Ask a Political Scientist

Previous articleNext article FreeAsk a Political Scientist: A Conversation with Michael S. Lewis-Beck about Vote Choice, Election Forecasting, and the 2022 MidtermsInterviewer: Mary Stegmaier and Interviewer: Charles TienInterviewer: Mary Stegmaier and Interviewer: Charles TienPDFPDF PLUSFull Text Add to favoritesDownload CitationTrack CitationsPermissionsReprints Share onFacebookTwitterLinked InRedditEmailQR Code SectionsMoreMary Stegmaier & Charles Tien:You have written extensively about voters, elections, and the economy in countries far and wide. We want to first ask you about the American voter. How have American voters changed or stayed the same over the last fifty years?Michael Lewis-Beck:When we co-authored The American Voter Revisited,1 an overriding question was whether the explanation of American electoral behavior, as articulated in The American Voter2 remained the same. We concluded that it did, as encapsulated in the classic funnel of causality, sorting determinants into long-term forces at the opening of the funnel, i.e., socio-demographics and party identification, and short-term forces at the tip of the funnel, i.e., issues and candidates. That explanation still fundamentally holds, in my view.The funnel can be sketched as a causal model, tracing out the direct and indirect paths to the vote choice. The diagram can be written as a set of equations to be estimated. Each equation contains conceptually the same variables over time. The slope coefficients of these variables can change, becoming stronger or weaker. The strength of the party identification coefficient has considerable interest, as it forms an indicator of the stability of the electoral system. However, the weights of the slope coefficients in the model are more inertial than commonly supposed, e.g., despite the presence of candidate Trump in 2016, Republican identifiers were no more likely to defect than Democratic identifiers from Clinton. Some of the weights do change, though. For example, social class seems to have less importance, as Democrats abandon their working-class base. Also, measurement of socio-demographics can change, even if the concept does not seem to. For example, take gender, where there is a move away from the simple dichotomous notion of male or female. Further, with regard to race and ethnicity, we see the increasing presence of Latinx voters is changing the measurement of that variable.MS & CT:The media environment has changed dramatically over the last fifty years. Has this impacted vote choice?MLB:The forms of media have changed greatly. For example, in surveys, media questions used to focus on newspapers, radio, TV, magazines. Now the whole internet apparatus has introduced various types of social media, e.g., Facebook, Twitter, Instagram, blogs. All this suggests big shifts in the sources of political information. Have these different sources of information made voters more knowledgeable about politics? The knee jerk reaction may be "yes." However, considerable survey research argues that voters are no more knowledgeable about US political institutions and electoral processes than they were fifty years ago. Furthermore, content analysis of the American National Election Study (ANES) open-ended questions, which allows voters to extensively express themselves about the political world, shows a limited amount of sophisticated content in how they connect the political dots, e.g., most don't really know much about what constitutes a "liberal" or a "conservative." The political framework of typical voters has loose logic and little connective tissue; in some ways their political brain works like a popcorn popper, e.g., random reactive kernels. To take an example, if one examines the open-ended responses of the 2016 ANES regarding negative opinions about Clinton, few mentions of ideology, issues, or demographics appear; often they just say, "I don't like her."MS & CT:Recent elections have taken place during a pandemic—the first pandemic in the modern era of public opinion polling. How has the pandemic impacted elections and voters? Will it have lasting effects?MLB:The electoral impact of COVID-19 has been pervasive, mostly hurting incumbents, who are blamed for the massive death and destruction. The Trump defeat in 2020 has been attributed to COVID, and more than one study has shown that his handling of that crisis cost him votes. However, at the same time, other key variables continued to operate per usual, e.g., Trump was also blamed for a faltering economy. Indeed, the impressive survey investigation by Clarke et al., shows that the two most important issues in the contest were COVID-19 and the economy.3 This is suggestive of the pandemic's role in the "causal funnel." COVID has emerged as an important issue, to be considered at the tip of the funnel. However, it has not driven out all other issues. Moreover, when it ceases to be an issue—and it will cease—its coefficient will become, effectively, zero.MS & CT:As you look toward the future, what do you think scholars of voting behavior and elections should focus on?MLB:In democracies, vote choice follows laws of electoral behavior. Since these laws are probabilistic, there may appear some fluctuation over time in the estimated weight, say, of the party identification variable. Nevertheless, the underlying process endures, e.g., Republican identifiers vote overwhelmingly for the Republican candidate, Democratic identifiers vote overwhelmingly for the Democratic candidate. The strength of this bond has survived serious challenge, e.g., in the 2016 presidential election, despite the apparent disparity of the candidates, party identification was an equivalent magnet for Democrats (87% for Clinton) and for Republicans (86% for Trump), according to our analysis of the ANES data.4 These fundamental processes deserve ongoing study, with the expectation that they are operating. More and more, the "journalistic fallacy" has become the dominant hypothesis. That is, observers assume the election at hand exists apart from past elections, with a de novo electorate moving into unchartered territory. That perspective may sell newspapers, but it tends to be false.MS & CT:How concerned are you that American elections will be less free and fair in 2022 and 2024?MLB:Voting means choosing your leaders. If potential leaders are not in competition, then choice doesn't exist. That's why organizations like Freedom House count "free and fair elections" as a critical component of a democratic system. Voters in democracies around the world count on free and fair elections as a presence and as a continued aspiration.When we say an election is "free and fair," that says there is no corruption or coercion, an ideal not always realized in the tumble of practical politics. Take the example credited, perhaps unjustly, to pols in Gary, Indiana, where the charge was to "vote early and vote often." In the United States, vote-buying or ballot box stuffing is not unheard of and sometimes there are unintentional errors, e.g., lost ballots or double counting. However, these deviations today are rare and, most importantly, they are scattered and local, e.g., in a precinct or town. At the margin, say a very tight mayoral race, the declared winner might actually have won with the help of faulty ballots. But these misfires are isolated, virtually never of a magnitude large enough to shift a national election. This healthy condition comes from the selection and integrity of the state officials chosen to set the rules. It's driven by trust on the ground level, in American communities, and