For much of the 1970s and 1980s, the study of parties as political institutions received relatively little attention—perhaps in part because of limited polarization in Congress, and from voters regularly showcasing mixed-partisan voting in elections. However, the recent move to an ideologically polarized party system has coincided with a marked increase in attention to political parties as institutions. Starting with a new wave of research on parties in Congress (Aldrich Reference Aldrich1995; Cox and McCubbins Reference Cox and McCubbins1993; Kiewiet and McCubbins Reference Kiewiet and McCubbins1991; Rohde Reference Rohde1991), partisanship and the presidency (Milkis Reference Milkis1993), and national party organizations (Goldman Reference Goldman1990; Klinkner Reference Klinkner1994) in the 1990s, recent scholarship has dealt with a variety of topics related to parties that were previously ignored.Footnote 1
One of these topics has been the study of national party organizations, the Democratic and Republican National committees (DNC and RNC). The DNC and RNC are the only institutions that represent their parties as national organizations, but traditional research has largely dismissed them as mere “service providers” (Cotter and Bibby Reference Cotter and Bibby1980). In this view, the DNC and RNC provide campaign and financial support to party candidates, but otherwise have no real power over their party since they cannot select candidates or force them to take on policy positions. Thus, as Daniel Galvin has summarized, national committees “play a supportive role, offering resources and services to candidates who seek their help” (Galvin, Reference Galvin2012 57) but are otherwise perceived as engaging in “politics without power” (Cotter and Hennessy Reference Cotter and Hennessy1964).
A different strain of research has challenged this view. Starting with Klinkner’s (Reference Klinkner1994) study on “out-party” committees—that is, parties out of the White House—a number of scholars have argued that national committees play a more important role than previously thought. They show that party leaders—including presidents, Congressional leaders, governors, and activists within the parties—fight for control over national party organizations and use this control to try shaping the party to their ideological liking (Galvin Reference Galvin2010, Reference Galvin2012, Reference Galvin2020; Conley Reference Conley2013; Rosenfeld Reference Rosenfeld2018; Hilton Reference Hilton2019). The DNC and RNC are particularly valuable due to their publicity role, which party leaders try to use for party branding purposes—that is, shaping the perception voters have of the party and which voting groups it prioritizes at a given moment (Heersink Reference Heersink2018a). National committees are not alone in trying to shape this brand, but the DNC and RNC—at least from the perspective of party leaders—are crucial in this regard because they alone represent the full national party and try using this position to create a nationally recognized brand.
Historically, the DNC and RNC have tried to do so in important ways. But the committees have not been constant in what voting groups they target, which image they promote, or even in how consistently they engage in such branding activities. In the 1950s, the DNC promoted the Democratic Party as supportive of civil rights (to the frustration of Southern Democrats within the party), and in the early 1960s the RNC targeted Southern white voters (against the wishes of Northeastern moderate Republicans). But after pushing for a clearly liberal Democratic image in the 1950s, the DNC in the decades that followed largely promoted the party as a centrist organization, hoping to regain support among white voters (Frymer Reference Frymer1999). And during the chairmanship of Bill Brock (1977–1980), the RNC focused much of its attention on appealing to Black voters (Wright Rigueur Reference Wright Rigueur2014). More recently, under the chairmanship of Howard Dean (2005–2008), the DNC initiated a series of programs aimed at winning support from white evangelical voters, while after the 2012 election loss, the RNC sponsored an analysis that called on the party to support comprehensive immigration reform and to soften its positions on gay rights.
Each of these moves occurred in response to electoral defeat in presidential elections and reflected the national committees’ attempts to alter their existing brand to improve their likelihood of future electoral success. But how to alter a brand is often controversial, and the branding activities the national committees engage in regularly place them in the center of major intra-party debates about what policies their party should promote, or which voting groups it should target. Because they can be controversial, studies have also argued that the DNC and RNC regularly decrease some of their activities—including branding (Heersink Reference Heersink2018a)—and whether a party holds the White House is a particularly important predictor of committee activity decline (Klinkner Reference Klinkner1994; Galvin Reference Galvin2010, Reference Galvin2012). Understanding how national committees operate and under what conditions they increase or decrease their branding role is important to our knowledge of the American political party system because it can explain how parties adjust their electoral coalitions, and when and how they change the policies they take on. Indeed, despite this recent wave of research, major questions about national party organizations and their influence on the broader American political system remain; for example, Nolan McCarty has noted that “we still have a poor understanding of the role played by political party organizations in producing more or less polarization” (McCarty Reference McCarty and Persily2015, 136).
In studying national committees, scholars thus far have found themselves relying exclusively on qualitative historical case studies. One likely reason for this is that neither party maintains a comprehensive archive. While archival collections that cover the records of individual political actors who worked at one of the national committees are accessible (often former national committee chairs), these records are spread across libraries in the United States, and do not always include the same type of documents. As a result, it is nearly impossible to collect consistent quantitative data covering national committees across a longer period of time. Such historical qualitative research has undeniable strengths—for example, the studies can provide context and evidence that can show the mechanisms underlying any causal claims made. However, such studies also come with potential constraints in terms of case selection, access to archival sources, and our ability to consistently compare outcomes across cases. Having the ability to test theories about national committees both qualitatively and quantitatively could alleviate these concerns and help deepen our understanding of the role these organizations play in American politics.
