A Preliminary Report on Open Seat House Nominations in 2014

PARTIES ON THE GROUND: A PRELIMINARY REPORT ON OPEN SEAT HOUSE NOMINATIONS IN 2014

by Kathleen Bawn Knox Brown Angela Ocampo Shawn Patterson John Ray John Zaller

UCLA

August 2014

The purpose of this paper is to organize and take stock of what we have learned in our initial interviews, with a view to asking better questions as we proceed to the main phase of the study. Please check with us before citing.

Do not post this paper on the Internet. We have promised our sources that nothing they told us will be publically reported before 2016.

1 Republican infighting in the run-up to the 2014 congressional primaries attracted unusual attention.1 The national Chamber of Commerce and the National Retail Association vowed to primary Republican members of Congress who threatened default on the national debt. Ideological groups like FreedomWorks, Tea Party Patriots, and Club for Growth supported these same House members and have tried to nominate more like them.

The Republicans’ current conflict over nominations is not unprecedented. Similar struggles occurred when evangelicals became active in Republican primaries in the 1990s (Cohen 2005) and when the Tea Party first contested nominations in 2010 (Zernike 2010, Skocpol and Williams 2012). Democrats, too, have their intra-party struggles: the movement into the party of civil rights liberals in the 1940s and 50s (Baylor 2013), of feminists, greens and anti-Vietnam War radicals in the 1970s (Kirkpatrick 1976; Carsey and Layman 2010), and most recently of LGBT advocates.2 Changes in the commitment of the party’s nominated candidates, often little noticed as they occur, moved the Democratic Party to the left between 1940 and 1970 and the Republican Party to the right in the 1990s, and they continue pushing both parties toward their extremes. One can see the roots of the contemporary polarization of the parties in the candidates that parties have nominated for federal office in recent decades.

No one doubts the importance of the battle between Democrats and Republicans to control Congress in general elections. But the intra-party contests leading up to primaries are, over the long run, equally consequential. The process of nominating candidates is the process that defines what a party stands for.

Yet congressional nominations are beneath the radar for most voters and even most political scientists. Hundreds of scholarly articles and scores of books have been written about general elections for the Senate and House of Representatives; much less about party nominations for Congress. Scholars have demonstrated how political ambition leads candidates to enter competition for political office (Fowler and McClure 1989; Lawless 2012.) Other work has focused on how the openness of the primary rules affects voting by the winning politicians in Congress or state legislatures (Gerber and Morton 1998, Burden 2001, Hirano et al. 2010, McGhee et al. 2013, Hill 2013.)

Our study aspires to a broader understanding of the nomination process: how candidates, groups and voters interact to determine whose interests and values get represented in the party’s chosen candidate. Understanding nominations will help

1 See, for example http://www.bloomberg.com/news/2014-08-06/republican-civil-war-costs-135- million-senate-odds-up.html, http://abcnews.go.com/Politics/republicans-tea-party-activists-full- scale-civil-war/story?id=21194296, and http://www.politico.com/story/2014/03/colorado-elections-2014-cory-gardner-mark-udall- 105082.html 2 These studies focus on Democratic activists in presidential nominations, but we suspect the same forces were at work in congressional elections.

2 us evaluate two competing theories of how the American party system works – one based on the incentives of ambitious politicians, another in which organized groups are the power players, politicians merely their agents. We focus on nominations for open seats because they are the most common way in which new members enter Congress (Gaddie and Bullock 2000.)3

House nominations are not easy to study. They take place at scattered times and locations; irregular district boundaries and sparse voter turnout make them difficult to cover in surveys; key influences like endorsements and informal support networks may leave little or no trace in public records; and nominations are barely reported on by the news media.

Given these challenges, our method is eclectic and relatively unstructured. We have made field trips to 32 districts in which nomination contests are underway, conducting open-ended interviews with leading participants and local observers. Our interview subjects range from local journalists and bloggers; party officials; activists; consultants; representatives of business, labor and advocacy groups; major donors and other people considered knowledgeable or influential. We note any activity by national groups such as Club for Growth, American Crossroads, EMILY’s List, and many others. Our goal is a broad understanding of the local political system in which each nomination takes place. The intention of this unstructured design is to detect behavior consistent with either the candidate-centered activity envisioned by most of the literature, or the more group-centric models proposed by Cohen (2005), Cohen et al. (2008), Masket (2011) and Bawn et al. (2012), and to be open to findings not predicted by either.

Our project is very much a work in progress and our findings as presented here should be regarded as preliminary. This report consists of four main sections. We first review the politician- and group-centered theories of political parties, noting implications for nomination contests. We then present background data on open seat House nominations in general, and the 2014 cycle in particular. The third section chronicles two nomination contests (one Democratic, one Republican) associated with special elections that occurred in the fall of 2013. The fourth section presents a semi-systematic catalog of “candidate assets” -- characteristics that are seen as advantageous to candidates in nomination contests across the country. Some of our preliminary conclusions are clearly in line with conventional wisdom: i.e., money is extremely important, as is candidate ambition. Our clearest finding, however, supports the group-based theory: strong candidates typically have longstanding and deep relationships with groups who regard them as “champions.”

PARTIES AND NOMINATIONS The nomination of candidates is often seen as a defining act of parties, but for almost all US House seats, voters make these nominations through primary

3 See Boatwright (2014) for a study of contested nominations involving incumbents.

3 elections, and have done so since the Progressive era. Rules vary across states, but the main pattern is that voters may participate in either the Democratic or Republican primary, and the candidate who gets the most votes in a party’s primary is that party’s nominee. One significant departure from this pattern is in the South, where most states have a runoff primary if no candidate gets 50 percent of the initial vote.4 A more significant departure is what is sometimes called a top-two primary, in which candidates of both (or all) parties run on a single primary ballot, with the top two finishers moving on to a runoff which serves as the general election. Louisiana has used variants of this system since the 1970s5, Washington state adopted it in 2004,6 and California in 2012. These states together contain 69 House seats, a non-negligible fraction (16%) of the total. For most scholars, the voters who participate in the primaries are not “the party.” That term is reserved for the set of elite actors who cooperate to influence and if possible determine the winner of the contest for primary votes. Who exactly these elite actors are and how they operate is a subject of contention. A dominant tradition, ably explicated in John Aldrich’s Why Parties?, focuses on politicians, activists, and formal party institutions. An alternative theory of parties, focusing on policy-demanding interest groups and activists, has been advanced by Bawn, et al (2012). Empirical studies in the dominant school assume that, for purposes of studying House nominations, the party consists of certain of its formal institutions: the congressional campaign committees of Democratic and Republican office holders in the House and Senate, and also state and county party committees in the 50 states, which consist mainly of activists. All agree that these entities have limited influence over House nominations. From this point scholars move quickly to the conclusion that House nominations are “candidate-centered.” As Gary Jacobson writes in the 8th edition of his classic textbook, The local party organization’s influence on congressional nominations varies but is typically feeble . . . National party leaders control more resources . . . but they use them sparingly in primaries and are regularly thwarted when they do. Usually, the nomination is not something to be awarded by the party but rather a prize to be fought over (when it seems worth taking) by freebooting political entrepreneurs (2013: 24). In the 6th edition of a second major textbook on congressional elections, Paul Herrnson takes much the same view:

4 http://www.ncsl.org/research/elections-and-campaigns/primary-runoffs.aspx 5 In Louisiana, the “primary” occurs on the date of the November general election, with a run-off held in December only if no candidate gets 50%. In CA and WA, the primary is held several months before, and the top two vote-getters compete on the November ballot regardless of primary vote share. 6 http://www.sos.wa.gov/_assets/elections/HistoryofWashingtonStatePrimarySystems.pdf

4 Most successful candidates are self-starters because the electoral system lacks a tightly controlled party-recruitment process… Because the system is candidate-centered, the desire, skills, and resources that candidates bring to the table in the electoral arena are the most important criteria separating serious candidates from those who have little chance of getting elected” (2012, p. 41). The “desire, skills, and resources” that candidates need to succeed loom very large in the literature on congressional elections, but they are typically measured with single dichotomous item intended to capture the “quality” of the candidate. That item is whether the candidate has held prior electoral office (Jacobson and Kernell, 1983; Jacobson, 1989; Cox and Katz, 1996). The elements of quality captured in this indicator are understood to be campaign skill, fund-raising ability, and the public support resulting from prior office (see especially Squire 1995; Squire and Smith, 1996; Jacobson, 2012, p. 50). The strength of a candidate’s ties to important interest or activist groups is notably absent from this simple measure. Krasno and Green (1988) develop a notably more complicated measure of candidate quality, which does give one point (out of seven possible) to candidates who have not held prior office but who meet the criterion of “political activist, party connection, political aide.” But this is a very limited step toward incorporating group relationships into the concept of candidate quality. Gaddie and Bullock (2000) showed that political experience, the simple proxy for candidate quality, is indeed associated with nomination success in open House seats. Other systematic research in this area has focused on factors that motivate the emergence of quality candidates in a nomination environment in which formal party institutions play little role. The personal qualities that stoke “political ambition,” the intense desire to hold elective office, have been documented (Fowler and McClure 1989). Gender has been a particular focus in these studies (Fulton, Maestas, Maisel and Stone 2006; Lawless and Fox 2010; Lawless 2012). Scholars also examine the strategic factors, such as a newly drawn district or national conditions that favor one’s party, that draw ambitious politicians into the fray (Stone and Maisel 2003, Maestas, Fulton, Maisel and Stone 2006.) These studies say little about voters and issues in nominations. But in Why Parties? (1995, 2011), Aldrich fills in this part of the argument. Candidates, as he argues, must obtain campaign resources from activists and donors who tend to be ideologically extreme, and they must obtain votes from primary and especially general election voters who are more centrist. This gives rise to the expectation that candidates will locate themselves along a right-left dimension so as best to achieve the ultimate goal of winning both the primary and the general election (Aldrich, 1995, 190; see also Groseclose, 2001). Aldrich treats the policy commitments of politicians as sincere. But as ambitious office seekers, their highest commitment is to their careers. Hence if changes in

5 political conditions warrant, they may adopt new positions in response. “Politicians are just as autonomous in office as they are in elections.”7 Political ambition thus can be seen as the driver of the candidate-centered process of nomination and election. Ambition motivates quality candidates to detect electoral opportunity, to take the optimal ideological location for winning office, and to make the heroic effort needed to raise money and mount a strong campaign. The alternative approach to party nominations begins by viewing parties not as formal campaign committees, but as coalitions of policy-demanding groups. These groups -- which may be as varied as farmers, “downtown business interests,” Tea Party Patriots, EMILY’s List, the Association of Federal State County and Municipal Employees, etc. – work within a preferred party to nominate candidates favorable to their policy demands. The first loyalty of a group is not to the party coalition, but to its own policy goals. Hence, groups may cooperate or compete within one another within parties, depending on the strategic situation. In competitive elections, especially for top offices like the Presidency but also sometimes for competitive House seats, groups may calculate that they do best by supporting the nominee most likely to unify the coalition and thereby prevail in the fall election.8 In less competitive elections, especially safe House seats, groups aim to maximize their particular interest. In particular, each group seeks an agent who, if elected, will proactively pursue its policy demands with the kind of effort documented by Hall’s (1996) study of participation in congressional committees. In a word that has come up spontaneously in many of our interviews, each group wants to nominate a “champion” of its particular policies. Champions embody a combination of skill, ambition, and deep commitment, a combination rare enough that not every group can find a champion in every nomination contest. Indeed, groups often cannot. When no suitable candidate has entered a race, groups either stay on the sidelines or make only token contributions. Groups may rely on politicians to coordinate their activities in nominations, as when they channel the money of their members to party campaign committees; in so doing, they delegate to politicians some independent power over party affairs. But more often policy-demanding groups prefer either to support their preferred nominees through a group association or PAC, or to decide individually which particular candidate to support. The parameters governing the costs and benefits of group participation differ greatly between primary and general elections. In general elections, the opposing candidate is from the other party and hence likely to be repellent in multiple ways; groups therefore have reason to donate to candidates from their side who are less than full champions. In seats safe for one’s party, there is even more reason to

7 Why Parties? p. 308, 2011. 8 Cohen et. al. (2008) develops this line of reasoning in the context of Presidential elections.

