Political Parties, American Campaigns, and Effects on Outcomes

Brian J Brox & Daron R Shaw. Handbook of Party Politics. SAGE Publications Ltd. 2006.

Introduction

For almost 150 years, two facts about American elections seemed incontrovertible. First, campaigns were an integral part of elections. It is the campaign that conveys information to voters who, in turn, use it to reach individual and collective judgments about the relative merits of candidates. The apparent rationality of election outcomes in the United States—candidates presiding over failing economies or unpopular wars lose, while candidates presiding over economic growth and popular wars win—strongly suggested that political information was reaching the public. Furthermore, it validated the perception that campaigns, as the most obvious conduit of this information, were important institutions.

Second, the role of parties in democratic processes was no less critical. The parties developed the capacity to contact individual voters, to advertise through partisan newspapers and pamphlets, to publicize and carry out events such as picnics, carnivals, parades, and rallies, and to print and distribute ballots. These capacities cannot be overemphasized. Candidates were recruited and controlled by parties. It is true, of course, that certain popular individuals had a greater say in how the party handled their candidacy and campaign. But it is equally true that parties dominated the relationship.

The perceived importance of campaigns and parties was largely unchallenged by practitioners, pundits, and scholars well into the 20th century. Three developments, however, called the conventional wisdom into question. First, as scholars collected data from surveys in the 1940s and 1950s, they began to realize that the American public was not nearly as informed about or interested in politics as they had assumed. This finding shook the broader assumption of voters as attentive observers of the day-to-day events and policy pronouncements of the election campaign.

Second, the development of broadcast technologies—especially the emergence and proliferation of television—fundamentally changed the way in which information is disseminated. In particular, by the 1960s television had allowed individuals to communicate and to develop personal connections with an audience without the human resources necessitated by face-to-face contact. This, obviously, had the potential to empower candidates at the expense of political parties.

Third, in the late 1960s and early 1970s, the American parties reformed their internal nomination processes to increase democratic input. The direct result was an almost complete reliance on primary elections to determine candidates. The indirect result was a forfeiture of party control over nominating processes. In the words of Alan Ehrenhalt (1991), candidates were asked to ‘nominate themselves.’ Furthermore, as candidates began to contest primary elections, they developed campaign organizations and expertise independent of the political party. These candidates not only were not beholden to the party when they won the nomination, but also often had personal campaign organizations and did not need help from the party as they turned their attention to the general election contest.

Together, these three developments led scholars to suggest that by the late 1960s we had entered a period of ‘candidate-centered politics’ in the USA (Wattenberg, 1991). The central idea is that candidates drove electoral and electioneering processes during this era. A corollary idea is that as candidates have been ascendant, parties have scrambled to remain relevant. Indeed, much of the recent literature on the American parties has emphasized the attempts of parties to recraft their functional and theoretical roles in light of more ‘personalized’ and candidate-driven politics.

At the same time, these developments have also prompted a number of interesting studies analyzing the effects of these candidate campaigns on voters and elections. More specifically, the persistent finding that voters do not know very much about politics and do not pay much attention to politics (or campaigns) has led scholars to look at the impact of other factors on elections. Indeed, the success of voting models that rely on factors such as incumbency, presidential approval, and economic performance has produced a sizable group of political scientists who view campaigns skeptically.

In this chapter, we review the classic literature on campaigns, elections, and voting behavior. This review, however, consciously attempts to recognize subtle and complex arguments on campaign effects. We will then discuss the sources of renewed interest in (1) campaign effects and (2) the role of parties in campaigns, before moving on to a delineation of the most recent findings produced by this renaissance. We close the chapter by discussing the prospects for continued party involvement with election campaigns. Our focus throughout is on presidential elections, although we comment from time to time on US congressional races.

What Classic Studies Tell us about Campaign Effects

A functioning democracy presumes voters have enough information to reward successful office-holders or to punish unsuccessful ones (e.g., Key 1966; Fiorina, 1981). The initial empirical work of political scientists casts doubt on even this low-level rationality, and the empirical findings underpinning this doubt became the focus of subsequent scholarship. The finding that voters may not have the requisite information to hold public officials accountable for performance in office creates a profound disconnect: if people are so ignorant, why are presidents who preside over economic recessions or unpopular wars or political scandals thrown out of office? Why does the system appear to function rationally in the aggregate if there is, in fact, no individual-level rationality?