I present the first quantitative study of DNC and RNC party branding activity by relying on a new data set, based on a measure of committee activity external to the organizations themselves: newspaper coverage of DNC and RNC activity. I collected all articles in the New York Times published between January 1, 1953, and December 31, 2012, that mention either the DNC or RNC. I subsequently coded each article to distinguish specific types of activities by either committee. On the basis of these codes, I created a monthly metric of DNC and RNC activities across different categories of operations, and I use this data to test whether the DNC and RNC decrease their branding role when their party holds the White House. The results suggest that this is true: New York Times references to branding operations decrease by more than 40% when a party has an incumbent president in the White House. This result holds for both majority presidents (those whose parties hold the White House and simultaneously have majorities in House and Senate) and minority presidents (those whose parties do not have unified control of government). However, there is no such decline when it comes to other activities—including fundraising, organizing national conventions, and other traditional service operations.
National Committees as Political Institutions: Service Providers or Party Branders
The DNC and RNC were both founded in the middle of the nineteenth century and initially focused their attentions almost exclusively on the organization of their party’s national conventions. Over time, both committees expanded their role to include a number of services for politicians running for office, starting with the organization of national presidential election campaigns in the late nineteenth century (Klinghard Reference Klinghard2010). The traditional political science view of the DNC and RNC—starting with Cotter and Hennessy (Reference Cotter and Hennessy1964) and Cotter and Bibby (Reference Cotter and Bibby1980)Footnote 2 —has seen service provision as the principal role that national committees play in American politics. National committees assist in organizing campaigns and raise money for candidates but in this view, they are also perceived to be largely powerless. While the financial and campaign support services the committees provide hold real value to candidates, the parties provide them in relatively even levels to candidates in competitive races. But since the DNC and RNC do not select candidates and cannot force them to take on specific policy positions, the national committees are depicted as being essentially powerless within their parties. This perception remains common; David Karol has recently noted the “limited importance of the formal leadership of national, state, and local party chairs” (Karol Reference Karol, Jenkins and Volden2018, 143) in American political parties.
To be sure, national committees do provide services, lack the power to select candidates, and cannot force issue compliance on those candidates. However, a different literature on national party organizations in American politics has raised questions about the extent to which the DNC and RNC are indeed mere bystanders. This scholarship has shown that a myriad of influential political actors within the parties—including presidents, Congressional leaders, governors, other elected officials, and party activists—pay considerable attention to their national committees, and often actively compete for control of these institutions (Klinkner Reference Klinkner1994; Galvin Reference Galvin2010, Reference Galvin2020; Conley Reference Conley2013; Rosenfeld Reference Rosenfeld2018; Heersink Reference Heersink2018a, Reference Heersink2018b; Hilton Reference Hilton2019).
One particular reason why party leaders appear to value their national committees is that party leaders try to control their national committee to change their parties’ policy positions.Footnote 3 For example, Sam Rosenfeld (Reference Rosenfeld2018) in The Polarizers shows that some national committee chairs attempted to push their parties towards more homogeneous ideological policy positions. One particular example is DNC chair Paul Butler (1955–1960) who, after the 1956 election, attempted to use the national committee to turn the Democratic Party into a clearly liberal national party, and to (simultaneously) minimize the power and influence of Southern conservative Democrats. As I argue (2018a), party leaders think national committees are important because they believe their publicity helps shape a national party brand. This party brand concerns the basic understanding that voters have of what the party represents to them—including its policy positions, and an assessment of the general quality of the party (Stokes Reference Stokes1963; Grynaviski Reference Grynaviski2010; Butler and Powell Reference Butler and Powell2014). Parties construct this brand in a variety of ways. One component, valence, may come about through the quality of the party’s governance and the presence (or lack) of scandals. Voters’ understanding of the policy component can come from a party embracing specific positions—such as opposing abortion or supporting a higher minimum wage—or by more broadly appealing to specific voting groups. Importantly, such information can affect not just the group that is being targeted, but also other groups in society: that is, a party appealing to conservative Christians may also indirectly inform other groups, such as LGBT voters, about its priorities.
Brands are important to parties and candidates because they believe a brand that appeals to a majority of voters will result in electoral victory, while a brand that does not can lead to defeat. Thus, what the brand looks like is crucial. There is, however, no single actor or institution in either party that can single-handedly create such a brand. Rather, a number of these actors—including Congressional leaders, presidents, presidential candidates, and governors—provide voters with informational cues that, when combined, shape that brand. The national committees—as the sole representatives of the national party—also try to do so through the “educational campaigns” (Klinghard Reference Klinghard2010) they began organizing in the late nineteenth century. Through publicity activities the committees try to convince voters that they should support their party. The DNC and RNC—in the eyes of party leaders—provide voters with crucial informational cues on the party’s brand, and those who control the national committee can use this power to try and change the party brand in their preferred direction. If changes in the brand result in electoral victory, party leaders may see this as a success on the side of their national committee, but if a changed brand results in defeat they may blame the committee. For example, Senate Minority Leader Everett Dirksen (R-IL) concluded after the dramatic Republican losses in the 1964 election that “we failed to present a clear-cut image and sell it to the voters . . . It was the fault of those whose business it was to project the true Republican image. It was the national committee’s business and it flubbed the job” (Heersink Reference Heersink2018a, 94).