6 donate: the candidate is quite likely to get into office where she will be in a position to do favors -- even if not the favors expected from a champion. In primaries, by contrast, any of several candidates may be acceptable to most party donors. The donors may then have little incentive to make any contribution at all; they can simply save their money for the general election, when stakes are higher. But the calculus changes if one of the candidates for nominations makes a credible offer of special benefit to a particular interest or factional group. This might be a pro-life or pro-choice activist soliciting support from churches or women’s groups, or an equity investor or public housing advocate seeking support from banks or civil rights groups. The groups in these examples would likely see more value to their causes in these sorts of committed candidates than in generic partisans, and gauge their contributions accordingly. Group champions would then enjoy a fundraising advantage over generalists at the primary stage. Some of the interests that play in nominations may have peak organizations capable of large donations. Examples are the U.S. Chamber of Commerce, EMILY’s List, and the Senate Conservatives Fund, for whom independent expenditures of $200,000 to a single candidate are not unusual. But the large majority of donors in party nominations are individuals whose contributions are limited to $2,600. How does the logic of the group-centric model apply to this typical spending pattern? Imagine a district that has two large interests, agribusiness and energy extraction, plus numerous smaller ones, none of which has a peak organization that makes contributions for the whole group. A state legislator who has specialized in agriculture issues, and who is perhaps a farmer besides, will regard farmers in the district, state, and even region as a fundraising base: a set of individuals who want roughly the same things from government that the politician/farmer does, and who are often friends or colleagues. Agribusiness donors, for their part, will regard the politician as someone who knows their needs and can be trusted to fight for them in Washington. They will give and encourage their friends and colleagues to give, a process called “bundling.” It is common for individuals to get 10 or more friends to contribute, which makes for a bundle of about $25,000. The candidate will, of course, also seek donations from non-farm interests, but she will have fewer and weaker personal ties to these potential donors, and will be able to make less persuasive arguments. The non-farm interests, for their part, will have less reason to donate to a politician they know less well and who has no particular knowledge of or commitment to their needs. They may contribute some money, but they are less likely to solicit “bundled” contributions from their friends. Fundraising from non-farm interests will therefore be slow going and inefficient. And what of the energy interests in this imaginary district? If another candidate in the race is as closely tied to energy as the above-imagined one is to agriculture, they would support their champion in the same way as agribusiness. But many potentially important groups do not have champions in most races. In the absence of

7 a candidate clearly tied to their interests, the energy business will likely try to stay on the good side of all viable candidates, refraining from deep involvement. The point, then, is that when contributions are not managed by peak organizations, politicians can still be clearly and accurately identified as “group candidates.” The logic of voter turnout favors group champions as well. Partisanship is a strong motive to participation in general elections, but most voters stay on the sidelines in primaries. Membership in a group – blacks, evangelicals, farmers, teachers, union members – may, however, motivate a decision to turn out in a primary for a champion of that group. Likewise for gun owners, veterans, environmentalists or other groups in the category Philip Converse called “issue publics,” that is, individuals with a strong interest in some particular issue. Even a small number of primary voters who turnout to support a group champion can make a difference in a low turn-out, plurality election with a crowded field of candidates. For both groups and voters, then, the logic of participation in a nomination favors candidates who specialize in service to a particular part of the coalition. Generalist politicians, relying on political skill more than deep political commitment, could certainly succeed in winning some nominations. But in open nominations for seats safe for one party – the most important entry point to Congress – champions will normally have the advantage. The group-based theory views policy demanders and politicians as being in a principal-agent relationship, focusing on how principals select, monitor and control the agents they nominate.9 This problem is severe for two reasons. First the work of politicians in the legislature is extremely difficult to monitor from the outside. Legislators’ most consequential actions are not their public votes, but behind-the- scenes committee work and coalition building (Hall 1996), both of which can be hard for policy-demanding principals to observe. Worse yet, in cases when policy demanders do become aware of faithless representation, the task of punishing a sitting member of Congress is daunting. Electoral challenges to incumbent members of Congress are extremely costly, with odds in the incumbent’s favor. Politicians are typically at their weakest, and policy demanding groups at their strongest, at the nomination stage of an open seat contest.10 Politicians do not yet enjoy the powers of the office to which they aspire, and their previous office, if any,

9 Candidate-centered theories also treat politicians as agents, but typically as agent of voters (Bianco 1994, Schwartz 1994.) Although the information asymmetry that creates the agency problem is arguably more severe when voters are cast as principals, the tone of this work is often upbeat, stressing the ability of voter/principals to learn from imperfect cues (Popkin 1991.) 10 Delegating politicians great power over nominations could result in the creation of an independent political class that would serve its own interests rather than those of the policy demanding groups. What Mayhew (1987) calls Traditional Party Organizations have achieved this status at certain times and places in American history, but they have always been exceptions rather than the rule in American politics, and they barely if at all exist today. In particular, Mayhew estimates that these organizations may, in their late 19th century heyday, have governed 40 percent of the U.S.

8 is rarely sufficient to generate resources necessary for a House nomination. As principals, groups have very limited ability to sanction moral hazard; their best recourse is to limit adverse selection at the nomination stage. That is, they should seek politicians who have demonstrated -- by past service, costly action, or actual group membership -- that they have internalized the group’s policy demands and can be trusted in the absence of monitoring and sanctions. This kind of politician is precisely the king of “champion” that groups most want to support. The importance of group support in nomination contests means that champions are systematically advantaged as candidates. This is the most theoretically distinctive and normatively important claim of the alternative theory of party: open House seat nominations tend to go to politicians who are most deeply committed to particular organized interests and groups within the broad constituencies they are nominally elected to represent.

Where’s the party?

Readers may wonder whether the parties theorized in the group-based theory really are parties at all. They bear little resemblance to the party machines of 19th and early 20th century America. They do not dispense patronage, or organize campaigns; most of the time, they do not formally recruit candidates. In nearly every detail, the parties of the group-based theory differ from traditional parties. But they share one theoretically central feature: they are heterogeneous coalitions that vote and act together across a series of social choice decisions.

The degree of cooperation, as our account emphasizes, varies by situation. In open seat primaries for safe House seats, groups tend to act in their own best interests, sometimes risking the coalition’s interest for the sake of theirs. But where no group will get much unless all groups pull together – as is the case in nominations for competitive offices and for control of legislatures – cooperation emerges and tends to remain stable over multiple candidates and issues and over long periods of time. A leading school of political science refers to this cooperation as a “long coalition” and views it as the defining feature of political parties (Schwartz, 1989). The long coalition idea was initially developed as a model of parties in the legislature, stable voting blocs formed by re-election-seeking legislators to offset the chaos of unstructured majority rule (Schwartz 1989, Aldrich 1995; see also Cox 1987). But the long coalition logic applies as well to groups seeking to nominate and elect champions (Bawn and Rosenbluth 2006, Cohen et al. 2008, Bawn et al. 2012).

Although the phrase “long coalition” was coined relatively recently and is associated with models grounded in social choice theory, it coheres with a long empirical tradition that views parties as coalitions of diverse actors, each wanting something from government, all working together to nominate and elect candidates who will work on their behalf (Schattschneider 1942).

9 Mid-20th century political scientists did not doubt that groups played a central role in the life of the political parties. But with the rise of candidate-centered scholarship, groups lost their centrality as scholars pushed them off into a subfield of their own. Bawn et al.’s theory returns groups to their former centrality, asserting that groups and their policy demands are critical to understanding politics at every level.

“Group,” like “party,” is also to be construed broadly. The concept of group includes traditional organized interest groups, but also communities of ideological activists, ethnic organizations, and religious groups. Groups may be national or local, formal or informal, large or small. They may be concerned about material benefit, rights, or simply the moral state of society. The essential feature of the group is that its members have solved a collective action problem to demand from government: enactment of policies, representation of points of view, articulation of values. Everything else is a matter of hypothesis, the most important of which is: groups enter politics to support the nomination of candidates who will serve as reliable agents.

The candidate-centered approach to nominations has tended to study the impact of groups separately from that of party. This is not unreasonable, if we can confidently assume that they work through independent paths. But it is virtually a truism that the Republican Party is a coalition of big and small business, evangelicals, gun advocates, and assorted ideological groups (libertarian, Tea Party, traditional conservative), and that the Democratic Party is a coalition of labor, greens, several kinds of civil rights groups, and varied liberal activists. Studying groups outside the context of a party coalition can lead us to underestimate the leverage gained from seeking policy benefits through membership in a long coalition.

Empirical analysis The opposing theories of parties and nominations sometimes seem rooted less in facts than in interpretation. Example: When a dentist-legislator successfully solicits a donation from the dental association, the candidate-centered view will tend to see it as a politician taking advantage of a personal network, whereas the alternative sees it as a principal using a credible cue (the legislator’s profession) to secure a reliable agent. These perspectives are not mutually exclusive, but they can still be evaluated in terms of which set of factors -- candidate skill and ambition, or group support -- is more central to understanding nominations. Specifically:

• The standard view holds that candidates succeed by taking optimal ideological locations along a single right-left dimension; the group theory focuses on taking positions optimal for appealing to particular groups, each concerned with its own issues. • The standard view plays up candidates’ skills in electoral politics and stops there; the group-centric view holds that groups want candidates who can not only win elections, but effectively represent them in the legislature and bureaucracy

10 • The standard view acknowledges the importance of policy demanding groups, but does not incorporate their strategies and goals. Groups are simply money spigots for skilled candidates to tap. In the alternative view, groups are strategic actors who organize themselves, give and withhold support as it suits their interests, and sometimes even recruit and train their own candidates. The most basic difference between the theories is where power lies in nominations and hence in the party system. The standard view treats politicians as the key players, winning office on the basis of ambition, skill, and self-location in ideological space. The alternative view casts policy-demanding groups as the power players. Power is a famously slippery concept to observe, but by studying the nomination environment in its totality, we can offer some relevant facts.

AN OVERVIEW OF HOUSE NOMINATIONS

We now turn to some basic features of House nominations. We begin with some evidence that open seat House nominations are, in the phrase of Gaddie and Bullock (2000) “where the action is.” We then present a summary of the open seats in the 2014 election cycle, and make some preliminary observations about the size candidate fields, the role of money, and turnout.

Table 1: Partisan Competitiveness of House Seats, 2014

Safe Democratic (PVI > 5) 148 34% Competitive Democratic (PVI < 5) 36 8% Even (PVI=0) 9 2% Competitive Republican (PVI < 5) 60 14% Safe Republican (PVI > 5) 182 42%

Table 1 shows that in most House seats, district partisanship is strong enough that winning a primary essentially guarantees a win in November. We classify seats as safe or competitive based on Cook’s Partisan Voting Index (PVI), which compares the district’s presidential vote in the last two elections to the national average. A PVI of D+5, for example, indicates that the district averaged 5 points more Democratic than the nation as a whole in the last two presidential elections. 11 The point of Table

11 The disadvantaged party occasionally does win a seat in which the other party is favored by more than 5 on the PVI, but this is rare. There are 5 such cases in the current House: GA-12, MN-7, NC-7, UT-4 and WV-3. All are Democratic MC’s in Republican districts. In two cases, the partisan-oddball incumbent has is retiringannounced their retirement (McIntyre of NC-7 and Matheson of UT-4) and

11 1 is that 76% of House races are quite safe for one party or the other, safe enough to be decided at the nomination stage.