I do not see that safety mechanism significantly changing; it's too planted in our soil. Cicero said something like, "whatever the character of the leader in the state, such was the character of the state." It's not the rules, it's the rule-makers that matter. Take the current example now, of partisan gerrymandering: while Republicans have had control over drawing more districts, the Democrats have still managed gains. That's the sort of give-and-take one could expect from a healthy democracy. Of course, threats are ever-present, e.g., courtroom challenges to the Voting Rights Act.MS & CT:These threats, do they worry you about the state of American democracy? What keeps you up at night?MLB:In the short-term, "yes," in the long-term, "no." Once a country firmly establishes a democratic form of government, it tends to have remarkable resiliency. Take the Latin American example, where democracy can be more fragile, in part simply because it has not been in place long enough.5 With respect to the US case, we have, for the first time in my memory, a large portion of the public who believe the 2020 presidential election was rife with ballot fraud, despite evidence to the contrary. That attitude, if persistent, will undermine the democratic functioning of our institutions. It will persist if the good people who are our leaders do nothing. As Jefferson said, "Give the people light, and they will show you the way." That means to me, as a political scientist, I should speak the truth as I know it. Our training, which has as its core the reasoned sifting of scientific evidence, gives us a special obligation in that regard. One piece of relevant empirical evidence seems especially encouraging: when liberal-conservative ideology is examined via the ANES surveys, the modal response remains "moderate"; moreover, we see that those at the political extremes of the spectrum remain small, e.g., from 2008 to 2020 the range was only 3–5%. One implication is that the cries of polarization are too loud. (See my recent blog post with Helmut Norpoth).6This brings me to something that keeps me up at night as a political scientist. I'm worried about the quality of the survey research that, as scholars of public opinion and voting, we routinely use. In the United States and the United Kingdom, to name two leading democratic examples, the vote intention polls pre-election have too often been grossly off, in terms of correctly identifying the winner. With respect to the US case, we recently had the surprise win of Trump in 2016 and the surprise narrow win of Biden in 2020. The prediction errors have been large and are driven by the widespread poor quality of the polls.7 The core problem rests with the increasing departure from the scientific standards of probability sampling laid down by Leslie Kish.8 In a few words, pollsters seldom sample randomly from a population list (or its equivalent), making their task one of educated guess work.MS & CT:Let's turn away from the United States for a moment. Your book Economics and Elections: The Major Western Democracies (1988) is the foundation for comparative research on voting behavior and elections. Since then, scholars have considered how political institutions, economic contexts, and social conditions impact the magnitude of the economic vote. In democratic societies, does the economy always matter for elections? How might scholars improve our comparative analysis of elections and voting?MLB:The economy is like the weather; it's always there. In democracies, the economy matters for elections.9 Of course, because the "law" of economic voting is probabilistic, its estimated economic voting coefficient may carry a somewhat different value across contests. However, the underlying process endures, e.g., the White House gains votes in the face of boom, loses votes in the face of bust. The fundamental stability of the economic vote remains a hypothesis that forever faces challenges, on the reasonable assumption that, under certain conditions, it can change significantly. With co-authors, I've looked at several of these challenges, which mostly involve specification of interaction effects. For one, we examined, comparatively, whether electoral rules, by blurring clarity of responsibility, altered the economic voting coefficient; to our surprise, it didn't.10 We also examined the notion that, because of declining party identification, the economic voting coefficient has changed; again, it didn't.11 In contrast, we did find that, contrary to conventional wisdom, the retrospective economic voter can have a long, rather than a short, time horizon, with voter evaluations "abiding."12 These "stability" challenges are worth pursuing, in the name of better theory. However, they can be hard to prove. As Blalock observed fifty years ago, at the empirical level, the linear, additive model generally serves as well as any.MS & CT:Increasing globalization reduces government control over their national economy. Do you think voters recognize this? Or will they continue to hold their national leaders responsible?MLB:Conventional wisdom suggests that globalization, by reducing government control over the economy, reduces the valence of the economic vote. This argument bases itself on the clarity of responsibility notion, which holds that unless voters see government as economically responsible they won't punish it at the ballot box. However, recent work on benchmarking, including that by Brandon Park,13 suggests that globalization has allowed voters to receive information about their "benchmark" competitors, e.g., the Poles can now easily read about how the Czechs are doing and so make a judgement about their own government's relative performance. Such findings are suggestive of how dynamic research in economics and elections continues to be.MS & CT:How has the rise in right-wing populism changed European elections and voters? How would you assess the health of democracy in Europe?MLB:About populism, there are two kinds—right-wing and left-wing. Too often, left-wing populists, a very progressive force in my view, are lumped together with right-wing populists. An ongoing concern in Europe, for some, is the growth the bureaucracy, especially the supra-national bureaucracy of the European Union. This is particularly true as the EU, the collective government apparatus informally called "Europe," intervenes more and more in domestic policy decisions, large and small, e.g., from monetary policy to how raspberry jam can be made. This type of intervention in the economy, such as budget limits and regulations, takes a form few Americans would be comfortable with. This "Europe" pressure on democracy has been much investigated by continental scholars, such as Marina Costa Lobo and the MAPLE project. Once, when Paul Krugman, Nobel Laureate, was lecturing in Spain, someone asked him what the Spanish government could do about their high level of unemployment. Krugman, a devout Keynesian, said "nothing," it's all controlled in Brussels. A political scientist might suppose, then, that democratic dissatisfaction is on the rise in Europe, out of a lost sense of participation and control. However, work I've coauthored suggests that the average European is as satisfied with the functioning of democracy in their country now, as they were ten or twenty years ago.14 In other words, no downward trend. That is surprising to me. However, it again indicates the value of comparative quantitative political science research and how it can speak authoritatively to important public policy issues.Consider the case of Denmark, in some ways a beacon of democracy