Voting, written by Bernard Berelson, Paul Lazarsfeld, and William McPhee, was published in 1954. It employed a panel study to examine the political opinions, attitudes, and candidate preferences of residents of Elmira, New York, during the 1948 presidential election campaign. The broad argument—that voters tend to get their preferences from contact with ‘opinion leaders’ within their social groups—is familiar to any college student who has taken a course on public opinion and voting behavior. What is less well known is that the authors explicitly acknowledge the fact that political campaigns can have an effect on both individual voters and aggregate outcomes.

For example, the authors estimate that 16% of their sample ‘wavered’ between the parties during the campaign, while an additional 13% ‘wavered’ between a party and neutrality. The shifts were particularly evident amongst Elmira’s small Democratic population, with 36% of these voters wavering between the parties and another 14% wavering between the Democrats and neutrality (Berelson et al., 1954: 16-18). More to the point, the whole of Chapter 12 of Voting analyzes the trend towards Truman that took place late in the 1948 campaign, arguing that the Democratic rally was due to previously disaffected Democrats (and Democratic-leaning groups) responding to the class issues emphasized by Truman’s ‘Fair Deal’ campaign. A decade later, survey researchers in Ann Arbor also acknowledged the potential for campaign effects. The reliance of The American Voter on party identification as an explanation for vote choice has led many to conclude that the Michigan scholarship did not consider presidential campaigns as significant. But this is to ignore the actual argument of the text. In Chapter 19 of The American Voter, the authors pointedly contend that party identification is one of several factors that determine vote choice. The specific argument is that attitudes towards the candidates, domestic issues, foreign policy issues, parties as managers of government, and group-related attitudes drive votes, with party therefore serving as a critical but non-omnipotent conditioning variable (Campbell et al., 1960: 531). This position should come as no surprise given that the elections serving as the backdrop for this analysis saw the minority party candidate wallop the majority party candidate. The potential significance of campaigns is even apparent in the funnel of causality, in which party identification screens the acquisition and acceptance of political information. In their schematic, Campbell et al. place factors other than party identification, including issues and candidate perceptions, closer to the bottom of the funnel, indicating that political context is a critical variable for understanding voting.

In ‘The nature of belief systems in mass publics,’ Philip Converse (1964) describes the American public as largely uninformed and unengaged, and uses this as the basis for arguing that persuasive information—a category into which campaign messages most certainly fall—faces significant partisan resistance (at the level of the ideologue) or falls on deaf ears (at most other levels of sophistication). Zaller’s The Nature and Origins of Mass Opinion (1992) demonstrates, however, that a sophisticated reading of Converse does not necessarily lead to a minimal effects perspective. Zaller, in fact, uses Senate election data to suggest that voters with ‘middle level’ awareness may be quite susceptible to information flows. So while Zaller himself is agnostic as to whether there are significant persuasive campaign effects in presidential elections (at least in this study), it is certainly a possibility given his understanding of Converse’s theoretical construct.

What Contemporary Studies Tell us about Campaign Effects

Despite scholarly fascination with non-campaign factors during the past sixty years, there have been significant studies of presidential candidate activities (e.g., Kelley 1983) and media influence (Iyengar, 1991; Iyengar and Kinder, 1987). But because the Columbia and Michigan schools estimate that only 10-15% of voters are persuadable; with net effects thus constrained to only a few points, political scientists have looked for subtle, less direct campaign effects, as well as for other causal explanations for variance in voting behavior.

Of course, alternative explanations were readily identified. The activation of party identification was developed as a dynamic explanation for aggregate- and individual-level movement over the course of a presidential campaign. Gelman and King (1993) observe that shifts in the fortunes of candidates over the campaign largely involve uneven sequences of partisan activation. Persuasion is confined to independents and some weaker identifiers, and tends to be driven by conditional and objective circumstances, such as the state of the economy and presidential job approval. Party identification thus determines the base vote a candidate can expect, with genuine (but limited) potential existing for significant improvement (Iyengar and Petrocik, 2000).

Aside from party identification, the role of economic variables in shaping candidate preferences has been a consistent theme in the voting literature. In the 1970s, political economists began modeling presidential elections as a function of macroeconomic factors such as economic growth and unemployment rates (see Fair, 1978; Tufte, 1978). During the 2000 election there were at least seven distinct presidential forecasting models.What is interesting is that the forecasting models do not universally posit that campaigns do not affect presidential voting behavior. Most of them, for example, offer presidential job approval as a predictor of the vote, and approval rates could clearly be affected by the campaign. Moreover, a few models rely on past vote totals to predict the upcoming race, leaving open the possibility that past campaigns might affect current elections. Even forecast models with no endogenous (or lagged endogenous) variables frequently admit that campaigns are necessary to educate voters about the external reality upon which their predictions are based. Furthermore, some modelers have even suggested that campaigns are not equally skilled at accomplishing this.

