This post looks at the opinions of Twitter users surrounding the first Presidential Debate. We look at content containing at least one debate hashtag, shared immediately before, during, and after the debate; and we determine the “stance” or opinion (for or against) of each tweet towards Biden and Trump.
The figure below shows the average proportion of expressed support or opposition for the candidate every minute of the debate from 8pm (20:00) to 11:30pm (23:30). A score above zero indicates a net positive stance towards the candidate. A score below zero indicates a net negative stance.
Presidential Debate 1: Stance of Candidates on Twitter
We see that in the hour before the debate begins, both candidates have a net negative stance. In other words, more opinions against each candidate are being shared than are opinions for each candidate. At around the 11 minute mark in the debate (roughly 21:11), pro-Biden expressions begin increasing, and continues to increase until the overall stance is in support of Biden. In contrast, around the same time, stance towards Trump decreases and continues to decrease for the first 10 minutes.
Over the course of the debate there are specific moments that help and hurt each of the candidates. When there is perceived bickering, there is usually a decline in stance for both candidates, although there are exceptions. The moment in which Trump received the most support was when he spoke about judges. Biden’s best moment was when he discussed race relations and the need to support black Americans.
By the end of the debate, the stance of Twitter discussion towards Biden had increased by 0.5 – a striking shift. He clearly benefited from the debate, at least in the short term amongst Twitter users. In contrast, the stance of Twitter discussion towards Trump decreased by approximately 0.2. Even as there was a good deal of opposition towards Trumps expressed immediately before the debate, there was even more negativity towards him at the end of the debate.
It is worth noting that within an hour of the debate the expressed stance towards Trump returned to pre-debate levels. These are decidedly negative, of course; but the additional negative impact of the debate on Twitter discussion of Trump may have been short-lived. The same is not true for Biden. The hours surrounding the debate saw a marked shift in expressed stance towards Biden, from by-minute averages that were anti-Biden to clearly pro-Biden. The shift is evident only 10 minutes into the 90-minute debate, and durable for the hour following the debate as well.
Twitter is by no means an accurate representation of public opinion more broadly – we must be sure to interpret these results as indicating the debate impact on Twitter discussion, not the public writ large. That said, where Twitter is concerned it seems relatively clear that Biden ‘won’ the debate.
Information about the analysis:
This analysis was conducted using approximately 1.3 million tweets that contained at least of the debate hashtags. We collect posts using the Twitter Streaming API. We use the core debate hashtags for this analysis, e.g. #debates2020, #presidentialdebate2020, etc. We determine if the tweet showed support, opposition, or neither for each candidate. For each minute, we compute an aggregate stance score as follows: Stance Score = (# Support – # Oppose) / (# of tweets that minute having a stance). To determine the stance itself, we trained a BERT fined tune model with a single layer on 5 million posts related to election 2020. We also had three people label 1000 tweets with stance to further improve our model.
This analysis was conducted by the Political Communications Election 2020 project of the Social Science and Social Media Collaborative (S3MC). The faculty involved in that project include Ceren Budak (University of Michigan), Jonathan Ladd (Georgetown University), Josh Pasek (University of Michigan), Lisa Singh (Georgetown University), Stuart Soroka (University of Michigan), and Michael Traugott (University of Michigan). The work is funded in part by National Science Foundation awards #1934925 and #1934494 and the Massive Data Institute. This project is a collaborative effort by the University of Michigan and Georgetown University to address how to harness the abundance of data from social media in order to understand social and political trends better. For the latest updates about this group’s research related to the 2020 Election, visit the project website: https://s3mc.org/political-communication/election-2020-project/. For information about the interdisciplinary methodology being developed by this group, visit: http://smrconverge.org/home/methodology/
Until recently, there has been little need to measure the electoral volatility, changes in vote shares between parties, in authoritarian regimes because most conventional authoritarian regimes were either one-party or no-party systems. In general, high levels of volatility are considered to be a sign of instability in the party system and show that the existing parties are unable to build connections with their constituencies.
New research by Wooseok Kim, Allen Hicken, and Michael Bernhard examines the ways that electoral volatility in democratic regimes may be useful for understanding competitive authoritarian regimes.
As a greater number of authoritarian regimes have permitted electoral competition and greater party autonomy, electoral volatility has become more salient. Multiparty elections in competitive authoritarian regimes are different from those in democracies, in that competition is more constrained and incumbents have the ability to manipulate the outcomes.
Electoral volatility can provide clues about the level institutionalization in the ruling and opposition parties, as well as the level of support for the authoritarian incumbent. Low volatility suggests a high level of stability and control in the ruling party institutionalization; high volatility as associated with weak party organizations, weak societal roots, and low levels of cohesion.
The authors tested the relationship between electoral volatility, which is the most commonly used measure of party system institutionalization, and the survival of competitive authoritarian regimes. To do this, they used a dataset that included authoritarian regimes in the post-WWII period that hold minimally competitive multiparty elections with basic suffrage, which are determined using indicators from the Varieties of Democracy (V-Dem).
