31 March 2006
Unlike those slackers at the Social Science Statistics blog, we political behavior types foolishly keep motoring on through spring break at Harvard. This week's PPBW paper by Adam Berinsky of MIT is titled "Group Attachments and Public Support for War." Berinsky suggests that group identifications are important determinants of attitudes toward American foreign policy, particularly during WWII but also in selected cases after the war.
23 March 2006
These days, Italians don't get to vote quite as often as they used to. For the first time in the country's postwar history, we are being called to the polls on April 9 to vote on a government that has accomplished the extraordinary feat of serving a full legislative term. Critics may say it is the only accomplishment the Berlusconi government is legitimately entitled to claim. And, indeed, holding together - for a full five years - a ragtag coalition of know nothings, post-socialists, post-fascists, libertarians, and prudish altar boys - joined in the occasion of the upcoming elections by the real fascists led by Alessandra Mussolini in persona - was no cupcake. It may not have been an endeavor worthy of Winston Churchill, Napoleon, or Jesus Christ - to whom Berlusconi humbly compared himself over the course of a single week - but it's not a bad thing to have on your cv if you are looking for a job in management. Certainly, it remains to be seen whether the center-left will be able to claim so impressive a success, should it fail to once again snatch defeat out of the jaws of victory. As it turns out, the Ulivo is itself a motley crew of anti-globalization activists, squatters, anarchists, communists, post-communists, and post-Christian Democrats scotch-taped together by the strange cult of personality of a career bureaucrat - aptly nicknamed "Mortadella" after the cheapest of Italian cured meats - who was once in charge of the government holdings company during the lost days of the Italian mixed economy. Italians better hold a wet handkerchief to their mouths and noses when they walk into those polling stations. The stench will be unbearable.
An interesting by-product of the increasing africanization of Italian politics is the increased malleability of the rules of the game, which now seem to have lost so much of their proverbial stickiness that even the most brazen of manipulations goes almost unnoticed. Armed with opinion polls demonstrating the growing collective queasiness over the Berlusconi government's redecoration of the country's legal code and its inability to defibrillate an economy with no pulse, the "governing" coalition used a little muscle this winter and dug deep into the endless reserve of Italian creativity to swiftly push through an electoral reform. The majoritarian component of the Italian mixed system - once the subject of a hard-fought campaign that won the support of 85% of the voters in a 1993 referendum - was summarily cashiered, as the government triumphally announced the improbable return of proportional representation. No one even bothered to cook up an excuse to justify the return to a system once held responsible for each and every evil of the First Republic. "Technical problems," Berlusconi says.
But as the eternal Giovanni Sartori was quick to point out, the new "proportional" system is not all that proportional.
In the Chamber of Deputies, the new law makes sure that the "winning" coalition - even if it fails to win an actual majority - will take home at least 55% of the seats (340 out of 630) thanks to a majority premium. The loot is then shared by the coalition's parties that received more than 2% of the vote, this time in actual proportion to their vote shares. This makes governing easier, Berlusconi has claimed. And one can't really argue with that, nor fail to appreciate the candor exhibited by our savior the "Jesus Christ of politics" (his words, not mine). The rest of the seats are given to the other, sub-plurality coalitions with more than 10%, to unaffiliated parties with more than 4% nationally, and to parties representing officially recognized linguistic minorities that received more than 20% of the vote in their constituency.
In the Senate, the law is even more complicated. This time, the majority premium is assigned within each region to the coalition with the largest number of votes locally. The rest of the seats, in each region, goes to coalitions with more than 20% of the votes and to unaffiliated parties that received at least 8% of the vote. In this case, the "it makes governing easier" explanation doesn't make as much sense. The law simply makes red regions redder and blue regions bluer.
20 March 2006
Last Friday at PPBW, Richard Lau of Rutgers University presented a paper co-authored with David Redlawsk entitled "How Voters Decide: Four Strategies of Voter Decision Making, and Their Consequences," which is part of a larger book project on the topic. The authors present four models of voter decision making: 1.) the rational choice approach, where voters gather as much information as possible about all the candidates; 2.) the confirmatory approach, where voters rely on longstanding predispositions like party ID; 3.) a "fast and frugal" approach, where voters gather information about all candidates but on just a few relevant dimensions, and finally 4.) the "intuitive" decision making approach, where voters start their search with one candidate chosen at random and then keep searching until an acceptable alternative is found. Somewhat unexpectedly, the rational choice approach seemed to fare the worse in helping voters make "correct" decisions.