Case studies I presented in previous research (2018a) suggest that national committees do indeed engage aggressively in party branding activities. These cases show that committees invested considerably in publicity programs, took on a policy setting role within the party, or attacked the opposite party, with the goal of helping shape and adjust voters’ perception of their party’s brand. For example, during the Eisenhower administration the DNC created the Democratic Digest—a magazine available through subscription services and at newsstands—which presented the Democratic policy positions and attacked the GOP. Similarly, in the 1960s, the RNC created Comment—a weekly radio and TV show that featured Republican politicians discussing salient issues. These programs were sent free of charge to radio and TV stations across the country, which could either play them in full or use clips from the program to supplement their political coverage in news programs. National committees also invested in subdivisions focused on setting policy positions for their party. After the 1956 presidential defeat, the DNC created the Democratic Advisory Council (DAC), which was given the right to set party policies for the Democratic Party as a national institution (Klinkner Reference Klinkner1994). Under Butler, the DAC had a clear liberal slant and pushed the party to embrace pro-civil rights and union positions (Rosenfeld Reference Rosenfeld2018). After the dramatic defeat of Barry Goldwater in the 1964 presidential election, the RNC followed the same path and created the Republican Coordinating Committee (RCC), which attempted to lure voters back to the GOP by “moderating” the party’s image (Heersink Reference Heersink2018a).
National committees have engaged in similar projects since then. For example, during the first Nixon term, the DNC’s Democratic Policy Council (DPC) helped set the party’s Vietnam policies, which called for a complete cut of funding for the war and a withdrawal of American troops regardless of whether a peace agreement could be reached.Footnote 4 Under chairman Bill Brock (1977–1980), the RNC invested heavily in reaching out to Black voters—including inviting civil rights activist Jesse Jackson to an RNC meeting, using the Black-owned political firm Wright-McNeil to help train Republican candidates to appeal to Black voters, and organizing its 1980 national convention in Detroit (Wright Rigueur Reference Wright Rigueur2014). During the 1990s, the RNC invested in a new television studio at the RNC headquarters (at a cost of $1.7 million), and produced Rising Tide, a weekly TV show which ran on 2,000 cable systems. In 1999, the RNC created e.GOP, an internet access service which allowed subscribers to receive direct correspondence from the RNC and watch the 2000 Republican convention online.Footnote 5 More recently, under chairman Howard Dean, the DNC engaged in a “fifty state strategy” aimed at increasing the number of states the Democratic Party could compete in after the 2004 election. Part of this strategy was to downplay social issues and reach out to white evangelical voters. For example, in the 2006 midterms, the Democratic Party in Alabama used DNC support to produce advertisements calling for mandatory bible literacy courses in public schools, a constitutional amendment banning abortion, and expressing opposition to same-sex marriage (Sullivan Reference Sullivan2008). And after the 2012 election the RNC created the Growth and Opportunity Project, aimed at investigating the causes of the party’s electoral losses. The resulting report—commonly referred to as the RNC’s “autopsy report”—concluded that the party would need to reach out to non-white voters, support comprehensive immigration reform, and become more open on issues related to gay rights.Footnote 6 Thus, since the 1950s both committees have regularly engaged in publicity activities with the goal of adjusting their party’s brand and appeal to different voting groups.
White House Control and Decline in Branding Activity
But such branding activities are not constant. Indeed, there have been notable periods where national committees appear to have decreased such activities. For example, the DNC, under the control of John F. Kennedy and Lyndon Johnson, saw far fewer of such activities. Indeed, the DNC ended many of the branding activities initiated during the 1950s—including the DAC and Democratic Digest—after Kennedy’s 1960 election victory, while Johnson further slashed DNC activities during his time in office (Galvin Reference Galvin2010; Heersink Reference Heersink2018a). And under Jimmy Carter, the DNC declined dramatically as well, with major layoffs to DNC staff and a sense in the Carter team that the DNC (as well as state party organizations) were “potential antagonists to be neutralized” rather than “potential allies to be nurtured” (Price Reference Price1984, 78).
Why, and under what conditions, would a national committee decrease its branding role? One explanation for why national committees may be more or less active over time is whether their party has won the most recent presidential election. Klinkner argues that parties that have lost a presidential election are more active than those that won since out-parties have “a powerful motivation for changing their personnel, organizational structure, internal party procedures, and platforms since past methods have failed to achieve victory” (Klinkner Reference Klinkner1994, 1). To be sure, Galvin (Reference Galvin2010) shows that some presidents—specifically, Dwight Eisenhower, Richard Nixon, Gerald Ford, Ronald Reagan, George H.W. Bush, and Bill Clinton during his second term in office—also were party builders. That is, they invested time and energy in expanding their party’s national organizations. On the other hand, presidents like Kennedy, Johnson, Carter, and Clinton during his first term in office largely ignored their parties. Building on this work, I argue (2018a) that party branding activities in particular decline when a party is in the national majority—meaning, when it has unified control of government.
Several considerations appear to form the mechanism that results in parties’ national committees adjusting their branding activities when in the White House. Presidents, when in office, have near complete control over their party’s national committee (Galvin Reference Galvin2010, Reference Galvin2020; Heersink Reference Heersink2018b). Thus, a national committee’s activities at a time when its party has a president in the White House is effectively dependent upon what that president wants it to do. As Galvin argues, presidents may have incentives to want their party to improve its electoral performance if it is in the minority in Congress as “[they] want to make the system work for them, but more than that, they want to have a lasting impact on that system” (Galvin Reference Galvin2010, 19). Minority party presidents face limits to the kind of legislation they can pass when a contra-party majority holds Congress. Additionally, even if they do achieve successes, those achievements may not survive long after them if their party cannot build “durable political majorities” (Galvin Reference Galvin2010, 20). Majority party presidents may lack a direct incentive to make such investments. With their party in the majority in Congress, they may be less concerned with “helping” their party’s candidates. Additionally, with the expansion of the federal government, presidents can use their own position independent of the national party to promote themselves or their policy positions (Milkis Reference Milkis1993).Footnote 7 Altogether, this results in the general neglect of the party’s national committees under presidents like Kennedy, Johnson, and Carter. While neglecting the party might not necessarily have been a wise decision in the long term, from the perspective of these incumbent presidents it may have been rational to focus less time and resources on their parties. Or, as political advisor Hamilton Jordan, advised Jimmy Carter during his presidency, investing in a “DNC operation that requires a lot of time and supervision from here is more of a problem than a help” (Galvin Reference Galvin2010, 209).