Table 2: Method of First Election to the House, Members of the 112th Congress

Open Seat, standard election timing12 58% Open Seat, special election 14% Defeated incumbent of other party 23% Defeated incumbent of own party13 5%

100%

Table 2 shows how House members of the 112th Congress were first elected to the body. As can be seen, 72 (58 + 14) percent of current members of Congress (MC’s) entered the House by winning an open seat either in a regularly scheduled election or a special election. An additional 23 percent entered the House by defeating an incumbent of the opposite party, 5 percent by defeating an incumbent of their own party. The current fraction of MC’s elected in open seats is almost identical to the 71.8% Gaddie and Bullock (2000) found over the period 1983-1995. Most members of Congress achieve that status by winning an open seat election.

Table 3 lists the open seats for the 2014 electoral cycle, including special elections held in the last quarter of 2013 or later.16 Six are nominations for special elections. Our observation that House nominations fly below the radar of political science holds even more for special elections.17 Yet, as noted above, a non-trivial fraction of House members first gain their seats via special election.

the winners of the Republican nomination (David Rouzer and Mia Love) are heavily favored to win the general election.d the odds of the Democrats keeping the seat are considered very low. 12 For members who served in the House, left and returned, we count the first election in their current sojourn in the House. For a member who served from 2000 to 2004, lost in 2006, but returned to the House in 2010, we based our classification on how they won their seat in 2010. If we based classifications on the very first election to the House without regard to breaks (2000 in the example), we would have an even higher fraction first elected in an open seat. 13 In most cases, “defeat incumbent of other party” means that the candidate defeated an incumbent in the general election and “defeat incumbent of own party” means the candidate defeated an incumbent in a primary. However, in states which use “jungle primaries” -- Louisiana, Washington and now California -- other outcomes are possible. Hence the more general language. Note that California alone accounts for 12% of House seats, so these non-partisan primaries are no longer . 16 Technically, TX-4, VA-7 and MI-11 are also open seats, but this is because the incumbent lost a primary challenge. Our focus is on truly open nomination contests in which the incumbent does not run in the primary. 17 For example, Anzia (2014) leaves special elections out of her study of election timing and turnout. Her general finding, however, that organized interests are advantaged in elections that take place at times other than November of even numbered years, seems quite likely to apply to special elections as well. We do note, however, that sometimes special elections do catch the interest of news media. For example, the November 2013 AL-1 race in which Bradley Byrne defeated Dean Young was covered by the New York Times and NPR.

12 Table 3: Open Seat Overview

District Location Incumbent PVI Reason for vacancy AL-6 Birmingham suburbs Bachus (R) R+28 Retiring TX-36 Houston suburbs Stockman (R) R+25 Running for Senate LA-6 Baton Rouge Cassidy (R) R+21 Running for Senate GA-11 Marietta Gingrey (R) R+19 Running for Senate UT-4 SLC suburbs Matheson (D) R+16 Retiring AL-1 Mobile Bonner (R) R+15 Resigned. Special 11/2013 AR-4 South and West Cotton (R) R+15 Running for Senate LA-5 Northeast/Monroe Alexander (R) R+15 Appt’d to state office. Special 11/2013 GA-10 Athens Broun (R) R+14 Running for Senate WA-4 Central Hastings (R) R+13 Retiring NC-7 Raleigh suburbs McIntyre (D) R+12 Retiring FL-19 Southwest coast Radel (R) R+12 Resigned. Special 4/2014 OK-5 Oklahoma City Lankford (R) R+12 Running for Senate CO-4 Eastern Plains Gardner (R) R+11 Running for Senate WV-2 Charleston Capito (R) R+11 Running for Senate MN-6 Minneapolis suburbs Bachman (R) R+10 Retiring NC-6 Greensboro Coble (R) R+10 Retiring GA-1 Savannah Kingston (R) R+9 Running for Senate AR-2 Little Rock Griffin (R) R+8 Retiring CA-45 Orange County Campbell (R) R+7 Retiring MT-AL Entire state Daines (R) R+7 Running for Senate WI-6 East central Petri (R) R+5 Retiring MI-4 North central Camp (R) R+5 Retiring CA-25 Antelope Valley McKeon (R) R+3 Retiring PA-6 Philadelphia suburbs Gerlach (R) R+2 Retiring MI-8 Lansing Rogers (R) R+2 Retiring VA-10 DC exurbs Frost (R) R+2 Retiring FL-13 Tampa Bay Young (R) R+1 Died. Special 1/2014 NJ-3 South-Central Runyan(R) R+1 Retiring NY-21 North Owens (D) Even Retiring IA-3 Des Moines + SW Latham (R) Even Retiring ME-2 North Michaud (D) D+2 Running for Governor NY-4 Long Island McCarthy (D) D+3 Retiring CA-31 Rancho Cucamonga Miller (R) D+5 Retiring IA-1 Cedar Rapids-Dubuque Braley (D) D+5 Running for Senate CA-33 LA Westside Waxman (D) D+11 Retiring NJ-1 Camden Andrews (D) D+13 Resigned PA-13 Philly + suburbs Schwartz (D) D+13 Running for Governor MA-5 Boston Markey (D) D+14 Elected to Senate. Special 10/ 2013 NJ-12 Trenton Holt (D) D+14 Retiring MI-12 Dearborn Dingell (D) D+15 Retiring CA-35 Inland Empire Negrete-McLeod D+15 Running for County Supervisor VA-8 Arlington Moran (D) C+16 Retiring AZ-7 Phoenix Pastor (D) D+16 Retiring CA-11 East Bay Miller (D) D+17 Retiring HI-1 Honolulu Hanabusa (D) D+18 Running for Senate NC-12 Charlotte/Greensboro Watt (D) D+26 Apptd to FHFA. Special 5/2014 MI-14 Detroit/suburbs Peters (D) D+29 Running for Senate Italics indicate that we have done interviews in the district as of August, 2014.

13 We use shading to indicate district partisanship, based on Cook’s PVI (as used in Table 1.) For the most part, we are generally not seeing active nomination contests when district partisanship favors the other party by 5 PVI points or more. For example, Republicans in Iowa’s 1st district struggled to attract an experienced candidate in a district that is D+5. And in California’s Orange County (CA-45), with just a slightly higher PVI of R+7, the clear consensus is that no Democrat has any chance of winning.19 Typically, however, an open seat features only one meaningful nomination contest because the winner of the dominant party’s primary is overwhelmingly favored to win the general election.

There are 48 open seats listed in Table 3. As of August 2014, we have made field trips, to 32 of them (italicized in the table.) We had initially hoped to make trips to all open seats during primary season or shortly thereafter. It is unlikely at this point that our resources will allow us to do so; we will probably need to base our findings on a subset. We visited all districts that came open before the end of 2013. Among the remaining districts, we gave some priority to those to which we could travel easily and stay with family or friends to economize on research expenses. We have been careful not to potentially bias findings by selecting districts based on features of the race: crowded primary fields, involvement by national groups, “interesting” local politics. The set of districts we have visited is not a random sample, but our selection criteria were not based political variables.

Before turning to preliminary findings from our field research, we present some descriptive statistics about the number of candidates in nomination contests, money spent and votes needed to win a primary.

Fields It is common for open seat nominations to attract a large number of candidates, especially in seats safe for one party or the other. As local informants in several districts reminded us “congressional seats don’t come open all that often.” Figure 1 shows the number of candidates from the majority party in open seat House primaries from 2004 to 2012.20

Figure 1 (below) shows that that primary voters must typically choose from large – sometimes very large – multi-candidate fields. Of course, not all of the candidates depicted here are actually serious contenders.

19 On the other hand, both MT-AL (R+7) and WV-2 (R+11) have serious Democratic candidates, Nick Casey, former West Virginia State Party Chair, and John Lewis, former State Director for retiring Senator Baucus. Both of these Democrats were unchallenged for the party’s nomination. Democrats in MT-AL did, however, have a lively nomination contest in 2012, when the seat was also open. 20 Runoff primaries, which are used in most southern states and always have exactly two candidates, are excluded.

14 Figure 1: Number of Candidates in Open Seat Nominations, 2004-2012

Figure 2 presents a more reasonable portrait of the size of the field in primaries, using data from 2012 only. It displays the “Effective Number” of candidates, proposed by Laakso and Taagepera (1979) and widely used in comparative politics. If there are n candidates and each candidate i receives vote share vi, then the 1 Effective Number of Candidates is n . The “Effective Number” construct is thus v2 ∑ i i=1 simply the reciprocal of the Herfindahl index, traditionally used in economics to measure market concentration. It is straightforward to show that if there are 2 candidates, each with 50% of the vote, ENC = 2, 3 candidates each with 33%, ENC = 3, etc.

Figure 2: Effective Number of Candidates in Open Seat Primaries 201221

Percent of cases

0 2 4 6 8 Effective Number of Candidates

21 Includes favored party in all districts and non-favored party when PVI is less than 5.

15 Figure 2 uses the Effective Number of Candidates to present a more nuanced perspective on primary fields in the 2012 election cycle. In 17 percent of the cases, the nomination was essentially uncontested -- the winner faced either no competitors (12%) or a much weaker opponent. Another 24% were two-candidate races. The remaining 59% featured larger fields of serious contenders. Most primaries do indeed feature the complicated dynamics that come with many candidates in a plurality election, but a sizable minority does not.

Observers often lament lack of strong candidates in primary races as evidence that voters are getting no choice in who will represent them. But our interviews suggest that, in at least some of these cases, lack of strong candidates is due in part to the judgments of potential candidates that they would be unable to win. This appears to have been the case in the Republican nomination in Minnesota’s 6th district. Tom Emmer, who nearly captured the governorship for his party in 2012, crushed his sole opponent in the 2014 primary. Two other candidates were in the race for a time but dropped out, and observers as someone who could have represented the district well mentioned another candidate. All of these candidates would probably have won some votes, which would have made the race more nominally competitive, but none would likely have beaten Emmer, a firebrand who seems a worthy successor to Michelle Bachmann in the district. Given this, voters may have been just as happy – and just as well off – to have a non-competitive race to pick the candidate that a more competitive race would also have picked.

Money About a fifth of the candidates depicted in Figure 1 raised so little money that they were not required to file FEC reports. Many did raise significant amounts, however, as shown below in Figure 3.

Figure 3: Primary Spending in Open Seat Nominations, 2004-2012

16 Average spending by first-place finishers is $950,000, suggesting that money plays an important role.$1.7 million. In the graphs below,! these dollar values have been converted to “spending shares.” Hence, Spilka’s $500K! is now reported as roughly 15 percent of all spending by Figure 4, which plots contribution share versus vote share for two special elections ! any candidate; Clark’s $1.7 million is now reported 22 as about 38 percent of all spending. held in Fall 2013, would seem to strengthen this suggestion.!

