Congressional Endorsements in the Presidential Nomination Process

Abstract

Examining congressional superdelegate endorsements in the 2008 Democratic presidential nomination, the authors show that changes in the political context affected the balance of factors in members’ decisions to endorse Clinton or Obama. Specifically, the national standing of the candidates became increasingly important—and local opinion less important—to Obama endorsements even as constituency views became a stronger influence over Clinton endorsements. The findings reveal how constituency considerations affect the elite endorsement choices that shape the presidential nominating process. In addition, the analysis highlights the ways in which members of Congress balance conflicting considerations in a changing political context when an issue plays out over an extended period.

Keywords

Congressional decision making presidential nominations superdelegates endorsements

You’ve got divided constituencies in [some senators’] states, and they’re going to hope this resolves itself with them not having to get in the middle of it. And if it can, then they haven’t necessarily hurt themselves with their own base.

Sen. Chris Dodd (D-CT)

Of course what my district says is important to me. But other factors are, too: electability, job performance.

Rep. Howard Berman (D-CA)

Three months ago, everyone in the district was saying how great it was to have these strong candidates. … Now, whenever I’m at a rally or somewhere else, I hear people saying, “I used to like Jason, but if he endorses the one I don’t like, I’m not going to vote for him.”

Rep. Jason Altmire (D-PA)

Barack’s impressive showing in our state is attractive to me.

Sen. Amy Klobuchar (D-MN)

I don’t know what I’m going to do.

Rep. Jim Moran (D-VA)¹

Between early 2007 and mid-2008, Democratic members of the House and Senate weighed in on the contentious choice of a 2008 Democratic presidential nominee. As superdelegates to the national party convention, all Democratic members of Congress were entirely free to choose among the presidential candidates, and they were free to publicly endorse a candidate before primaries began, during the primary season, or at the convention. Recent research has demonstrated that these choices are quite consequential for the presidential nomination process (Cohen, Karol, Noel, & Zaller, 2008), but we know little about how elected party elites make these crucial decisions.² The choices of the 2008 Democratic congressional superdelegates provide an ideal window into several crucial questions about these choices. How did members reach these decisions? How much did constituency considerations drive their calculations, and were other goals at work? And did the dynamics of the nomination process affect the way members weighed their options?

We argue that members of Congress faced considerable uncertainty and conflicting pressures in the endorsement decision, a situation that parallels members’ legislative choices. In the legislative context, members’ multiple goals sometimes point neatly in the direction of a single choice, but members frequently must make choices in which the potential consequences are unpredictable or in which advancing one goal threatens another goal (Kingdon, 1989). When a decision process is stretched out over a long time frame, members’ efforts to reconcile competing considerations are further complicated. In these cases, members’ strategic calculations influence both final decisions and the timing of those decisions (Box-Steffensmeier, Arnold, & Zorn, 1997; Caldeira & Zorn, 2004), and the nature of the choice can shift with the changing political context (Fenno, 1986).

Anecdotal evidence, illustrated by the quotations above, points toward the complex and conflicting considerations in congressional superdelegate endorsements. Members encountered pressures from activist partisans, a well-defined and electorally salient subconstituency (Bishin, 2000). At the same time, members recognized that the identity of the presidential nominee could affect their own electoral margins. For members with power goals and progressive ambition, the choice to align with Hillary Clinton or Barack Obama held the possibility of future reward—or punishment. Meanwhile, members also weighed the potential effect of the candidates’ positions and political prospects for their policy objectives.

In this article, we examine the influence of electoral, policy, and power goals on endorsement choices, finding evidence that all three were relevant to members’ decisions but that the relevance of these factors depended on the changing political context in which members made their decisions across the nomination season. Our findings inform the substantial literature on the role of elite behavior in the postreform nominating process (e.g., Cohen et al., 2008; Steger, 2008) by elaborating the interplay between local considerations and national campaign dynamics in shaping the crucial choices of party elites. The results also add to a small body of research showing that important choices outside of the legislative process are shaped by the competing pressures present in roll-call decisions.

