Unaggrandizing the Partisan Gerrymander

I explore the relationship between partisan votes and partisan seat allocation in U.S. state lower-house elections. Specifically, I measure the representational form (the rate of partisan seat changes given particular partisan vote changes) and partisan bias (asymmetry in the seats-votes relationship) of 441 lower-house state legislative elections in 46 states from 1968 to 1987. I then test a number of hypotheses that have been advanced to explain variation in representational form and partisan bias. Values for representational form and partisan bias are generated by creating simulations from actual election results. I simulate seat gains made by Republicans given one percent uniform party vote swings across all districts and assuming Republicans would win between 35% and 65% of the mean district vote. After generating 31 data points for each election year, I use a logit equation to operationalize representational form and partisan bias for each election year in each state. These data then become dependent variables in pooled, cross-sectional time-series analyses used to explain variation in representational form and partisan bias across time and across states. As in previous studies, I find that representational form is declining over time. I also find that representational form is a function of party competition across election districts. In elections having a large number of competitive districts, there is a rise in the value of representational form. The size of election districts (by population) as measured by Taagepera's Index has a positive but substantively weak effect on representational form. Effective district magnitude (the existence of multimember districts) also has a positive but substantively weak impact on representational form. It was thought that partisan bias would result from partisan gerrymandering during redistricting. While party control of redistricting does have the hypothesized effect in eight of the nine even-numbered election years, only in 1970, 1976, and 1982 did gerrymandering effects reach statistical significance. The results for partisan bias support recent studies that suggest that gerrymandering at the state level is not pervasive but does occasionally occur.

Recent studies revealed more increasingly political polarization in the distribution and consumption of political news. Political polarization demonstrates the disagreement between people aligned with different ideologies or political parties (e.g. left vs. right, Democrats vs. Republicans). Increasingly political polarization can have negative effects on our society; for example, extreme cases influenced by the left/right ideology can lead to massive bombing or shooting incidents. Thus, through four different research streams this thesis will help people to understand the roles of humans, algorithms, and cyborgs in political polarization. In terms of humans, prior research has shown that people are mainly consuming news conforming to their pre-existing beliefs. People also prefer to have homophilous social interactions. Both of these lead to political polarization. Thus, to help inform the political slant of news people consume, we develop a lightweight and scalable news slant measurement using Twitter. Moreover, utilizing this method to estimate each Twitter user as a Republican or Democrat, we analyze political discourse on Twitter communications in the combination of three aspects including political affiliation, personality perception, and policy discussion around several main candidates from both parties during the 2016 U.S. presidential election. In terms of algorithms, researchers have recently started to question whether algorithms create distinct personalized experiences for users, which can unintentionally contribute to a more polarized society. Thus, it is important to study the roles of personalization algorithms employed by search engines and social media in reinforcing pre-existing biases. To this end, we examine the personalization of Google News Search based on the users' browsing history, especially when it comes from the users with different political biases. In terms of cyborgs, there have been numerous reports of widespread misinformation campaigns during the 2016 U.S. presidential election. Of particular notes, there are reports which identified the efforts to manipulate social media (e.g. Twitter, Facebook) by the Russian state-sponsored accounts. These external manipulations by cyborgs cause significant pressure on social media services to mitigate spam, abuse, and political polarization on their platforms. Specifically, Twitter publicly acknowledged the exploitation of their platform and has since conducted aggressive cleanups to suspend the involved accounts. To shed light on Twitter's countermeasures, we conduct a postmortem analysis of about one million Twitter accounts who engaged in the 2016 U.S. presidential election but were later suspended by Twitter.

Significant efforts have been made to analyze the political stance or bias in news articles, especially as political polarization intensifies over the years. Recent advancements in machine learning have enabled researchers to develop various bias prediction models, which typically learn features not only from the text of the news articles but also from external knowledge. However, when training these models, the political bias label assigned to a news article is often based solely on the news source which published it. This approach can be problematic, as a news outlet with a particular political stance might publish an article that reflects a different political perspective. To address this issue, we first identify distinct text patterns associated with specific news sources or publishers, that are minimally relevant to predicting the political bias of a news article. We then conduct comprehensive experiments to investigate (i) whether existing models trained to predict political bias can also accurately predict the source, and (ii) whether these models change their predictions when a distinct pattern from a source with a different political stance is incorporated into a news article. Our experimental results reveal that all existing models tend to predict the source, even when trained solely to predict bias. Based on these findings, we propose a new deep learning model for political bias prediction that avoids learning source-indicative patterns specific to a given news source.