! vv" ! " Figure 4: Money Share and Vote Share in Two Early Primaries " Massachusetts)5th)District) Alabama)1st)District)

" " " Total Individual Receipts + Independent Expenditures (in shares) ! " How tightly the points are packed ! around the line – which is the same as goodness of fit In these races, with spending " amounts converted to spending ! shares and used to " explain vote and r-squareshares --, the slope is close to 1.0 in both cases. is the same in both! sets of graphs. That is, each one percent So the visual impression is similar. This of spending share is associated with an additional one percent of vote shar"is because the x-variables are just linear transformationse. of one another. " Of course, there are many nomination contests in which the relationship between " money and votes is not so pretty."But similar is not same. For example, The dollarsthe Alabama case graph makes in Figure 5 clear at led to a a glance that the winning run"candidate-off in which inthe front the Arkansas-running candidate outspent his opponent race (Hill) spent fewer dollars by about 10 to (about $800K) than top 1, but won at the ballot box by only 52 to 48. " "candidates in the MA and VA races. This information is completely lost in the lower So"graph.metimes, in fact, What standsmoney and votes have a negative association out instead there is the Hill, as had Figure 5 a higher shows. share of overall spending In PA"than-13, the relatively low the candidates -inspending winner, Brendan Boyle benefitted from union MA and VA, which is also important to know. get" -out-the-vote activity on election day, and from being the only candidate from Philadelphia" while three other candidates split the Montgomery County vote. Both of these factors "One can readily-- the interaction of geography and field and the importance of see why these differences exist. Hill spent about $800K in Arkansas groups in a candidate’s groundyyagainst" opponents who game paid-- have been important in many other races.out very little (about $200K), which gave Hill a big share of ! ! overall spending. But the same !$800K is a lower share of spending in Massachusetts, where several candidates spent $800K! or more. It seems intuitive that spending relative to the field, as captured by the share ! measure, would better capture money’s possible

22impact Massachusetts on data election based on 3outcomesrd quarter FEC reports than. dollarsData from the Alabama election are also in of spending, regardless of what anybody else is need of adjustment.spending.

Another advantage of the share measure is that it generates more17 readily intelligible coefficients. Consider the coefficient of .018 in the upper Massachusetts graph, where money is denominated in dollars. What this coefficient says is that candidates get an additional .018 of vote share for each unit of $1000 they spend. So if they spend, say, 100 thousands ($100K), they will get 1.8 percent of vote share (100 X .01 = 1.8). That is, they get 1.8 percent of the vote for every $100K they spend. This is an intelligible claim, but it takes time to get the calculation right, and it’s still not clear whether the effect is big or small. Now consider the share coefficient for Massachusetts in the lower panel, where it is .82. It should be immediately clear that .82 is a big effect; this is because .82 is almost the biggest the coefficient could plausibly be, which would be 1.0 (a coefficient of 1.0 means that each one percent of money share gets one percent of vote share). So you can just look at a coefficient based on shares and see what it means.

Figure 5: Another Take on Money Share and Vote Share The biggest spender doesn't always win

Iowa-3 (Republicans) Pennsylvania 13 (Democrats)

40 40 Boyle

30 30

Margolies Zaun Cramer 20 Schultz 20

Vote Vote Share Shaw Vote Share Leach Young Arkoosh

10 10

0 Granadette 0

0 10 20 30 40 0 10 20 30 40 Spending Share Spending Share Note: Winning candidate, Zaun, spent $100K Note: Boyle spent $751K; Arkoosh spent $2.1M

As a side comment on Figure 5, it is worth noting that although Zaun won the Iowa- 3 primary (left panel in Figure 5), he received less than 35% of the vote, thus leaving the final nomination decision to a district convention. After 5 ballots, the convention ultimately chose Young. Several other states call conventions when primary results are inconclusive, a larger number use run-offs, and in many states conventions are among the options for special elections. The point here is that primary institutions are often complicated in consequential ways. Political scientists who have studied primary institutions have focused almost exclusive on rules governing voter participation. Studies have mostly found no difference between open and closed primaries, but our preliminary research suggests that other institutional features -- vote thresholds and multi-round processes -- are more likely to have systematic impact.

Figure 5 also gives an indication of the diversity across races in the amounts of money spent in primaries. In PA-13, health policy advocate Val Arkoosh spent over $2 million to come in fourth with 15% of the vote -- roughly the amount spent by all candidates together in IA-3.

Money is always to some degree endogenous in electoral politics, so that simple correlations overstate its causal impact. This is known to be a serious problem in House general elections. Donors want access to winners, and may give to candidates who are likely to win for other reasons, of which incumbency is most prominent. Incumbents build up significant war chests, increasing their (already very high) odds of reelection (Jacobson 1987); they raise even more when seriously challenged. Challengers, especially primary challenger, generally raise little money

18 except when it is generally believed that the incumbent is weak and can probably be beaten.

But incumbency is not an issue in open seat nominations. Moreover, the standings of candidates are often so fluid and ambiguous that no one really knows who is ahead. Moreover, some donors contribute to candidates who are unlikely to win in order to maintain existing relationships -- candidates who are currently serving in the state legislature often receive donations of this type. The endogeneity of money therefore seems less severe in open seat nomination contests. Nonetheless, the desire on the part of donors to establish or maintain a relationship by giving to the eventual winner remains an issue.

In our interviews with local participants in and observers of nomination contests across the country, money is inevitably a major subject – so much so that our case studies of “parties on the ground” sometimes seem like studies of “fund-raising in House nominations.” Yet money is by no means the only important determinant of outcomes. Even in cases in which it does appear to be dominant, the reasons are usually political, that is, they have to do with the relationship of the candidate to policy-demanding groups.

Turnout Turnout is low in congressional primaries, especially in non-presidential years. Kamarck (2014) reports average statewide turnout in 2010 primaries as 18.7% of the voting age population, but notes that statewide turnout numbers are generally driven by senatorial and gubernatorial races, and thus overstate participation in House primaries. Following Brady, Han and Pope (2007), Kamack calculates actual average participation in contested House primaries in 2010 as 7.5% of VAP. But for our purposes, the 7.5% figure is probably too low. It includes all contested primaries, including those of the disadvantaged opposing party in races with incumbents, where participation will typically be quite low. Our interest here is in contested open seat primaries for parties that have a chance to win the general election. In this set of primaries in 2012, the average number of votes cast (for all candidates) was 62,504, roughly 11.4% of the voting age population.

Another way to frame the turnout question is to ask how many votes the primary winner received. Figure 6 shows the distribution of this quantity for 2012 open seats (including only contested primaries, only for advantaged parties and those disadvantaged by a PVI of 5 or less.) The average number of votes cast for primary winners in 2012 was 28,189.

More relevant is the number of votes needed to win a primary in safe districts, in which winning the dominant party’s primary is equivalent to winning the general election. Recall that at least 75% of districts are in this category Again, using data from 2012, in safe districts (PVI>5), the average number of votes cast for the primary winner was 31,452 -- roughly 6% of voting age population in a typical district.

Figure 6: Votes received by primary winners 2012

Fraction

0 0 20000 40000 60000 80000 Votes Cast for Primary Winners 2012 Excludes non-partisan "jungle" primaries (CA and WA).

It is clearly extremely difficult to get voters to participate in congressional primaries. Campaign activities that turnout relatively small numbers of voters can have a large impact on the result.

ANALYSIS OF TWO NOMINATION CONTESTS It is not clear how to best organize and present the extensive, detailed and diverse information we are obtaining in our field trips. One method is to present each nomination as a case study, developing generalizations across cases. We might, for example, find that group champions win in some conditions and generalist politicians in others. Another approach is to systematically evaluate the basic features of each candidacy – that is, create variables – and discuss the patterns that appear in them. We might, for example, tabulate the percentage of all viable candidates who qualify as group champions. In this preliminary report, we employ both of the basic methods more or less independently, beginning with case studies of two early nomination contests. Massachusetts 5th District: A Democratic Free-for-All MA-5 is one of the most liberal districts in one of the most liberal states in the nation. Seven Democratic candidates entered the special primary election held October 15, 2013 to nominate a replacement for Ed Markey who had been elected to the Senate seat vacated by John Kerry’s appointment as Secretary of State. Four of the candidates were current members of the state legislature and one was the elected sheriff of the largest county in the district. Two non-office holding

20 candidates also entered the race. All observers saw winning the Democratic nomination as equivalent to winning the seat. Fractured nomination contests like this appear to be the norm in Massachusetts. Markey had won his first nomination in 1976 by getting 22 percent of the vote in a 12-candidate field. A recent nomination in a neighboring district had only one candidate, but one of our interview subjects volunteered that this could only happen when the candidate’s last name is Kennedy.

The official Democratic Party keeps a low profile in both district and state politics, but activist organizations are important. Among them are Mass Alliance, a coalition of 24 progressive groups; Emerge Massachusetts, which promotes the candidacies of women; and Mass Equality, a pro-gay rights group. These activist groups run candidate schools to train promising activists in the arts of running for office, supporting their graduates for election for as long as they need help and remain faithful to the group line.

When asked who picks winners and losers in nomination contests, an activist leader replied, "Well, I guess we do that in state elections… and so do a lot of other organizations that are part of our coalition." With the formal party so weak, "you need organizations who can make distinctions between Democrats." Said the leader of another activist organization: "We look for champions -- folks who will not only vote for us, but will be actively vocal about their support ... and make efforts to encourage their colleagues to support." The winner of the 5th district Democratic nomination, Katherine Clark, went through Mass Alliance’s screening and candidate school, as did the third place finisher Carl Sciortino.

These activist groups focus their candidate training and recruiting efforts mainly on state and local office, but their culture pervades federal elections as well. Each of the five serious MA-5 candidates, we were told, would base their House campaigns on their network of activists from prior statehouse elections. If we looked at the congressional race in isolation, these networks seem candidate-centered, but they are built in a candidate training environment dominated by policy demanders

Conducting interviews about a month before the election, we were told repeatedly that, while the race was too close to call, it would be decided by the candidate with the strongest activist-based GOTV organization. As a party official said: “Nobody's coming in here and writing a $2 million personal check, there's no super PACs coming in, at least so far … This is a group of [candidates] who have a remarkably effective case to make, and they go to the ground, and they knock on doors, and they build networks … I really think it is going to be whoever is going to pull out whatever base they have, and issues, and get them to the polls … You know what they're all doing: They're knocking on doors like fiends.”

This widely shared assessment remains plausible, but as noted above in Figure 5, vote shares in the race were almost perfectly predicted by spending shares.

21 Moreover, differences in spending were driven by out-of-district group support. EMILY’s List made an independent expenditure for winner Katherine Clark increasing her voter communication spending by about 40 percent. The amount was $200,000 rather than $2 million, but it was nonetheless a substantial help. EMILY’s List also provided Clark with consultant services. Clark won the race comfortably, with 32 percent of the vote in the seven-candidate field, 10 percentage points ahead of her closest rival.

The support Clark received from EMILY’s List reflects her longstanding relationship with feminist causes and groups. According to one interview source, she worked with local activists to bring other women into politics. In 2013 she received the Legislator of the Year award for her work on women’s issues from the Women’s Bar Association. In her legislative campaigns, she was also supported by NARL Pro- Choice Massachusetts and the Planned Parenthood Advocacy Fund.

The candidate who finished second in the race, Peter Koutoujian, also had substantial support from a group, in his case, Armenian-Americans. We found more than 50 articles in two regional Armenian newspapers that covered Koutoujian’s rise in Massachusetts politics over the last two decades. An ethnic organization, Armenian National Committee of America, endorsed Koutoujian, organized a phone bank, and helped him fundraise. An excerpt from an ANCA press release captures the attitude of this group toward Koutoujian:

“We have a chance to elect someone who has the same story we all have; who speaks not just as our friend, but as one of us," explained ANCA National Board Member Richard Sarajian at a weekend meeting of activists and supporters from throughout the Eastern United States. "I have one task for you. I want you to call every single friend, relative or other person that you know in Peter Koutoujian's district and get them out to vote."

[While in the legislature] Koutoujian championed causes important to Armenian-Americans, including providing trade incentives for countries that recognize past genocides and penalizing companies that do business with countries known to be committing them. He was also the lead sponsor of annual Armenian Genocide Commemoration.24

FEC filings show that about 40 percent of Koutoujian’s contributions came from persons having names that are common in the Armenian sections of Europe, most often names ending in “ian.”25

24 “Every vote vital to Koutoujian victory in Massachusetts Special Election.” ANCA press release. http://www.anca.org/press_releases/press_releases.php?prid=2325 25 This is probability an undercount because some married women who contributed may have given up their family name.