Presidential Endorsements and Member Decision Making

Democratic members of the House and Senate have played a formal role in the presidential nomination process since the early 1980s, when dissatisfaction with the McGovern-Fraser rules led Democrats to carve out delegate slots for “party leaders and elected officials” (PLEOs), including a subset of Democratic MCs (Mayer, 2009, pp. 88-89). Since 1996, all congressional Democrats have automatically participated as unpledged free agents at the nominating convention (Mayer, 2009, p. 93). Although the superdelegates’ direct influence on candidate selection has been limited (Herrera, 1994; Mayer, 2009), the superdelegates’ endorsements, along with those of other elites, matter for the dynamics of the nomination process. Recent empirical work finds that elite endorsements are particularly important to the process of converging on a nominee (Cohen et al., 2008; Dowdle, Adkins, & Steger, 2009; Steger, 2008). As a result, early endorsement decisions are stronger predictors of delegate share than pre-Iowa polling (Cohen et al., 2008, p. 292). In short, the parties use the “invisible primary” to converge on acceptable party candidates, and the endorsement decisions of elites are central to that process. According to Cohen and colleagues, the process has become “the equivalent of bargaining at party conventions” in the pre-McGovern-Fraser system (p. 187).

The Democratic party’s PLEO rule gives particular focus to superdelegates as elite endorsers and provides all House members and senators with a stage for public action to influence presidential politics. Not all members will choose to issue a public preconvention endorsement, but for those who do, the direction and timing of the choice is a calculated decision with potentially significant consequences for members’ multiple goals. As elected officials with strong incentives to respond to local constituencies, Congress members’ endorsement choices are more complex than they are for some other party elites. Cohen and colleagues (2008) acknowledge that this is a “complicating factor” for the argument that elites will use endorsements to move quickly toward a consensus nominee:

Consider an officeholder whose base is some factional group—a religious or racial group, a union, a political movement like the nuclear freeze of the early 1980s, or an ideological wing of the party. This politician can support the candidate favored or perhaps even revered by members of her political base—someone, for example, with the standing of the Rev. Jesse Jackson in the African American community—or she can support a candidate more likely to unite the party. If she goes with her group’s candidate, few members of her group are likely to complain. Even if the factional candidate flops in the primary contests, no one can blame her if she has been on the correct side. If, on the other hand, the politician supports the candidate with broad acceptability within the party as a whole, she may be accused of being a turncoat or disloyal to the group, an outcome that could endanger her standing with her political base. (p. 174)

In the 2008 nomination contest, with its close competition between Hillary Clinton and Barack Obama, endorsement choices received heightened scrutiny from electoral constituencies. After the first several contests, Obama’s margin of pledged delegates from state contests settled in at about 150 and held near that level through the end of the primary/caucus phase in June. During that time, the campaigns stepped up their efforts to lobby the superdelegates (Hearn, 2008a, 2008b; Parnes & Hearn, 2008). Congressional superdelegates found themselves facing a difficult choice. Although the superdelegates would not, in the end, reverse the outcome as determined by the delegates chosen in primaries and caucuses, House members and senators were still required to engage in a very public balancing of their goals. In this environment, Democratic senators and House members faced several choices: whether to endorse, when to endorse, and whom to endorse. Although these choices were outside the formal legislative process, we expect that members approached the decision in a way that reflects the calculus on salient legislative choices.

