Join John Sides and Lynn Vavreck for a free online discussion and Q&A on The Gamble [Change in Date!]

Event logoJoin and political scientists John Sides (GWU, The Monkey Cage blog) and Lynn Vavreck (UCLA) for a free online talk about The Gamble: Choice and Chance in the 2012 Presidential Election followed by an audience Q&A session.

Date: Friday, October 7, 2013 [Change in date!! this was originally scheduled for September 27, but is postponed to October 7]

Time: 3:00 PM EST

Place: Your computer — all that’s needed is a fast internet connection and access to an internet browser

Sides and Vavreck will reveal their Moneyball approach to campaign analysis and discuss the writing process for The Gamble, a book praised by Nate Silver as “the definitive account of what really happened and what really mattered in the campaign.” Sides and Vavreck specialize in bringing hard data to bear and casting doubt on a lot of commentary and conventional wisdom. As a result they inject a dose of much-needed reality into a discourse too often dominated by speculation and folklore.

You can learn more about Sides, Vavreck, and The Gamble at the book’s dedicated web site:

Check out the event page at Shindig: Let us know if you’ll be there by RSVP’ing below, though this is not really necessary — you can just show up if you want.

John Sides on How the 2012 Election Was Good for Political Science

In late September, I was involved in an email exchange in which a historian stated that “Someone should do a piece cataloging down all the poli sci consensi being undone this season.”  Now I can write with some confidence that the findings of the political science canon were largely confirmed by the 2012 election. And those findings deserve some plaudits alongside the polls, the forecasters, and the “nerds” at the heart of the winning presidential campaign.

In our book, The Gamble, Lynn Vavreck and I are attempting to show how those lessons can inform our understanding of the 2012 election.  Here is a list of findings that I think hold up reasonably well, with citations to representative studies and findings from our book where possible.

1. In presidential primaries, party leaders work to coordinate on a candidate even before the first caucuses and primaries are held. The candidate backed by the most party leaders is likely to win the presidential primary (Martin Cohen et al., The Party Decides).

In 2012, like in 2008, consensus was harder to achieve than in some previous years.  Many Republican party leaders did not publicly endorse any candidate, as Lynn and I show in our chapter “Random, or Romney?”   At the same time, among those who did endorse, the vast majority endorsed Romney.  Here is our graph of endorsements by Republican governors and members of Congress before the Iowa caucus.

Some party leaders also worked to oppose certain candidates — most publicly, Newt Gingrich and Rick Santorum.  Consider this headline for example: “Governors look to Santorum with dread.”  The behavior of party leaders was one reason why I was confident that Romney would win.

2. Views of presidential primary candidates can changed sharply in the wake of events judged to be dramatic or unexpected, and covered as such by the news media (Larry Bartels, Presidential Primaries).

Just because Romney was favored to win didn’t mean that he would always lead in the polls.  Lynn and I also track the sharply changing fortunes of the Republican candidates and show how a burst of favorable news coverage could give them a boost in the polls, but less favorable news coverage tended to have the opposite impact.  A good example is Herman Cain.  Here’s our graph from the “Random, or Romney?” chapter, drawing on data on media coverage and tone from General Sentiment.

3. In the general election, incumbent presidents running amidst even modest economic growth are  likely to be reelected.

I summarized this here.

4. The vast majority of people identify with one of the major political  parties and vote loyally for that party in presidential elections (Campbell et al., The American Voter). 

In a post-election YouGov poll, conducted Nov. 10-12,  89% of self-reported voters identified with or leaned toward the Democratic or Republican party.  Rates of party loyalty were extremely high — 93% of Democrats voted for Obama and 94% of Republicans voted for Romney.  The same was true in the exit poll, which does not ask whether independents lean toward a party. Polls since at least April had suggested that party loyalty would be strong, despite discussion of the “divided” Republican party, Obama’s “rebellion on the left,” etc.  Party identification has become more strongly associated with voting behavior since the 1970s, and this shows no sign of weakening.

5. Voters tend to have stable preferences about the two major-party presidential candidates (Lazarsfeld et al., The People’s Choice).

The cite above is to perhaps the earliest quantitative study of a presidential election. In this study of the 1940 election, Paul Lazarsfeld and co-authors found that the campaign “served the important purposes of preserving prior decisions instead of initiating new decisions.”  The power of party identification is likely one important reason for this stability.  For example, consider YouGov respondents who were interviewed first in December 2011 and then again the weekend before the Election Day — almost 11 months later.  Of those who said in December that they would vote for Obama in an Obama-Romney race, 95% still preferred Obama on the election’s eve.  Of those who preferred Romney in December, 94% did so again in November.