22 The candidate who finished third in the race, Carl Sciortino, presented himself as the “true progressive” in the race. Sciortino is openly gay and ran an amusing TV advertisement that made clear his sexual preference and attracted national attention. He was endorsed by Mass Equality and the national LGBT Victory Fund. As with Clark and Koutoujian, we hope to establish what fraction of Sciortino’s funding came from donors who contributed to other candidates endorsed by these groups.

The fourth-place candidate was Will Brownsberger. In contrast to the other top candidates, Brownsberger is not associated with any pre-existing group or ideological faction. He cultivates the style of the iconoclastic truth teller who wants to get policy right, regardless of politics. He maintains an active blog in which he explained the reasons for his policy choices, which were usually but not always leftwing -- e.g., he is a defender of the Citizens United decision -- and is eager to engage with constituents who disagree with him. Our interview subjects regarded Brownsberger as odd but well meaning. As one person observed, "If you say, 'Hey, Will, I think motherhood and apple pie is good,’ and he thinks apple pie sucks, he'll tell ya: ‘You're wrong, … let me show you the four studies that explain why blueberries are better for your health, taste better, and da de da de da…’ he's not going to be, like, ‘Listen, I appreciate it, I like apple pie too.’ So he's non-traditional in that sense, and he's got this passionate following of people who really expect that."

Brownsberger clearly does have a passionate following, which is how he got elected to the state Senate. Much more than any of the other MA-5 candidates, he fits the candidate-centered model of politics, which stresses the skill of politicians in building their own networks. Were it not for the contributions Clark received from EMILY’s List Koutoujian from Armenian Americans, and Sciortino from the LGBT community, Brownsberger might have been competitive. But as the race actually played out, he received just under 15% of the vote.

In fifth-place, last among the office-holding politicians, was Karen Spilka. She entered politics to press for a more favorable allocation of state funds to the school her children attended and built a regional coalition of like-minded moms that actually succeeded in changing the funding formula. She ran for the state Assembly, moved up quickly to the state Senate, and joined the party leadership team. She became particularly active in moving labor legislation and is now seen as clearly aligned with unions. We received different accounts of Spilka’s state legislative service. According to the leader of an activist organization, she became more conservative as required by the pressures of her leadership position; according to a top state party official, she worked hard, stayed late, and did the unglamorous grunt work necessary to pass the labor agenda. Both comments could well be true. Spilka’s 5th place finish may indicate that labor is weak in MA-5; alternatively labor may have muted its support for a trailing candidate.

What of ideology? The leaders of the activist organizations perceived some significant differences among the candidates, with two – Brownsberger and Spilka –

23 seen as more moderate. However, others said that Brownsberger was more off- dimensional than moderate, and that Spilka had only one or two “not awesome” votes. None of the interview subjects, however, said that ideological location played an important role in the election. There was also a consensus that none of the candidates would turn out to be a Bernie Sanders style socialist or a Blue Dog; all would be Massachusetts liberals, voting comfortably for what the Democratic Party put onto the floor of the House. 26

Analysis There are elements of this nomination contest that seem to fit the candidate- centered model well. In a district in which activists are considered important, four of the five candidates were solid liberals, as would be necessary to draw activist support. The fifth candidate, Brownsberger, was not liberal, but he managed to build a strong personal following based on his own view of good policy, also consistent with the candidate-centered model.

But other, arguably more significant, elements favor the group-centered model. Two months before the election, we were told by several sources that the leading candidates were closely matched and that the best GOTV effort would carry the election. We did observe GOTV effort, but, as noted above, the two candidates, Clark and Koutoujian, who finished well ahead of the others were the two candidates who spent the most, and were able to do so because of the support they got from national groups. One must wonder whether the locally organized GOTV efforts mainly cancelled one another out, leaving national groups – EMILY’s List, Armenians, perhaps also national LGBT networks – to push their candidates to the top.27

It is also important to note that activists are not self-organizing. They need to be identified, trained, equipped with the right messages for the voters they contact, and deployed efficiently both before and on Election Day. It is likely that Katherine Clark, the best funded of the candidates, had a substantial advantage in organizing activists because she alone employed a large field staff over the two months before the primary. Thus, while door-knocking and GOTV may have been critical to Clark’s success, these activities may have been simply another reflection of her general fund-raising success, which was in turn due to her connection to women’s groups, both nationally and perhaps locally.

26 Analysis of the roll call votes of these legislators in the Massachusetts statehouse does not shed light on their relative ideological locations. Clark, Spilka, and Brownstein served in the state Senate, which is so skewed – 37 Democrats and 3 Republicans in 2013 -- that Shor and McCarty do not attempt to calculate ideology scores for members. Koutoujian and Sciortino, who served in the state Assembly, have scores that are close to their party medians. 27 It is also possible that our sources -- many of whom are activists -- hold exaggerrated views of the value of their own efforts. The claim that “this is a shoe leather district” (which we have heard in other races as well) may be a reflexively-repeated exhortation (along the lines for “your vote makes a difference”) more than an accurate assessment of impact.

24 This nomination contest, then, was not decided by which candidate could mobilize the largest personal network. It turned on which of three contending groups – feminists, Armenians, LGBT – could contribute most to a politician who was not only a good Democrat but also a champion of their cause.

From the candidate-centered perspective, one could argue that Clark, Koutoujian, and Sciortino won their group support with superior fund-raising effort, by having the foresight to add important national groups to their personal networks and the ambition that led them to put in the hours and cultivate the relationships needed to raise the necessary money. Clearly, the talent, energy and smart political choices of these candidates affected their vote share. But who had the real power in these critical fund-raising relationships that determined the outcome of this race – candidates who “raised” the money or the groups that gave it?

Each of the top three candidates was, to quote the Armenian National Committee for America, “one of us” in the eyes of the group that supported her or him. Each had, moreover, rendered important service to the group over the course of his or her political career. Thus, the candidate had not simply asked for support; he or she had engaged in costly actions to demonstrate that she would be a reliable agent. In making such credible demonstrations of faithful agency, candidates seem not so much to be exercising power, but rather to be acknowledging the power of groups as principals.

Alabama 1st District: In the Shadow of the Shutdown The Republican primary for the open seat in Alabama’s 1st Congressional District took place in the long shadow of Washington politics – that is, two weeks after House Republicans refused to extend the debt limit, leading to a shutdown of federal government services and a threat of default. National business leaders who usually supported the Republican Party were angry about the possible default. The U.S. Chamber of Commerce announced it would “primary” congresspersons who endangered the nation’s credit rating and pledged “at least $50 million” for this purpose.28 But FreedomWorks, Tea Party Patriots, and Club for Growth, national conservative groups that had pushed for the confrontation over federal spending, applauded the shutdown and vowed to defend members of Congress who had brought it about. The South Alabama congressional race quickly became Ground Zero in what the media called the “Republican Civil War.” “In Alabama, a Test of Business Efforts to Derail Tea Party,” headlined the New York Times. “GOP Internal Divide Plays Out in Alabama Campaign,” said the Associated Press. 29

28 “Report: Chamber to hit Tea Party with $50M.” By Cameron Joseph, The Hill, December 26, 2013. 29 October 31, 2013; November 2, 2013.

25 At least one candidate in the AL-1 Republican primary welcomed the national attention. “I’m the Ted Cruz guy,” said candidate Dean Young, referring to Senator Cruz’s high profile obstruction of the continuing resolution needed to prevent a shutdown.30 If elected, he promised to be a “Ted Cruz congressman.”31 Going beyond even Cruz, Young said he would never vote to raise the federal debt ceiling until the national budget was actually in balance. 32 For the other candidate in the Republican runoff primary, Bradley Byrne, the government shutdown presented a challenge. As the business-backed candidate, he could scarcely match Young’s enthusiasm for Cruz. But in the competition for the votes of South Alabama Republicans, he could give no ground in abhorrence of federal spending. He staked out a “moderate” position, as follows: He approved the government shutdown that had just occurred, but said that in the future he would vote to raise the debt ceiling as part of a plan to balance the budget without raising taxes.33 That difference, slight as it was, may have been the biggest policy disagreement between Young and Byrne. But the underlying differences were much larger. The major theme in Young’s campaign was the need to bring constitutional and Godly principles into government so as to halt the decline of “a western Christian empire.”34 Foremost among the needed principles was uncompromising opposition to gays who wanted, as Young said, to pretend that they were married. Discussing the kind of congressman he would be, Young said that, if he were elected, his supporters should get “a big ole thing of popcorn and a Big Super Gulp and lean back and turn on C-SPAN. Because I promise you, I will stand on the floor of the House and stand for the principles that we believe in that made this nation great.”35 This comment, picked up in the national media, suggests that Young's representational style would have centered on articulation of religious values perceived to be threatened by a secular modern society. Young had never held elected office, but he had a record of standing for religious principle in public life. He entered politics working for Roy Moore, the Chief Justice of the Alabama Supreme Court who was forced from office in 2003 for refusing to remove a stone sculpture of the Ten Commandments from the public courthouse. In 2011, with Young as a political consultant, Moore was re-elected to his previous Chief Justice position. As a sitting judge, Moore could not formally endorse Young in

30 “Alabama’s Tea Party champion cast off by his core as special election nears.” Paul Lewis, The Guardian. Posted online, October 30, 2013 31 “Dean Young vows to be ‘Ted Cruz congressman,’ but pro-Cruz PAC backs Byrne.” Brendan Kirby. All Alabama. Posted November 4, 2013. 32 “GOP internal divide plays out in Alabama campaign,” Bill Barrow. Associated Press. Posted November 4. 2013. 33 Ibid. 34 “GOP congressional candidate told gay citizens to go ‘back to California’” Tim Murphy, Mother Jones. Posted Nov. 4, 2013. 35 “GOP establishment frets over Alabama contender.” Alex Isenstadt. Politico. Posted November 4, 2013.

26 the congressional race, but he did write a letter praising Young and appeared with him on the stump. If the need for Godly principles was the central message of Young’s campaign, Byrne’s was that he was a “work horse, not a show horse.”36 He had been a member of the state Board of Education and of the state Senate, where, as a conservative reformer, he often clashed with the state teachers’ association. He was later appointed Chancellor of the state’s community college system, where he led a restructuring of the system in the aftermath of a corruption investigation. Byrne also worked with the state Chamber of Commerce on development issues. Byrne’s career of institutional service earned him the support of the state’s business establishment in the 2010 gubernatorial race, in which he quickly became the consensus frontrunner. But teachers were still angry. The Alabama Education Association took out ads against him that were judged by fact-checkers to be false. Though Byrne had been called the frontrunner throughout the Republican primary, he wound up losing, and the teachers were widely seen as part of the reason. “[H]e really declared war on us,” said the leader of the Alabama Education Association. “[W]e had no choice but to get involved. It obviously had some effect on the outcome."37 Byrne continued his policy advocacy even in defeat. With money left over from the Governor’s race, he founded a group called Reform Alabama to continue pushing his favored policies, including charter schools, merit pay for teachers, greater transparency in campaign finance, and Medicare reform. In work with the state Chamber of Commerce, he pressed ideas for economic development. Most of this activity consisted of speeches, TV appearances, and newspaper columns. One op-ed piece, for example, urged that the state whose college football teams had won three national championships in a row should work to become first in the nation in education as well. It is not clear that that this activity had much real impact on policy, but it validates Byrne’s claim that he is a work horse, and it contrasts sharply with Young’s invocation of Godly principles as the solution to the country’s problems. In a first stage race with nine mostly poorly financed candidates, Byrne finished first with 32 percent of the vote and Young was second with 22 percent. But the runoff was expected to be much closer. Although Byrne maintained his spending advantage over Young, three polls taken shortly before the election showed the race to be essentially a dead heat. Then national forces kicked in. The U.S. Chamber of Commerce, which had pledged to oppose insurgent right candidates at the nomination stage, made a direct expenditure of $200,000 for Byrne. The End Spending PAC put down an additional

36 “Will it be ruin or resurgence for the Tea Party in Alabama race?” Jessica Taylor. MSNBC. Posted October 29, 2013. 37 “Update: AEA admits role in runoff ads attacking Bradley Byrne (with video).” Associated Press. Posted july 22, 2010.