The importance of constituency in such choices is well established in the congressional literature (e.g., Bishin, 2000; Fenno, 1978; Mayhew, 1974), and we know that other goals are considered in the context of constituency constraints (Kingdon, 1989). For members considering the 2008 endorsement decision, contemplating the constituency took on a clear focus. The preferences and reactions of an identifiable subconstituency—active Democrats in the district participating in the presidential nomination process—should have been a central consideration, especially as those constituents were activated by the intense primary campaign. In Arnold’s terms, this subconstituency constituted a highly attentive public with strong preferences who were also part of the member’s core primary constituency (Arnold, 1990, p. 84; Fenno, 1978). When this public was attentive and its presidential nomination preferences clearly on one side, a member’s endorsement decision should have been fairly straightforward. But the decision was not this clear for many members. Some members faced divided constituencies in which a sizable attentive public was activated on both sides of the presidential nomination question. For some, the constituency question was further complicated by calculations about how the identity of the Democratic nominee might affect their own close general election races (Hearn, 2008b; Parnes & Hearn, 2008).

Members also considered the effects of their choice on objectives other than their own potential primary or general election fates. To the extent that Democrats perceived Clinton and Obama as representing not only stylistic but also substantive differences, the endorsement choice had direct and significant implications for members’ personal policy preferences. In a related way, Senators Clinton and Obama had built connections with their fellow legislators, and personal loyalty to one candidate—and the potential for future career advancement that comes along with it—weighed on the endorsement decision as well.

To further complicate matters, the changing dynamics of the long campaign season affected the balance of constituency, policy, and power considerations. The long and changeable 2008 nominating process presented members with a shifting political context that may have altered how members viewed the endorsement choice over time, especially as the identity of the apparent consensus nominee changed across the invisible primary and primary/caucus seasons. By viewing members’ endorsements with an eye toward the influence of campaign dynamics on endorsement choice and timing, we can gain insight into how members balance the broader political considerations that the nominations literature emphasizes against the factors that shape members’ individual choices.

Measuring Member Endorsements

To collect the endorsement data for the Democratic members of Congress, we sought to determine whom each congressional superdelegate supported and when the endorsement was announced using primary sources (press releases) where possible and secondary sources when primary sources were unavailable.³ When a member’s endorsement was announced by the presidential campaign rather than by the member, we used keyword searches on Lexis and Google News to verify the endorsement and its date in press coverage. In several cases, we were unable to find endorsement announcements by members of Congress before June 7, 2008, the date Clinton withdrew from the race (see below). In these cases, we assumed that the superdelegate did not endorse a nominee. We included Representatives and Senators as long as they served in Congress at any point between March 1, 2007, and June 7, 2008. When members departed or joined Congress in this time frame, the data set reflects those entries and exits.⁴

Figure 1 shows the number of congressional superdelegate endorsements over time. Lacking any commonly agreed-upon date of the “start” of the 2008 presidential campaign (Joe Biden announced his candidacy before the 2006 congressional elections; Obama and Clinton formally announced their campaigns in February 2007), we marked the beginning of our data set as March 1, 2007.

Figure 1.

Congressional superdelegate endorsements from March 1, 2007 to June 7, 2008

By that time, 24 members of Congress had endorsed Hillary Clinton and 14 members of Congress had endorsed Barack Obama.⁵ Clinton slowly accumulated congressional endorsements throughout 2007, ending the year with 76 endorsements, 43 more than Obama. As states began to hold their contests, Obama began to accumulate endorsements rapidly. After winning the Iowa caucuses, Obama took an early lead in pledged delegates⁶ and began to catch up with Clinton in congressional superdelegate endorsements. The rate of endorsement slowed considerably during March and April as Clinton won some key primaries. On May 6, Obama won North Carolina by almost 15 percentage points. This surprise started a surge of superdelegate endorsements, and 9 days later, Clinton lost her lead in superdelegate endorsements. On June 3, 2008, Obama won enough delegates (both superdelegates and pledged delegates) to win the nomination. On June 7, 2008, Clinton endorsed Obama. In our models, we include congressional superdelegate decisions through June 7.⁷

Influences on the Endorsement Decision: Goals and Context

For a theoretical framework for understanding elite endorsement choices, we turn to research on legislative decision making. Although presidential endorsement decisions are outside of the lawmaking context, we hypothesize that this highly publicized choice for congressional superdelegates activated the same individual-level factors that shape legislative choices (Arnold, 1990; Kingdon, 1989). These include electoral considerations, power and career factors, and policy and personal preferences. We also test hypotheses about how the national political context both shaped the endorsement choice and altered the influence of individual-level factors, elaborating on legislative research that suggests a relationship between when members make a decision and how they make it (Fenno, 1986).