6. Campaign events can move the polls. 

Nothing about #3-5 means that presidential general election campaigns have no impact whatsoever — though it sometimes seems hard for commentators to grasp this nuance. During the 2012 campaign, campaign events had effects largely in line with previous research.  For example, there was predictable movement after the party conventions, mainly the Democratic National Convention.  I wrote about the research on convention bumps here.  Similarly, candidate debates during the general election can affect preferences, but tend not to propel the underdog to victory.  Romney’s gains after the first debate were real, but ultimately not enough.  Here was my post-mortem on the 2012 debates, in light of my earlier piece.

7. The outcome late in the election tends to reflect the polls.

I wrote about this here, drawing on Robert Erikson and Christopher Wlezien’s Timeline of Presidential Elections.  Perhaps the most interesting thing about 2012 is that the polls tended to under-estimate the frontrunner’s margin of victory (that is, Obama’s), whereas usually they over-estimate it.  This may be because the Obama campaign was particularly good at minimizing the “no-shows” — people who prefer the frontrunner but ultimately fail to show up and vote.

Finally, let me cite two other findings that I suspect will hold up, but for which we do not yet have definitive evidence from the 2012 campaign.

8. Presidential television advertising can matter at the margins, but it typically does so only when one candidate can outspend the other significantly. However, the effects of ads appear to dissipate quickly, within a week.  I cited some of this literature here.  From my analyses of the ad data at Wonkblog — here was the last one before the election — it seems as though neither Obama nor Romney ever got enough of an edge in advertising volume for it to make a difference.  But more thorough analysis is needed.

9. A variety of get-out-the-vote tactics do in fact increase turnout — especially person-to-person contact (Green and Gerber, Get Out the Vote).  The social science research on GOTV was central to the Obama campaign. Hopefully there will be ways to measure the effect of the “ground game” on the outcome.

Of course, I’m happy to acknowledge that political science findings aren’t ironclad. But little about this season strikes me as undoing its central findings about presidential elections.

For much more on this topic, stay tuned for more from The Gamble.

Are Mitt Romney’s Wealth and Taxes Taxing His Campaign?

Back in January, Lynn Vavreck, Joshua Tucker, and I asked this question: what would happen if people knew more specific details about Mitt Romney’s income and tax rate?  We conducted a simple experiment the exposed people to information about these topics.  In this post, I will report on a new iteration of this experiment, which seems particularly timely in light of this Obama ad, for example.

In our first study, we found that Americans tended to think that Romney did not pay his “fair share” of taxes.  In a July 28-30 YouGov poll, the same was true, and perhaps even more so.  Compare perceptions of Romney and Obama in January vs. July:

During these six months, the percentage who had no opinion about what Obama and Romney paid has declined.  For both candidates, the percentage saying that they paid less than their fair share has increased (by about 5 points).  Romney’s disadvantage on this issue remains: a slight majority of Americans now believe that he does not pay his fair share.  And this is not something that only Democrats believe.  Almost half (48%) of “pure” independents believe this as well.

For comparison, here is how respondents felt about the taxes paid by Newt Gingrich, Bill Gates, “most people,” and “people like yourself.”  This shows, for example, that perceptions of Romney are less favorable than those of Gingrich and Gates.

After July respondents had answered these questions, a half-sample was assigned at random to see one of the following:

  • Information about how the average American’s income compares to Romney’s: “The Census Bureau has estimated that the average American household earned about $50,000 in 2010.  In August, Mitt Romney disclosed that in 2010 he and his wife had earned somewhere between $7 million and $40 million.”
  • Information about how federal tax rates for different income levels compare to Romney’s: “In this country, a person making under $20,000 each year pays about 2% of their income in federal taxes.  A person making $60,000 pays about 13% of their income in federal taxes.  A person making $250,000 pays about 20% of their income in federal taxes.  Last week, Mitt Romney suggested that he paid about 15% of his income in federal taxes.”
  • Information about average income and tax rates, with no mention of Romney: ““The Census Bureau has estimated that the average American household earned about $50,000 in 2010.  In this country, a person making under $20,000 each year pays about 2% of their income in federal taxes.  A person making $60,000 pays about 13% of their income in federal taxes.  A person making $250,000 pays about 20% of their income in federal taxes.”