27 $100,000 for Byrne.38 Notably, however, the groups that had supported the government shutdown did not put down money for Young. As a journalist reported, While the GOP cavalry is riding in to protect Byrne, national Tea Party and conservative groups haven’t shown up for Young–and he’s clearly frustrated and puzzled by their absence. “Where are y’all? Because we’ve got the classic battle taking place down here,” Young said on Wednesday. “We’ve got the people; we just don’t have the resources [Byrne has] with establishment groups giving him money and helping him. That’s very difficult to swallow.”39 And it got worse. Despite Young’s embrace of Cruz, one of the PACs that had supported Cruz came in for Byrne. When asked why it was not supporting the pro- Cruz candidate, the PAC leader replied: “It's a great talking point to say you'll be like Ted Cruz. But Ted Cruz is a very smart, a very thoughtful person. And we think Bradley Byrne is better on all of these points."40 National groups did not reach their conclusions about whom to support in a vacuum. As the Guardian reported: Last month, Byrne flew to Washington for a meeting with Club for Growth, shortly after Young made the same trip. Alex Schriver, Byrne’s 25-year-old campaign manager, said there is “continuing outreach” with Club for Growth and groups like it. An attorney and close confidante of Byrne was more explicit: "Young goes up there to encourage them to get involved," he said. “We go up there to try to explain why they shouldn't. We don't have any notion that they would help us – we're up there trying to defuse."

Young did get a contribution from a national Tea Party group, but the big donors never came through. One story reports Young calling up a television station to see if any independent expenditures had been scheduled for him; none had. According to FEC data, the Byrne campaign and its PAC friends spent around $1 million in the runoff; a pro-Young PAC spent about $50,000 With that advantage Byrne was able to prevail on Election Day by a margin of 52.5 to 47.5. While national politics affected the balance of candidate spending in the race, it by no means dominated campaign discourse. More important were religious concerns, especially same sex marriage, which both candidates opposed but Young spoke about more often. Another was Biblical literalism. When Byrne was running for governor, he said on a questionnaire that “there are parts of the Bible that are meant to be literally true and parts that are not.” Attacked for this by Young, Byrne then

38 Federal Election Commission. 39 “Will it be ruin?” Work cited. 40 “Dean Young vows to be ‘Ted Cruz congressman,’ but pro-Cruz PAC backs Byrne.” Brendan Kirby. All Alabama. Posted November 4, 2013.

28 said he believed “the Bible is the word of God, and it is 100 percent true.”41 Young then called Byrne a liar for denying his past views. Byrne, however, managed to go on the offense with a campaign ad that said that when Young was managing Moore’s Supreme Court campaign, he raised money “from good Christians” but kept most of it through payments to his own consulting business. Young cried foul, saying that he spent most of the money on ads and paid himself only the standard commission, and neutral fact checkers, after checking the records, agreed.42 But Byrne, and later a national PAC that supported him, continued to make the bogus charge. Although the runoff was a special election, several debates, advertising, and significant news coverage kept up public interest. Turnout actually rose from 52,344 in the first stage Republican primary to 72,684 in the runoff.43 One possibility is that Young’s strong showing was a direct response by voters to the heated public campaign. Another possibility is successful GOTV effort by evangelical groups. Analysis As in most of our races, money made a big difference in the Alabama First. On a more level playing field – one on which Young could raise as much campaign cash as Byrne -- Young would very likely have won. But to say that Young lost because he raised less money than Byrne is not much of a conclusion. Why did Young have less money? A candidate-centered explanation might be that Young raised less money because he is a less skilled and energetic fundraiser. That could be true. But Young was trying to raise money from evangelicals, a notably low wealth group, whereas Byrne was raising money from business, a notably high wealth group. The financial advantage of the groups active on Byrne’s behalf is an important part of the imbalance in candidate spending in this race. The resource advantage of the business community is a recurring, systematic feature of Republican nomination contests; it should be treated as more than an epicycle attached to the candidate-centered model. Yet despite being heavily outspent, Young almost won, possibly because of non- monetary resources that reflect the comparative advantage of his support groups, possibly because of his direct appeal to voters. We doubt, however, that candidate positions on the government shutdown had much direct effect on vote decisions. It was simply not much discussed in the campaign, either in ads or on news clips we could find. Candidate positions on the government shutdown came up in response to a question during a TV debate and, as far as we could tell, that was all. Modest attention to this potential campaign issue is consistent with other research showing

41 Taylor, MSNBC, cited above. 42 Who’s fooling whom in Alabama?” FactCheck.org. Posted October 23, 2013. 43 Young’s campaign manager charged that Democrats, who are allowed under Alabama law to declare a party affiliation at the polls, came into the runoff for Byrne. However, a reporter looked for telltale signs of Democratic involvement, i.e., suspiciously high support for Byrne in Democratic precincts, and found little. “Did Democratic voters tip congressional GOP primary runoff to Bradley Byrne?” Brendan Kirby, All Alabama. Posted November 7, 2013.

29 that voters rarely have enough information to vote on policy issues in primary elections (Lenz, Hirano, Pinkovskiy, and Snyder, 2013). Even for Washington elites, the government shutdown issue had an unexpectedly muted effect. It did motivate the U.S. Chamber of Commerce to spend against Young. But the big groups on the other side took a pass, except for one that crossed factional lines to support Byrne. This was a remarkable occurrence. Club for Growth and FreedomWorks, well-heeled and sophisticated players in Republican primaries, had an excellent opportunity to pick up a vote for their side of the party’s big factional struggle, yet they stayed on the sidelines. What this shows, we believe, is that issue positions can be less important than expectations about how a candidate will perform if elected. Despite having never held elective office, Young gave clear indications about the kind of agent he would be if elected. Young’s stated preference for high profile obstructionism -- speeches on CSPAN and the like-- created the worry that he might become a loose cannon, an embarrassment to his political allies. It may be rare for a candidate to come as close to winning as Young did only to fail a test about expected performance in office. But we think it likely that many candidates fail this test early on in nomination contests and then get no attention at all, including from the political scientists who construct theories of nominations. Young’s failure on the test of political style could be seen as a rare indicator that expected efficacy is an important factor in the politics of party nominations. None of this is to deny that the nomination contest had policy content. Religion, in particular, was discussed in several guises. One of the most common was Young’s promise that he would bring Godly principles to government. Young’s deeply authentic religiosity undoubtedly brought him support from the less well heeled but numerous and mobilizable evangelical community. Indeed, Young may have simply had the more popular position on this issue. The Cook Report said in July, before any polls had been done or the field even set, that Byrne was the favorite in the race, but that Young’s rhetoric would have appeal in the conservative south Alabama district. “If Young makes it into a runoff with Byrne, many expect the Mobile business community to get very nervous and close ranks behind Byrne, but it could still be close.”44 The Cook Report saw Young as a Tea Party candidate rather than an evangelical, but otherwise the report was spot on. It is unlikely that Young wound up losing this race because his rhetoric had less appeal than Byrne’s; more likely he lost because he lacked money to communicate it. Young lacked money less because he was a poor fundraiser, more because the groups that give money in Republican primaries preferred another candidate. More specifically: 1) Young’s core group of evangelicals has limited resources, 2) local business was afraid of Young and made large contributions to oppose his candidacy, 3) national business strongly did not want another “Ted Cruz congressman” in Washington and so made large contributions against Young, 4) national

44 Cook Political Report, July 25, 2013.

30 conservative groups, even though actively looking to strengthen their standing within the House Republican caucus, were unwilling to accept Young’s offer of help, presumably because it cares about aspects of representation other than how an MC votes on pubic bills. Thus, group preference, more than voter preference or candidate skill, determined this nomination. Special primaries for two other vacant seats, the Florida 13th and the Louisiana 5th, were held in the early period of our study. Democrats cleared the field to nominate a former gubernatorial candidate, Alex Sink, to run for the highly competitive Florida seat; Republicans nominated a former staff member of the longtime GOP House member who had died. In the first stage of Louisiana’s all-party primary, a self- financed outsider barely made the runoff but then trounced the Republican Party insider to win the seat. We mention them here simply to show the wide variety of patterns that turn up.

CANDIDATE ASSETS ACROSS OPEN SEAT NOMINATION CONTESTS We have interviewed more than 200 local observers and participants in 32 nomination contests. We have deliberately kept the interviews unstructured. We are interested how local insiders think about the races, and have tried to avoid any kind of leading questions. In most of our interviews, we used some variant of this simple probe:

What are your impressions of the candidates for the [e.g.] Republican nomination in the 2nd Congressional District? How is the race going?

Interviews typically last around an hour, depending on the subject’s availability. As part of our informed consent procedure, we let our subjects know that we do not expect to publish our results before 2016, that we will not attribute information to them by name without explicitly seeking permission, and that we will not identify them or their organization by name. We email these ground rules, as well as a study description, to subjects in advance of the interview and go over the important points orally before the interview. Most of our sources allow us to record the interviews, so that we have digital audio files of most of our interviews.

The section presents a preliminary and impressionistic attempt to distill some basic themes across races, intended as a foundation for more systematic analysis in the future. Specifically, we catalog a set of “assets” mentioned repeatedly by our sources as signs of a strong nomination contender.

Some commonly mentioned assets relate to candidate ambition: energy, willingness to work hard, to prioritize the nomination contest above any competing demands. A source associated with a nationally active group emphasized that serious nomination candidates need to be willing to quit their day job and work full-time at being a candidate -- that is, to work full-time at fund-raising. One candidate complained that the process was “brutal,” and required “a full 9-to-5 schedule of begging for money.” Warned one group strategist, “If they’re not willing to sit down

31 in that chair and make those calls, then the race is over.” A long list of past donors and a proven record as a fund-raiser are regarded as essential candidate assets in all of the races we are studying. None of our informants have contradicted, and many affirmed, the basic claims of ambition theory: successful politicians must be strongly motivated to win elections.

Beyond ambition, however, there are other, more political, assets consistently mentioned by the political insiders we have interviewed. In going through the full set of comments, we developed seven categories that seem particularly meaningful across races. The first is membership in a group or faction As noted above, many candidates have a group of core supporters who regard the candidate “one of us.” Being “one of us” establishes the candidate as trustworthy and authentic. The group is assured the candidate is committed to its goals, that its policy demands will receive high priority and serious effort as opposed to mere lip service, that the candidate will fully understand the group’s stake in technical and obscure policy details. Because groups have limited ability to sanction a misbehaving agent, they want to be confident that their candidates not only understand current issues, but also would come to the “correct” position on any future issue. This confidence increases the more they know their agent shares their values and philosophy. Rather than filtering on policy positions, groups screen candidates based on fundamental beliefs that help form positions. Evangelical groups trust the authenticity of preachers, ethnic communities trust co-ethnics who have been active in their communities, unions trust union members, and business groups trust those who “know what it means to meet a payroll.” Being considered “one of us” by a powerful group is a significant asset for a candidate.