Besides party identification and economic variables, political scientists have continued to develop the sociological framework established by the Columbia school. Specifically, political communication scholarship has explored the interpersonal networks through which people acquire their political information. The key findings from this literature are that (1) opinion leaders exist and are critical to informing the less aware members of a group, (2) communication differences between and among groups appear to be a function of the distribution of political awareness throughout a particular group, and (3) interpersonal communication remains vital, even as television has come to dominate the broader dissemination of information (Huckfeldt and Sprague, 1995; Johnson and Huckfeldt, 2001; see also Putnam, 2000).

The upshot of these studies is that understanding elections and voting does not require an understanding of campaigns. Though not irrelevant, presidential campaigns are epiphe-nomenal. The minimal effects perspective is therefore not a direct attack on campaigns. Rather, it is an inferred perspective; an attributed position based on its emphasis of non-campaign factors in studies of voting and elections. There have been almost no serious scholarly analyses suggesting that campaigning does not influence voters in congressional or local elections. In fact, the ability to raise and spend funds is a large part of the explanation for incumbency advantages in the US House and Senate (Jacobson, 1983; Mayhew, 1974). The minimal effects inference is confined to research on presidential voting. This, of course, makes sense. There are particular circumstances surrounding the presidential election that make it especially unlikely that a campaign will be decisive.

More specifically, the minimal effects scholarship does not contend that no one is persuaded by the presidential campaign, but rather that the net effect is typically incidental to the election outcome. The broader theoretical point of the minimal effects perspective should not be misconstrued, however. Most scholars writing from this point of view seem to believe that campaigns are important. First, presidential campaigns serve as exemplars for citizen responsibility and control over political power. Second, and perhaps more pragmatically, they mobilize support for the two major party candidates. Indeed, this mobilization process might not occur without prompting by the parties. In addition, the way in which campaigns mobilize voters (the particular appeals, the commitments made, the understanding of their own coalition) could be a critical factor for understanding subsequent governance and public policy decisions. Still, scholars who emphasize non-campaign factors typically argue that differential mobilization effects between the parties are unlikely, and this severely limits the chance that campaigns will determine who wins the presidency.

Despite its reasonableness and scholarly foundations, political pundits and casual observers of politics—both of whom tend to see presidential campaigns as decisive—show disdain for this view. Perhaps more interestingly, political communication scholars are somewhat perplexed by this because a slightly different minimal effects debate has already been resolved in their field.

As with the initial empirical studies of election campaigns, early analyses of news media had a difficult time finding effects. In their watershed article on agenda setting, McCombs and Shaw (1972) point out that voluminous research up to that time revealed precious little correlation between the tone and content of reporting on a given subject and the attendant nature of public opinion. In fact, research up until the early 1970s showed that citizens were quite capable of reading newspapers and watching television without much effect on their opinions and attitudes.

The suspicion that news media effects exist persisted, however, and led political communication scholars to posit and investigate more subtle influences. McCombs and Shaw (1972) presented persuasive evidence that the media’s influence is not in telling people what to think, but rather what to think about. The idea that media effects occur primarily through ‘agenda setting’ turned the minimal effects perspective upside down and paved the way for other, more subtle understandings of impact. Iyengar and Kinder (1987) used extensive empirical evidence of public opinion and news media coverage surrounding the Iran-Contra affair to contend that the media ‘prime’ citizens to use certain criteria when evaluating a particular figure or issue.Iyengar (1991) also explored the possibility that the ‘frame’ used by the news media to present a given story can create politically significant connections in voters’ minds. This research tends to be dominated by experiments, which allow greater control over (and isolation of) stimuli and effects. It has unquestionably transformed the nature of the debate on media effects and leaves many political communication scholars wondering what all the fuss is about when it comes to campaign effects. Surely the debates among campaign scholars could be resolved by a more sophisticated conceptualization of effects and greater flexibility and subtlety in research design.

New Perspectives on Campaign Effects

Given the array of studies questioning the significance of campaigns in US elections, is there any reason to cling to a more traditional perspective? We argue that there is. Recent research provides evidence that campaigns may, in fact, be more influential than heretofore believed. In particular, we point to four areas in which clear gains in our understanding of campaigns have been made: (1) estimating the net effects of campaigns, (2) measuring information effects from campaigns, (3) gauging the effects of specific campaign activities, and (4) identifying how candidates and campaigns approach the campaign. This section considers each of these in turn.