This experiment was designed before Romney had released some details about his income and tax rate, but the information in the experiment corresponds closely to what he released, as we noted in the original post.  For the sake of strict continuity, I opted to keep the experiment the same rather than change it to reflect what Romney released.

Below I compare the key findings from the January experiment — quoting from the earlier post — with new analysis from the July experiment:

…respondents who saw information about either Romney’s wealth or tax rate were less likely to believe he “cares about people like me”—provided they already believed he wasn’t paying his fair share of taxes.

In other words, a group that didn’t like Romney very much to begin with liked him a little less after seeing this information.  In the July experiment, however, even this modest finding did not emerge.  Neither piece of information affected perceptions of Romney on this dimension.

…being told Romney’s income increased the percentage who said that “cares about the wealthy” describes Romney “very well”…

The same is true in July.  In general, Americans now perceive Romney as more concerned about the wealthy than they did in July.  In July, 50% said that “cares about the wealthy” described him very well. Now, 60% say that.  In both experiments, specific information about his income increased that percentage.  Compared to someone who saw no information about Romney, someone who learned that he had made millions of dollars in 2010 was 12 points more likely to say “very well” in the January survey and 10 points more likely to do so in the July survey.

We previously found that believing Romney cares about the wealthy is correlated with believing he doesn’t care about “people like me.”  In this experiment, this correlation is significantly larger when told either about Romney’s income or tax rate than when told neither piece of information.

In the July experiment, we found that information about Romney’s income (although not his tax rate) had this same effect.

What is the upshot here?  First, on dimensions related to wealth and empathy, Romney is perceived less favorably than Obama.  That’s been the upshot of several of my posts with Lynn (e.g., here) and we will be updating that analysis soon.  Other public polls, such as those of the Washington Post and Gallup, confirm this.

Second, specific information about Romney’s wealth and income may worsen perceptions of him on these dimensions, but it is not a game-changer.  As we noted in our earlier post:

The information about Romney’s income or tax rate did not affect how respondents evaluated Romney on other dimensions, such as his willingness to stick by his positions, his honesty, or his trustworthiness.  It didn’t make respondents more likely to describe him as personally wealthy (most already do so anyway).  And it didn’t change whether they believed he cares about the poor or middle class.  When the information does move opinions, the shifts aren’t large.  Many respondents may already have heard about Romney’s income or tax rate or simply don’t consider those facts germane.  The Obama team may find that a campaign that implicitly or explicitly characterizes Romney as a plutocrat isn’t a slam dunk.

That’s still true now, and it’s worth keeping in mind when ads like the one linked above debut.  At the same time, our experiments are just words on the screen during a survey interview, and may not have the impact of a political ad, with its richer palette of image and sound.

Third, even if Romney faces disadvantages on dimensions related to wealth and empathy, it’s not certain whether those dimensions will be the most important ones in November.  Empathy is not the one true key to victory.  More important may be perceptions of which candidate will best improve the economy — something on which Romney has the advantage.

Welcoming new regular guest blogger on Technology and the Election, John McGinnis

John McGinnis, author of the forthcoming Accelerating Democracy: Matching Governance to Technological Change will be taking to Election 101 to blog each week about technology and the election.  Anyone with a twitter account and an iPhone knows that technology has advanced at an explosive pace in recent years. We have, after all, elected the first ‘social media’ president, viral marketing has never been bigger, and prediction technology often seems instantaneous.  Yet democratic governance has not yet caught up with all the advancements that have taken place. Technology can, of course, be a boon to democracy in many ways–while social planning was once a top-down enterprise, now we have a stream of information connecting us to the issues everywhere in record time. But technological advancement is not without dangers, and information often moves faster than it can be harnessed to impact public policy. John McGinnis will be dedicating his weekly posts to these issues, and the ways in which the government must keep pace with technological change. Read on for his introductory post:



Predicting the Effects of Policy at Election Time

John McGinnis


In an election season, politicians promise that their policies will deliver wonderful results. But their proposed policies often conflict.  Will more government spending or tax cuts generate economic growth?   Will charter schools improve educational achievement or is it better to spend more on existing schools?  To determine which policies to follow, democracy needs better information about their likely results.