But regarding continuing national committee branding activities, all presidents may be cautious. Presidents generally can be expected to want to legislate and govern, but a national committee does not have a clear role in this regard. More importantly, national publicity programs can be disruptive, as national committees may be emphasizing issues on which there is no consensus within the party. I argue (2018a) that because of this, if the party has unified control of government, national majority parties may be hesitant to “rock the boat.” For example, the DNC’s attempts at branding the Democratic Party as a liberal, pro-civil rights party after the 1956 election caused major conflicts between the committee and Southern Democrats—including Congressional leaders Sam Rayburn (D-TX) and Lyndon Johnson (D-TX). Similarly, the RNC’s decision in the early 1960s to prioritize its appeals to white Southern voters alienated moderate Republicans. A party that already is in the national majority might prefer to limit such controversy.
However, presidents—regardless of whether their party is in the majority or the minority—may have an additional reason for why they would be cautious to use their national committee for branding purposes, as they often have to rely on votes from the opposition party in the House and, more commonly, in the Senate to actually pass their legislation. Even if a topic or target of national committee branding activity is not controversial within their party, it could still antagonize the opposition, or frame current legislation in an entirely partisan way, thereby limiting the likelihood of legislative success. For example, historian Sean J. Savage argued that Johnson—who as president had large Democratic majorities in both House and Senate—slashed DNC activities because he “perceived a strong, national, party organization with regular publicity . . . activities emphasizing partisan differences to be a threat to the suprapartisan, centrist consensus that he wanted to develop for his presidency” (Reference Savage2004, 160). But this logic applies even more for minority party presidents. After all, such presidents, by definition, cannot pass legislation without support of the opposition party. We therefore might expect any incumbent of the White House to limit branding activities by their national committee in comparison to when a party is the out-party.
Methodological Hurdles in the Study of National Party Organizations
To be sure, I show (2018a) in a case study of the Nixon-Ford era (during which Republicans held the White House but were in the minority in House and Senate) that the RNC continued to engage in at least some branding operations—including through a publication called Monday, which was used to criticize Democratic opponents of Nixon, particularly in the runup to the 1972 election. But while the RNC might have engaged in branding activity, on the basis of such historical qualitative case studies alone it is difficult to make any clear comparison across cases in terms of quantity. This is a common problem in the study of national party organizations. While recent scholarship has presented a strong claim regarding the more influential role that the DNC and RNC play in American politics, each of the studies discussed earlier rely exclusively on qualitative historical case studies to test their theories.
This methodological approach has benefits but also potential limitations. A clear benefit is that these types of historical case studies provide context that not only shows what the DNC or RNC were doing at a given moment in time, but also why they were doing it. That is, by presenting in-depth historical case studies—based on correspondence between political actors, transcripts of meetings, newspaper accounts, or interviews with political actors about their activities—scholars can show the kind of arguments political actors were making at the time to explain their choices. This provides qualitative scholars with the ability to provide more specific evidence supporting their theories: that is, these studies can show not just whether certain outcomes occurred when a theory predicts they should, but also that political actors were actually using the theory’s underlying logic in making their decisions—something that is much harder to do through quantitative analysis alone.Footnote 8
However, qualitative research may also come with limitations. Scholars frequently are constrained in the number of cases they can present within any individual study.Footnote 9 Extending the analysis to additional cases can help, but often it is not possible to cover all cases or to cover them equally, especially regarding national party organizations, whose archival collections are limited in scope, and are spread out geographically. Thus, certain periods for specific parties are more easily studied, but other periods are harder to cover. Additionally, it is often difficult to compare cases to each other in a consistent manner. While we may know from archival sources that a party engaged in some types of activities during a specific period, we may lack a stable measure for comparison across different cases.
The obvious solution would be to test the theories presented in qualitative historical studies through quantitative analysis. However, the same issue with accessing archival resources from within the DNC and RNC that affects qualitative research has even bigger implications for attempts at creating quantitative assessments of these institutions. Neither national committee maintains an archive covering its full history that is open to scholars. While there are many collections that include some components of the DNC or RNC’s papers, such collections are often based on papers donated by a specific individual (usually a former national committee chair) to an institution of their choosing. Thus, for specific time periods and for a single party, scholars can find original documents in these collections. But scholars face major difficulties in their ability to produce a comprehensive data set for quantitative analysis covering both parties consistently from such records. We thus face a clear hurdle in studying American political parties in general, and the role of national party organizations in particular. While a number of studies have argued that these political institutions have played a major role in the party system, methodologically we are constrained in further testing through more diverse methodological approaches when and how the committees behave in this regard due to a lack of access to internal committee data.
New York Times Coverage of National Committee Activity: Data Collection and Coding
To help overcome this problem, I propose an alternative indirect approach to measuring national committee activity by relying on media reports. DNC and RNC activities are regularly reported in the news media, and since such reports are published consistently and are publicly available, they can be used to produce a consistent metric for both parties and throughout a lengthy time period. I use this data to test whether national committees of parties that hold the White House reduce their branding activities, and if so, whether there is any difference between majority and minority presidents in this regard. To do so, I collected New York Times reports of committee activities between January 1, 1953, and December 31, 2012. I use the New York Times because it is both the “paper of record” and is consistently available online throughout this period—as well as in years before—allowing for an extension of the project beyond the period presented here.
