Reflections on the American National Election Studies: An Interview with Nicholas Valentino

Reflections on the American National Election Studies: An Interview with Nicholas Valentino

Nicholas ValentinoThis year marks the 75th anniversary of the American National Election Studies– the definitive study of American political attitudes and behavior. The ANES has run national surveys of citizens before and after every presidential election since 1948, providing a rigorous, non-partisan basis for understanding contemporary issues as well as change over time. According to the National Science Foundation, the ANES provides “gold-standard data on voting, public opinion, and political participation in American national elections.” The Center for Political Studies interviewed ANES Principal Investigator Nicholas Valentino to gather his reflections on the history of the ANES, what we’ve learned from it, and where we’d be without it. This interview has been edited and condensed.

It’s the 75th anniversary of the ANES. What do you want people to know about it?

VALENTINO: It’s really important to remind ourselves about how important the ANES has been in terms of a gold standard survey, both methodologically and substantively, that other polling outfits use as a benchmark. We don’t often talk about those things, because we assume it is well known how important the ANES is for the entire discipline. But it is important that the ANES collects the highest quality samples; we care a lot about measurement, and we don’t change measures unless we know that the alternative is a better measure of the concept that we’re trying to tap. Every year, hundreds of dissertations and publications are written using ANES data. That means ANES data appear in thousands of scientific works per decade, and tens of thousands over the life of the project.

This is the most centrally important study of American public opinion and political behavior in existence. 


What’s unique about the ANES survey?

VALENTINO: With the rise of other less expensive samples and sampling methods, especially volunteer internet samples, entire generations of students are now able to collect data on their own rather than rely on infrastructure projects like the ANES. This is no doubt a great innovation, because each scholar can learn a lot about new topics and perform survey experiments for which the ANES is not well suited. But these new samples and sampling techniques can’t match ANES’s data quality both in terms of measurement practices and the sampling quality. They can’t provide valid and reliable estimates of opinions that the population holds because they do not collect representative samples of the country. They can’t replace address-based samples and the face-to-face interviewing methodology that the ANES has employed (with the COVID-year exception in 2020) for 75 years. And, perhaps equally importantly, with its time series questions, the ANES allows one to measure population level change in opinions over time. The ANES is very unique for both of these reasons, and so we think it has made good on the hopes of its founders 75 years ago. 

Let me say a bit more on how the ANES represents truly invaluable infrastructure for the entire discipline. The Department of Astronomy at Michigan recently invested in the instrumentation program for the Extremely Large Telescope (ELT), which is a consortium of universities and partners building the largest optical telescope in the world. Few universities can afford to build their own telescopes; instead, they buy into a measurement strategy that is fundamentally collaborative. The National Science Foundation supports these big projects, because they know that scholars from universities around the country can use the telescope regardless of the research budgets on their home campus. Well, that’s what the ANES is doing for the study of public opinion and political behavior. We include questions on the survey that have been asked before because we want continuity of measurement, and that are interesting to the broadest array of users in the community. The user community has several mechanisms for influencing the content of the ANES, including specific proposals and by serving on the large and diverse Advisory Board.


Where would we be today were it not for the ANES?

Valentino: One example is that we are currently very concerned about affective polarization or the polarization of society along lines of party. It was once the case that Democrats and Republicans didn’t dislike each other very much at all. In fact, if you ask Democrats and Republicans how warmly they felt toward their own party and toward the out-party, they once gave both roughly the same score. If you track this in the ANES over time, we see that after the 1980s there was a dramatic, continuous increase in how much Americans disliked members of the other party. That discovery has triggered a wide ranging set of inquiries about whether it was driven by elite preferences, differences in the party platforms, or by something that was changing in the electorate. We’re still trying to answer this question about why. But we wouldn’t really have known how big the problem was, or when it started, if not for the ANES. The ANES tells us when it started, and how fast it grew. And those pieces of information are very critical for understanding why it happened. 


What kinds of questions have been asked continuously over the history of the ANES? 

Valentino: It must have been very hard in 1950 to know which questions to ask, especially if one wanted to begin a time series that would last 75 years. You can imagine how different the world was when Warren Miller and his colleagues started thinking about what questions they should ask. One example of a question we have asked continuously over time is, “How much do you trust the government to do what’s right?” If you look up the trend on government trust on the accessible web page– the ANES Guide to Public Opinion and Electoral Behavior– you will see a dramatic decline in the percentage of people who say they trust the government to do what’s right. This trend is one of the things people point to when they worry about democratic backsliding and the deterioration of democratic norms. Back in the late 1950s, in the Eisenhower era, almost 80% of Americans said they could trust the government “most of the time or just about always.” Now that number has decreased to about 20%. Another question has been asked for quite some time on this topic: “Are governmental officials crooked?” At one time the vast majority of Americans responded “not many,”or “hardly any.” Since 2008, there has been a massive increase in the share of the electorate from either party who say “quite a few” elected officials are crooked. So that’s a massive shift and a decrease in trust that we’re aware of because of the ANES asking the same questions consistently over time. Americans’ trust in government hit its nadir in 2020– making the future of the ANES particularly valuable for understanding how democracies react to declines in public confidence.