My forthcoming book, Accelerating Democracy: Matching Governance to Technological Change, argues that we can make better policy by bringing democracy within the domain of our digital information revolution.  We are all aware that our lives are being transformed as the devices created by exponential increases in computational power connect us to information and to one another faster than ever before.  But we have not systematically considered how this technological revolution can improve democracy in our day, as the printing press did in its time.

New or improved information technologies, like empiricism, prediction markets, and dispersed media, can refine our evaluation of past policies and sharpen our predictions of future policy results.  But they can do so only if we change laws and political structures to permit the information revolution to wash through our politics.  In this post and those to come, I will use events in our current election to describe in more detail how we can improve democratic outcomes by embracing in contemporary politics our accelerating technological future.

I begin with the question of how we can better predict the consequences of an election.  Experts predict these consequences all the time.  For instance, the economist Joseph Stiglitz stated that Governor Romney’s election would “significantly” raise the risks of a recession.  If this statement were true, it would be a reason to vote for the incumbent.   But Professor Stiglitz is not only a Nobel winning economist, but also inhabits the left side of the political spectrum.  It is not impossible that he is biased against the presumptive Republican standard bearer.

Predictions markets could help us test Professor Stiglitz’s claim.   We could make markets in the likelihood of a recession conditional on victory by Governor Romney and conditional on  victory by President Obama.  One advantage of markets is that they offer powerful incentives against bias. Participants are rewarded for accurate assessments, not ideological frolics. They must put their money where their mouth is.   And, not surprisingly, we have evidence that prediction markets are more accurate at evaluating the future than alternatives.  They have proved better at predicting the winners of elections than polls.

Prediction markets help answer a fundamental question of democracy.  How can America know what Americans know? Prediction markets help pool individual assessments for the collective good. Sadly, United States law greatly impedes the operation of prediction markets because they are seen as internet gambling.  The most important such market, Intrade, is thus run out of Ireland.  By relaxing our legal prohibitions, we could help generate more innovate prediction markets on a wider range of subjects.  Indeed, so important is the information that they can provide, the government should consider subsidizing experiments with these markets rather than banning them.  Information is a great public good and government restricts its effective provision at our peril.

John O. McGinnis is the George C. Dix Professor of Constitutional Law at Northwestern University.


Gay Marriage by the Numbers–a guest post by Lynn Vavreck and Ryan Enos

YouGov is in the field each week on Saturday mornings with a nationwide, 1,000-person survey. Today, I’m going to look at the surveys before and after President Obama’s ABC interview during which he came out in support of same-sex marriage.

The interview made news for days — and was the topic of endless speculation about whether black voters would “follow obama” on gay marriage. Others wondered: Why was he doing it? Why now? And, of course, what effect would it have on the 2012 election?

The questions stirring in the journalistic community were intense and the conversation from coast to coast, among those interested in politics, relentless.

And yet – very little of interest has happened to public opinion as a consequence of Obama’s revelation. The week before his announcement, 49% of Americans in the YouGov poll supported gay marriage. The week after — 48%, a change far too small to take seriously. Essentially – opinion in general did not move at all after Obama’s interview.

Pooling all 2011 pre-announcement interviews and comparing them to the one post-announcement survey (so we have 1,427 blacks), support for gay marriage changed not at all – 41% before and 41% after. Because our data are a panel, we can compare people’s opinions in December of 2011 to their opinions during the week in which they were interviewed in 2012 (everyone gets two interviews). When we look at how many people are changing – moving off of their initial position to a different one – we find no change in the number of African Americans moving pre and post announcement. About 12-15% are switching their positions regardless of whether we interview them after Obama’s support of the policy or before. But – and here’s the most interesting tidbit – if you changed your opinion before the announcement, there was a 50-50 change you were moving in either direction. But of those who changed after his announcement, 85% moved toward the more supportive position. While there are very few cases to evaluate in the one week after the interview, the result is still statistically distinguishable from zero.

Now, let’s look at white opinion. Before the interview, 54% of the changers (people moving from their December position on gay marriage) were moving toward support. And after the interview, that share drops to 47% — too small to call it different in statistical terms, but an order of magnitude smaller than the change among blacks! To be clear, most people did not change their opinion after Obama’s support of gay marriage, but among those who did, blacks were far more likely to move toward Obama’s position than were whites.

And here’s one more way race is playing a role: Whites with low levels of racial prejudice were more likely to change their position on gay marriage after the announcement than those with high levels of racial prejudice. This result holds even if we control for other things that might drive a reaction to Obama’s announcement (and positions on gay marriage) like party identification and ideology.