Electoral Considerations

We expect that members considered the effect of their endorsement decision on their own electoral security. Anecdotal evidence suggests that members recognized the potential for the choice to become an issue in their own future primaries and that the identity of the party’s presidential candidate could affect the level of support members received at the polls in November 2008. Strong pro-Obama or pro-Clinton sentiment in members’ Democratic subconstituencies indicates an active and intense public (Arnold, 1990). As a result, we hypothesize that higher levels of district support for Obama increased the likelihood of a member endorsing Obama and decreased the likelihood of a member endorsing Clinton. We include the difference between Obama’s vote share and Clinton’s vote share for each superdelegate’s congressional district. We assume that members had a reasonably good sense for their districts’ preferences even in advance of the state’s primary or caucus.⁸ However, as we argue below, there is greater certainty about district preferences after the state primary. In other words, context will affect the nature of this relationship.

Power and Career Factors

The presidential choice had potential implications for members’ political ambitions beyond reelection. For members with political connections to one of the candidates, the outcome of the nomination contest would have an effect on the member’s prospects for career goals of higher office and/or power within Congress. Intuitively, we expect that members with closer political ties to Obama were more likely to endorse Obama, while those with closer political ties to Clinton were more likely to endorse Clinton. We measure political ties in two ways. First, we measure the campaign–finance connections between each member and the two presidential candidates. We code the total amount each presidential candidate’s PAC contributed to each congressional superdelegate across the 2006 and 2008 election cycles.⁹ Over these two election cycles, Obama was more aggressive in contributing to other members. Obama gave contributions to 35% of Democratic MCs, whereas Clinton gave to 14%. Obama’s average 2006-2008 contribution was US$6,900 to Clinton’s US$6,500. As a result of these two disparities, the average congressional superdelegate received about US$1,500 more from Obama than Clinton. As a second measure of political ties, we use member seniority to capture the less visible connections that Clinton had relative to Obama. Clinton’s nearly four decades of involvement in national Democratic politics and her own tenure in the Senate dating to 2000 likely gave Clinton closer political ties to senior members of Congress.

Policy and Personal Preferences

Individual members may have been affected by ideological considerations in the choice over the 2008 nominee. The overall ideological differences between the two major candidates were “negligible,” with Barack Obama “at most marginally more liberal” than Hillary Clinton (Carroll et al., 2008). However, because of their differences on a few particularly salient policy issues—notably the Iraq war, which had a strong influence on perceptions of the candidates (Jacobson, 2010)—liberal members may have perceived Obama to be the candidate more in line with their policy preferences. We expect, as a result, that liberal Democrats were more likely to endorse Obama than conservative Democrats. We measure ideology using first dimension DW-NOMINATE scores, recoded so that higher scores indicate greater liberalism.¹⁰

Race and gender were factors in the Democrats’ 2008 nomination process, and female and African-American members in particular may have considered the historic opportunity to advance the presidential candidacy of a woman or an African-American. Thus, we expect that African-American MCs were more likely to endorse Obama and female MCs were more likely to endorse Clinton. However, since these two factors together produce greater uncertainty, we expect that the effect of race and gender on the endorsement choice was diminished for African-American women in Congress. To capture these effects we include variables for African Americans, women, and an interaction between race and gender for African American women.