Estimations of the Net Effects of Campaigns

Over the past fifteen years, there have been several estimates of the overall magnitude of presidential campaign effects. Moreover, these have tended to be fairly conservative. Steven Finkel (1993) uses the Major Panel Survey of 1980 to analyze individual-level movement in presidential preferences. He finds that while many respondents change their preferences, the net movement is 2-3 points at most. At the aggregate level, Gelman and King (1993) demonstrate that there is significant volatility in pre-election survey estimations of presidential preferences, but that net campaign effects are almost zero because the vote tends to converge on a predictable point on or around election dayErikson and Wlezien (2001) use time series estimation techniques to calculate an aggregate preferences shift of about 5 points in recent presidential campaigns. Unlike Gelman and King, however, they attribute the considerable preference volatility over the election cycle to campaign factors.

In addition to studies of preference shifts, some have suggested that general campaign effects can be understood as the residual variance from multivariate models of the presidential vote, presuming that those models contain only exogenous variables. Bartels (1993) uses this logic to estimate that presidential campaign effects are typically on the order of 2-3 points. The forecasting models discussed earlier can be viewed in this light, with the mean error estimates—which generally run between 1 and 4 points—serving as estimates of campaign effects.

Campaigns as Information Sources

Despite the continued prominence of articles and books on the presidential campaign’s influence on votes, some scholars have argued that the focus on vote choice is an overly narrow way to consider campaign effects. In particular, a number of analyses focusing on how campaigns affect voters’ information have been produced since 1990. In addition to the information processing models proposed by Zaller (1990) and Lodge et al. (1995), several scholars trace the path of campaign information. Building on studies of Alvarez (1997), Lupia and McCubbins (1998), and Popkin (1991), William Bianco (1998) finds that voters in Senate elections can fulfill the expectations of both rational choice scholars and political psychologists by using information readily provided in the early stages of political campaigns. Kahn and Kenney (1997: 1173) go one step further; after examining the impact of intensity in 97 Senate races between 1988 and 1992, they contend that:

Intense campaigns encourage individuals to rely more heavily on both sophisticated criteria and simple decision rules when forming impressions of candidates. As campaigns become more hard-fought, people are more likely to consider policy and ideology as well as partisanship and retrospective evaluations of the president and the economy. While the campaign setting clearly affects citizens’ decision-making processes, different types of people react differently to the intensity of the campaign. As races become more competitive, novices begin to rely more heavily on issues, sociotropic assessments, party identification, and presidential approval, whereas political experts are less affected by changes in the campaign environment.

Based on these studies of how campaigns affect the information levels of voters, political scientists have recently taken to estimating the ‘informed preferences’ of voters to determine if a fully informed electorate would elect the same candidates as the actual electorate. In his study of information effects in presidential elections, Larry Bartels (1996: 194) contends that:

At the individual level, the average deviation of actual vote probabilities from hypothetical ‘fully informed’ vote probabilities is about ten percentage points. In the electorate as a whole, these deviations are significantly diluted by aggregation, but by no means eliminated: incumbent presidents did almost five percentage points better, and Democratic candidates did almost two percentage points better, than they would have if voters had in fact been ‘fully informed.’

Scott Althaus (2001) expands Bartels’ analysis by including non-voters in his study of how full information affects congressional vote preferences. Like Bartels, he finds differences between informed and uninformed voters, although Althaus does not find the same systematic party differences at the congressional level that Bartels finds at the presidential.

In addition to these innovative designs, there have also been a few experimental studies investigating the kinds of information that voters want to access about candidates and how that information affects the vote decision. Richard Lau and David Redlawsk (1997) conducted a series of computer-based experiments investigating these questions during the mid-1990s. They found that voters favor biographical information over hard issue information, and that information containing an affective component tends to be more influential than issue-based information. This corroborates other recent analyses arguing that emotion plays a considerable role in the presidential voting decision (e.g., Marcus and MacKuen, 1993).

Specific Campaign Effects

While analyses of political information and campaigns have helped us understand what presidential campaigns do, analyses of specific types of campaign activity have sharpened our understanding of how (and how much) campaigns influence voters. This trend toward disaggregating the specific manifestations of presidential campaigning has been matched by a tendency toward more innovative data and research designs. Consider the following aspects of electioneering that have received substantive empirical treatment over the past ten years.