Whites with low-levels of racial prejudice who changed their position after the interview were more likely than those with high-levels of prejudice to move toward supporting Obama’s position. The announcement itself had no effect on the positions of these switchers – it was the combination of the announcement and their level of racial prejudice that motivated the movement.

Lynn Vavreck is associate professor of political science and communication studies at UCLA. Ryan Ennos is assistant professor at Harvard University.

Check your References — Public Opinion Polls

As part of Election 101, we are posting exclusive content from The Concise Princeton Encyclopedia of American Political History on subjects related to Election 2012.

In this article Sarah E. Igo reveals the origins of the “straw poll” and other fascinating tidbits from the long history of trying to predict the winners in American politics. Here’s a short excerpt:

Starting in the nineteenth century, rudimentary political surveys, whether for entertainment or electoral gain, were undertaken by reporters, party loyalists, and ordinary citizens. In the 1820s, partisan newspapers began conducting straw polls as a means of both calculating and swaying political contests. “Straws,” named for the way a straw held up in the wind could determine which way it was blowing, were haphazard instruments for gauging opinion, with passengers on a train or people encountered during a phase of a political campaign polled as the entire sample. Regardless, these quantitative surveys were popular news features into the twentieth century, encouraging a “horse race” approach to reporting elections that continues to the present.

The current American political process is nearly unimaginable without public opinion polls. These surveys measure not only the relative standing of candidates for office but also citizens’ views on myriad social and political issues. Attempts to record political opinion are as old as the nation. But such polls, at least in their contemporary guise, date only from the mid- 1930s, when George Gallup, Elmo Roper, and Archibald Crossley championed the ability of scientific sampling methods to reveal the “pulse of democracy.” The modern public opinion poll— along with its ramifications for U.S. politics and civic life— was born of their success.


Read the complete article here:


The preceding is an excerpt from The Concise Princeton Encyclopedia of American Political History, edited by Michael Kazin, Rebecca Edwards, and Adam Rothman. To learn more about this book, please visit Copyright © 2011 by Princeton University Press. No part of this text may be distributed, posted, or reproduced in any form by digital or mechanical means without prior written permission of the publisher.

Forty-nine Percent of Gingrich Voters Choose Romney over Santorum, guest post from Lynn Vavreck

Back in January, many people wondered whether Mitt Romney’s support in the Republican primary (about 25%) was a floor or a ceiling.  The question was whether that quarter of the Republican electorate was made up of solid supporters who would be a foundation from which Romney’s support could grow or whether they were the only Republican primary voters who would vote for Romney.  Through our relationship with Model Politics, John Sides (George Washington University) and I have been asking 1,000 people a week to tell us about their political choices.  We now have 11 weeks of data about Romney, his challengers, and how voters choose among them.

These data provide a great snapshot of the GOP electorate at any point in time, but more importantly, YouGov, which fields the Model Politics surveys, did a large political profile of all the people who answer their polls in December of 2011, right before the nomination process officially started — and wait for it – everyone who has been interviewed in one of the weeks in 2012 was also interviewed in December of 2011.  This means that we have repeated data on each respondent in our weekly surveys, which means we can track changes over time.

What do we learn from these changes over time?  As you will see, there is very little evidence to suggest a strong “anyone but Romney” trend, but there does seem to be a small set of Republican primary voters holding out on switching to Romney.  More interestingly perhaps, this pattern is not unusual.  Simon Jackman (Stanford) and I found a similar phenomenon when we analyzed the Hillary Clinton-Barack Obama contest in 2008.

The first number worth reporting is from these data is 40%.  That is the percentage of likely Republican primary voters who have changed their minds at least once between December 2011 and when they were interviewed.  This has a bit to do with how many candidates have dropped out of the race.  With the exits of Michele Bachmann, Rick Perry, and Herman Cain, some voters are forced to change their minds and move to another candidate.  When we take away these “forced” movers and recalculate the rate of change, we find about 28% of Republican primary voters changed their minds at least once (among people whose initial preference in still in the race).  And if we isolate change further looking only among voters whose initial preference was either Newt Gingrich or Rick Santorum, we find about a third (32%) changed their minds at least once to date. The comparable number for initial Romney voters is 24%.