Contextual Factors

With their central focus on reelection, members of Congress depend on their ability to anticipate and react correctly to political dynamics. In 2008, Democratic members were closely following the changing dynamics of the nomination race and the prospects of the major candidates. Particularly for MCs who faced reelection in 2008, the outcome of the nomination contest promised to be an important external factor in their own political fortunes—multiple goals gave members reasons to value being associated with the winner of the presidential nomination. And, as elite endorsers, congressional superdelegates would sense the need to form a “united front” around a consensus nominee (Cohen et al., 2008, p. 173). As a result, we expect that members became more likely to endorse Obama as Obama’s likelihood of securing the nomination increased relative to Clinton’s. Similarly, members became more likely to endorse Clinton as Obama’s likelihood of securing the nomination decreased relative to Clinton’s.

Campaign dynamics are measured using Iowa Political Market (IPM) data (see Forsythe, Nelson, Neumann, & Wright, 1992). The IPM works by allowing investors to buy shares of a candidate. When the nomination is finalized, shares of the winning candidate can be cashed in for US$1 per share whereas losing candidate shares are worth nothing. The price at any given time therefore reflects the market’s expectation that a candidate will win the nomination. Figure 2 shows the daily closing share price for Clinton and Obama. IPM prices are better measures of campaign dynamics than measures like accumulated delegates because the prices are based on expectations. Thus, they simultaneously include information about past events (winning contests) and what current observers expect to happen in the future. Accumulated delegate counts are only retrospective, but IPM data can capture the current informed impression of a candidate’s chances of winning. Furthermore, IPM data have consistently outperformed traditional opinion polls in predicting election outcomes (Berg, Nelson, & Rietz, 2008).

Figure 2.

Iowa political market prices for Obama and Clinton from March 1, 2007, to June 7, 2008

In addition to changes in the perceived likelihood of Obama’s success, we argue that there were changes in the interpretive context as well. Political decisions can have different meanings at different points in the arc of an issue’s consideration, as legislative research shows (Fenno, 1986). For a congressional superdelegate, a critical change in the decision environment occurred when the member’s state held its primary or caucus. We expect the home-state results to demarcate a key change in the superdelegates’ decision context. Specifically, the importance of subconstituency opinion (Obama vote margin) and the national political context (likelihood of Obama victory) may change once the district has formally registered its preference. We test these expectations using interactions between an indicator for whether the home-state contest has taken place and the vote margin and IPM variables.¹¹

Results

Our main model of congressional superdelegate decision uses a dependent variable that divides members into three categories based on their initial endorsement: Obama, Clinton, or no endorsement/other candidate. Because the choices made by superdelegates are not ordered and violate the assumption of the independence of irrelevant alternatives, we estimate a multinomial probit (MNP) model of this choice, using a data set in which each observation is a member-day during the time period from March 1, 2007, to June 7, 2008.¹² Observations on each MC end in the data set on the date when that MC first endorses either Clinton or Obama.¹³ The model includes a variable for the logged duration of the MC in the data set at time t in order to account for duration dependence in the data (Box-Steffensmeier & Jones, 2004). The baseline category for the model is endorsement of neither Obama nor Clinton (i.e., no endorsement or endorsement of a third candidate at t).

The first column of Table 1 shows the estimates for the covariates’ effects on endorsing Clinton at t relative to a baseline of endorsing neither Obama nor Clinton; the second column reports the estimates of the covariates’ effects on endorsing Obama at t relative to the baseline. Turning first to power and career factors, a member’s personal connections had a measurable effect on the probability of issuing a Clinton or Obama endorsement. As expected, higher levels of contributions from Obama increase a member’s probability of endorsing Obama but decrease the probability of endorsing Clinton, with the reverse effect appearing for contributions from Clinton. The impact of these effects is substantial; a member who received at least US$10,000 in contributions from Obama within the past two election cycles (which was the case for 30 congressional superdelegates) was more than one and a half times more likely to endorse Obama than a member who didn’t receive any money. The effect for Clinton was even greater. A member who received at least US$10,000 from Clinton (which was the case for 15 superdelgates) was more than 6 times more likely to endorse Clinton than a member who did not receive any money. Unlike financial connections, the expected link between seniority and endorsement choice is not supported in this model.

Table 1.