for Western Europe, not to say the rest of the advanced democracies in the world.15 In what ways does democratic performance shine in that country? Denmark has managed to maintain a clear transmission of the political will of the people, as embodied in the decisions of its elected elites, in the face of tough issues such as immigration, economic growth, and climate change. This has not always been so in other Western European democracies. Take the example of France, where the political elites have tended to favor the European Union, even when important sectors of the public have hesitated to embrace it.16MS & CT:You are one of the founders and leading lights of election forecasting. What has surprised you about forecasting in political science? Why do you think it is important that political scientists make election forecasts?MLB:When I began submitting election forecasting papers in the early 1980s, scholarly journals did not take the enterprise seriously (including one editor simply returning the submission). That world no longer exists. Election forecasting has taken its place as respectable science within the discipline, becoming a growth industry. This success has had a downside, in that certain commentators, not trained in the ways of social science, have jumped on the forecasting bandwagon. It is highly important that political scientists continue to lead the forecasting way since, unlike journalists, they follow the scientific of e.g., relevant scholarly and public a critical role in in political about are As we it to some it has e.g., the is made to the A tough of a or is to it the and see if it This is how the Take as an example the role of in the 2008 presidential election. Some observers that, was a free and fair race would not be a in the election. However, Charles and I from an election forecasting that cost about in the & you think the media are too on the issues enough during question is at if you about which more and more political scientists and voters The media role then, because they are the leading public information turn to media to see who is and political scientists count on media to their scientific A is that the media, driven by the race, never science which is what we at the big errors made in media election in the US in the 2016 and 2020 presidential contests. At one may be that the of public opinion has become seriously this is something that I've been with Charles and comes down to the quality of the A response to this is forecasting which I have investigated with Charles and Mary for a long time. forecasting works as well if not better vote intention and, believe it or it does not probability forecasting has been to democracies the e.g., the United and it's been to the 2022 & CT:The 2022 are What can we terms of we are in the time where key are to performance on key political and economic issues, with the of for the party in the White Take the as an We that the Republicans would likely have a serious of in to the Trump had a rather and the economy, despite his was returning only average With respect to we have the of rather Biden However, the economy pre-election has had its e.g., the good growth the of COVID-19 the first of with the ongoing In contrast, continues to Add to this the recent of by the may be key by the time we out our in it's for the & to your on elections, also for your and of research analysis and How would you your to research good scientific in causal What it or or or something The for this comes from doing and I would have a say, "I don't to take a because I had it I don't it you do it over and your to you about the world, has to to the about causal from the sample to the These are because most political scientists don't do like But even to where comes from when you do an which means random of to by the say, to one and not With that you much have we do that in political science, e.g., we make a we have to have more than one on the of the We in other in to control for the they would the of control is to to the true value of impact of in the the in your gives you a way to think about what is on (or in the political In my I doing of while always about model specification and what it means on the especially in the face of the & been to research with so we know the of the Lewis-Beck seem to have an to with and your with one clear you believe your first with that in a at a time, then a at a time. a of the by that lead the you are it follows from the The same with the two when one will but no more than simple with to the of the big the in your in words, not just If the look it out in equations or causal That should help the the to a the on the first the research question and its in In the the key At the of the an example, such as a or an or then on to from That should the In the say something and say it and last of political science, e.g., It's all about a with and a of or What is the & you have when is How do you course, some are than I to If I'm on one political science I to political science I to a or so at the same time. I in political science I to a or a I've that the science the and the the I'm not I like to and political magazines. If you want to be a good be a good most political I like for my to lead to The for me, is to follow the of at the work in a serious including a serious of If you a which you that out as as you be too about the was on a sampling i.e., an of only two or for of to is it was a good but I'm not you which & CT:You a ago, you have continued to in the What you to and on I'm in the social or A scientific comes up and I want to it good especially in the of with an It's a I to see the first or equation In up the I I'm an explanation someone to I look to submitting the for and I out journals I have not just for the of it. are good also and and there are ten what they are in political these good Lewis-Beck holds the of Political at the of are comparative elections, election political economy, and quantitative Lewis-Beck has or co-authored over and including Economics and The American Voter The Latin American Elections: and The and has as of the American of Political and the in quantitative is past of of and current of In to his at has various more of Paul at the of at for and of at Political of can be at Stegmaier is for and in the of and at the of research elections, voting behavior, and forecasting. work has been in journals including the of Forecasting, Political and Political & can be at is of Political at and the was a in American at in recent have in of The and can be at Michael S. Helmut and The American Voter of and The American Voter of and Trump The and Voting in the 2020 Michael S. Lewis-Beck and of on Michael S. and Latin American Elections: and of Helmut and Michael S. may be but most Americans are and policy Michael S. Charles in the 2016 US Vote Vote of and and Michael S. the Vote in the 2020 What with the in and Political in the 2020 and Leslie & Michael S. Lewis-Beck and Mary of of Political and Michael S. and the of European and Michael S. Vote in Western European Political and Michael S. the of European of Political Brandon to and relative economic and Michael S. less European does not with Michael S. and The Democratic and of Michael S. and Charles and Michael S. A Political and Charles and Michael S. in the 2016 US presidential vote vote of Michael S. Lewis-Beck and Charles as a of election of Michael S. Lewis-Beck and Mary voters see the and Michael S. blog Charles and Michael A Trump The Lewis-Beck and Michael S. An Previous articleNext article by The of the Political on this are from 2022 2022 Political All the this and Charles

