The ground war 2000-2004: strategic targeting in grassroots campaigns.

Panagopoulos, Costas ; Wielhouwer, Peter W.

Following the epitaphs written for American political parties in the wake of political and legal changes during the post-World War II era, a resurgence of sorts has occurred that would make George Washington Plunkitt proud. The political parties have long shown themselves to be adaptable in an uncertain political environment, making and remaking themselves as conditions permit or demand. They have, in a sense, seen opportunities and taken advantage of them. The emphasis on grassroots politics that once was the bread and butter of the local political parties described by the preeminent scholar V. O. Key gave way in the 1980s to technocentric parties that seemed to have forgotten the party-in-the-electorate. Political scientists were quick to ascribe all manner of political pathologies to the decline of political mobilization by entities such as parties, and, truth be told, there was something to the argument.

By the late 1990s, however, evidence pointed to the rediscovery of the mass base of political decision making. Grassroots mobilization efforts, by all accounts, were making a comeback. The virtually unfettered ability of individuals and organizations to disseminate information and facilitate mobilization efforts in the first few years of the twenty-first century was but one indicator of the trend. The political parties, too, expanded anew their personal contact campaigning efforts, particularly in light of ambiguous evidence regarding the effects of mass media advertising on turnout, but particularly helpful and precise evidence on the efficacy of a more personal approach to politics (e.g., Gerber and Green 2000).

We examine the elections of 2000 and 2004, two years that saw the highest levels of personal contact campaigning since the American National Election Studies began asking about such tactics in the 1950s, Our goals are twofold. First, we test hypotheses, derived from statements of top campaign operatives from the presidential election campaigns in those years, that the parties shifted their grassroots campaign strategies from largely swing-voter-focused operations to more mixed-base and swing strategies. Our results present a modest degree of support that the parties accomplished such goals.

Second, we examine personal contact campaigning in a way that, to our knowledge, has not been possible to this point. Because of the rather low levels of respondents in national surveys who reported campaign contacts, the ability to rigorously assess their correlates has been hindered. The much greater degree of such reports in the last two cycles enables us to examine the determinants of unique Democrat and Republican contacts--those people who report either Democrat or Republican party contacts but not both. It is our contention that the kinds of people each party seeks out as part of a base-centered strategy are likely to be quite different from the kinds of people that both parties seek to influence in efforts to expand their electoral coalitions in search of political victory. In this case, our results are interesting.

Political Parties and the Resurgence of Grassroots Politicking

Conventional conceptions of political parties view them as multidimensional linkage institutions between the mass electorate and elected officials in the government (Baer and Bositis 1993; Eldersveld 1982; Sorauf 1967), Parties exist as organizations, with some degree of structure, varying divisions of labor, and some number of full-time employees and, in the government, with officials (actual and potential) standing for election under party labels. Eldersveld noted that political parties are a central type of linkage structure in the modern American political system. As "intermediary organizations," they "help produce positive action and effective decisions in the face of fragmentation, conflict, and mass involvement. These structures are groups that engage in activities and organize initiatives that make cooperative behavior possible" (1982, 4).

In the wake of political and legal changes to elections in the last half of the twentieth century (see, e.g., Beck 1997; Bibby 1998; Ladd and Hadley 1978), many of the services traditionally provided to by political parties were co-opted by the more amorphous concept of the election campaign. It has been argued, for example, that political parties declined in their importance to American elections, replaced by candidate-centered elections (e.g., Aldrich 1995; Wattenberg 1991, 1998) that are perceived as being dominated by an even more amorphous group: political consultants (Medvic 2001; Thurber and Nelson 2001).

Regardless of how one perceives the relative strength of political party organizations, their activities, or political consultants in elections, the campaign remains an integral part of the American political landscape. There is skepticism that campaigns matter for political outcomes, borne particularly out of the influential work of Columbia University sociologists Paul Lazarsfeld and Bernard Berelson in The People's Choice (Lazarsfeld, Berelson, and Gaudet 1944) and Voting (Berelson, Lazarsfeld, and McPhee 1954). As Lazarsfeld put it in 1944, "In an important sense, modern Presidential campaigns are over before they begin" (1944b, 317).

While there are models that attempt to predict electoral outcomes sans campaigns (see, e.g., Campbell and Garand 2000; Cuzan, Heggen, and Bundrick 2003; Fair 2002), the people and organizations that are interested in winning elections seem absolutely committed to campaigning. Spending on campaigns appears to bear this out, with Bush-Cheney and Kerry-Edwards spending more than $650 million in 2004, and the two major national parties spending nearly $1.6 billion (see http://www.opensecrets.org). Analysis of Federal Election Commission spending data by Raymond La Raja (2005) shows that the 100 state parties spent nearly $400 million in 2004. The rate at which money was spent by the state parties on voter mobilization and grassroots campaigning was more than double that of the previous three presidential election cycles (state Republican grassroots spending increased from about 16 percent to about 36 percent of total spending, while state Democratic grassroots spending increased from about 13 percent to 35 percent of total spending), partially as a function of changed campaign finance guidelines.