The authors designed experiments in which they could monitor how voters picked their candidates in mock presidential primary and general elections. Voting "correctly" was defined as when voters chose candidates that were aligned with their pre-experiment policy predispositions. Models 3 and 4 - which the authors classify as "boundedly rational" approaches - seem to fare the best in the four candidate primary experiment. Model 2 - in which voters rely on party ID - also did quite well in the general election. Model 1, however, did poorly, which seems to suggest that more information is not always better.
Questions were raised about whether these four models are mutually exclusive, whether voters might use different strategies at different times, and also how voting "correctly" can be defined outside of the policy predisposition context. Nevertheless, these results are sure to cause debate, particularly among those who view the rational choice model of voting more favorably.
Posted by Ian Yohai at 12:48 AM
19 March 2006
Political scientists often try to assess how well the government responds to public policy preferences. Friday's CAPS lecture by Martin Gilens of Princeton University pushes this research in a new direction - asking who the American government is most likely to respond to.
Gilens matches public opinion regarding proposed policy changes to whether these were actually enacted. As we might expect given the structure of American government, the status quo tends to win out and few potential changes that surveys ask about are implemented. Fundamental notions of equal representation, however, fare less well in the empirical analysis. As the abstract of Gilens' recent POQ article says,
When Americans with different income levels differ in their policy preferences, actual policy outcomes strongly reflect the preferences of the most affluent but bear virtually no relationship to the preferences of poor or middle-income Americans.
While understandably concerned with the normative implications of these findings, the audience responded positively. My major question about this research concerns the "black box" mechanism which translates the preferences of the rich into the policies the government enacts. Is it because they participate more in politics? (Bartels says not). Is it because politicians themselves tend to be drawn from the higher end of the income distribution, and so personally hold the same opinions? Or because the wealthy fund parties and candidates, allowing them more influence than someone who only has the power of voting? I have no answers, but would love to see more research in this area.
Posted by Phil Jones at 9:35 PM
16 March 2006
To help the scholarly community shape upcoming surveys, the (A)NES has launched the Online Commons. As the oveview puts it, "The goal of the Online Commons is to improve the quality and scientific value of each of our data collections, to encourage the submission of new ideas, and to make such experiences more beneficial to and enjoyable for individual investigators." Planning for the 2006 pilot study, 2007-2009 panel study, and 2008 general election study are underway. In the spirit of this and other blogs, the Online Commons experiment could turn out to be extremely useful in making those studies more effective.
Posted by Barry Burden at 9:52 AM
13 March 2006
This Friday there are two talks in the Government Department that are worth checking out. At Noon is a PPBW talk by Richard Lau of Rutgers titled "How Voters Decide: Four Strategies of Voter Decision Making and Their Consequences".
The Lau and Redlawsk paper makes the counterintuitive argument that rational voter decision-making is in fact the least rational strategy. Here's the abstract:
Most extant voting models are based on the untested assumption that voters engage in cognitively complex processes as they receive information about candidates during an election campaign. Whether it be by the many variants of rational choice theory or updated versions of The American Voter, people are generally assumed to be largely passive recipients of campaign information which they subsequently process by making complicated tradeoffs between good and bad attributes of the competing candidates or parties. No complete decision making framework has been proposed that addresses both the information search activities and the cognitive limitations that citizens face in trying to decide how to vote. This paper reports a significant step towards the development of just such a framework. Using a dynamic process tracing methodology, we examine the decision strategies used by voters to make sense of an election campaign. Those strategies, broadly categorized as rational, confirmatory, fast and frugal, and intuitive (which we label Models 1 through 4, respectively) are tested against a normative measure of decision quality â€“ correct voting. We find that voters consciously structure their information search in response to the campaign environment, and that counter to the promises of neoclassic economics, following a â€śrationalâ€? decision strategy is usually the least effective means of reaching a correct decision. Thus the widespread assumption that voters should employ some high-information rational procedure in making their vote decisions is called into question.
Posted by Barry Burden at 6:20 PM
In a recent post on Political Arithmetik, Charles Franklin provides additional evidence that the electorate learns about presidents as they govern. Specifically he shows that the percentage of respondents saying "don't know" to the presidential approval question declines over time within each administration. In a recent paper discussed here, Sunshine Hillygus and I argue that the regularity of this pattern provides support for the notion that "passive" learning (as we call it) decreases uncertainty about the incumbent. In contrst, "active" learning contributes to polarization around the incumbent, and both forms are associated with incumbent reelection.