A candidate who is “one of us” is more likely to be a faithful agent of the group, but this is not guaranteed. We have observed several Republican candidates whose background as business-owners gives them an acknowledged understanding of the policy demands of business. But their relative inexperience in public office leaves doubt as to their competence and efficacy in political settings. Conversely, some candidates, have demonstrated themselves to be faithful agents of a group despite not, strictly speaking, being “one of us.”

Another frequently mentioned asset is a reputation for hard work and willingness to pay attention to policy details. Many candidates are approvingly identified as “workhorses” or as “having carried water” for groups or causes. These judgments are typically based on the candidate’s past service in the state legislature, other elected office, or leadership in community and volunteer organizations. Legislative workhorses have cultivated reputations as people who can get things done in political environments Workhorses know how to propose effective legislation, how to shepherd bills through the legislative process, how to manage the negotiation and coordination necessary to pass laws. A reputation as a workhorse is a clear asset in attracting the support of policy-oriented groups. If a group is concerned about whether a candidate will be a good agent, a record as a workhorse demonstrates competence, energy and effectiveness, the ability to deliver policy.

Being seen as “one of us,” a faithful agent or a workhorse are three candidate assets that have come up repeatedly in interviews; all are directly relevant to the claim that groups seek to support champions. Four additional factors, less germane to the champion question are (1) whether the candidate has a local network, (2) whether the candidate receives substantial support from national groups, and (3) whether the candidate has a name recognition advantage (4) whether the candidate established a position as frontrunner. Table 4 presents a rough tabulation of these seven assets for all serious candidates in 13 nomination contests that began early. Candidates are listed by district and districts are listed by PVI, going from most Republican to most Democratic. Our assessments of candidate assets are preliminary and impressionistic: Table 4 is not a systematic coding, but hopefully a step toward that end.45

“One of us” and/or a Faithful Agent As noted above, being “one of us” is a good indicator of being a faithful agent, but is neither necessary nor sufficient. We thus tabulate these qualities separately. The first column of Table 4 is shaded black for candidates who are clearly members of their listed core support group: that is, business candidates who have run a business or had careers in business sectors, members of ethnic or identity groups, etc. Gray shading indicates partial or ambiguous cases.

It is interesting to note that none of the union champions listed here (Boyle, Murphy, Spilka) have been union members, although Boyle and Murphy are both sons of union members, and lawyer Spilka’s mediation practice focused on labor issues. However, in later races not listed in the table (e.g. Donald Norcross in NJ-1), the union candidate has actually been a union member.

In contrast, most of the business candidates in Table 4 have business experience, either running their own business (pharmacist Buddy Carter of GA-1, mustard maker Kevin Raye of ME-2) or in the financial sector (French Hill of AR-2, Tommy Moll of AR-4, Donna Sheldon of GA-10, John McCallum of GA-1, Mimi Walters and Pat Maciariello of CA-45). Lawyers Bradley Byrne of AL-1 and Ed Lindsey of GA-11 are partial exceptions. In Byrne’s case, long active involvement in the Mobile Chamber of Commerce seems to have established him as an authentic member of the business community. Also, as chancellor of the junior college system, Byrne facilitated a program in which junior colleges were used to prepare students for skilled labor, providing state industry with a trained workforce, and demonstrating Byrne’s competence and his friendliness to business. In Lindsey’s case, a law practice serving the Buckhead business community served as a springboard to a strong record of representing business in the Georgia state legislature.

45 We recognize that Table 4 does include assets that candidate-centered theory of parties considers important, such as an ideological position appropriate for the district’s primary electorate, or the ability to forge a coalition. These have not been mentioned as such in our interviews, but we need to consider whether they can be ascertained in other ways.

33 Table 4: Candidate Assets

CD PVI Candidate Core support One of Us Faithful AgentWorkhorse National SupportLocal Ties Name Rec. Frontrunner GA-11 R+19 Bob Barr Libertarian Tricia Pridemore Business Ed Lindsey Business Barry Loudermilk Evangelical AL-1 R+15 Bradley Byrne Business Chad Fincher Business (forestry) Quin Hillyer Conservative Dean Young Evangelical AR-4 R+15 Tommy Moll Business (finance) Bruce Westerman Conservative GA-10 R+14 Michael Collins Business Jody Hice Evangelical Donna Sheldon Business (finance) MN-6 R+10 Tom Emmer Populist Conservative GA-1 R+9 Buddy Carter Business (pharmacist) Jeff Chapman Populist Conservative John McCallum Business (finance) Bob Johnson "outsider" AR-2 R+8 Ann Clemmer Conservative French Hill Business (finance) Conrad Reynolds Tea Party CA-45 R+7 Pat Maciariello (quit) Business (finance) John Moorlach (quit) Conservative Greg Raths Populist Conservative Mimi Walters Business (finance) ME-2 R D+2 Bruce Poliquin Conservative Kevin Raye Business ME-2 D Emily Cain Progressive Troy Jackson Unions IA-1 D Swati Dandekar Indian-Americans Anesa Kajtazovich Progressive Pat Murphy Unions Dave Obrien Trial Lawyers Monica Vernon Cedar Rapids PA-13 D+13 Val Arkoosh Health/Liberal Brendan Boyle Unions Daylin Leach Progressive Marjorie Margolies Women MA-5 D+14 Will Brownsberger Katherine Clark Women/Liberal Peter Koutoujian Armenians Carl Sciortino LGBT/Progressive Karen Spilka Unions

Note: Primary winners in Bold.

34 Many candidates have “one of us” status with respect to ideological groups, indicating that the candidate emerged from an ideological or activist network. For example, Ann Clemmer, a college professor in Little Rock, was a party activist for some 20 years before being urged by friends to run for the legislature. Pundits such as AL-1’s Quinn Hillyer can be regarded as “one of us” based on their records ideological leadership.46 It seems reasonable to accept statements by members of the activist community that the candidate is one of them, but this is not verifiable in the way that ethnicity and professional background are.

Finally, there are a few candidates who do not have a clear core group: Bob Johnson (GA-1), a doctor ran as an outsider, and appealed to “Tea Party” voters. Our local sources regarded Johnson was an unknown quantity, and he remains so to us. If he was part of an existing network, it is not one that is well known to district insiders. He did however receive substantial support from Club for Growth and FreedomWorks. Monica Vernon (IA-1) was first elected to the Cedar Rapids City Council as a Republican in 2007,47 and her husband remains active in the Iowa Republican party -- facts mentioned by many sources in the district as potentially disadvantageous. Not only was she not “one of us” to Democratic groups, her core network was in the other party. Yet her efforts in helping Cedar Rapids rebuild after a devastating flood she established her as a faithful agent of the community. Will Brownsberger (MA-5), as discussed above, has loyal followers, but they are not a pre-existing group. Marjorie Margolies (PA-13) does not seem to be regarded by any group as an authentic agent.

Almost all of the candidates listed in Table 4 are regarded as “one of us” by their core supporters; most of these are also regarded as faithful agents. Several candidates who are “one of us” fail to be perceived as faithful agents because they have no experience in elective office and are untested in political environments (Collins, Moll, McCallum, Maciariello.) Conversely, two former members of Congress (Bob Barr and Marjorie Margolies) seem to be regarded as poor agents in part because their high name recognition gives them the ability to exercise independent judgment.

Workhorses A large fraction of the open seat nomination candidates have prior experience in the state legislature, but only some of these have achieved a “workhorse” reputation for effectively influencing policy decisions. A few other candidates are classified as partial workhorses in Table 3 based on efficacy outside the legislature. For example, Monica Vernon played an important role on Cedar Rapids City Council in rebuilding after a flood devastated the city in 2008. Observers report that Vernon’s effectiveness during this crisis lead to support from local unions in the current nomination contest, including an early endorsement from the Building Trades Union of Cedar Rapids. Other candidates developed workhorse reputations serving on

46 See Noel 2013 on the role of pundits in shaping ideological coalitions. 47 http://www.bleedingheartland.com/diary/6274/ia01-democrat-monica-vernon-is-in

35 county and state boards (Bradley Byrne, John Moorlach), in appointed positions (Bruce Poliquin) or even as politically active private citizens (French Hill, Val Arkoosh).

In at least one case, a candidate’s hard work and efficacy in the legislature won her enemies as well as friends. Swati Dandekar of IA-1 (D) was a ringleader of the “Iowa six-pack,” a group of moderate Democratic State senators who voted as a bloc and made [Assembly leader] Pat Murphy’s life “a living hell,” blocking a union-backed prevailing wage bill, opposing gay marriage and promoting nuclear power. Similarly, workhorses Bruce Westerman and Mimi Walters have had their conservative credentials called into account for behind-the-scenes work involving potential compromises on Obamacare and taxes.

The evangelical candidates we have studied so far (Barry Loudermilk, Dean Young, Jody Hice) all fit the model of a “show horse” rather than a workhorse. They are inspiring speakers, but not legislative technicians, policy experts, or coalition builders. The current incumbent in GA-10, Paul Broun, also strongly tied to the evangelical community has a similar profile: focused more on articulation of values than technical policy details. In contrast, business candidates tend to be workhorses. One might speculate that this difference reflects differences in what these groups want: in the case of evangelicals, the public articulation of religious values perceived to be threatened by modern secular society; in the case of business, specific tax, regulatory and spending policies that improve bottom lines.

National Support National support denotes the candidate’s ability to raise money from outside the district. This can take the form of donations from a national network of co-ethnics (Koutoujian, Dandekar), personal networks forged from Washington (former Members of Congress Barr and Margolies,) national networks based on particular policy demands like gay rights (Sciortino), healthcare (Arkoosh) or free market principles (Palmer in AL-6).

A particularly powerful form of national support comes from ideological groups: for example EMILY’s List or Club for Growth. The primary value of such an endorsement is the money it will bring in from the group’s national mailing lists, websites, etc. These two particular groups endorsed Emily Cain (D) and Bruce Poliquin (R) in the two races in ME-2 which, at D+2, is competitive for both parties. Cain and Poliquin both easily beat moderate local candidates with long careers in the state legislature (Kevin Raye and Troy Jackson).

Run-offs, common in the South, present particularly good opportunities for out-of- district money, due to a compressed time period and clearer choices. With one exception, these Republican run-offs have pitted a business establishment candidate against a more militant conservative. Out-of-district groups on both sides (national and state Chambers of Commerce, Club for Growth, Senate Conservatives Fund,

36 Madison Project) have chosen to spend significantly in some run-offs and to stay out of others. As detailed in the discussion of AL-1 above, the Chamber of

Commerce spent heavily for Bradley Byrne in the run-off while FreedomWorks and other conservative groups did not come in for Dean Young. The relative closeness of the vote, juxtaposed with the extreme imbalance in spending suggests that both decisions (one side to come in, the other to stay out) were critical.

Out-of-district spending seems critical in some nomination contests; in others it is conspicuously ineffectual. Val Arkoosh in PA-13, for example, spent $2 million (mostly raised out of district) to come in fourth with 15% of the vote. National support also sometimes creates resentment among local stakeholders, as in the case of Marjorie Margolies.

An earlier iteration of Table 4 included national support in the form of endorsements from ideological blogs. Tommy Moll’s candidacy in AR-4 received a potential boost when Erik Erikson of RedState wrote, “In Arkansas’s 4th Congressional District, there is only one candidate conservatives can support with a clear conscience. That would be Tommy Moll.”48 RedState also published a sympathetic interview with Greg Raths of CA-45. But in neither case did the ideological endorsement seem to have much impact on outcome or race dynamics. Out of district support seems to be effective mostly when it takes monetary form.