The spending on grassroots mobilization continued a trend begun in the early 1990s. While personal contact campaigning, as measured by the American National Election Studies (ANES), generally increased between 1956 and the early 1980s, these rates dropped off through the early 1990s. The rather anemic grassroots efforts of the 1980s gave way to widespread personal contacting by both parties by 2000, and the campaign "ground wars" have been at levels unseen m the post-World War II era (Wielhouwer 2006). Overall, 43 percent of respondents reported contact by at least one of the major parties during the 2004 election cycle, up from 35 percent in the 2000 election, which was the highest level since the ANES began asking this question in 1956. Personal contact campaigning increased substantially over the past six decades, more than doubling between 1956 and 2004. In 2004, 31 percent of respondents reported contact by the Democratic Party, compared with 28 percent reporting GOP contact.

The renewed emphasis on grassroots in 2004 generated excited proclamations from both parties (for summaries of campaign, party, and independent claims of grassroots campaigning rates, see Bergan et al. 2005). The Democratic National Committee claims that in 2004, Democrats "recruited over 25,000 trained precinct captains, conducted 530 Organizing Conventions across the country, mobilized 233,000 volunteers, knocked on 11 million doors and made 38 million volunteer phone calls and 56 million paid calls" (DNC 2005). The Republican National Committee claims that the party's "[g]rassroots get-out-the-vote activities in 2004 surpassed all of the RNC's expectations ... 2.6 million Team Leaders and volunteers, and 7.5 million e-activists rook action on behalf of the party and its candidates ... 9.1 million doors were knocked on, and 27.2 million phone calls were made" (RNC 2005).

Moreover, many 527 organizations and interest groups (including union, environmental, and religious organizations) were very active in grassroots mobilization efforts. One analysis noted that 2004's campaign funds "purchased record amounts of television and radio advertising, phone calls, person-to-person contacts, and direct mail pieces. ... [S]ophisticated marketing techniques helped campaigns identify voters who sometimes received more than a dozen contacts.... [P]olitical parties and interest groups devoted more money to the ground war than ever before.... and targeted a hard money bonanza into ground war activities and independent expenditures" (CSED 2005). In this context, turnout of the voting eligible population increased to 60.3 percent in 2004 (McDonald 2004), the second-highest turnout since 1968. The U.S. Census Bureau reports a citizen vote rate of 63.8 percent, and voting age turnout of 58.3 percent, both about average for elections since the 1960s (U.S. Census Bureau 2005).

A New Look at the Data

Data from the American National Election Studies are the most common source of information about the dynamics of party direct-contact efforts. In most presentations of these data (e.g., ANES 2006; Rosenstone and Hansen 1993; Wielhouwer and Lockerbie 1994), Democratic and Republican contacts are considered collectively. That is, respondents are queried as to whether they were personally contacted during the election campaign by a campaign; respondents reporting contacts are probed to determine which party's representatives made the contact. What generally emerge are variables that measure reported contact by Democrats or Republicans; importantly, respondents may have been contacted by representatives of both parties. (1)

Analyses that treat these personal contacts as dependent variables (e.g., Gershtenson 2003; Rosenstone and Hansen 1993; Wielhouwer 2003) do so in order to find evidence of strategic targeting by the parties. Generally, findings confirm the notion that parties contact people predisposed to participate in politics and are members of their respective political coalitions. The former set of findings fuels the "parties exacerbate political inequalities" accusation (see, e.g., APSA 2004; Verba, Schlozman, and Brady 1995; but see Wielhouwer 2007).

What is not generally analyzed, however, are unique contacts by each party. Strategic targeting arguments would suggest that there are some people who might be contacted by both parties, such as swing or independent voters; at the same time, there is reason to speculate that there are some people whom Republicans are very unlikely to contact in campaign information distribution or mobilization efforts, and the same could be said for the Democrats. We thus examine people who reported Republican contacts only, Democratic contacts only, and contacts by both major parties. It seems reasonable to hypothesize that the kind of person likely to be targeted by both parties might be qualitatively different in politically important ways from the kind of person likely to be targeted by only one of the parties.

[FIGURE 1 OMITTED]

When it comes to unique contact by the campaigns, Figure 1 suggests that the parties have long been roughly balanced, including in 2000 and 2004. In 2000, respondents reported slightly more Republican contacts, but in 2004, the Democrats held a slight advantage. For the GOP, the 2000, 2002, and 2004 elections represent the highest levels of contacts in the time series, whereas the Democrats, increasing steadily since 1990, finally surpassed their post-World War II high mark in 2004. These data suggest that each party contacts about the same percentage of the electorate, and we will shortly assess whether the people each party contacts are unique in politically interesting ways.

Turning to respondents whom both parties contacted, Figure 2 shows that the proportion of the electorate for whom the parties genuinely compete through grassroots campaigning has virtually skyrocketed over the last two decades. In 1990, a mere 4 percent of the electorate reported contact by both parties; by 2000, that rate had tripled to 12 percent and had more than quadrupled by 2004 to about 17 percent. Though it is possible that this sharp increase suggests that contacts were not as targeted as the anecdotal evidence may lead us to believe, it may also be that the campaigns had acquired a degree of consensus about the voters for whom they should be competing. The trend suggests an increase in the campaigns' efforts at persuasion, though some analysts suggested that 2004 was not a year in which as much persuasion was attempted. Our analysis of whom the contested-for citizens were may give us evidence as to the kinds of persuasion that were being attempted by the campaigns.

