Search

Do campaigns matter? Do negative TV advertisements affect vote outcomes in elections? The scholarly conventional wisdom suggests that negative advertisements have a limited impact on election outcomes. In contrast, based in part on analyzing the incentives of candidates and campaign consultants, we theorize that negative campaign advertisements affect vote outcomes in elections. Theoretically, negative advertisements should matter, yet nearly all empirical work showing that they do not has been based data where causality has been difficult to assess. In concert with a statewide political campaign in Pennsylvania, we conducted a field quasi-experiment of TV campaign advertising. We find that negative campaign advertising leads to about a three percentage-point vote share increase for the ad-sponsoring candidate in treated geographic areas when compared to those geographic areas where no ad was run. [ABSTRACT FROM AUTHOR]

In a recent APSR article, Hajnal (2009) explores the question of who loses in American democracy, and finds that African Americans are more likely than whites to vote for a losing candidate. This finding is extremely troublesome, as it suggests that African Americans' voices are systematically being shut out of the political process. We revisit this question and show that Hajnal's analyses are too broad in their scope, and that they do not account for the sampling procedures unique to each national exit poll. When we disaggregate his results by year, weight the data according to each exit poll's probability sampling weight, cluster standard errors to take account of the first-stage precinct-wide sampling frame, and examine a longer time period, we arrive at different conclusions. We find a surprising number of elections where minority voters are more likely than whites to vote for a winning candidate. However, we also find many elections where minority voters lose, and we find that Hajnal overestimates the extent to which Latinos and Asian Americans win in some U.S. elections. The picture of "who loses" in American democracy may not be as bleak as Hajnal suggests for African Americans, but is somewhat less positive than he suggests for Latinos and Asian Americans. [ABSTRACT FROM AUTHOR]

We offer a theory of presidential campaign persuasion as a valence dimension. From 1976-2004, we find that presidential oratorical skills cause moderate voters to vote contrary to issue preferences, while more extreme voters vote based on issues. ..PAT.-Unpublished Manuscript [ABSTRACT FROM AUTHOR]

Do non-policy valence advantages that incumbent members of Congress possess affect spatial position-taking? Are legislators who deliver substantial amounts of federal largesse more likely to diverge from their constituency medians? Are legislators who are perceived as more competent than their opponents more likely to diverge from the preferences of their constituents? I argue and empirically demonstrate that valence advantages such as the distribution of "pork" projects and legislator competence allow incumbent legislators to deviate from the policy preferences of their constituents (in some instances). Formal valence theories of position-taking are examined in this paper, and I argue for an expansive definition of valence advantages that includes both valence policies and non-policy valence characteristics. I show that valence advantages sometimes cause legislators to converge to their constituents' preferences and to sometimes diverge. I test the expectations of valence theories of congressional position-taking by examining an example of a valence issue (distributive policy) and a valence characteristic (an incumbent's perceived competence relative to his or her challenger). These empirical tests are conducted with original data on senators' divergence from their states' median voters during the 104th-107th Congresses (1995-2002). One key contribution of this paper is the creation of ideal point estimates of legislators and constituency medians on a common scale using Bayesian MCMC ideal point estimation techniques (similar to the popular NOMINATE scores, though unlike NOMINATE, these scores include measures of constituents and senators). The findings are that valence theories of position-taking are demonstrated when examining incumbent divergence from the constituency median. In sum, senators with no valence advantage diverge from their constituents; senators with small valence advantages move closer to their constituents; and senators with large valence advantages are able to deviate far off of their constituents' preferences. The implications of these results are that senators who deliver very large amounts of federal outlays to a state or senators perceived as very competent relative to their campaign challengers are able to vote closer to their own personal preferences than to their constituents' preferences. ..PAT.-Unpublished Manuscript [ABSTRACT FROM AUTHOR]

This paper answers the following question: Under what conditions are states more likely to repeal disfranchisement provisions pertaining to the voting rights of persons convicted of a felony? We also address two broader questions: (1) Do parties seek electoral gain by changing the electoral system and voting laws? And (2) do parties override the ideological preferences of constituents in order to advance their electoral agenda? The answer to both questions is ‘yes’. We find that disfranchisement provisions are more likely to be repealed under a unified Democratic state government, but that states where electoral support for Democrats is generally weak are also more likely to repeal such provisions. We find no evidence that citizen ideology effects changes in voting rights. The implications of our analysis for students of parties and voting rights are clear: Rules are changed by those who stand to benefit, and parties matter at the state level. [ABSTRACT FROM AUTHOR]