Phones and Direct Mail

The most notable works in this area have been the ‘field experiments’ conducted by Alan Gerber and Donald Green (2000, 2001). During the 1998 elections in Oregon, Gerber and Green randomly selected voters from statewide voter lists, assigning them to control and treatment groups. The treatment groups received either (1) campaign mail from a candidate but no phone calls, (2) campaign phone calls but no direct mail, or (3) direct mail and phone calls. The control group received no campaign contacts. The authors took pains to ensure that their mail and phone calls were as realistic as possible, using genuine campaign consultants to design the materials. Controlling for a host of factors, Gerber and Green found that direct mail increased the candidate’s vote share 10% beyond what would otherwise be expected, but that phone calls actually had a negative impact on aggregate vote share. They also tested the effects of face-to-face contacting, which they found had a highly significant and positive impact on vote share.

Mobilization

Rosenstone and Hansen (1993) offer one of the most ambitious claims of all the recent campaign analyses when they contend that the decline in party mobilization efforts is a significant cause of the decline in aggregate turnout in the USA. This result is corroborated by Brady et al. (1995), who argue that party and candidate mobilization efforts can substantially reduce the costs of voting and make it easier for people with limited social capital to overcome the impediments to voting. More recently, Endersby and Petrocik (2001) argue that mobilization is perhaps the critical component to contemporary presidential election campaigns. They use National Election Study and exit polling data to build a compelling empirical case that while persuasion is minimal in presidential elections, the mobilization efforts of parties and candidates are critical to activating partisan predispositions.

Television Advertising

This is where the renewed interest in campaign effects has been most evident. One of the first of the ‘modern’ works was Darrell West’s study of the nature and effects of television advertising in federal elections. West (1983) essentially upheld the conventional wisdom that TV advertisements elicit minimal effects, but he also observed that campaigns do not expect these advertisements to persuade a large proportion of voters. Narrow, targeted effects are what campaigns seek and, West admits, we have little relevant evidence on their effectiveness. But it was Ansolabehere and Iyengar (1995) who revolutionized the study of campaign and political advertising with their experiments on TV advertisement effects in California during the 1990 and 1992 elections. They directly confronted the conventional wisdom of minimal effects by demonstrating that campaign advertisements significantly correlate with changes in candidate appraisals as well as the likelihood of turning out to vote. In particular, they argue that negative advertising mobilizes partisans but depresses turnout among independents.

The Ansolabehere and Iyengar experiments have prompted a slew of challenges. For example, Finkel and Geer (1998) take issue with Ansolabehere and Iyengar on the question of campaign tone and turnout. Using aggregate turnout rates and evaluations of campaign tone, they contend that negative campaigns tend to be coincident with relatively higher turnout. Wattenberg and Brians (1999) examine individual-level survey data and ultimately side with Finkel and Geer’s claim that negative advertisements increase turnout. Interestingly, although there is debate concerning the effects of negativity, all of these studies find effects.

Candidate Appearances

Several studies have updated the influential work of Stanley Kelley (1983) on the effects of candidates’ visits on local preferences. For example, Bartels (1985) estimated the pattern and impact of Jimmy Carter’s travel in the 1976 election. He argues that effects are not substantial, but that this is understandable because appearances are motivated by multiple factors, some of which are unconcerned with improving the candidate’s trial ballot standing. Shaw (1999a, 1999b) has examined both the pattern and effect of presidential candidate appearances from 1988 to 1996. He argues that three extra visits to a state are worth approximately one point in the polls. Two current projects, one by Thomas Holbrook and the other by Scott Althaus, Peter Nardulli, and Daron Shaw, are recreating candidate travel from presidential elections going back to 1948. The availability of more reliable data on candidate schedules and public opinion from the libraries of presidential candidates may allow us to calculate precise estimates of appearance effects.

In 2000, the scope of inquiry expanded to include appearance effects in primary elections. Using data from the New Hampshire primaries, Vavreck et al. (2002) demonstrate that personal contact with the candidate can do more than mobilize; it can actually persuade people to support a candidate. Voters who had met a particular candidate were significantly more likely to support the candidate. The authors argue that the effect holds even controlling for the fact that one is more likely to meet a candidate for whom one is predisposed to vote.

Campaign Events

Thomas Holbrook (1994, 1996) finds that conventions and presidential debates are the proverbial 800-pound gorillas of campaign events; both clearly influence voters’ preferences. This contention is backed by specific studies of campaign events by Campbell et al. (1992), Geer (1988), Lanoue (1991), and Shelley and Hwang (1991). Holbrook’s estimates of the effects of other events are much more ambiguous, suggesting that other campaign event effects are inconsistent and contextually dependent.