[FIGURE 2 OMITTED]

Overall, then, the 2000 and 2004 election years present us with interesting questions about the parties' strategic targeting, A picture has emerged of an electorate in which about 10 percent to 15 percent are contacted only by the Republicans, about 10 percent to 15 percent are contacted only by the Democrats, and about 15 percent are contacted by both parties. Given what we know about the social and demographic bases of each party's coalition, we certainly would expect the composition of their respective unique contacts to be quite different. By the same token, it is not entirely clear at this point what the people who were contacted by both parties looked like, whether they were more partisan or more independent, whether the campaigns were attempting to convert the opposition, or whether there were any important differences within this group at all.

The Logic of Strategic Grassroots Campaigning

Theoretically, we understand campaigns as organizations that are formed and managed with the explicit goal of winning elections. In order to maximize the probability that this goal will be accomplished, campaigns must first influence voter preferences in ways that advantage their own candidate, and second, do what they can to ensure that the supporters of their candidate are the predominant source of votes on election day. In general, the successful achievement of each of these goals requires generating messages and communicating them to one or more subsets of the electorate to persuade, to activate, or to sensitize potential voters. Thus, specific audiences are targeted for specific messages during the campaign, often based on voting predispositions and preexisting party coalitions; a key tactic involves personal contact campaigning with members of those audiences.

The scarce resources in the vast majority of elections impel many campaigns to coordinate activities with major political party structures, taking advantage of their institutional memories. The campaign (i.e., those running the campaign operations), in developing strategy, makes assumptions about its prospective polity and subsets within it and then selects communication techniques on the basis of those assumptions. Although those assumptions may be based on an understanding of aggregate voting behavior (Shea and Burton 2001, chap. 4), the basic campaign mind-set relates to the individual voter. Still, even the best funded campaigns must make resource allocation decisions. Those decisions must be implemented with organization and tactics accordingly.

The aspects of strategy that concern us here primarily involve electoral coalitions and campaign messages and their delivery. The parties' electoral coalitions, though potentially malleable, are also clearly distinct. Each party has its base coalition groups, as well as groups that are generally reliable party voters but not necessarily part of their "core" mass constituencies (e.g., Petrocik 2006; Stanley and Niemi 2006). Moreover, there are "swing" coalition groups that may vote for either party in any given election, given issue salience, economic situations, and conversion or activation efforts by campaigns. Prospective coalition groups may be targeted for campaign messages or may be avoided, depending on campaign resources and strategy. And there are a range of campaign communication approaches that can be implemented, ranging from the very expensive (such as campaign advertising) to the very grassroots (personal contact campaigning).

For this research, we expect that the presidential campaigns had the capability to distinguish among different coalition groups in their campaign message delivery processes. There is a long history of scholarship that distinguishes among different kinds of groups as they relate to campaigns. For example, Lazarsfeld's (1944) research on campaign effects identified efforts to reinforce potential voters' partisan predispositions, activate interest in the campaign beyond latent political predispositions, and convert opposition partisans. (2)

Support for the hypothesis that the 2000 and 2004 campaigns attempted to target different constituency groups is readily available. Campaign principals for Bush-Cheney, for example, suggest shifts in targeting strategies between the two election cycles. The 2000 strategy was apparently more focused on swing voters, whereas the 2004 strategy emphasized base activation in addition to swing voter persuasion. Matthew Dowd, the Bush-Cheney chief campaign strategist, noted,

 Obviously, as [we] approached 2000, motivating Republicans was
 important, but most of our resources [were] put into persuading
 independents in 2000.... structurally ... the campaign had not put
 enough resources into motivating and turning out those folks [the
 religious base]. Some of that was done. But ... one of the things
 we did wrong was not have enough person-to-person contact and
 on-the-ground staff and people to motivate folks. It wasn't just a
 broad, national message. It was people to talk to people in their
 neighborhoods. And there was a concerted effort to analyze that,
 this whole 72-Hour Task Force that we put together to figure out
 what we did right and wrong on turnout. And that's one of the
 things that we discovered that we tried to fix in 2004. (Frontline
 2005a)

And Mark McKinnon, media advisor for Bush in both years, discussing the 2004 "base strategy," commented,

 It struck me as a political consultant as something radical,
 because for years we had always talked about that persuadable
 middle electorate, and that's what it was all about. You ignored
 everything else. All your resources went into that persuadable
 vote. But that vote was typically 20 percent of the electorate. And
 when you look at the history, which Karl [Rove] and Matthew [Dowd]
 did very closely, they looked at it and said: "This share of the
 pie is getting pretty thin. It's getting down to, like, 7 percent
 of true swing voters." So if that's the case, it means two things:
 one, that 7 percent is more important than ever; and two, [the]
 other part of the pie, we better pay attention to that, because if
 it's only 7 percent that's persuadable, we've got to make sure that
 we get these people out and vote. And that's what this campaign put
 a real focus on, was paying attention to the 7 percent for certain,
 but also making sure that there was a lot of attention paid to this
 other 42 percent. (Frontline 2005b)

There is also, of course, a geographic component to campaign activities. At the presidential level, the winner-take-all electoral college vote allocation system that most states use makes state-level analysis worthwhile. Daron Shaw (1999), for example, has shown that the competitiveness of states in prior presidential elections is a significant factor in predicting the strategic allocation of campaigns fiscal resources, media spending, and candidate appearances. As Tad Devine, a Kerry-Edwards media consultant, put it,