I collected each article published in this period that mentioned either the Democratic or Republican National Committee or its chairs using the ProQuest Historical Newspaper database. In total, this produced a collection of 21,202 articles.Footnote 10 As can be seen in figure 1, which shows the yearly number of articles collected for both the DNC and RNC between 1953 and 2012, there is clear variation across time and party in the amount of coverage each committee received. However, this in and of itself does not necessarily tell us much about the extent to which either committee was active, since the collection process inevitably includes a considerable number of irrelevant articles that mention the committees (or their chairs), but in a context that is not relevant to the study’s purposes. For example, the articles collected also include obituaries of employees who previously worked at either national committee, or articles that reference former DNC or RNC chairs. Additionally, there are specific news events that result in a dramatic increase in coverage that are not related to anything the committee is doing. Some of the clear spikes in figure 1 are due to New York Times coverage of issues related to the national committees, but not due to coverage of those committees’ actual activities. For example, a much higher amount of coverage than normal occurred for the RNC in 1964 and for the DNC in 1972–1974. In both cases, the cause of this increase—respectively the vice-presidential candidacy of former RNC chair Bill Miller on the 1964 Republican ticket, and coverage of the Watergate break-in at the Democratic National Committee and subsequent investigations—is not an accurate reflection of an increase in activities by the national committees themselves.
Total yearly New York Times articles mentioning the Democratic or Republican National Committee or its chairs, 1953–2012.
To address this issue, and to allow for analysis of the different types of committee activities across time, I coded each article to assess whether they reported on DNC or RNC activity within the year the article appeared, and what type of activity (or activities) was reported. If it did, I coded for any of the activities defined. If an article did not report on any of the activities, I dismissed the article as irrelevant. Specifically, I looked for three broad categories of operations—branding, service, and otherFootnote 11 —and a subset of more specific types of activities that make up these broader categories.
Regarding branding, and relying on my earlier work (2018a), I identified three specific types of activities:
Attack on the Opposite Party: Does the article report on the relevant national committee (as an institution or through its chair or other staff members, or by releasing statements on behalf of other political actors) criticizing politicians or policies of the opposite party?
Publicity Programs: Does the article report on the relevant national committee creating, investing resources, or continuing a publicity program (including, but not limited to, magazines, TV or radio shows, radio or TV broadcasts of speeches by politicians sponsored by the national committee, newsletters, advertising campaigns, etc.)?
Policy Position: Does the article report on the relevant national committee (as an institution or through its chair or other staff members) publicly taking a specific position on a policy issue (including support for policies by the administration of a president from the party) or participating in trying to set one in cooperation with other party leaders (for example through policy commissions like the Democratic Advisory Council or the Republican Coordinating Committee)?
These activities cover different ways of shaping a party brand, by identifying policy positions for the party, attacking the opposite party (what Lee describes as “nonideological appeals accusing the other party of corruption, failure, or incompetence” (2016, 2)) which can further help differentiate the party from the opposition, and investing in, and using, the tools they rely on to present voters with those policy positions or attacks.
In terms of service activities, building on Galvin’s (Reference Galvin2010) description of party activities of committees under presidential control, I identified the following four activities:
Campaign Service: Does the article report on the relevant national committee providing campaign support for individual candidates—including presidential candidates, candidates for Congress, gubernatorial candidates, etc.—or the party as a whole, such as providing candidates’ campaigns with money, opinion polls, training, strategic advice, organizing campaign appearances by the national committee chair or other party leaders, targeting voting groups, and mobilizing those groups through voter registration activities and Get Out The Vote drives?
Human or Capital Development: Does the article report on the relevant national committee training future candidates in campaign schools, hiring new staff members, or investing in its real estate or technology?
Recruitment: Does the article report on the relevant national committee engaging in candidate recruitment activities—that is, attempts by the national committee to convince potential candidates to run for office?
Fundraising: Does the article report on the relevant national committee engaging in fundraising activities—either on behalf of the committee itself or by having the national committee chair engage in fundraising activities on behalf of other party organizations or candidates?Footnote 12
Finally, to fully cover the spectrum of national committee activity, I also coded for several other activities that do not fit in either the branding or service category, but which the national committees do engage in. While I do not present analysis of these activities separately in this paper, articles that reference any of these activities are included in the data set. Specifically, I coded for the following:
Patronage: Does the article report on the relevant national committee engaging in the division of patronage—that is, positions within the federal government—by managing job applicants and discussing job candidates with the administration?
Organization of National Conventions: Does the article report on the relevant national committee engaging in organization activities for an upcoming or ongoing national convention - including the selection of the convention city, setting rules for delegate selection and distribution, and the actual execution of the national convention?
Generic: Does the article report on the relevant national committee engaging in any type of activity that does not fall in the outlined categories outlined—including, but not limited to, the national committee chair holding meetings without additional agenda information, the committee chair resigning, a new committee chair being appointed, the announcement of staff retirements, the committee chair or the national committee as an institution expressing condolences, the committee chair expressing basic support for candidates of the party running in general elections, the committee chair presenting basic political positions in the media that do not fall in the category of policy positioning or attacking the opposite party, the committee chair providing predictions of election results, the committee chair discussing previous electoral strategies, etc.?
In coding each newspaper article, I relied on a dichotomous approach: if an article mentioned one or multiple activities that fell into the same category, I coded that article as “1” for that category, and “0” if it did not. In total, 10,408 of the articles collected reported at least one (and often, multiple) of the committee activities listed here. As can be seen in figure 2, there is a notable difference in the total number of articles collected and those “relevant”—that is, those that mentioned one or more of the listed activities—which eliminates some of the more extreme variation across time and party. Still, as can be seen in figure 3, there does remain considerable variation in coverage if we look at just these relevant articles. This variation in New York Times coverage makes it possible to test claims about whether national committees change their behavior depending on particular historical developments, including their party’s electoral success.
