies in
the United States Constitutional Political Economy, 2026 DataverseAbstract (click to expand)Combating communicable diseases produces public goods, and public health
institutions play a crucial role when transaction costs are high. Over
the past decade, populism has surged globally, shaping public attitudes
toward these institutions. Rooted in anti-elitism and a Manichean
worldview that divides society into “the good people” and “the corrupt
elites,” populists often target individuals while undermining the
institutions that constrain them. Using data from the 2020 American
National Election Studies, this study examines how individual-level
populist attitudes are associated with views of democratic institutions,
public health institutions and officials, and COVID-19 related policies
in the United States during the COVID-19 pandemic. The findings show
that populist attitudes are associated with opposition to democratic
institutions for checks and balances, support for unconstrained
executive authority, lower trust in both the federal government and
people in general, negative perceptions of public health organizations
and officials, including the CDC, the WHO, Dr. Fauci, and scientists,
and opposition to expert-recommended pandemic policies, even when for
anti-intellectualism, party identification, liberal–conservative
ideology, education, race, sex, and income. By contrast, although more
educated respondents also express skepticism toward the federal
government and negative perceptions of the CDC and the WHO, they tend to
show higher levels of interpersonal trust and stronger support for
democratic institutions as well as checks and balances. This contrast
suggests that skepticism toward government does not necessarily entail
targeting individual actors or opposing the universalistic institutional
constraints that sustain democratic governance.
Working
Papers in Preparation (Pre-prints available on request)
Concentrated but Misaligned: Trends in Voter Agreement on
Party Ideology in the United States, 1972–2020 (Under
Review) PreprintAbstract (click to expand)While Americans increasingly recognize differences between the major
parties, it remains unclear whether they agree on each party’s
ideological image. Using cumulative ANES data from 1972 to 2020, this
study finds that partisan bias in perceptions of party ideology was
moderate before 2008 but became dominant thereafter. First, voter
agreement on party ideology peaked in 2000, declined sharply by 2008,
and rose again in subsequent years. Second, consistent with research on
asymmetric ideological clarity, partisans in both camps agreed more on
the Republican Party’s ideology than on the Democratic Party’s from 1978
to 2004. Since 2008, however, partisans have shown greater agreement on
the opposing party’s ideology than on their own. Third, although the
partisan gap in perceived extremity has persisted, it remained moderate
before 2000 but has grown steadily since 2008—doubling by 2020. These
findings suggest that partisan bias has intensified, increasingly
outweighing shared perceptions across party lines.
Formal and Agent-Based
Modeling
Peer-Reviewed Publications
Deservingness heuristics drive redistributive
choices, but weights on recipient effort vary (with Reuben Kline) Humanities and Social Sciences Communications, 2025 oTree Program CodeDataverseNoBeC TalkAbstract (click to expand)Building on the deservingness heuristic–evaluating recipients based on
need and effort–from evolutionary psychology, this study integrates it
with the conditional altruism model from political science and economics
to understand individual similarities and differences in redistributive
preferences. We found that, when a recipient’s effort is known, most
participants’ choices could be explained by the need and effort effect
derived from the model. Furthermore, participants mainly fall into three
categories: highly responsive to effort, less responsive, and
self-interested. These categories reflect how much weight individuals
place on recipient effort in the utility function. However, when effort
information is indirect and partial, income becomes the primary factor,
even if effort can be partially inferred.
Working
Papers in Preparation (Pre-prints available on request)
Belief Updating and the Strength and Stability of
Norms (with Cristina Bicchieri)Abstract (click to expand)New norms often spread through social learning, as individuals update
their beliefs about others’ behaviors through sequential interactions.
Consequently, learning mechanisms—specifically, how beliefs are
updated—play a key role in norm strength and stability. This study uses
agent-based modeling in dyadic coordination games to compare three
stylized learning modes: (1) perfect Bayesian updating, (2) limited
memory, and (3) confirmation bias. We find that norms are more
persistent when maintained by confirmation-biased agents, especially
when they form the majority. Minority groups influenced by confirmation
bias are also more likely to shift prevailing norms. Even when
expectations are inaccurate, confirmation bias anchors behavior and
sustains norms. When both groups rely on confirmation-biased learning,
norm convergence slows or fails entirely. These findings suggest that
persistent misperceptions and norm divergence may result from cognitive
biases, highlighting the need to account for learning dynamics when
designing policies to change behaviors shaped by social norms.