 Strategically, our orientation was to win electoral votes in states
 that we considered battlegrounds. That's where we spent our time.
 That's where we sent our most precious resource--our candidates.
 That's where we spent our money on television. We did not have
 national cable buys. We did not do national radio buys. We
 approached the election differently. So the fact that John Edwards
 was spending time on our targets, that we developed on a very
 sophisticated model ... reflected our strategic orientation towards
 the race. The fact that you wouldn't see him in some other places
 was a deliberate choice and judgment that we made strategically to
 try to win 270 electoral votes. (Institute of Politics 2006,
 153-54)

Data Analysis

Given the assertions of shifts in strategic thinking by at least one of the campaigns, and the likely responsiveness of each campaign to the strategy and tactics of the other, we analyze the personal contact campaign patterns of the parties in both 2000 and 2004. This permits us to make several sets of comparisons. We can compare the differences in the parties' contact patterns to assess the extent to which those patterns match received knowledge about their electoral coalitions. We refer to these as base mobilization patterns, although the people targeted in these patterns need not be part of the party's base; they could be segments of the electorate that a party is attempting to recruit (or avoid) and that the other party is not. We can also compare the contact patterns of each party in the two years to assess changes in targets, such as the GOP shift from an emphasis on independents in 2000 to an emphasis on independents and the Republican base in 2004; we call these strategic changes. Finally, we can compare the patterns of personal contact campaigning for those citizens who were contacted by workers from both parties in order to assess those groups considered to be "swing" voters in the two years; we call these coalition competition. Appendix Table A-1 shows selected social, demographic, and political variables for each dependent variable.

Table 1 shows the results of multivariate dichotomous probit analyses that use the three personal contact types as dependent variables. Before considering the three types of comparisons, we first point out one set of results that at first glance may seem anomalous. The first row shows the effects of living in a battleground state on the likelihood of being contacted by the parties. In general, battleground residents were not more likely to be contacted by only one party (as in columns I and 2); instead, they were much more likely to be contacted by both parties, as seen in column 3. This reflects the geographic strategic reality that the parties are largely competing in the same states.

It also bears mentioning that, with one exception, prior voters were more likely to be personally contacted by campaigns. These voters are well-known for being targets, and their prior voting experience makes them excellent prospects as future voters (e.g., Brady, Schlozman, and Verba 1999; Green and Shachar 2000). This set of finding confirms that, across the board, campaigns target prior voters. The exception in these years was the 2000 Republican campaign, which did not particularly target prior voters; this confirms an internal GOP analysis that made such a critique and recommended more aggressive voter registration and new registrant mobilization as part of the 72-Hour Task Force strategy (72 Hour Task Force, 2002). Let's now turn to the parties' base mobilization efforts and base mobilization strategic changes between the two years.

Considering first Democratic contacts, in 2000, Democratic Party identifiers were just as likely to be contacted as independents (the excluded category), whereas Republicans were much less likely to be contacted. In 2004, the reverse was the case: Democratic identifiers were significantly more likely to be contacted by their party than were independents and Republicans (though the Republican identifier coefficient approaches conventional significance levels). The party thus pursued a partisan base mobilization and independent persuasion strategy in 2000, which was narrowed in 2004 to partisan base mobilization. In both years, the party was generally effective at avoiding Republicans, confirming the observation that little partisan conversion was attempted.

Respondents' age, gender, and income, and regularity of church attendance made no particular difference for Democratic contacts in either year. Black respondents were significantly more likely to be targeted in 2000, but were not especially so in 2004. Finally, Democrats targeted citizens based on their educational levels, with better-educated respondents reporting significantly higher levels of personal campaign contacts than less-educated respondents. This may reflect a strategy of targeting for mobilization people who are predisposed to vote, but it may also reflect an outreach effort to improve party voting performance among better-educated persons, given that the party has a constituency that is, on the whole, less educated than that of the Republicans. (3)

Turning to the Republican contacts (column 2), the party effectively avoided both Democrats and independents; although Bush-Cheney operatives argued that the campaign shifted from an independent mobilization strategy to a base plus independent mobilization strategy between 2000 and 2004, it appears that in both years, Republicans ended up contacting a significantly higher proportion of their own partisans than independents. In 2000, the GOP contacted women at a greater rate than men, and blacks at a significantly lower rate than nonblacks in both years. The latter is consistent with a base mobilization strategy, whereas the former is likely a coalition expansion strategy given the degree to which the gender gap has advantaged the Democrats over the last generation. By 2004, women were just as likely as men to be contacted. In short, both parties show evidence of base mobilization strategies and coalition expansion, as well as strategic shifts in their base mobilization strategies between the two years.

Table 1, column 3 shows the results of probit analysis predicting who reported contact from both political parties. This begins to paint a portrait of whom the campaigns perceived as the voters they were competing for at the grassroots level. In 2000, the parties were competing most actively in battleground states, for previous voters, and for better-educated citizens. These all reflect efforts to maximize the electoral returns on campaign investments, as the election would be decided in those states; prior voters are reliable voters; and better-educated people are also reliable voters (though it is not clear that better-educated respondents are the most responsive to campaign messages; see Wielhouwer 2005).