Holbrook’s research is consistent with Shaw’s (1999a) work on the matter, with a few addenda.First, Shaw finds that gaffes or mistakes are strongly correlated with changes in candidate preference. Second, Shaw finds that scandals are not especially significant for vote change (for a contrary view, see Fackler and Lin, 1995). Third, Shaw finds that messages (or policy initiatives) tend to be uncorrelated with contemporaneous shifts in candidate preference. Fourth and finally, Shaw’s research indicates that not all event effects persist; some efforts are durable over a period of ten days while others fade and still others grow. Put another way, the functional form of campaign effects depends on the nature of the event.

Media Effects

Several studies show that media exposure, while not influencing candidate preferences per se, influences a range of other political attitudes and impressions (Freedman and Goldstein, 1999; Brians and Wattenberg, 1996). Collectively, these analyses suggest that (1) we have been looking at the wrong variable when considering campaign effects, and (2) news media coverage matters because it affects impressions of candidates and issues and these, in turn, influence vote choice.

We should add, somewhat belatedly, that while there is no consensus that the news media have an ideological slant (but for a contrary view see Goldberg, 2003), a plethora of recent studies have empirically considered this possibility. Most notably, several studies of the 1992 presidential election show a significant anti-Bush tone to coverage (see, for instance, Sabato, 1993; Kerbel, 1995; Lichter and Noyes, 1995). More specifically, they show that economic coverage was far more negative than the objective condition of the economy and that this was the primary frame used to portray Bush and his administration (Hetherington, 1999; Lichter and Noyes, 1995). It is also the case that Bush received unfavorable coverage even when he was ahead in the polls (up until late June 1992), so it is difficult to blame the horserace for the tone of media coverage. Clinton, on the other side of the ledger, received positive coverage but only after he took the lead in the presidential preference polls just before the Democratic Convention. No such slant was discernible in 1996, at least not after controlling for Clinton’s large and persistent advantage over Dole in the race. Internal studies of broadcast and print media conducted by the Bush campaign indicate that coverage of the 2000 race was mixed, essentially following the polls. All of these suggest news media coverage is influenced by professional biases (see Robinson and Sheehan, 1983; Sigal, 1973), and these tend to produce favorable coverage for frontrunners and unfavorable coverage for underdogs. These biases, however, have not been connected to support shifts among voters.

Candidate and Campaign Approaches

In addition to these advances in the study of specific manifestations of the presidential campaign, there have been changes in the way we view both candidates and voters, and how they interact. These new conceptualizations, in turn, have affected our view of what campaigns are about.

Arguably, the most intriguing conceptual advance in the past decade’s studies of presidential elections is John Petrocik’s notion of ‘issue ownership.’ Petrocik (1996) posits that candidates use election campaigns to convince voters that their issues are more important than the opposition’s issues. Campaigns do not compete for the median voter along some summary left-right issue dimension; rather, they fight to set the agenda, knowing that Democratic and Republican candidates have different credibilities on different issues. Democrats, for instance, want to make elections about health care and the environment, while Republicans want to make them about taxes and defense. This comports with common sense, but it is quite different from how political scientists have traditionally conceived of electoral competition and (consequently) campaigns.

Another intriguing area of research focuses on the role of gender and ethnicity in how candidates are perceived and how voters react to candidates and campaigns. Two studies in this area merit particular attention, the first because of its impact on subsequent research and the second because of its innovative research design. The first study is Kahn’s (1993) analysis of gender differences in campaign messages and voters’ reactions. She finds that gender does indeed matter to both candidates and voters. Female candidates are more likely than males to emphasize ‘nurturing’ issues such as health care and education. Moreover, voters perceive female candidates as more credible and empathetic on these issues, irrespective of the actual positions or personalities.

The second study focuses on the effects of racial priming in news media coverage of issues such as crime. Nicholas Valentino (2001) uses experiments in which issues and images are altered slightly to determine if racial cues are being primed by the local news media’s presentation of certain issues. More importantly for this study, he extends the analysis to claim that such priming can affect candidate evaluations (presumably to the detriment of Democratic candidates) by raising the salience of racially charged subjects. While the evidence for Clinton evaluations in 1996 is weak, the connection posited by Valentino is interesting, particularly in light of the corroborative work by Tali Mendelberg (2001), who contends that racial priming has been a (successful) feature of Republican candidate advertising in recent elections.