Yearly total and relevant New York Times articles mentioning the Democratic or Republican National Committee or its chairs, 1953–2012.
Yearly relevant New York Times articles mentioning the Democratic or Republican National Committee or its chairs, 1953–2012.
In the remainder of this paper, the dependent variable is the number of monthly New York Times references to either specific activities or the combined number of activities within the broader category of operations.Footnote 13 To create this measure of committee activity, I added up the number of times articles referenced at least one of the activities. For example, if in April 1960 there were three articles that referenced the RNC engaging in fundraising and two articles referencing the committee engaging in campaign activities, the number of service references for that specific month and party is counted as five. Note that the same newspaper article can report on multiple activities, perhaps referring to a national committee holding a fundraiser and using the money raised to buy network time for a television special. If so, the article is coded for reporting both fundraising and publicity activities. The resulting data set has an N of 1,440 (that is, sixty years times twelve months for both parties).Footnote 14
It is important to note that relying on New York Times coverage of national committee activity as a measure of such activity represents an indirect metric with potential biases.Footnote 15 Fundamentally, the dependent variable is not the actual activity of national committees but news media coverage of it. To be sure, the DNC and RNC often engage in activities with the explicit goal of achieving news coverage to amplify their message. This is particularly true for branding activities—the core focus here. For example, when the DNC created the Democratic Digest in 1953 it intended it not only as a tool to reach those people who bought and read the magazine, but also as a way to encourage news media to cover the items included in the magazine as a form of “trickle-down public opinion.”Footnote 16 Thus, we might expect there to be strong correlation between a national committee increasing its publicity activities and New York Times coverage of such activities increasing as well.
Yet it is possible that news media may adjust their coverage regardless of whether the committee is in fact changing its activities. For example, it is possible that as an in-party focuses on legislative activity, reporters stop paying attention to its national committee, regardless of whether the committee is in fact decreasing its activities. As a basic assessment of whether coverage of the New York Times adjusts in line with changes in national committee behavior I present two descriptive tests in the online appendix. These tests are based on findings in Klinkner (Reference Klinkner1994), which note specific time periods in 1956–1992 when a national committee behaved notably different than it did in every other period. I find that, in line with Klinkner’s findings, there is a clear increase in New York Times coverage of DNC policy activity in the period 1957–1960 (during which Klinkner argues the DNC was uniquely focused on policy setting activity), and DNC procedural activity in the period 1969–1976 (during the period when the DNC engaged in major reorganization of its national conventions and presidential nomination processes). While this does not guarantee that New York Times coverage always increases or decreases in line with actual changes in committee activity, it suggests that the Times is responsive at least in these particular cases.Footnote 17
The main independent variables in this analysis are whether a party held the White House, and whether a president was a majority or minority president. In line with my earlier work (2018a), I rely on Goldman’s (Reference Goldman1990) definition of national majority parties as having
majority status in at least four places simultaneously: (1) the electoral college, derived from pluralities in a sufficient number of states, that is, the party-in-the-electorate; (2) the presidency; (3) the Senate; and (4) the House of Representatives.
(Goldman Reference Goldman1990, 569)Thus, a president with unified control of government will be considered a majority president, while a president whose party controls only one or zero houses of Congress, is considered a minority president.
I include multiple control variables that could plausibly also cause difference in reported committee activity across time and party. Both national committees help organize and execute election campaigns, and we might therefore expect increases in branding operations during presidential and midterm election years. Therefore, I include a dummy variable for whether articles were published in presidential election and midterm election years. Note that because national committees do not wait until the end of a calendar year to begin incorporating the effects of elections, I apply the effects immediately to my coding. That is, if a party wins or loses the White House or majority status in the House or Senate, I use the day after the election that determined this as the tipping point—not the later point in time at which the elected officials are sworn in. I also use this approach in coding for whether the activities described take place in an election year.Footnote 18
Additionally, it is possible that one party structurally receives more coverage than the other. To control for this, I include a dummy variable identifying whether the article concerns coverage of the DNC. It is also possible that the amount of space in the New York Times fluctuated across time. To control for this, I include a continuous variable of the number of articles published on the first weekday of the year that include the word “and” as a metric of the amount of space available in the paper—assuming that more articles found this way indicates that there was more space in the Times at that time. National committees can also find themselves facing major scandals concerning either their activities as an institution or concerning their chair. When these scandals occur, they might affect the committees’ ability to function or the likelihood that the New York Times decides to publish stories about any DNC or RNC activities unrelated to the scandal. To control for this, I include a measure of the number of monthly articles mentioning scandals. The coding process followed the same setup as described earlier, and I defined these as:
Scandal: Does the article report on the relevant national committee (or its chair or other staff members) being involved in a scandal (including, but not limited to, financial improprieties, criminal or Congressional investigations into alleged crimes, sexual scandals, etc.)?
Finally, in some of the models I include year and month fixed effects to control for any additional unobserved variation—both in terms of consistent increases or decreases in newspaper coverage or committee activity at set moments across the year, or in specific years in which coverage or activity was notably higher or lower because of other factors. In the models that include fixed effects, I leave out control variables that are consistent across the year—that is, whether the year was a presidential or midterm election year, and the measure of New York Times size—since these would perfectly correlate with the specific year fixed effect.
Results
To reiterate, the goal of this study is to assess whether national committees of parties holding the White House are less likely to engage in branding activities than those that are out of the White House, and whether we see a difference in that regard between majority and minority presidents.