The outbreak of COVID‐19 had a huge global impact, and non-scientific beliefs and political polarization have significantly influenced the population's behavior. In this context, COVID vaccines were made available at an unprecedented time, but a high level of hesitance has been observed that can undermine community immunization. Traditionally, anti-vaccination attitudes are more related to conspiratorial thinking than political bias. In Brazil, a country with an exemplar tradition in large-scale vaccination programs, all COVID-related topics have also been discussed under a strong political bias. In this paper, we use a multi-dimensional analysis framework to understand if anti/pro-vaccination stances expressed by Brazilians in social media are influenced by political polarization. The analysis framework incorporates techniques to automatically infer from users their political orientation, topic modeling to discover their concerns, network analysis to characterize their social behavior, and the characterization of information sources and external influence. Our main findings confirm that anti/pro-stances are biased by political polarization, right and left, respectively. While a significant proportion of pro-vaxxers display haste for an immunization program and criticize the government's actions, the anti-vaxxers distrust a vaccine developed in a record time. Anti-vaccination stance is also related to prejudice against China (anti-sinovaxxers), revealing conspiratorial theories related to communism. All groups display an ``echo chamber" behavior, revealing they are not open to distinct views.

The outbreak of COVID‐19 had a huge global impact, and non-scientific beliefs and political polarization have significantly influenced the population's behavior. In this context, COVID vaccines were made available in an unprecedented time, but a high level of hesitance has been observed that can undermine community immunization. Traditionally, anti-vaccination attitudes are more related to conspiratorial thinking rather than political bias. In Brazil, a country with an exemplar tradition in large-scale vaccination programs, all COVID-related topics have also been discussed under a strong political bias. In this paper, we use a multi-dimensional analysis framework to understand if anti/pro-vaccination stances expressed by Brazilians in social media are influenced by political polarization. The analysis framework incorporates: techniques to automatically infer from users their political orientation, topic modeling to discover their concerns, network analysis to characterize their social behavior, and the characterization of information sources and external influence. Our main findings confirm that anti/pro stances are biased by political polarization, right and left, respectively. While a significant proportion of pro-vaxxers display haste for an immunization program and criticize the government's actions, the anti-vaxxers distrust a vaccine developed in a record time. Anti-vaccination stance is also related to prejudice against China (anti-sinovaxxers), revealing conspiratorial theories related to communism. All groups display an ``echo chamber" behavior, revealing they are not open to distinct views.

Political polarization in America has intensified beyond mere disagreement to what scholars characterize as sectarianism—a condition where partisan identity fundamentally shapes moral judgments. A key marker of sectarianism is asymmetric moral standards for violence, where aggression against political opponents is considered more justified than identical violence targeting one’s own group. Using a survey experiment featuring a realistic political rally scenario, we find compelling evidence in support of such sectarianism: partisan bias in the US extends to evaluations of political violence. By manipulating the partisan affiliations of perpetrators and targets, as well as provocation severity, we find that both Democrats and Republicans exhibit substantial and symmetrical partisan bias. This double standard is particularly pronounced among strong partisans, who are nearly three times more likely to justify violence against the opposition than violence targeting their own party. These results extend sectarianism theory beyond policy preferences to physical violence, suggesting that partisan identity now functions as a powerful perceptual filter that can legitimize political aggression when directed at opponents.

This study leverages a survey experiment in the lead up to the 2016 U.S. presidential election to evaluate how partisan biases, poll results, and their methodological quality interact to shape people’s assessments of polling accuracy and electoral expectations. In a nationally representative sample, we find that individuals disproportionately find polls more credible when their preferred candidate is leading. Partisan biases are mitigated when the polls themselves vary in objective indicators of quality: while more educated respondents are more likely to identify high-quality polls accurately, low education respondents’ bias was reduced when they encountered polls with varying methodological quality. Finally, these moderators influence respondents’ electoral expectations as well. We discuss the implications for journalistic coverage of polls, public opinion, and political polarization.

Political polarization in various forms have increased, as affective processes dominate political discourse the proliferation of misinformation and emotional language in partisan news media have exacerbated the issue. Politically motivated numeracy describes the tendency for an individual to discern statistical information in a manner that confirms their political bias. Recent literature has expanded information in this area, and lead to discrepant findings. Specifically, the Identity Protective Cognition (IPC) account suggests that when an individual is faced with information that does not confirm their political bias, they are likely to perceive the information as a threat to their political identity. This leads the individual to discern politically relevant statistics inaccurately, particularly if they score high on cognitive reflection assessments. Conversely, the Cognitive Sophistication account has suggested that individuals who are high in trait cognitive sophistication are likely to use their cognitive facilities to discern information accurately, regardless of bias. In order to ameliorate these discrepant findings, I sought to investigate if affect might be implicated in politically motivated reasoning. In a single study, 61 participants (Mage = 19.72; Women = 46, Men = 12) were instructed to assess 8 politically laden numeracy tasks as well as one neutral task. Each politically laden task contained a contrived study on a politically relevant issue or policy, with statistics that either confirmed a liberal or conservative bias. In support of CSA, the trending results suggest Cognitively sophisticated liberals seem to show less bias and Extreme partisans showed less bias as a function of increased numeracy whereas moderates did not. On the other hand, trending results also suggest that cognitively sophisticated conservatives seem to show more bias, supporting IPC.