In 2004, however, the competitive targets were somewhat different. Beyond battleground state residents and prior voters, the parties were competing for older and nonblack citizens. It may be that both parties believed that the George W. Bush administration initiatives affecting older Americans (such as Social Security and Medicare reform) propelled more extensive campaign communication efforts toward that segment of the population. The emphasis on the nonblack population also is intuitive, in that there is little meaningful competition at the aggregate level for the African American vote, certainly compared with the competition for Hispanics and various white electoral subgroups.

Finally, in order to make this analysis more directly comparable to earlier research on the determinants of parties' contacts (e.g., Gershtenson 2003; Rosenstone and Hansen 1993; Wielhouwer 1995, 2003), Table 2 presents equations for the overall correlates of being contacted by each party (the dependent variables merge people contacted by both parties with those contacted uniquely by each party, respectively). As we have already seen, residence in a battleground state and voting in the previous election clearly influenced the likelihood of contact for both parties in both election cycles. For respondents residing in a closely contested state, the probability of being contacted by the Democratic Party in 2000 was 8 percent higher than for residents on nonbattleground states, Similarly, Republicans were 11 percent more likely to contact residents of battleground states. By 2004, voters in battleground states were more highly coveted by both parties. Republicans were 23 percent more likely to reach out to voters in battleground states, and Democrats were 24 percent more likely to do so.

Prior voting history, information readily available to campaigns and useful as a targeting criterion, also helps explain parties' mobilization targets in both 2000 and 2004. Democrats were 19 percent more likely to target a voter in 2000 if he or she had voted in 1996; Republicans in 2000 were 11 percent more likely to contact those who had voted in the last presidential election. In 2004, Democrats were 13 percent more likely to mobilize voters who had voted in 2000, compared to those who had failed to vote in 2000, and the likelihood of contact by the Republican Party in 2004 was 14 percent higher for those who had voted in 2000.

These equations provide evidence of shifting party mobilization strategies between the two elections. Democrats in 2000 were much less likely to target Republican sympathizers than independents, but they were equally likely to reach out to Democrats and independents (the omitted category). This suggests that Democrats may have pursued some base mobilization in 2000, but they also engaged in persuasion. The Republican Party strategy in 2000 is somewhat elusive, suggesting that Republicans didn't effectively distinguish among their own partisans, independents, or Democrats. One interpretation of this finding is that Republican targeting efforts in 2000, controlling for battleground residence and prior voting, were not especially fine-tuned based on the party-in-the-electorate. This pattern is conceivably one aggregate outcome that matches the party's own evaluation of problems in that year.

In 2004, both parties' mobilization strategies became more focused. Democrats targeted their base more than independent voters and effectively avoided Republicans. Though there remains evidence of persuasion efforts in 2004, Democrats clearly emphasized base mobilization; Democrats in 2004 were 7 percent more likely to be contacted by the Democratic Parry than were independents. Similarly, as we have already surmised, Republicans appear to have honed their mobilization targets in 2004, reaching out to their own partisans much more effectively. Republican Party identifiers in 2004 were 9 percent more likely to be contacted by the Republican Party than were independents. However, the data also suggest the Republican Party engaged in persuasion in 2004; the results reveal that Republicans were just as likely to contact independent voters as they were to contact Democrats.

Conclusions

To summarize, our findings provide mixed evidence regarding the degree to which the Republican and Democratic parties shifted their targeting strategies between 2000 and 2004. Analyzing total party contacts reveals apparent strategic shifts by both parties, with each emphasizing base mobilization to a greater degree in the later year. And among people contacted only by the Democrats, this pattern was also seen. Among people contacted only by the GOP, however, the party successfully targeted its own party identifiers in both years in comparison to independents and Democrats. The Republicans also more effectively targeted prior voters in 2004, while the Democrats did so in both years. Neither parry appears to have relied on mass partisanship when both were competing for the same group of people. Instead, they appear to have targeted based on educational attainment in 2000 but on age and ethnicity in 2004.

While the stark contrast in campaign strategies portrayed by the Bush-Cheney team did not emerge from this analysis, there was some adjustment in both parties' targeting strategies. Securing the base became much more prevalent in 2004 compared with 2000, though independents retained some modest degree of attention. Those swing voters that were the objects of targeting by both parties shifted between the two elections, and this is an intuitive result. While the specific rationale for the shift is not clear here, it is not surprising that the groups considered to be the deciding factors in an election would vary from year to year. The strategy of message development in any campaign has at its core the development of a knowledge base regarding what is going to matter in the current campaign. Electoral history is a guide to the future, but in the final analysis, each election must write its own history.

TABLE A-1
Social, Demographic, and Political Correlates of
Partisan Campaign Contacts, 2000 and 2004