A good deal of the changes in preferences come from people switching among candidates still in the race and not only from supporters of those candidates who have dropped out.  And perhaps more interestingly for the “anybody but Romney” narrative is the fact that more initial Gingrich and Santorum voters switch than initial Romney supporters – and Romney is picking up some of those switchers.

You can examine the rate of switching for each candidate among all likely GOP voters in the first column of the table below; and their rates of gains from switchers in the second column.



It is easy to see that Santorum has been a draw for switchers.  Half of his current support is made up of voters who have switched to him after initially preferring someone else. The fact that Santorum’s initial base of support was so low in December is not irrelevant in calculating this number, but it should not take away from the important pattern: people are moving toward him.  His pattern of draw resembles what political scientists call momentum.

But Romney’s numbers are telling, too.  Of his current composition of supporters, a third of them have switched from other candidates.  Among voters who have changed their minds, 28% have ended up with Romney and 37% with Santorum (Gingrich gets 24% and Paul 11%).  But where have these voters come from?  Is Romney picking up anyone of set of people we would think of as “anybody but Romney” voters?

In the figure below I present the distribution of initial Bachmann, Perry, Huntsman, and Cain (and other) voters.  One thing is clear from the figure:  Romney picked up a good deal of support from Bachmann’s voters, almost a third of Perry’s supporters, nearly half of Huntsman’s voters, and a quarter of Cain voters.  Romney’s support is made up of some voters who preferred one of the more socially conservative candidates at one point in time.  Some of these voters will vote for Mitt Romney.

The question is, will any of the people voting for Gingrich now switch to Romney if Gingrich leaves the race? When we examine a head-to-head choice between Romney and Santorum for voters who currently prefer Gingrich and even split emerges.  Forty-nine percent of Gingrich voters choose Romney in a direct choice between Romney and Santorum.  Fifty-one percent choose Santorum.

There is a set of voters in this nominating contest who seem unwilling to vote for Romney, but there are plenty of voters who will vote for him – even if they have preferred one of the more socially conservative candidates previously.

It would be wrong to believe that all or even most of Gingrich’s support would flow to Santorum if Gingrich left the race.



Editor’s note — this is the second post in a series of guest posts by John Sides and Lynn Vavreck that assess ongoing polls through the Presidential Election. This article is cross-posted at Model Politics.

Earlier posts:

John Sides on his forthcoming book with Lynn Vavreck

Over at The Monkey Cage, John Sides does a terrific job explaining the unique publishing initiative he and co-author Lynn Vavreck are undertaking, along with their political sciences editor Chuck Myers.

“We are going to try something unusual with this book—at least by academic standards,” writes Sides. How unusual? Well, Sides and his colleagues are tackling unusual from both sides — both in the writing and publishing of the book. The book will be based on articles they are writing for various blogs about YouGov‘s election survey data (you can read some of their posts here at PUP). They are encouraging dialogue with readers via comments so they can refine their text.

Regarding the publishing process, they are also trying something completely new. At an academic press, projects are subject to lengthy peer review processes but Sides and his colleagues have come up with a novel solution. “Blind” readers normally read a complete manuscript to provide feedback, but in order to speed up the publishing schedule for this book, the editorial team “has recruited reviewers who will read and comment on chapters as they are written.”

The publishing strategy for this book is also quite different. We will publish short ebooks — parts of the whole — prior to the print publication. “Princeton will also be publishing the first 2 chapters in electronic form in August 2012. These chapters will deal with the (1) political and economic landscape as the campaign got under way, and (2) the GOP primary. Hopefully they will serve to generate some interest and discussion,” writes Sides.

This process is an experiment in publishing as well as an attempt to meet Princeton University Press’s mission “to disseminate scholarship (through print and digital media) both within academia and to society at large.”

While the team has thought through all the logistics of this process, they are at the mercy of the candidates and the news. As Sides notes, somewhat tongue in cheek, “if the GOP primary doesn’t end soon, we can’t promise anything.”

An Agent Based Model for Election Buffs

Michael Laver of New York University has written a truly geek-tastic computer program in NetLogo that invites you to experiment with the 2012 US presidential election campaign, using the agent-based model of party competition set out in his new book cowritten with Ernest Sergenti, Party Competition: An Agent Based Model.

  • Does increasing voter alienation keep the two candidates apart in the presidential election as they fear alienating their “base”?
  • Does increasing the number of independents in the primaries help Romney?
  • What happens if Ron Paul becomes a rampant vote seeker who doesn’t care at all about policy?