We investigate whether weak executive federalism was beneficial or damaging for COVID-19 management in the USA. We formulate a policy response model for subnational governments, considering the national government’s preferred policy, in addition to other factors, with incomplete and with complete information. The hypotheses derived are tested using econometric techniques. Our results suggest that ideological and political biases were more influential in a situation of incomplete information than in one of complete information. As such, weak executive federalism allowed more agile policy responses in Democrat-led states when information was incomplete, thus reducing the rates of incidence and mortality. When information was complete, ideological and political biases were found to be of no relevance at all.

The COVID-19 pandemic affected people's health behaviours and health outcomes. Political or affective polarization could be associated with health behaviours such as mask-wearing or vaccine uptake and with health outcomes, e.g., infection or mortality rate. Political polarization relates to divergence or spread of ideological beliefs and affective polarization is about dislike between people of different political groups, such as ideologies or parties. The objectives of this study are to investigate and synthesize evidence about associations between both forms of polarization and COVID-19 health behaviours and outcomes. In this systematic review, we will include quantitative studies that assess the relationship between political or affective polarization and COVID-19-related behaviours and outcomes, including adherence to mask mandates, vaccine uptake, infection and mortality rate. We will use a predetermined strategy to search EMBASE, Medline (Ovid), Cochrane Library, Cochrane COVID-19 Study Register, Global Health (Ovid), PsycInfo (Ovid), Web of Science, CINAHL, EconLit (EBSCOhost), WHO COVID-19 Database, iSearch COVID-19 Portfolio (NIH) and Google Scholar from 2019 to September 8 2023. One reviewer will screen unique records according to eligibility criteria. A second reviewer will verify the selection. Data extraction, using pre-piloted electronic forms, will follow a similar process. The risk of bias of the included studies will be assessed using the JBI checklist for analytical cross sectional studies. We will summarise the included studies descriptively and examine the heterogeneity between studies. Quantitative data pooling might not be feasible due to variations in measurement methods used to evaluate exposure, affective and political polarization. If there are enough relevant studies for statistical data synthesis, we will conduct a meta-analysis. This review will help to better understand the concept of polarization in the context of the COVID-19 pandemic and might inform decision making for future pandemics. PROSPERO ID: CRD42023475828.

ABSTRACTPolitical polarization and generational politics are important topics in contemporary political science classrooms. This article presents an approach to teaching political polarization in an introduction to politics course. Coauthored by two Generation Z students from the course and their Boomer Generation professor, the article provides conflicting views of young people and politics as found in the work of Robert Putnam and Russell Dalton. The article presents survey data on affective and issue political polarization from the course, including discussion by the two student coauthors of the survey results interpreting their generation’s political polarization. The course approaches the introductory politics course using cognitive psychology concepts including confirmative bias, motivated reasoning, and other cognitive biases. Teaching from this micro-level perspective helps students to reflect on their own political biases. The article provides concepts and readings for political science professors to use in replicating the course.

This paper investigates whether political polarization of the TV audience is emerging also in a typical democratic corporatist system. The study is motivated by the claim put forward by several US scholars, who argue that in today’s high choice information environments, partisans tend to see mainstream media as ‘hostile’ and therefore seek out and select broadcasters who confirm and deepen their worldview (Arceneaux and Johnson, 2013; Iyengar and Hahn, 2009; Tewksbury and Riles, 2015). This demand, they argue, expands the market for partisan TV and contributes to growing political polarization. We ask if there is evidence of a politically polarized Norwegian TV audience, by exploring the relationship between partisan preferences, perceived political bias and selective exposure to TV news. We find that many Norwegians believe that both the public broadcaster and the leading commercial broadcasters are politically biased. Consistent with the “hostile media hypothesis”, people on the right accuse the broadcasters of favoring the parties on the left, whereas people of the left tend to see the broadcasters as favoring the parties on the right, albeit not to the same degree. By using a survey experiment, our study also demonstrates that given the opportunity, the audience does select news stories consistent with their political beliefs from a politically ‘friendly’ broadcaster. However, they also choose news stories consistent with their political beliefs from a perceived hostile news source over politically inconsistent stories from a friendly source. This suggests that ‘friendly’ content triumphs perception of broadcaster bias. Despite widespread perceptions of partisan favoritism in the Norwegian TV market, we find few traces of a politically polarized audience. The main reason for this is that the public broadcaster still draws a wide audience across the political spectrum, as even critics consider this news source as too important and relevant to be ignored.