Do party preferences or constituency preferences affect legislator positions in support of the president? Spatial theories have long implied that legislators will locate themselves at the median of their district. However, most empirical evidence related to roll-call support of the president’s agenda in Congress indicates the opposite, showing that party and ideology often cause a legislator to deviate from the district median (and that constituency preferences thus have a negligible or muted effect on presidential support). We answer our research question and explain this paradox by showing that legislators’ platforms are different from legislators’ votes on roll-calls. Spatial election theories are based upon the platforms that legislators take when running for office. The platforms communicated to constituents may be related to roll-call voting records, but clearly are not one and the same. Our findings show that constituency preferences are the only significant predictor of legislative support for the president when analyzing legislators’ platforms, and that party preferences have no effect. However, when analyzing roll-call votes, our analysis supports the large body of research on executive-legislative relations, suggesting that party and ideology are the key predictors of presidential support. We analyze legislators’ platforms and the messages they send to their constituents with newsletters and other mass mailings from the 107th Congress. The data were collected from the Legislative Resource Center and coded by the authors. [ABSTRACT FROM AUTHOR]

Some of the scholarly conventional wisdom regarding voter turnout and participation, especially in the U.S. South, has been that race and ethnic background can be important predictors of turnout. However, the disproportional effect of felon and ex-felon disfranchisement laws on racial/ethnic minorities in some states suggests that the effects of race on aggregate voter turnout will be underestimated unless these laws are also considered. This electoral institution-the disfranchisement of felons and ex-felons-is an important omitted variable in models of state-level aggregate voter turnout, which might result in biased inferences pertaining to minority citizens' level of participation, especially in recent elections. Looking at statewide elections in the U.S. South from 1984-2000, we hypothesize that this electoral institution helps explain lower levels of voter turnout in states with high rates of felon disfranchisement. We also demonstrate, once felon disfranchisement is controlled for, that southern states with high minority populations will have higher levels of turnout in recent elections where minority voter mobilization has been on the rise. [ABSTRACT FROM AUTHOR]

Can a candidate's non-policy valence advantages over his or her opponent cause voters to vote contrary to their issue preferences? We theorize that both issues and valence affect voting behavior, and, more importantly, their effects depend on each other. We use insights from the valence models of voting to explore the nature of this relationship: we argue that the effect of issues and traits are contingent upon both (1) how large the valence or non-policy trait advantage of one candidate over the other is; and (2) how far away the voter is from one candidate relative to the second candidate. We test our hypotheses with a unique data set from the 2006 Cooperative Congressional Election Study (CCES). We employ an experimental design in which some respondents view trait advertisements for their two U.S. Senate candidates, while others do not. Our results suggest that the effects of valence traits and issues are indeed contingent on each other. Voters extremely close to one candidate, though distant from the other, vote entirely based on issue preferences. Voters only slightly closer to one candidate than the other, so-called "Stokes voters," vote, instead, on traits. One of the implications of our results is that voters have distinct preferences for candidates with better traits and for candidates with congruent issue preferences; and not that traits are used simply as low-information information shortcuts. Our theory and findings move beyond the strict assumption of pure rational choice-based theories of voting behavior based on issues; it also moves beyond behavioral vote models that overemphasize the weight of traits by failing to realize that traits matter in relation to issue congruence. ..PAT.-Unpublished Manuscript [ABSTRACT FROM AUTHOR]

Are legislators who present inconsistent voting records truly mavericks or are they simply responding to uncertainty they have over the preferences of their own constituencies? We investigate the impact of these uncertainty-causing and uncertainty-reducing factors on the (in)consistency of senators’ voting records for the period 1915-2004. We theorize and show that uncertainty about constituents’ preferences explains inconsistent voting behavior by legislators nationally. We also show that for southern senators, the interplay between party and the influx of black voters presents a particularly interesting case of legislator response to a changing environment. ..PAT.-Conference Proceeding [ABSTRACT FROM AUTHOR]

We examine candidate emergence patterns in the U.S. House in the American South from 1992-2000. This period of elections is distinguished by the rapid ascent of Republicans and the concomitant decline of Democrats. There are many explanations for why the GOP was so successuful in the South for this electoral period, but we look at the role of political geography, an understudied component of candidate emergence. An uncharacteristically large percentage of newely elected Republicans had no previous officeholding experience and we find that this is tied to the fact that so many Republicans state legislators represnted districs that mostly overlapped with congressional districts already represented by the GOP. This geographic bottleneck deterred many politically experienced Republicans from running for Congress. Additionally, we demonstrate that if the overlap of a state legislative distric is substantial with a congressional district held by a representative of the opposing party, then the likelihood of emerging for Congress is much higher for both Democrats and Republicans. [ABSTRACT FROM AUTHOR]