Specifically, the authors measure two types of electoral volatility in competitive authoritarian regimes: type-A volatility and type-B volatility. Type-A is volatility measures the exit and entries of parties from the system. Type-B volatility measures the reallocation of votes or seats from one party to a competitor.
Electoral authoritarian regimes are more stable when they tightly control the party system and the opposition is disorganized. The authors conclude that type-B volatility promotes authoritarian replacement, while type-A volatility is associated with a greater likelihood of a democratic transition. In addition to considering measures of party system institutionalization in authoritarian regimes, future case studies may shed more light on the link between electoral dynamics and outcomes.
This year the American National Election Study (ANES) will conduct its 19th time series study of a presidential election. In every U.S. presidential election since 1948, the ANES has conducted pre- and post-election surveys of a large representative sample of American voters.
Since 1977, the ANES has been funded by the National Science Foundation. It is used by scholars as well as high-school students, college students, and journalists. The data are made publicly available online for free as soon as it is processed after the election; principal investigators of the study do not receive privileged access to the survey data.
The ANES aims to answer two fundamental questions: how do citizens select the candidate they vote for? Why do some citizens participate in politics (e.g., vote, work on campaigns, etc.) while others do not? These questions are answered with nationally representative survey data.
The value of the ANES comes not only from the care and precision brought to designing questions, but also from the way the study balances continuity and innovation. In order to achieve this balance, the ANES asks identical questions over time about vote choice, turnout, party identification, ideology, political information, and attitudes about candidates. But even as questions are preserved over time, new questions are added about issues as they arise. The investigators and board members solicit public input on new questions and determine which ones will add value.
Recent data trends
Professor Hutchings outlined findings from some of the questions that were recently added to the ANES, including questions about the Black Lives Matter movement and police misconduct.
Respondents to the 2016 ANES were asked to rate the Black Lives Matter movement on a 0-100 “feeling thermometer” scale. Ratings 50-100 degrees signal favorable feelings toward the group; ratings 0-50 degrees signify unfavorable feelings. Respondents would rate the group at the 50 degree mark if they don’t feel particularly warm or cold toward the group.
Hutchings points out that there are important partisan and racial divides in the results shown above. For example, Black Republicans have warmer feelings toward the Black Lives Matter movement than white Democrats in 2016. This question will be repeated in the 2020 study, giving researchers a way to track changes in perceptions of the movement over time.
Attitudes toward the Black Lives Matter movement were a very strong predictor of the candidate a respondent would vote for in 2016. As Hutchings showed using the graphic below, voters who supported the Black Lives Matter movement were much more likely to support Hillary Clinton for president.
Similarly, perceptions of police violence were correlated with voter preference. Those respondents who believed that whites were treated better by the police were much more likely to support Hillary Clinton than respondents who believed that police are unbiased.
The value of the ANES
Professor Hutchings concluded his talk by reflecting on the value of the ANES. “It allows us an opportunity to assess the health of our democracy,” he said. “We can assess levels of trust in government, levels of perceived corruption in government, levels of racial animus, levels of religious and gender intolerance. We can assess how things have changed – or how things have not changed – over time. And we can only do this as a consequence of this study.”
A transcript of Michael Traugott’s remarks follows.
There’s been quite a bit of research about voting by mail. I actually participated in a research project in Oregon in 1995 the first all-mail election and there is no indication that mail-in voting produces any kind of fraud. For that matter, we have almost no fraud in American elections.
Having a vote-by-mail election is a complicated enterprise. Any election is an audit process in which the security of the ballots has to be maintained. Vote-by-mail elections actually cost more than a machine-based election because it requires more staff, the votes come in over a longer period of time, they have to be secured, and then counted. So it’s just as safe and secure, with proper preparation and with sufficient funding, as any other machine election.
One thing that might be going on is that the President is trying to run out the clock, in the sense that in order to have a secure vote-by-mail election, we probably have to have the funding in place and the local election administrators have to be organized by September. So there’s really only four or five weeks left in order to prepare for our mail election or to have a large number of absentee ballots printed and available.
It’s actually a kind of a fable or a myth that we have national elections in the United States. We really have a series of state and local elections held on the same day. But all of the rules about how you register, how you can get an absentee ballot, how many precincts there are, all of this is regulated by local officials. So while each local official is responsible for the election in their own jurisdiction. It takes a lot of coordination to get the votes counted, for example, at the state level.
Congress passed a law in 1845 as a way of regularizing the Electoral College procedures and they said that federal elections will be held on the Tuesday after the first Monday in November in even numbered years, and that has set the calendar for all of our elections. They have never been altered or postponed. Sometimes under unusual circumstances a local election has been postponed, for example a storm or hurricane or something like that. But there is no way that the president of the United States can change the date of an election. It requires an act of Congress.
I think the tweets are strategic. Donald Trump uses these tweets to distract journalists, for example, from covering other important elements of the news of the day. They also have purpose in appealing to his particular base but they don’t serve any useful function for the general public. And in fact, I would be concerned that tweets about the quality of voting in the United States or the need to postpone election day would increase distrust in the public about how our government functions. That’s clearly a bad thing.
I think that the Trump administration is trying to question the validity of the election in November, the accuracy of the vote count and other related factors. It’s all of a kind of debilitating message to American democracy.