Research purpose. The main objective of this study is to assess the political bias present in the news programmes broadcast by two private TV stations – TV Markíza and TV JOJ – and the public service TV station Jednotka during the campaign period leading up to the 2023 parliamentary election in the Slovak Republic. This research is motivated by the fundamental importance of objectivity in both media practice and media analysis. Design / Methodology / Approach. A total of 93 broadcasts were examined. We conducted a quantitative content analysis of the main evening news programmes, using a coding framework focused on formal parameters and categorisation of election news items. Each news item was independently coded by two coders. To measure the media visibility and exposure of each electoral subject, the Media Visibility and Media Exposure Index were used to qualify the proportion and frequency of subjects’ appearances in the news programme. The Overtone Index was used to assess the tone of coverage. These three indicators were synthesized into the Media Political Bias Index, which allows for synthetic evaluation of the level of political bias in each analysed news item, as well as a comparison of the political bias in each television station’s news programmes over time. Findings. The analysis showed that TV JOJ had the highest share of election news (17.25%), followed by TV Markíza (12.7%) and the lowest share was held by the public broadcaster Jednotka (8.21%). Meanwhile, the public broadcaster showed the highest level of neutrality (MPBI = 0.08), while TV Markíza showed the highest political bias (MPBI = 0.38), with private TV channels being more likely to reflect regional preferences and having a more negative tone towards most of the relevant parties. The exception was Progresívne Slovensko, which was the only political party to achieve a positive overtone index. Parties with low preferences were virtually ignored in the private media. These findings suggest that while the public broadcasters maintained a commitment to neutrality, the private broadcasters tended to reflect the prevailing public sentiments. Originality / Value / Practical implications. This study provides a contribution by measuring the political bias in Slovak television news programs during a critical electoral period, using a comprehensive media bias index. Its findings highlight the disparity in objectivity between public and private broadcasters, offering valuable insights into how media bias can influence public opinion and electoral outcomes. The research also underscores the importance of media neutrality. These results can inform policymakers, media regulators, and broadcasters on the need to promote balanced reporting standards. Furthermore, the study’s methodology could serve as a model for future analyses of media bias in other countries and contexts.

We consider two symmetry metrics commonly used to analyze partisan gerrymandering: the mean-median difference (MM) and partisan bias (PB). Our main results compare, for combinations of seats and votes achievable in districted elections, the number of districts won by each party to the extent of potential deviation from the ideal metric values, taking into account the political geography of the state. These comparisons are motivated by examples where the MM and PB have been used in efforts to detect when a districting plan awards extreme number of districts won by some party. These examples include expert testimony, public-facing apps, recommendations by experts to redistricting commissions, and public policy proposals. To achieve this goal we perform both theoretical and empirical analyses of the MM and PB. In our theoretical analysis, we consider vote-share, seat-share pairs ( V , S ) for which one can construct election data having vote share V and seat share S , and turnout is equal in each district. We calculate the range of values that MM and PB can achieve on that constructed election data. In the process, we find the range of ( V , S ) pairs that achieve M M = 0 , and see that the corresponding range for PB is the same set of ( V , S ) pairs. We show how the set of such ( V , S ) pairs allowing for M M = 0 (and P B = 0 ) changes when turnout in each district is allowed to vary. By observing the results of this theoretical analysis, we can show that the values taken on by these metrics do not necessarily attain more extreme values in plans with more extreme numbers of districts won. We also analyze specific example elections, showing how these metrics can return unintuitive results. We follow this with an empirical study, where we show that on 18 different US maps these metrics can fail to detect extreme seats outcomes.

This chapter examines the extent to which partisan redistricting creates long-term distortions in congressional elections compared to other types of redistricting. The Supreme Court's failure so far to agree on a coherent and definitive test with which to adjudicate the issue of partisan gerrymandering has been predicated in large part on the absence of evidence of pervasive and long-lasting effects sufficient to meet the Davis v. Bandemer standard. It is thus necessary to determine exactly how effective partisan gerrymandering has been in terms of its long-term benefits to the gerrymandering party. This chapter considers the effects of control of redistricting on aggregate electoral disproportionality and partisan bias, as well as on the outcomes of elections in individual congressional districts. In particular, it discusses the probability that the Democratic Party candidate will win the election in a House district in a given year. The results suggest that partisan gerrymandering can produce a small but sometimes persistent bias in favor of the party that implemented the redistricting plan.