New Perspectives on the Role of Parties in Campaigns

It is clear that parties today are stronger than they have been in the last thirty years, though by no means as strong as they were during their machine-politics heyday. Yet parties have adapted to remain relevant to elections by working with candidates and voters alike. While it is unlikely that, short of a serious upheaval in the political system, parties will deviate from their current status as service organizations, it seems probable that they will seek to exert greater influence over elections. Based on current trends in party electoral activity, we see three areas in which the academic understanding of parties needs to focus: (1) the developing role of parties in campaign finance, (2) advances in service provision, and (3) efforts to influence nomination politics.

Developments in Party Financing of Campaigns

Several scholars have already begun to explore new avenues for party fundraising, avenues that will certainly expand under the Bipartisan Campaign Reform Act (BCRA) of 2002. Herrnson (2000) notes the rise in ‘party connected contributions,’ or campaign contributions that occur from members, former members, or leadership political action committees (PACs) established by current members of Congress. Leadership PACs, in particular, have become an increasingly important way for parties to influence campaigns. In the 1999-2000 election cycle, contributions from leadership PACs of both parties were ten times the amount contributed in the 1983-84 election cycle and two and a half times the amount contributed in 1995-96 (Potoski et al., 2003). As Potoski et al. note, the implications of the BCRA suggest that leadership PACs may play an even greater role in the future, given that PACs can contribute $5000 per campaign (whereas individual contributions are limited to $2000 per campaign), and members can control multiple leadership PACs.

Of particular interest will be research into the use of new and existing committees by parties seeking creative means to cope with the implementation of the BCRA. Members of Congress and the parties have already established new ‘shadow’ committees designed to get around BCRA restrictions by accepting the ‘soft’ money that once went to national party committees (Edsall, 2002). In addition, Senate candidates in 2000 began creating ‘victory committees’—joint fundraising committees that were operated by the candidate and the party. These victory committees would raise both hard and soft money, the latter being transferred to the national party which would send it on to various state and local parties—though frequently these funds would be transferred back to the state of the candidate involved in the joint effort (Dwyre and Kolodny 2002). The role of the Hill committees will also be important.Existing studies suggest that the Hill committees are quite active, raising and spending both hard and soft money (Dwyre and Kolodny, 2002) and channeling resources with the goal of maximizing seats rather than encouraging party support in Congress (Damore and Hansford, 1999).With BCRA’s soft money ban, researchers will have to reassess the role of Hill committees as the latter reinvent themselves to work only with hard money.

State parties stand to win in the wake of the BCRA, as soft money finds its way to state party organizations, especially the parties in the 14 states that do not impose limits on corporate contributions and those in the 19 states that do not impose limits on contributions from labor unions (Dunbar, 2002). The role of state parties in financing state legislative campaigns varies greatly by state (Gierzynski and Breaux, 1998). State parties do play an important (if not overwhelming) role in financing some federal campaigns; in the case of campaigns for the US Senate in 2000, the Democratic Party even entrusted the responsibility for making coordinated expenditures to the state parties (Brox, 2004). State legislative committees, similar to the Hill committees, are also emerging as a force in campaign finance (Gierzynski and Breaux, 1998).

Advances in Party Service Provision

Research into the service role of parties has expanded as parties have become more active and more valuable to the candidates they serve. And though television advertising is likely to remain the dominant form of party campaigning in the near future, parties are expressing renewed interest in applying the shoe-leather techniques of an earlier era, enhanced with advances in technology (Balz and Allen, 2003; Nagourney 2002). For example, the Bush campaign and the Republican National Committee are using the Internet to recruit volunteers, and they are creating mobilization strategies that incorporate early voting programs and a ‘72 Hour’ plan for election day get-out-the-vote drives (Balz and Allen, 2003). In addition, both the Republicans and the Democrats are implementing technology in their search for voters; in 2004 the parties will be using advanced software to target likely voters. This software incorporates extremely large—160 million records—voter lists augmented with political, demographic, consumer, and personal data to help the parties coordinate email, phone, and direct mail efforts for both fundraising and mobilization (Theimer, 2003).