To do this, I rely on a data set that covers the number of monthly references to different types of DNC and RNC activities as reported by the New York Times between January 1953 and December 2012. Since the dependent variable of this study concerns a count of monthly references to branding and service operations in the New York Times, I rely on a negative binomial regression for most of the analyses presented here.
The results of such a model are shown in table 1, demonstrating the effect of a party being in the White House on the expected log count of the number of monthly New York Times reports on national committee branding operations—that is, the combined number of references to a national committee attacking the opposite party, using existing or new publicity programs or presenting policy positions on issues. All models presented in this paper include a one-month lag on the DV to allow for the possibility that national committees may need some short period of time to respond to a change in their party’s majority or minority status.Footnote 19 I present four versions of the main model testing the branding theory: in Model 1 and 2, the IV is whether a party is in the White House or not. Model 3 and 4 include the type of president—majority or minority—as separate variables. Models 1 and 3 include the control variables discussed in the previous section, while Models 2 and 4 exclude some of those controls in favor of year and month fixed effects. In all models, the effect of a party being in the White House is negative and significant at the 0.001 level.Footnote 20 Regardless of whether a party has unified control of government or not, New York Times coverage of branding activity declines considerably when a national committee’s party holds the White House.
Negative binomial regression of presidential and party status on monthly (lagged) New York Times coverage of national committee branding operations, 1953–2012
† p < 0.10,
* p < 0.05,
** p < 0.01,
*** p < 0.001.
While a negative binomial model is appropriate given the nature of the dependent variable, interpreting the substantial meaning of its coefficients is complicated. To provide some easier to interpret assessment of the practical implication of the size of the decline in New York Times coverage of branding operations, I present the results of an OLS regression that uses the same variables as Model 3 in table 1. The results, presented in figure 4, suggest that monthly New York Times references to branding operations by majority presidents decline by –1.32, while for minority presidents the decline is –1.17. To put these numbers in context, the mean number of branding references by month for the period 1953–2012 was 2.97, so this effect represents, respectively, a 44% and 39% decrease in reported branding coverage.Footnote 21
Linear regression of majority and minority presidential status on monthly (lagged) New York Times coverage of national committee branding activity, 1953–2012.
The results presented in table 1 and figure 4 suggest that national committees of parties that hold the White House see a clear decrease in New York Times coverage of their branding operations. However, it is possible that these results indicate that the committee of such parties are actually decreasing all their activities—that is, branding operations as well as service operations. While this could still indicate a decline in national committee activity and influence, this would not be entirely in line with the branding theory. More problematically, it could also mean that the Times is decreasing coverage of all activities, not because committees are less active but because the Times is less interested in covering their ongoing activities once a party holds the White House.
To assess whether this is the case or not, I present an assessment of different types of operations national committees can engage in. Table 2 includes the results from Model 3, table 1, as well as similar models for all coded operations, and the combined service operations. The results show that parties in the White House see a decline in all reported activity but reports of service operations (the combined references to campaign activities, candidate recruitment, human and capital development, and fundraising) are not significantly lower for either majority or minority presidents.
Negative binomial regression of majority and minority presidential status on monthly (lagged) New York Times coverage of national committee operations, 1953–2012.
† p < 0.10,
* p < 0.05,
** p < 0.01,
*** p < 0.001.
To see whether any specific branding or service activities are more or less likely to be affected, I run the same model as presented in Model 3, table 1 for each individual activity coded for. The results, presented in figures 5 and 6, show that in terms of branding activities, parties in the White House are less likely to be reported to engage in attacks on the opposite party or to take policy positions. These are notable results since arguably these are relatively easy activities for a national committee to engage in: the actual cost of attacking the opposite party or of taking a position on a policy issue is very low and there is no financial or organizational reason why a national majority party would have to decrease its activity in this regard. But these types of branding activities are also those that might be most disruptive, and presidents may prefer to avoid them lest they interfere with their legislative activities. However, the effect on references to publicity is not statistically significant. This may be because publicity investments are not as controversial, though it is also possible that this might reflect an inherent flaw in the reliance on newspaper articles as a measure of ongoing committee activity. That is, the DNC or RNC might engage in ongoing publicity programs but the New York Times might not consistently report on them every time a new issue or episode is released. In contrast, if a national committee attacks the opposite party or takes a position on a policy issue this may be newsworthy each individual time it occurs.Footnote 22
Negative binomial regression of majority and minority presidential status on monthly (lagged) New York Times coverage of national committee branding—attacks, publicity, and policy positioning—activities, 1953–2012.
Negative binomial regression of majority and minority presidential status on monthly (lagged) New York Times coverage of national committee service—campaign, human or capital development, candidate recruitment, and fundraising—activities, 1953–2012.
In terms of service activities, as can be seen in figure 6, majority presidents are neither more nor less likely to be reported to engage in campaign activities, candidate recruitment, or fundraising. The one exception concerns human and capital development reports—activities whereby the national committee hires new staff members, trains candidates, or invests in its organizational structure. The effect here is negative and significant, suggesting parties in the White House are less likely to invest in their party’s organizational structure. The results for minority presidents are similar throughout, with one notable exception: the effect of a minority president on reports of national committee fundraising is positive and significant (at the 0.05 level), suggesting minority presidents in particular prioritize fundraising activities on behalf of their party—a finding that is in line with Galvin (Reference Galvin2010).