Congressional Superdelegate Endorsement of Clinton or Obama

	Clinton	Obama
Electoral considerations
Obama margin in district (precontest)	−0.010***	0.005**
	(0.002)	(0.002)
Obama margin in district (postcontest)^a	−0.026***	0.005
	(0.006)	(0.003)
Power and career factors
Obama PAC contributions (in $1000s)	−0.077***	0.040***
	(0.023)	(0.012)
Clinton PAC contributions (in $1000s)	0.081***	−0.048***
	(0.020)	(0.017)
Seniority	−0.005	−0.001
	(0.005)	(0.004)
Policy and personal preferences
Liberal ideology	0.038	0.974***
	(0.296)	(0.255)
Woman	0.331***	0.045
	(0.114)	(0.114)
African American	0.797***	0.078
	(0.179)	(0.179)
Woman × African American	−0.546**	−0.415
	(0.246)	(0.327)
National political context
Obama IPM advantage (precontest)	−0.706**	1.182***
	(0.348)	(0.165)
Obama IPM advantage (postcontest)^b	−0.351	4.671***
	(0.731)	(0.810)
Control variables
Postcontest dummy variable	0.665	−2.460***
	(0.427)	(0.688)
Endorsed other candidate_(t-1)	−0.480**	0.007
	(0.243)	(0.120)
Log duration	−0.307***	−0.108^*
	(0.049)	(0.065)
Constant	−3.150***	−4.189***
	(0.205)	(0.358)
Observations	76,405
Log likelihood	−1523
Number of individuals	269

Note: Cell entries are multinomial probit coefficients; robust standard errors are in parentheses. The baseline category is endorsing neither Clinton nor Obama.

Coefficient is the sum of the Obama margin coefficient and the coefficient for an interaction term for after the state contest (–0.016 for Clinton; –0.0005 for Obama). Standard error is calculated as √((var(b1) +X22var(b3) + 2X2cov(b1b3)) where b₁ is the main effect of Obama margin and b₂ is the interaction coefficient between Obama Margin and the dummy variable for postcontest. Since χ² is a dummy variable for postcontest, all χ² are set to 1.

Coefficient is the sum of the IPM coefficient and the coefficient for an interaction term for after the state contest (0.354 for Clinton; 3.488 for Obama). Standard error is calculated using the same formula as for Obama margin (see Note a) where b₁ is the main effect of IPM and b₂ is the interaction coefficient between IPM and the dummy variable for postcontest.

p < .10. **p < .05. ***p < .01 (two-tailed tests).

Ideology, race, and gender each shaped members’ choices over which candidate to endorse. Ideology has a significant and sizable effect on the probability of endorsing Obama but not on the probability of endorsing Clinton relative to endorsing neither candidate at time t. For the most liberal Democrat, the probability of endorsing Obama is almost 9 times as large as the probability for a Democrat at the ideological center of the party. Although ideology had an independent effect on the probability of endorsing Obama but not Clinton, the race and gender of congressional superdelegates was related to their decision to endorse Clinton but did not significantly affect the decision to endorse Obama. White female MCs were 56.6% more likely to endorse Clinton than white males, and African-American men were 85.4% more likely to endorse Clinton than white men. For African African-American, though, the effect of race and gender is more complicated. African-American women were 45.7% more likely to endorse Clinton than white women. But African-American women were 60.9% less likely to endorse Clinton than African-American men.

Electoral considerations affected the decision to endorse Clinton, and they strengthened once the district held its state-level delegate selection contest. Prior to the contest, the district’s level of Obama support is associated with a lower likelihood of endorsing Clinton (a decrease of about 42% in the predicted probability as Obama’s district margin ranges from a 10-point lead by Clinton to a 10-point lead by Obama).¹⁴ After the subconstituency’s preferences were formally recorded in the primary or caucus, the negative relationship between the level of Obama support and the probability of endorsing Clinton strengthens (a decrease of about 69% in the predicted probability as Obama’s district margin ranges from a 10-point lead by Clinton to a 10-point lead by Obama).