The literature on parties is only beginning to assess the impact of these efforts. Gerber and Green (2000, 2001) have cleared a path for a number of innovative studies (many using natural experiments) to gauge the effectiveness of these mobilization strategies. Their latest work suggests that face-to-face efforts are effective at stimulating turnout in local elections (Green et al., 2003). The vanguard of this work seeks to discover the effects of these mobilization efforts on particular demographic groups. Elizabeth Bennion (2003) finds that personal contact using a non-partisan get-out-the-vote message was (somewhat) effective at mobilizing young voters in South Bend, Indiana, during the 2002 campaign. Melissa Michelson also explores the effectiveness of personal contact mobilization, focusing on a Latino population in California. She finds that face-to-face canvassing was effective at mobilizing Latino voters for a school board election in 2001 (Michelson, 2002), though that canvassing effort did not translate to increased turnout in 2002 among those subjected to the mobilization treatment in 2001 (Michelson, 2003). And Wong (2003) finds modest effects of telephone and mail mobilization efforts on Asian-Americans in Los Angeles County, California, with the effects varying by ethnic group. Future research will continue to tease out how party mobilization efforts vary based on the groups targeted, the type of appeals offered (partisan vs. non-partisan) and the mode of contact (face-to-face, telephone, mail, internet).

Another unique feature of parties that is being developed in the literature is an understanding of parties as brokers of services (Herrnson, 1986b), or as liaisons between candidates, consultants, and PACs. Herrnson (2002) reports that parties help candidates by facilitating contact between candidates and consultants and by guiding candidates toward PACs for contributions. From the point of view of the PAC, parties provide election information, guidance regarding which candidates to support, and opportunities to meet and greet candidates and elected officials. And consultants also use the parties as brokers; during election years, they benefit from the contacts with candidates that the parties make possible, and during non-election years parties often hire consultants to assist with long-range planning (Herrnson, 1988, 2002; Kolodny 2000; Sabato, 1988).

The ability of parties to perform their service role has been enhanced over the last decade by continued institutionalization and nationalization of the Democratic and Republican parties. In terms of institutionalization, national party organizations have more money, more staff and better infrastructure, and they are more involved with PACs and with state and local party organizations (Herrnson, 2002). As parties become more institutionalized, their role in elections becomes more relevant. Large amounts of hard and soft money have allowed national party organizations to have greater (though by no means complete) control over the content and strategy of their candidates’ campaigns (La Raja, 2002). Further, parties have nationalized, with national party organizations using their financial resources to influence the activities of state and local party organizations (Bibby 1998), effectively making the latter into branches of the former (La Raja, 2002).

Renewed Efforts at Influencing Nominations

Earlier we noted that the parties have generally lost the ability to control which politicians get to run for office. Yet new research suggests that parties are attempting to regain some of their previous power with respect to candidate selection. Maisel et al. (2002) find that party officials are playing an increasingly important role in candidate recruitment through contacting potential candidates. Buchler and La Raja (2002) find that party activity and incentives (such as primary endorsements) increase not only the likelihood of recruiting a candidate for the US House, but also the quality of that candidate—but only for Republican state parties. Though they are not as important as they once were (Jewell and Morehouse, 2000), endorsements also help parties play a role in candidate selection. Cohen et al. (2001) find that presidential candidates who are broadly endorsed by party elites are more likely to win the nomination. Dominguez (2003) also looks at the impact of endorsements by party elites; she finds that ‘party loyal’ donors react to elite endorsements when making contributions during the primaries.

Discussion

We believe the pendulum is swinging back on the campaign effects argument, and part of this is driven by the increasing relevance of parties. Having settled into their roles as service organizations, they have proceeded to expand their influence in elections through innovative use of campaign funds and the implementation of unique technologies that enhance the value of the services they provide.

Looking to the future, we see two points that ought to be kept in mind. First, parties are probably going to remain relevant for the foreseeable future. Despite recent efforts at campaign finance reform that seek to limit issue advocacy and eliminate soft money, parties will continue to be a stable conduit for the large sums of money that will inevitably find their way into politics. State parties appear to be in a prime position to take up much of the slack resulting from the BCRA’s soft money ban at the national level. Leadership PACs run by members of the party in government are also likely winners if the reforms are kept in place. Parties will continue to provide services to their candidates and will continue to be vital to the identification, registration, and mobilization of voters.

Second, campaigns and parties throughout the world are going to look increasingly like those in the United States—if not ideologically, then structurally and in terms of their strategies and outreach. We have already seen political consultants from the United States going abroad, using their expertise to help devise strategy for campaigns in Israel, the former Soviet Union, Europe, and Latin America (Arterton, 2000; Harman, 1999; Beamish, 1994). In addition, both major American parties send staff members abroad to help developing democracies establish party systems (Holley 2003; Dobbs, 2001). Campaigns are ultimately about helping candidates talk to voters. Parties help candidates undertake that communication effort, and they help make sure voters hear the message. It is likely that the trends and developments taking place in the United States will spread to other parts of the world; as a result, the United States is no longer ‘exceptional,’ it is at the vanguard.