Conclusion
By relying on New York Times coverage of activities by both national party organizations in the period 1953–2012, I have circumvented the lack of consistent quantitative data covering national committee activity from within these organizations. I have used this new data set to test whether the DNC and RNC were less likely to engage in branding operations when their party was in the White House, and whether majority and minority presidents act differently in this regard. The results presented here suggest that New York Times coverage for national committee activities does indeed decrease when a party has an incumbent president, regardless of whether that party also holds majorities in Congress. But crucially, this is not true for all types of operations. While Times coverage of branding operations is notably lower, coverage of service operations is not. In particular, there is a clear (and statistically significant) decline in New York Times coverage of in-party national committees attempting to create a party brand by attacking the opposite party, and by stating policy positions when their party has unified control of government. There is no similar decline in coverage of the committees engaging in campaign service or recruitment of candidates, and committees of minority presidents see an increase in coverage of fundraising activities.
I contribute to the ongoing assessment of the role of national party organizations in American politics and add to the argument that the DNC and RNC, as the sole national institutions within each party, try to shape a national party brand. However, the extent to which they fulfill this role is dependent upon the electoral context in which the party finds itself: while a party out of the White House has clear incentives to engage in branding activities to improve its performance in future elections, an in-party will be less inclined to do so due to a lack of interest of its incumbent president or due to the potential costs of branding for intra-party unity and legislative effectiveness.
This study is the first to use a quantitative, though indirect, measure of DNC and RNC activities across a substantial period of time—suggesting that future research on American political parties as organizations can rely on quantitative analysis, even if data from within the national committees remains difficult to collect. To be sure, the method proposed here has limitations. The metric does not directly measure national committee activity itself, but newspaper reports of such activities. And not all committee activities are likely to receive similar amounts of newspaper coverage. For example, candidate recruitment—a core party-building activity identified by Galvin (Reference Galvin2010)—likely occurs mostly behind the scenes, and thus means there may be little or no press attention to these types of activities.Footnote 23 However, campaign activities, fundraising, attacks on the opposite party, and promotion of policy positions on behalf of the party are all public activities, are newsworthy, and are often explicitly designed to attract media attention. Additionally, the coding process is dichotomous—that is, an article either mentions a type of activity or it does not—making it a somewhat blunt tool to measure reported activity. Finally, the metric presents a measure of quantity of coverage, not of the actual content or importance of the branding activities, and the analysis presented provides a comparison of the average level of activity across different types of presidents and parties. While the results show a decline in coverage of branding activity for parties in the White House, branding never ceases entirely. What the remaining activities looked like, with what purpose they were used, and how influential they were remains a relevant question worthy of historical qualitative assessment.
While the metric thus has clear limitations, it can nonetheless help provide insight into national committee behavior, in line with Schickler’s argument that historical social science research should “draw on diverse types of evidence and methodological approaches in order to gain insight into a question that is not ideally suited to isolating the causal effect of a single variable” (Schickler Reference Schickler2016, 17). The approach is also promising for future research because newspaper coverage can open up the study of party activity for time periods in which archival collections are not currently available. This is true for the period prior to 1953, which—with notable exceptions (Goldman Reference Goldman1990; Klinghard Reference Klinghard2010; Galvin Reference Galvin2012)—has been largely ignored in the recent wave of scholarship on American political parties as independent organizations. But it also applies to more recent periods for which archival sources may become available over time, but that currently have not yet been opened to the public.
This is particularly relevant given the changing political context in which the national committees have been active in recent years. Traditionally, the two parties often found themselves in a position where one was dominant and the other not for a substantial period of time. But as Frances Lee (Reference Lee2016) notes, since 1980 it has become much rarer for a party to hold on to a majority in Congress. To be sure, there are exceptions, such as the four-year period between 2002–2006 when Republicans had unified control of government. But the Clinton, Obama, and Trump administrations all saw just two years of unified control of government. And while Democrats had unified control of government at the start of the Biden administration, their majorities in House and Senate are very slim and may not survive the 2022 midterms. In Lee’s assessment, the lack of a consistent majority party has produced a perpetual campaign mentality and more confrontational party politics in Congress. It is possible that presidents and national committees are adjusting their behavior similarly and that the potential negative effect of branding on legislative effectiveness is less of a concern now.
At the same time, national committees are also functioning in a much different media and communication landscape than before. In the bulk of the period covered here, committees had a position of power within their parties due to their ability to produce publicity in a way that individual political actors could not, or not as easily: reaching a large audience required publicity tools that were expensive to access or difficult to create. National committees had the resources to make these investments, and their ability to produce magazines, radio and TV broadcasts, and other types of publicity provided them with a level of control over what policies they could focus on, which politicians they could ask to appear in the party’s publicity, and what groups to target. But the rise of ideologically oriented cable news networks, conservative talk radio, and most of all, the internet, has dramatically lowered barriers to such branding activity. While both committees have tried to keep up, it is likely that in recent years it has become harder for any individual party institution to break through to voters as had been possible in the past. The combination of these changes presents major new questions, not just about how the DNC and RNC may have influenced the creation of the modern, polarized American party system, but also how they are functioning within that system now.
Acknowledgements
Heersink thanks Julia Azari, Heath Brown, Michael Hankinson, Alexander Hertel-Fernandez, Robert Hume, Jeffery Jenkins, Katherine Krimmel, Matthew Lacombe, Sidney Milkis, Michael Miller, Jordan Carr Peterson, Richard Valelly, Yamil Velez, and Christina Wolbrecht for their feedback. He is particularly grateful to Jeffrey Cohen, Daniel Galvin, Brenton Peterson, and the anonymous reviewers and editor for their extensive comments, and to Paul Alongi, Jr. and Christian Decker for their help as research assistants.
Supplementary Materials
Data Collection and Coding
Descriptive Assessment of New York Times Articles as Measure of National Committee Activity
Main Models, Robustness Checks, and Alternative Analyses
To view supplementary material for this article, please visit http://doi.org/10.1017/S1537592721000025.