Local Ties/Geography We also note whether candidates have a local ties: a network of local people who will serve in key campaign roles, and encourage others in their networks to donate and to volunteer. In interviews, local ties have not necessarily been cited as an advantage, but their absence has been noted in two ways. First, candidates who are not known in the district are seen to be at a distinct disadvantage with regard to voters and local organized support, that is local donors and activists. That is, voters do not recognize their names, and there is no local network to support a campaign.

Second, observers of nomination contests frequently note the extent to which a candidate’s previous district (state legislative or otherwise) overlaps with the congressional district. A candidate like Mimi Walters, whose California Senate district almost perfectly overlaps Congressional District CA-45, is widely perceived as advantaged thereby. Other candidates (Karen Spilka, Daylin Leach, Ed Lindsey, Donna Sheldon) have been perceived as disadvantaged by geography because the overlap between their state legislative districts and the congressional district was minimal. Geography can also interact with the candidate field in significant ways: two or more candidates from the same county may be seen as splitting the same vote. In PA-13, for example, Brendan Boyle benefitted from being the only candidate from the city of Philadelphia, while the vote of suburban Montgomery County was split among his three competitors.

48 http://www.redstate.com/2013/11/12/i-support-tommy-moll-for-congress/

Name recognition Name recognition is a truly candidate-centered asset, a feature of the strategic environment to which groups strategically react. Name recognition can come from having previously been elected to Congress (Margolies, Barr) or from unsuccessful campaigns for high profile office, such as Tom Emmer’s and Bradley Byrne’s unsuccessful runs on their party tickets for governor. Name recognition is also enjoyed by candidates who are related to long-serving prominent officials, like GA- 10’s Michael Collins, son of former Representative Mac Collins, or NC-6’s Phil Berger Jr., son of State Senate leader Phil Berger Sr. In some cases, name recognition reflects a significant policy achievement, as in the case of CA-45’s John Moorlach who still retains credit for having forecast Orange County’s 1994 bankruptcy and for coping with its aftermath. Among the candidates mentioned in this paragraph, however, only Byrne and Emmer won their nominations. Moreover, we have observed candidates rise from virtually zero name recognition to win the nomination (Mark Walker in NC-6, Gary Palmer in AL-6.)

Frontrunners Frontrunner status often reflects other assets. It can derive from candidate-centered assets like name recognition (Barr, Margolies, Emmer) or from having the ambition, foresight and connections to make a decisive early entry. Mimi Walters (CA-45) exemplified the latter in announcing her entry and an extensive list of endorsements within hours of the previous incumbent’s retirement announcement. But frontrunner status can also be a product of effective coordination by strategic groups.

Frontrunner status becomes as asset in its own right when it discourages the entry of potentially strong candidates, and/or discourages groups from supporting other candidates. This seems to have occurred in the cases of Emmer and Walters as well as some other cases not depicted here: Mia Love UT-4, Donald Norcross NJ-1, Alex Sink FL-1, Staci Appel IA-3, John Foust VA-11. It seems likely that Table 4, which focuses on seats that opened early, and therefore generally enjoyed long campaign periods before the primary, understates the frequency of frontrunner dynamics in House nominations.

Analysis: Voters or Groups? Among the assets presented in Table 4, “faithful agent” and “workhorse” stand out as particularly relevant to the principal-agent problem that Bawn et. al (2012) assert is at the core of relationships between groups (principals) and politicians (agents). But one might argue that “workhorse” and “faithful agent” are used not primarily by groups, but rather by voters as indicators of likely performance in office. Perhaps Clark’s association with women’s issues gained her votes directly, that the resource effect of EMILY’s list support was secondary. And perhaps, Koutoujian did well because more because local Armenian American voters recognized him as one of their own, not because of the money raised by national

38 Armenian networks. Indeed, Watertown, which is part of the district, is known to have a large Armenian community.

This alternative is one to which we must attend carefully going forward. Trust and authenticity have been noted by other scholars as important in congressional elections (Fenno 1983, Bianco 1994.) They are conventionally understood as applying to the relationship between candidates/politicians and voters. But, as articulated by our local observers, credible candidates in nomination contests are generally faithful agents of organized groups. This distinction is highlighted by a nomination contest not part of the current election cycle, thus outside the formal scope of our project, but nonetheless germane.49

In 2012, CA-41 (Riverside) was a newly created majority-Latino district. The area leans Democratic, but union membership is weak with only 14% of workers unionized. It has record of social conservatism, reflected in 63% local support for a 2008 ballot initiative to ban gay marriage. One might naturally expect the large Latino community to elect “one of its own” and to demand a Latino candidate whom they would trust as an authentic representative of voter interests. But no Latino candidate emerged. The eventual winner was openly gay Japanese-American Mark Takano, a strong supporter of unions. Authenticity and trust were indeed critical for Takano -- his status of “one of us” in the gay community is seen by local observers as critical for the money he raised, as was his strong relationship with unions. Latino voters in CA-41 would quite probably have preferred to vote for one of their own, but the absence of politically active organized Latino groups in this district made the prospect a non-starter.

CONCLUDING REMARKS Our study of open seat nominations is a work in progress. Any attempt draw definitive lessons would be premature. The purpose of this paper has been to organize and take stock of what we have learned so far.

But four points clearly emerge. First, Mark Hanna’s famous quip about the two things that matter in politics (“Money is the first and I can’t remember what the second thing is”) applies quite well to open seat nominations. Local observers assess candidates’ chances of winning the nomination in terms of their ability to raise funds. Money is not sufficient to win, but it is necessary.

Second, we have seen ample evidence to support the claim that candidate ambition matters. Running for an open seat nomination is a full-time job. The inability to leave one’s existing career and put fundraising effort ahead of family obligations was cited repeatedly as a reason why many seemingly promising candidates stayed out of the race. We have no disagreement with the claim that personal ambition is essential to winning a seat in Congress (Fowler and McClure 1989, Lawless 2012.)

49 We have interviewed local observers about this race from previous election cycle.

39 Ambition is clearly necessary. But it too is not sufficient, and the incentives that ambition responds to arise in a nomination environment structured by groups.

The role of groups forms our third point. There are a handful of characteristics that are regarded as valuable across all of the districts we have studied so far: authentic membership in, or commitment to, a core group (being “one of us” and/or a faithful agent), reputation as a workhorse, national and local support, name recognition, and frontrunner status. Some of these assets -- name recognition in particular -- fit a candidate-centered model of nomination politics. As the Margolies case particularly shows, candidate name recognition may win votes even in the absence of group- funded persuasion or mobilization. On the other hand, initially unknown Dean Young’s strong showing against a former gubernatorial candidate shows the limits of this asset. Indeed, while Margolies was the top vote getter among the suburban candidates, she lost to Brendan Boyle whose union supporters were observed systematically getting out the vote on election day. The other assets: local ties, national support, etc. clearly favor the group-based model. Authenticity and workhorse reputation are critical for a policy-demanding group considering investing in a candidacy. Groups want to be assured that they are electing a reliable political agent.

Fourth, among the candidate assets we have observed, one clearly stands out: the more a group perceives a candidate to be a champion, the stronger its support. Almost every serious contender in the races we have observed so far has a clear connection to a group that expects the candidate not only to vote in its favor if elected, but also to actively pursue the group’s policy demands at every stage in the legislative process.

If past patterns hold, the winners of the 2014 open seat nominations will go on to long careers in the House. But in most of our nomination contests, who wins will not affect the pattern of roll call votes in Congress: Democratic winners will vote with their party and Republican winners with theirs. But it will affect the content of what is voted on: which issues are championed, what policies are enacted. It could also affect the ideological locations of the parties: More Dean Youngs and fewer Bradley Byrnes would result in a more radically conservative Republican House caucus; more Daylin Leaches and fewer Margorie Margolies’ would mean a more aggressively liberal Democratic House caucus. This is so despite the fact that neither of the candidates in these pairs differ much if at all from one another on issues that were important in the campaigns. Understanding who is powerful and who is passive in these nomination contests is essential to understanding the agency relationship between politicians, voters and groups.

40 References

Aldrich, John. 2011. Why Parties? A Second Look. University of Chicago Press.

Anzia, Sarah. 2014. Timing and Turnout: How Off-Cycle Elections Favor Organized Groups. University of Chicago Press.

Baylor, Christopher. 2013. “First to the Party: The Group Origins of the Partisan Transformation on Civil Rights, 1940–1960.” Studies in American Political Development 27(2): 111-141.

Bawn, Kathleen, Martin Cohen, David Karol, Seth Masket, Hans Noel, and John Zaller. 2012. “ A Theory of Political Parties.” Perspectives on Politics 10(3): 571-597.

Bianco, William. 1994. Trust: Representatives and Constituents. University of Michigan Press.

Boatwright, Robert. 2103. Getting Primaried. University of Michigan Press.

Brady, David, Hahrie Han and Jeremy Pope. 2007. “Primary Elections and Candidate Ideology: Out of Step with the Primary Electorate?” Legislative Studies Quarterly 32(1). 79-105.

Cohen, Martin. 2005. Moral Victories: Cultural Conservatism and the Creation of a New Republican Congressional Majority. Ph.D. Dissertation. UCLA.

Fenno, Richard. 1978. Homestyle: House Members in Their Districts. Boston: Little, Brown and Co.

Fowler, Linda and Robert McClure. 1989. Political Ambition Yale University Press.

Fulton, Sarah, Cherie Maestas, L. Sandy Maisel, Walter J. Stone. 2006. “The Sense of a Woman: Gender, Ambition, and the Decision to Run for Congress.” Political Research Quarterly 59(2). 235-48.

Gaddie, Ronald Keith and Charles S. Bullock, III. 2000. Elections to Open Seats in the U.S. House. Rowman and Littlefield.

Green, Donald and Jonathon Krasno. 1998. “Salvation for the Spendthrift Incumbent: Reestimating the Effects of Campaign Spending in House Elections.” American Journal of Political Science 32:884-907.

Hall, Richard L. 1996. Participation in Congress. Yale University Press.

Jacobson, Gary. 1987. “Running Scared” in Congress: Structure and Policy, Mathew McCubbins and Terry Sullivan (ed.) Cambridge University Press.

Jacobson, Gary, 2001. The Politics of Congressional Elections. 5th ed. Boston: Allyn and Bacon.

Karmark, Elaine. 2014. “Increasing turnout in Congressional Primaries.” Brookings Institution

Kirkpatrick, Jean. 1976. The New Presidential Elite. Russell Sage.

Lawless, Jennifer. 2012. Becoming a Candidate. Cambridge University Press.

Lawless, Jennifer and Richard Fox. 2010. It Still Takes a Candidate: Why Women Don’t Run for Office. Cambridge University Press.

Layman, Geoffrey and Thomas Carsey, John C. Green, Richard Herrera, Rosalyn Cooperman. “Activists and Conﬂict Extension in American Party Politics.” 2010. American Political Science Review. 111-141.

Maestas, Cherie, Sarah A. Fulton, L. Sandy Maisel, Walter J. Stone. 2006. "When to Risk It? Institutions, Ambitions, and the Decision to Run for the U.S. House." American Political Science Review 100(2) 195-208.

Masket, Seth. 2009. No Middle Ground: How Informal Party Organizations Control Nominations and Polarize Legislatures. University of Michigan Press.

Shor, Boris, and Nolan M. McCarty. 2011. The Ideological Mapping of American Legislatures. American Political Science Review 105 (3): 530-551.

Skocpol, Theda and Vanessa Williamson. 2012. The Tea Party and the Remaking of Republican Conservatism. Oxford University Press Stone, Walter and L. Sandy Maisel. 2003. “The Not-So-Simple Calculus of Winning: Potential U.S. House Candidates’ Nomination and General Election Chances.” Journal of Politics 65:4; pp. 951-77.

Zernicke, Kate. Boiling Mad: Inside Tea Party America. Macmillan.