Experiment with the sliders to see how they affect the result. For those who want to get technical, roll up your sleeves and reprogram the model, or use NetLogo’s built-in “Behavior Space “ feature to have 1000 reruns of the presidential election under subtly different conditions. Fun stuff!

Check out the details after the jump.

Download the code for this model here: Presidential Election Model from Michael Laver (this is a zipped .nlogo file)



Presidential Election Model

Michael Laver



The election has two phases. First, there is a primary contest between the four Republican challengers still standing on 1 February 2012. Second, the winner of the Republican primary fights a general election against Democratic incumbent President Barack Obama.

CANDIDATES in each election compete with each other by setting out the policy positions they hope will appeal to as many voters as possible. Different VOTERS have preferences for different policy positions and are divided into three distinct groups: Republicans, Democrats, and Independents.

Voting in the primary is confined to Republicans plus a proportion of the Independents that can be set by the user. Obviously, all three groups of voters may vote in the general election. A successful Republican challenger must first set policy positions that win the primary, then must set policy positions that win the general election.

The winning candidate at the end of the general election is the one with the most popular votes – there is no electoral college to contend with.



Key policy positions in the election, both for voters and for candidates, are described using two dimensions of ideology.  The horizontal dimension describes economic policy positions on a left-right scale; the vertical dimension liberal vs conservative positions on “social” issues such as abortion and gay rights.

VOTERS in each group (Democrats, Republicans, Independent) tend to have distinctive distributions of opinion on these important matters. Each group of voters is characterized by: its size relative to the other two groups; the policy preferences if its average member; the diversity of opinion within the group.

These key population parameters all have default values that you see in the “Population Designer” panel on the interface when you fire up the program. You can change any of these using the sliders to design a voting population you find more realistic or interesting.  Indeed one thing you should definitely explore is the effect of different assumptions about, to take just one example, how left- or right-wing are the views of typical Republican voters on economic policy.

You can set the proportion of Independents voting in the Republican primary using the “indeps-in-primary” slider on the interface. This is something else to play with.

CANDIDATES, as we have seen, compete with each other by offering policies to potential supporters – so we expect their policy positions to change over the course of the election. The default starting policy position of each candidate can be seen on the interface, but you can change these using the sliders to any setting you find realistic and/or interesting.  This is something else you should definitely play with.

CANDIDATES use one of two decision rules to choose which policy position to offer voters. Some candidates never change policy position no matter how few votes this attracts; their main objective is to state their case. They are using what the model calls a Sticker rule. Other candidates continuously adapt their policy positions in search of more votes; their main objective is to win the election, whatever policy will help them do this. They are using what the model calls a Hunter rule. The decision rule used by each candidate also has a default setting, but you can also change this on the interface to play with the effect of different candidates different decision rules. (For example you might decide that the incumbent President is “stuck” with his record in office and cannot change position at all, in the eyes of voters, during an election campaign.

VOTER ALIENATION.  Many people do not vote. There are many reasons for this, but the model lumps all of these together under the heading “voter alienation”.  The electorate is divided into some who are so “alienated” they do not vote, and others who do indeed vote. Electors become alienated if there is no candidate with a policy position “close enough” to them in the election – where “close enough” is a parameter set using the “voter-alienation” slider on the interface. This has a default value, but you can slide it to any value you find interesting or realistic, and once more are encouraged to play with this. Alienated voters are colored brown on the interface; the shading of other areas reflects the density of voters with ideal policies in the area in question; white-hot areas have the most voters.



SETUP sets up candidates, supporters and system parameters as specified above.

GO starts the simulation using current parameters. It first runs a 100-day primary, waits a couple seconds for you to look at the result of the primary and get over the excitement, then kills off the losing primary candidates and runs a general election.



Experiment with the sliders to see how they affect the result as described above, or use NetLogo’s built-in “Behavior Space “ feature to have 1000 reruns of the presidential election under subtly different conditions.



Download the model here: Presidential Election Model from Michael Laver (this is a zipped .nlogo file)

As with all the models in the Laver-Sergenti book, this model is programmed in Netogo. To run it you will need to download and install the excellent NetLogo agent-based modeling environment. This is free for personal use from:

You can run the model with defaults simply by hitting “setup” and then “go” but the point is obviously to play with the effects of  the various parameters set by sliders in the interface.

There is a brief description of everything, mostly repeating what is set out above, on the “information” tab. The model’s code, should you want to play with this, is on the “procedures