Electoral factors matter for the Obama choice as well, with stronger subconstituency support for Obama corresponding to a higher probability of endorsing Obama on any given day. However, the effect of district-level Obama support is statistically significant only before the state-level results are recorded, suggesting that Obama endorsers tended to perceive and respond to constituency preferences earlier in the campaign season, perhaps to gain both electoral advantages and nomination-contest influence from their early endorsement.

The national political context had a substantial impact on endorsement choice. Prior to their district primary or caucus, members became less likely to endorse Clinton as Obama’s relative IPM price increased; a shift from a 10-cent lead by Clinton to a 10-cent lead by Obama translates to a 32% decrease in the predicted probability. Members’ likelihood of endorsing Obama similarly varied with his IPM value (an increase of about 77% in the predicted probability as Obama’s IPM price ranges from a 10-cent lead by Clinton to a 10-cent lead by Obama). After the state-level contest, the impact of the national political context on superdelegate decisions decreases for Clinton and increases for Obama.

It appears that the state-level contest changed the weight that congressional superdelegates placed on constituency opinion and the national political context. But this change was different for the decision to endorse Clinton and the decision to endorse Obama. Before the state contest, both expectations about the district decision and reading of the national political context mattered for making an endorsement. After the state contest, the district decision was more important than the national context for those who endorsed Clinton. For those who endorsed Obama, the national context was more important than the state contest results.

Another way to explore the impact of context on decisions is to divide the entire primary season into distinct periods. As Figure 2 indicates, the campaign unfolded in three distinct phases. The first is the period before the Iowa Caucus (January 3, 2008). During this time, Clinton was seen as the clear favorite to win the nomination. Her average price advantage over Obama was 28.4 cents. Obama’s win in the Iowa Caucus marked the beginning of a new period in which Clinton’s lead was questioned. Yet, up until Super Tuesday (February 5, 2008), Clinton maintained a lead over Obama in expectations (her price advantage was 13.8 cents on average). After Super Tuesday, expectations were in Obama’s favor (his price advantage was 61 cents on average).

The influence of electoral considerations varied as the primary season evolved. Figure 3 shows the instantaneous probability of endorsing Obama and Clinton by Obama’s electoral margin in the district for three phases in the nomination contest.¹⁵ Figure 3a shows that the overall probability of endorsing Obama or Clinton before the Iowa Caucus is relatively low (compared to later periods). There is also a general bias in favor of Clinton, shown by the point where the probability of endorsing Obama becomes higher than the probability of endorsing Clinton (around 15% of net Obama support). Thus, prior to the Iowa Caucus, Clinton was the most likely choice for superdelegates, unless the superdelegate represented a district or state that was strongly supportive of Obama.

Figure 3.

Predicted probability of endorsing Obama and Clinton by Obama’s electoral margin for three phases of the nomination season: (a) Before Iowa caucus, (b) After Iowa Caucus and before Super Tuesday, and (c) After Super Tuesday

Figure 3b shows the predicted probabilities after the Iowa Caucus but before Super Tuesday. To make the graph easier to read, we include only the predicted probability curves for endorsing Clinton after the state contest and for endorsing Obama before the state contest. For each candidate these reflect the highest probabilities across the range of election outcomes. That is, superdelegates were more likely to endorse Clinton after the state contest than before; the opposite is true for Obama. During this period, superdelegates whose districts favored Clinton were likely to endorse her. With expectations still on Clinton’s side, superdelegates with districts that were only slightly in favor of Obama were still more likely to endorse Clinton.

Figure 3c shows the probabilities after Super Tuesday. As in the prior period, superdelegates were more likely to endorse Clinton after their state’s nomination contest and to endorse Obama before the contest, so the figure contains only those two probability curves. The critical point where the Clinton and Obama curves intersect has shifted dramatically. After Super Tuesday, superdelegates were more likely to endorse Obama, unless their district supported Clinton by more than a 25% margin.