Experts from the Center for Political Studies are available to discuss current topics related to elections, politics, international affairs, and more. Click here to find experts.
Words That Matter is the collaborative work of eight authors: Leticia Bode, Ceren Budak, Jonathan M. Ladd, Frank Newport, Josh Pasek, Lisa O. Singh, Stuart N. Soroka, and Michael W. Traugott. The authors have expertise in a range of disciplines including public opinion, communications, public policy, and computer science, and they take different approaches to the study of campaign media. As a result, the book is nuanced in its handling of news content, social media posts, and survey responses.
There are a number of reasons that the 2016 presidential campaign was exceptional. The media landscape has changed dramatically in recent years, with many people accessing and sharing news through social media. The authors find that news coverage during the 2016 campaign “was more negative than in recent previous presidential campaigns, consistent with these candidates being the most personally unpopular nominees in polling history.”
Words That Matter guides readers through the media’s process of producing information, how that information gets to voters, and what information voters actually absorb. The authors argue that advances in media technology call for new ways to measure the information environment. They address this challenge through innovative surveys and content-analytic research techniques.
This figure highlights the changing topics that Americans remember about Clinton since July 2016. The x-axis shows the date and the y-axis the fraction of responses that fall into a particular topic.
A key finding of the work is that the largely negative campaign played out differently for the two major party candidates: Donald Trump was confronted with a shifting but largely uninfluential series of scandals, whereas Hillary Clinton faced a single, stable, and influential scandal involving her use of a private email server. The authors show that the long-standing nature of the email scandal made it especially sticky in the public mind. They write “Even when there was other news about Hillary Clinton, the public thought about ‘her emails’—for months and months—indeed, starting before the election campaign was even underway.”
Some scholars are skeptical that the media have the power to influence votes, whereas others believe that campaign messaging can have a large effect. The authors show that not all voters are equally open to influence. The most politically-engaged voters are steadfast, while the least engaged are difficult to reach at all. “The fact that middle- and low-engagement voters are the most susceptible to influence,” write the authors, “also helps us understand why the topics given heavy attention in the media environment can be consequential.”
News stories that are repeated over a long period of time are the most likely to be noticed by people who are not highly engaged with politics. The authors also find that telling people how to vote is less effective than simply changing the subject. Voters who don’t follow the news carefully may not remember the details of various scandals, but they do tend to notice if one specific issue garners sustained coverage. Those sustained scandals stand out as more important when voters make their choice.
The authors conclude that media content can indeedshift voter behavior for some voters, and that in a close election like the 2016 presidential election, these effects can be of real consequence.
Post developed by Katherine Pearson and Mai Hassan.
States can exert powerful social control over citizens. In her newly-published book, Regime Threats and State Solutions, Mai Hassan demonstrates how leaders use their authority to manage bureaucrats to advance their policy and political goals.
By controlling which bureaucrats are hired, where they’re posted, how long they stay in a post, and who gets fired or promoted, leaders can induce the bureaucratic behaviors that will help keep them in power.
Focusing on Kenya since independence, Hassan uses qualitative and quantitative data gleaned from archival records and interviews to show how the country’s different leaders have strategically managed the public sector. The data show that the strategic management of bureaucrats existed under the one-party authoritarian regime beginning with Kenya’s independence in 1963, and continued after Kenya’s transition to an electoral regime in 1991. Under both regime types, leaders were able to co-opt societal groups that are needed for support and coerce the groups most likely to challenge the regime.
Haasan examines how leaders rely on bureaucrats to manage popular threats against the leader such as protests and strikes. First, she argues that leaders assign bureaucrats with deep social bonds to those areas where the leader needs to co-opt the local population. These deep social bonds compel bureaucrats to work on behalf of the area. But in areas that need more coercion, the leader tends to prevent the posting of bureaucrats with deep local roots because those who have deep roots will be unwilling to coerce locals.
Second, she finds that the parts of the country that are most strategically important for the leader — and thus, the areas of the country where bureaucratic compliance is needed most — are staffed by the most loyal bureaucrats, those who are most willing to help keep the leader in office. Leaders can also neutralize the risks of disloyal bureaucrats by carefully managing where potentially disloyal officers are posted and how long they stay in their posts.
Why would a leader hire or promote disloyal bureaucrats in the first place? Hassan addresses this question by showing that most state bureaucracies are not actually packed with the leader’s in-group members, who tend to be the most loyal. Elite threats, such as coups, tend to be more pressing than popular ones. Leaders can appease rival elites by hiring and promoting bureaucrats who are loyal to elites other than the leader. Strategically posting and shuffling bureaucrats allows the leader to recruit potentially disloyal bureaucrats in order to temper elite threats, while still relying on loyal bureaucrats to prevent popular threats where they are most likely to emerge.
Overall, Hassan’s analysis shows how even states categorized as weak have proven capable of helping their leader stay in power. Her work demonstrates how the strategic management of bureaucrats solves both elite and popular threats, and in doing so, highlights why bureaucrats must be taken seriously. States may assert power, but states do not act: bureaucrats do.