Posted by Barry Burden at 10:56 AM
10 March 2006
Cara Wong presented her research, "Place & Politics: Understanding the Effects of Racial Context" at the Center for American Politics Workshop this afternoon. Her research addressed the following questions: 1) by what mechanism does the racial context (defined as percent black) affect political attitudes and 2) Do people perceive the racial context of their neighborhoods accurately? Wong uses data from the 2000 GSS and from an original pilot study to answer these questions. Her findings from both the GSS and the pilot study indicate tht people misperceive the racial composition of the country and of their local communities. Interestingly enough, however, these misperceptions have no effect on negative stereotypes of blacks in models that control for standard demographic characteristics such as age and education. Objective measures of racial context from the census do predict stereotypes of blacks, such that people in more diverse communities view blacks more negatively.
The audience responded positively to Wong's research. Most comments pointed to alternative data sources and surveys that Wong could use to measure the effects of racial context. Also, audience members pointed out the endogeneity of racial attitudes and place of residence (such that people with negative attitudes toward blacks might choose to live in areas with fewer blacks) and discussed ways that Wong might overcome this problem.
This afternoon Cara Wong of the University of Michigan will give a CAPS talk entitled "Place and Politics: Understanding the Effects of Racial Context."
Posted by Barry Burden at 9:55 AM
3 March 2006
Today, the Political Psychology and Behavior Workshop sponsored a talk by Jens Hainmueller (Harvard â€“ Department of Government) and Holger Lutz Kern (Cornell â€“ Government) on â€śParty Incumbency as a Source of Contamination in Mixed Electoral Systems.â€? The paper examines propositions derived from the vast literature on incumbency advantage and the more recent research on â€ścontaminationâ€? â€“ which designates the interaction between the majoritarian and proportional components of mixed electoral systems â€“ employing data from the elections of the German Bundestag. Hainmueller and Lutz Kern address the methodological problems faced by the â€ścontaminationâ€? literature with an innovative Regression Discontinuity design. They show that - in marginal districts - party incumbency boosts a partyâ€™s nominal and list votes by about 1.5 percentage point. Simulations estimate that party incumbency effects lead to net shifts in Bundestag elections that range from 3 to 9 legislative seats.
Posted by Federico Ferrara at 10:34 AM
At this weekâ€™s American Politics Research Workshop, Prof. Barry Burden presented a paper co-authored with Prof. D. Sunshine Hillygusâ€™, â€śThe Devil You Know: Voter Learning, Polarization, and the Reelection of George W Bushâ€?. The authors argue that voter learning about candidates occurs throughout a presidentâ€™s term in office, and not just during the first term in office. As a result, low information voters become less uncertain about the incumbent president, which results in voters considering the incumbent a less â€śriskyâ€? choice. On the other hand, knowledgeable voters respond to this increased knowledge differently. High-information voters actually become more polarized in their evaluations of the president relative to his opponent. Paradoxically, this increased polarization appears to lead to be an important component to incumbent presidential victories.
Burden and Hillygus show that their basic pattern is consistent over the course of six modern presidencies, but focus particularly on the presidency of George W Bush. They argue that the events of September 11 focused the publicâ€™s attention on the president, and this increased learning about the president and his traits may have played a strong role in President Bushâ€™s reelection. This learning manifested itself in two ways. First a time series composed of Gallup polls showed that the proportion of the population expressing â€śdonâ€™t knowâ€? responses to approval questions had a huge drop after the terror attacks. In addition, and contrary to his pledge that he would serve as a uniter and not a divider, President Bush also polarized the American public more than any other president in their survey. But according to Burden and Hillygus, Bush is not culpable for this polarization. Instead it appears to be a pattern the public experiences time and again with different presidents.
Beyond their primary theoretical argument, Burden and Hillygus address a large gap in the behavior literature. How voters learn about presidents and other politicians is an important issue. Yet, little in the literature addresses learning processes, and when learning is discussed it is rarely tied to the outcome of real political events. In short, this research agenda offers the promise of casting light on mass behavior in a new different way, and directly ties these findings to political outcomes.
Posted by Justin Grimmer at 8:57 AM