Ask price · Conversation · Economics · Media studies · Political science · Politics · Sociology · Communication · Electoral Systems and Political Participation · Fiscal Policies and Political Economy · Law · Populism, Right-Wing Movements

• Stopping the Steal

  Mathilde Allain, Marco Lorenzo Allain et al.•The Professional Geographer•2024

  Add to list

• Stop Wars

  Alyson Cole, Robyn Marasco et al.•Polity•2022

  Add to list

• Did Covid‐19 Kill Trump Politically? The Pandemic and Voting in the 2020 Presidential Election

  Open Access•Harold D Clarke, Marianne C Stewart et al.•Social Science Quarterly•2021

  Add to list

• A changing economic vote in Western Europe? Long-term vs. short-term forces

  Open Access•Ruth Dassonneville, Michael S Lewis-Beck•European Political Science Review•2019

  Add to list

• More ‘Europe’, less Democracy? European integration does not erode satisfaction with democracy

  Open Access•Ruth Dassonneville, Alan Jabbour et al.•Electoral Studies•2021

  Add to list

• Rules, institutions and the economic vote

  Ruth Dassonneville, Michael S Lewis-Beck•West European Politics•2017

  Add to list

• The Hillary Hypotheses

  Open Access•Michael S Lewis-Beck, Stephen Quinlan•Perspectives on Politics•2019

  Add to list

• The abiding voter

  Open Access•Dieter Stiers, Ruth Dassonneville et al.•European Journal of Political…•2020

  Add to list

• Economic Determinants of Electoral Outcomes

  Michael S Lewis-Beck, Mary Stegmaier•Annual Review of Political Science•2000

  Add to list

• Compared to what? Media-guided reference points and relative economic voting

  Open Access•Brandon Beomseob Park•Electoral Studies•2019

  Add to list