 I II
 Democrat Republican
 Contact Contact

 2000 2004 2000 2004

Total 10.1 15.3 12.6 13.1
Democrats (incl leaners) 13.2 21.4 9.2 6.2
Independents 7.9 13.4 5.5 9.3
Republicans (incl leaners) 6.9 8.9 20.5 20.9
Income Level
 0-16 percentile 6.9 12.9 4.8 8.0
 17-33 percentile 8.8 11.8 13.1 8.6
 34-67 percentile 12.1 15.6 11.8 14.2
 68-95 percentile 10.6 20.5 16.8 19.0
 96-100 percentile 8.3 20.0 22.2 13.7
White 9.9 15.1 15.2 15.9
Black 12.6 19.0 5.0 3.1
Hispanic 9.2 6.5 6.1 4.3
Church attendance
 Every week or more 11.5 16.6 14.6 17.0
 Almost every week 10.6 9.8 21.8 16.7
 Once or twice a month 10.6 15.2 11.5 10.1
 A few times a year 9.9 23.3 12.1 11.0
 Never 8.5 13.3 10.2 9.6
Union members 12.7 25.5 10.9 9.8
Union household 14.6 23.7 10.2 10.2
Male 10.9 13.5 12.4 11.1
Female 9.5 16.8 13.7 13.9
Education
 Eighth grade or less 11.5 10.0 5.8 10.0
 Grades 9-12 6.3 14.1 12.6 1.6
 High school diploma 8.0 10.4 10.1 13.0
 or equivalent
 Some/junior college 8.9 14.7 13.9 11.8
 Bachelor's and advanced 13.7 21.1 15.9 15.5
 degrees

 III
 Contact by
 Both Parties

 2000 2004

Total 13.1 16.7
Democrats (incl leaners) 13.6 16.0
Independents 11.6 20.6
Republicans (incl leaners) 12.8 16.7
Income Level
 0-16 percentile 6.9 7.4
 17-33 percentile 10.9 19.9
 34-67 percentile 13.2 13.2
 68-95 percentile 17.1 21.5
 96-100 percentile 16.7 24.2
White 14.3 20.2
Black 9.4 9.2
Hispanic 6.1 7.5
Church attendance
 Every week or more 14.1 20.6
 Almost every week 18.8 17.4
 Once or twice a month 10.2 20.3
 A few times a year 14.8 8.9
 Never 10.6 14.7
Union members 18.2 20.6
Union household 18.4 22.6
Male 13.7 17.1
Female 12.6 16.3
Education
 Eighth grade or less 3.8 13.3
 Grades 9-12 10.5 17.2
 High school diploma 10.1 16.7
 or equivalent
 Some/junior college 14.8 16.8
 Bachelor's and advanced 15.5 16.7
 degrees

Source: 1948-2004 American National Election Studies
Cumulative Data File (October 31, 2005, release).

Number of cases: 2000, 1,517; 2004, 1,049.

Note: Democrat and Republican contacts reflect unique
contacts by each party, with no overlaps.

References

72-Hour Task Force. 2002. RNC PowerPoint presentation (author's personal copy).

Aldrich, John H. 1995. Why parties? The origin and transformation of political parties in America. Chicago: University of Chicago Press.

American National Election Studies (ANES). 2006. ANES guide to public opinion and electoral behavior: Political involvement and participation in politics: Mobilization (Tables Series 6C). http://www.umich.edu/~nes/ [accessed February 22, 2008].

American Political Science Association (APSA). 2004. American democracy in an age of rising inequality. Perspectives in Politics 2: 651-66.

Baer, Denise L., and David A. Bositis. 1993. Politics and linkage in a democratic society. Englewood Cliffs, NJ: Prentice Hall.

Beck, Paul Allen. 1997. Party politics in America, 8th ed. New York: Longman.

Berelson, Bernard L., Paul F. Lazarsfeld, and William N. McPhee. 1954. Voting: A study of opinion formation in a presidential campaign. Chicago: University of Chicago Press.

Bergan, Daniel E., Alan S. Gerber, Donald P. Green, and Costas Panagopoulos. 2005. Grassroots mobilization and voter turnout in 2004. Public Opinion Quarterly 69: 760-77.

Bibby, John F. 1998. Party organizations, 1946-1996. In Partisan approaches to postwar American politics, edited by Byron E. Shafer, 148-85. New York: Chatham House.

Brady, Henry E., Kay Lehman Schlozman, and Sidney Verba. 1999. Prospecting for participants: Rational expectations and the recruitment of political activists. American Political Science Review 93: 153-68.

Campbell, James E., and James C. Garand, eds. 2000. Before the vote: Forecasting American national elections. Thousand Oaks, CA: Sage.

Center for the Study of Elections and Democracy (CSED). 2005. New campaign finance law (BCRA) pushed parties, candidates to raise dramatically more money from individuals in 2004: Changes the way money is raised and spent. News release, February 7. http://csed.byu.edu [accessed March 3, 2008].

Cuzan, Alfred G., Richard J. Heggen, and Charles M. Bundrick. 2003. Voters and presidents: A fiscal model. Philadelphia: Xlibris.

Democratic National Committee (DNC). 2005. DNC 2001-2005: Mobilizing, modernizing and building the Democratic Party. http://www.gwu.edu/-action/2004/parties/dnc05facts.html [accessed March 3, 2008].

Eldersveld, Samuel J. 1982. Political parties in American society. New York: Basic Books.

Fair, Ray C. 2002. Predicting presidential elections and other things. Stanford, CA: Stanford University Press.

Frontline. 2005a. Karl Rove--The architect. Transcript of interview with Matthew Dowd. www.pbs.org/ wgbh/pages/frontline/shows/architect/interviews/dowd.html [accessed February 22, 2008].

--. 2005b. Karl Rove--The architect. Transcript of interview with Mark McKinnon. www.pbs.org/wgbh/pages/frontline/shows/architect/interviews/mckinnon.html [accessed February 22, 2008].