What have we learned from the ANES about how voters select their candidate?

Valentino: One of the early discoveries of the founders of the ANES was that psychological attachments to parties themselves are group identities.

Party identification is the core predisposition shaping political beliefs and behaviors. In other words, it may be that votes are driven most strongly by those partisan identities rather than a citizen’s individual preferences on a bundle of policies. That discovery can have a very scary set of implications.

Most definitions of democracy insist that citizens have the ability to freely choose candidates who represent their issue preferences and hold elected officials accountable when they fail to produce results consistent with their campaign promises. So what if pre-existing attachments to parties drive vote choices even when candidates are not delivering on issues or economic performance? In that system, leaders could pursue policy interests divorced from the majority’s will. It reduces the impact that issues, trends, and performance have on democratic elections. If so, elections increasingly resemble sports events, in which it is forbidden to criticize the home team. 

The ANES has documented a long-term shift in party identification, and especially a decline in Democratic dominance from the 1950s to the present. Over time, there is a lot of variation in partisan attachments that is not predicted by social group memberships. You might look at individual-level psychological forces to explain, for example, how white, blue-collar Democrats were pushed toward the Republican party after the Civil Rights movement. That is a dynamic you couldn’t understand by looking solely at social group memberships or by considering material interests. The early scholars of the ANES were among the first to discover that, contrary to popular belief, people don’t vote with their personal pocketbooks. People are motivated less by their own financial security than by how they think the country is doing as a whole. Americans, at least until very recently, support the incumbent if the country is doing well, even if that candidate is from the other party. The ANES was central in discovering this pattern. It will also be central for understanding how these long-term linkages between economic performance and support for incumbents may be breaking down.


What have we learned from the ANES about why some voters participate in politics and others don’t?

Valentino: It is often remarked that political science has few “laws” – theories that are so powerful and elegant that they have reached the status of a settled scientific explanation for important phenomena. One exception is the Civic Volunteerism model of participation proposed by Verba, Schlozman and Brady (1995). This theory claims that people turn out because they want to, because they’ve been asked to, or because they can. Those forces– engagement, recruitment, and resources– are all measurable, and the ANES measures them over time. The model has done a fantastic job explaining who will participate in politics in a variety of ways.

But short-term forces also matter. For example, people with a lot of resources participate in some elections but not others, and some people who have very few of these advantages participate, even though the model might predict that they wouldn’t. The short-term forces are also measurable, and many political psychologists are working on those. We have, for example, identified short-term emotional dynamics that can drive people to the polls– sometimes, but not others– even holding resources constant. You need to ask people about the emotional intensity they are experiencing in a given election, the intensity of competition between the candidates, about campaign negativity, and the tone of political discourse. These factors have also been measured by the ANES, and so our theories of participation are getting stronger and more comprehensive.


How and why are new questions added to the ANES?

Some parts of the ANES every election are about mapping long-term trends, but about 30% of the content on the ANES is concerned with the contemporary moment. People think that the ANES is only about asking the same questions, 75 years straight, but that is a profound misconception. The project solicits input from the user community every cycle, and relies on its diverse Advisory Board to identify contemporary issues it must explore. The ANES actively cultivates new questions about very new issues that might be driving the dynamic in a given election year, but may or may not be permanent parts of our landscape. And we try to balance those new inquiries against the space needed in order to measure long-term trends. Only then can political science as a discipline understand how we got to this point, and predict where we might be going.



  • The ANES is a centrally important study of American public opinion and political behavior.
  • The ANES provides gold-standard data on voting, public opinion, and political participation in American national elections, priding itself on sampling quality and measurement quality.
  • Looking at long-term trends, we currently stand at a low-point for Americans’ trust in government and an all-time high for affective polarization. These trends are understandable because the ANES’s time series questions track change over time.
  • The ANES has demonstrated that political identity plays a strong role in determining how voters choose candidates.
  • People participate in politics when they are interested, asked to participate, and have the resources to participate. But short-term factors like the emotional intensity of an election also matter. The ANES survey measures all of these factors.
  • In recent cycles, about 30% of the survey has been dedicated to new questions investigating contemporary issues based on input from its community of users.
  • Tens of thousands of dissertations and publications have been based on data from the ANES over the project’s history.

Nicholas Valentino is Professor of Political Science at the University of Michigan and a Research Professor affiliated with the Center for Political Studies at the Institute for Social Research. He received his Ph.D. from the University of California, Los Angeles, in 1998. His research focuses on political campaigns, racial attitudes, emotions, and social group cues in news and political advertising. His current work examines the intersection between racial attitudes and emotion in predicting political participation and vote choice, as well as the sources of public support and opposition to immigration in the U.S. and cross-nationally. 

The ANES is currently a collaboration of Duke University, the University of Michigan, Stanford University, and the University of Texas at Austin, with funding from the National Science Foundation.

This post was developed by Tevah Platt.