Gerber, Alan S., and Donald P. Green. 2000. The effects of canvassing, telephone calls, and direct mail on voter turnout: A field experiment. American Political Science Review 94: 653-63.

Gershtenson, Joseph. 2003. Mobilization strategies of the Democrats and Republicans, 1956-2000. Political Research Quarterly 56: 293-308.

Green, Donald P., and Ron Shachar. 2000. Habit formation and political behaviour: Evidence of consuetude in voter turnout. British Journal of Political Science 30: 561-73.

Institute of Politics, Harvard University. 2006. Campaign for president: The managers look at 2004. Lanham, MD: Rowman & Littlefield.

Ladd, Everett Carl, Jr., and Charles D. Hadley. 1978. Transformations of the American party system: Political coalitions from the New Deal to the 1970s, 2nd ed. New York: W. W. Norton.

La Raja, Raymond J. 2005. State party spending in the 2004 federal elections. Paper presented at the State of the Parties Conference, Bliss Institute of Applied Politics, University of Akron, October.

Lazarsfeld, Paul F. 1944. The election is over. Public Opinion Quarterly 8: 317-30.

Lazarsfeld, Paul F., Bernard Berelson, and Hazel Gaudet. 1944. The people's choice: How the voter makes up his mind in a presidential election. New York: Duell, Sloan and Pearce.

McDonald, Michael P. 2004. Up, up and away! Voter participation in the 2004 presidential election. The Forum 2(4). www.bepress.com/forum/vol2/iss4/art4 [accessed March 3, 2008].

Medvic, Stephen K. 2001. Political consultants in U.S. congressional elections. Columbus: Ohio State University Press.

Petrocik, John R. 2006. Party coalitions in the American public: Morality politics, issue agendas, and the 2004 election. In The State of the parties: The changing role of contemporary American politics, 5th ed., edited by John C. Green and Daniel J. Coffey, 279-97. Lanham, MD: Rowman & Littlefield.

Republican National Committee (RNC). 2005. Rising tide: The magazine of the Republican National Committee. http://www.rnc.org/News/RisingTideRead.aspx?ID=89 [accessed September 19, 2005].

Rosenstone, Steven J., and John Mark Hansen. 1993. Mobilization. participation, and democracy in America. New York: Macmillan.

Shaw, Daren R. 1999. The methods behind the madness: Presidential electoral college strategies, 1988-1996. Journal of Politics 61: 893-913.

Shea, Daniel M., and Michael John Burton. 2001. Campaign craft: The strategies, tactics, and art of political campaign management, rev ed. Westport, CT: Praeger.

Sorauf, Frank J. 1967. Political parties and political analysis. In The American party system: Stages of political development, edited by William N. Chambers and Walter D. Burnham, 33-55. New York: Oxford University Press.

Stanley, Harold F., and Richard G. Niemi. 2006. Partisanship, party coalitions, and group support, 1952-2004. Presidential Studies Quarterly 36: 172-88.

Thurber, James, A., and Candice J. Nelson, eds. 2001. Campaign warriors: The role of political consultants in elections. Washington, DC: Brookings Institution Press.

U.S. Census Bureau. 2005. Reported voting and registration by race, Hispanic origin, sex, and age groups: November 1964 to 2004. Historical time series Table A1. http://www.census.gov/population/www/socdemo/voting.html/.

Verba, Sidney, Kay Lehman Schlozman, and Henry E. Brady. 1995. Voice and equality: Civic voluntarism in American politics. Cambridge, MA: Harvard University Press.

Wattenberg, Martin P. 1991. The rise of candidate-centered politics: Presidential elections of the 1980s. Cambridge, MA: Harvard University Press.

--. 1998. The decline of American political parties, 1952-1996. Cambridge, MA: Harvard University Press.

Wielhouwer, Peter W. 1995. Strategic canvassing by political parties, 1952-90. American Review of Politics 16: 213-38.

--. 1999. The mobilization of campaign activists by the party canvass. American Politics Quarterly 27: 177-200.

--. 2003. In search of Lincoln's perfect list: Targeting in grassroots campaigns. American Politics Research 31: 632-69.

--. 2005. Political parties and participation equality, 2004. Paper presented at the State of The Parties conference, Bliss Institute of Applied Politics, University of Akron, October.

--. 2006. Grassroots mobilization. In The electoral challenge: Theory meets practice, edited by Stephen Craig, chap. 9. Washington, DC: CQ Press.

--. 2007. Changes in the effects of personal contact campaigning on participation inequalities. Paper presented at the National Conference of the Midwest Political Science Association, Chicago, April.

Wielhouwer, Peter W., and Brad Lockerbie. 1994. Party contacting and political participation, 1952-90. American Journal of Political Science 38: 211-29.

COSTAS PANAGOPOULOS

Fordham University

PETER W. WIELHOUWER

Western Michigan University

(1.) We acknowledge the criticisms of this measure, particularly regarding its underlying assumption that ANES respondents can accurately recall whether they were contacted and which party's workers did the contacting. Readers should see Wielhouwer (1999, 185-86; 2003, n. 4) for discussions of the measure's strengths and weaknesses.

(2.) We would note that informal personal contacts and influence among personal friends were central to their findings, though these effects were informal, unstructured, and unconnected to formal campaigning.

(3.) The ANES data show, for example, that in 2000, the mean level of education for Democrats was 3.95 on a six-point education scale, compared with a GOP mean of 4.41 (the difference between the two was statistically significant, p < .000). A similar pattern was the case in 2002, but by 2004, Democrats apparently had made up ground, and the educational attainment difference between the two sets of party identifiers was not significant.

Costas Panagopoulos is an assistant professor of political science and director of the Center for Electoral Politics and Democracy and of the graduate program in Elections and Campaign Management at Fordham University.

Peter Wielhouwer is an assistant professor of political science at Western Michigan University.

TABLE 1
Personal Contact Campaigning: Unique and Joint
Party Contacts, 2000 and 2004

 Contacted by
 Democrats Only

Independent Variables 2000 2004

Battleground state .03 .11
 (.13) (.11)
Voted (previous .97 *** .29 **
 pres. election) (.22) (.13)
Democrat (inc. lean) -.26 .27 **
 (.21) (.14)
Republican (inc. lean) -.58 ** -.27
 (.23) (.15)
Age .03 -.03
 (.02) (.02)
Age squared -.00 .00
 (.00) (.00)
Female -.21 .10
 (.13) (.11)
Black .41 ** .02
 (.18) (.15)
Income -.06 .08
 (.09) (.07)
Education .09 ** .11 ***
 (.05) (.04)
Attends church .04 .08
 (.14) (.11)
Constant -2.85 ***-1.54 *** -2.71 ***-2.65 ***
 (.62) (.40)
N 893 1084
Log likelihood -241.80 -399.40
Pseudo R-squared .11 .06

 Contacted by
 Republicans Only

Independent Variables 2000 2004

Battleground state .23 .08
 (.13) (.11)
Voted (previous .17 .43 ***
 pres. election) (.17) (.15)
Democrat (inc. lean) .23 (.11)
 (.25) (.18)
Republican (inc. lean) .61 ** .67 ***
 (.26) (.17)
Age .00 .02
 (.02) (.02)
Age squared .00 -.00
 (.00) (.00)
Female .36 *** .14
 (.14) (.12)
Black -.54 ** -.42 **
 (.26) (.22)
Income .06 -.04
 (.09) (.07)
Education .01 .03
 (.05) (.04)
Attends church .19 .12
 (.14) (.12)
Constant -3.14 ***-3.23 ***
 (.57) (.49)
N 893 1084
Log likelihood -253.19 -343.70
Pseudo R-squared .09 .10

 Contacted by Both
 Major Parties

Independent Variables 2000 2004

Battleground state .41 *** .89 ***
 (.12) (.11)
Voted (previous .64 *** .40 ***
 pres. election) (.19) (.12)
Democrat (inc. lean) .04 .05
 (.22) (.15)
Republican (inc. lean) -.18 -.15
 (.23) (.15)
Age .01 .06 ***
 (.02) (.02)
Age squared -.00 -.00 **
 (.00) (.00)
Female -.O1 -.19
 (.13) .11
Black -.06 -.41 **
 (.20) (.19)
Income -.02 .01
 (.08) (.07)
Education .11 *** -.03
 (.04) (.04)
Attends church .23 .10
 (.14) (.12)
Constant
 (.56) (.51)
N 893 1084
Log likelihood -269.14 -375.10
Pseudo R-squared .11 .16

Note: Estimates obtained using probit analysis.

Source: American National Election Studies.
*** signifies statistical significance at the p < .01 level;
** p < .05 level; * p < .10 level.

TABLE 2
Combined Party Contacts, 2000 and 2004 (Probit)

 Contacted by Democrats

Independent Variables 2000 2004

Battleground state .32 *** .70 ***
 (-.11) (.09)
Voted (previous pres. election) .94 *** .42 ***
 (.16) (.11)
Democrat (inc. lean) -.14 .23 **
 (.18) (.12)
Republican (inc. lean) -.49 *** -.28 **
 (.20) (.13)
Age .03 .01
 (.02) (.01)
Age squared -.00 -.00
 (.00) (.00)
Female -.14 -.05
 (.11) (.06)
Black .24 -.22
 (.16) (.14)
Income -.05 .05
 (.07) (.06)
Education .13 *** .06 **
 (.04) (.03)
Attends church .17 .14
 (.12) (.10)
Constant -2.93 *** -2.01 ***
 (.50) (.37)
N 893 1084
Log likelihood -380.86 -565.64
Pseudo R-squared .14 .12

 Contacted by Republicans

Independent Variables 2000 2004

Battleground state .42 *** .71 ***
 (.10) (.09)
Voted (previous pres. election) .47 *** .50 ***
 (.14) (.12)
Democrat (inc. lean) .15 .08
 (.19) (.13)
Republican (inc. lean) .26 .31 **
 (.20) (.13)
Age .01 .05 ***
 (.02) (.02)
Age squared .00 -.00 **
 (.00) (.00)
Female .20 -.04
 (.11) (.10)
Black -.31 -.53 ***
 (.18) (.16)
Income .02 -.03
 (.07) (.06)
Education .08 ** -.00
 (.04) (.03)
Attends church .26 ** .14
 (.12) (.10)
Constant -2.78 *** -2.90 ***
 (.47) (.41)
N 893 1084
Log likelihood -395.54 -523.76
Pseudo R-squared .12 .15

Source: American National Election Studies.
*** signifies statistical significance at the p < .01 level;
** p < .05 level; * p < .10 level.