Discussion and Conclusion

Our findings demonstrate the interplay of individual goals and changing political contexts in elite presidential endorsement decisions. Different factors drove members who endorsed Obama and those who endorsed Clinton, and the balance of considerations changed as the political environment shifted. District and national factors can pull members in different directions, and those two forces can be mediated by a member’s own personal preferences. Starting with the personal preferences, we find very much what we would expect in a static political environment; women were more likely to endorse Clinton (though, so, too, were African Americans, all else equal), liberals were more likely to endorse Obama, and both presidential candidates reaped the rewards of their PAC largesse. Before the caucus or primary is held in a member’s district, Clinton and Obama endorsers were influenced similarly by electoral and contextual factors; members in districts more supportive of Clinton (Obama) were more likely to endorse Clinton (Obama). As Obama’s lead in the IPMs grew, members became less likely to endorse Clinton and more likely to endorse Obama.

The separation between Clinton and Obama endorsers emerges after the state contest is held. After the contest, district-level factors appear to have a particularly strong pull on Clinton endorsers; the effect of Clinton’s relative vote margin in their districts strengthens after the state contest, suggesting that members who represented districts where Clinton did well were more likely to endorse her after she performed well in their districts. At the same time, Obama’s national momentum has no effect on those members who decided to endorse Clinton after the state contest. Electoral and contextual factors operate very differently on the decision to endorse Obama after the local primary or caucus. In the later phase of the campaign, it is not Obama’s performance in the local election but rather his momentum in the overall campaign for delegates that shapes the decision to endorse his candidacy.

This postcontest separation offers an important insight into the interplay of localized and national considerations in endorsements. When the two point in the same way (and, in particular, when they coincide with a member’s personal preferences), the decision is quite straightforward. The challenge arises when those forces pull in different directions. Our analysis suggests that the broader political dynamics add a dimension to the choice. At the same time, that added dimension offers opportunities for members to maneuver. That is, members are given an additional piece of information they can use in both calculating and justifying the decision they make. Clinton endorsers, after state contests, could point to the preferences of their district. Obama endorsers, after state contests, could increasingly point to national momentum and make a case for party unity behind the candidate they viewed as the likely winner. The added dimension of the campaign dynamics, then, can actually offer members more freedom to make a defensible decision.

Our empirical findings enhance and elaborate the presidential-selection literature’s argument about the key role of party elites. For elites who are very visible and face regular electoral accountability from defined constituencies—especially in a contentious contest like 2008—the interest in converging on a consensus party choice is weighed against constituency and other interests. As we demonstrate, strong constituency signals can outweigh national considerations for party elites in this position; at the same time, we provide evidence that many members who endorsed the emerging frontrunner were strongly influenced by the national party considerations as the nominations literature would predict.

Although our explanatory focus has been on members of Congress as elite endorsers, we also note that the findings speak to a small body of literature that examines member behavior outside of the legislative context (e.g., Mayhew, 2000). Presidential endorsement decisions provide an excellent example of a visible and consequential nonlegislative choice made by a large subset of members in a defined time frame. This data-rich example establishes that members mediate between personal, power, and particularly electoral goals in making these extralegislative decisions, balancing the representational pull of the electoral connection against more personal objectives in ways that parallel legislative decision making. These findings complement and extend the limited existing work on member decisions outside of the lawmaking process (see, in particular, Green & Harris, 2007; Highton & Rocca, 2005). Moreover, this case elaborates the role of context in shaping the interplay of competing goals in member decisions. Viewing members through the lens of their involvement in presidential selection politics allows us to see both the relevance of multiple goals and the time-dependent nature of members’ efforts to mediate between them.

Footnotes

Author’s Note

This article reflects the views of the authors and does not necessarily reflect the views of the Administrative Office of the U.S. Courts. We thank Jan Box-Steffensmeier, Chris Ellis, Staci Rhine, Sarah Treul, and Margie Williams for valuable comments and conversations. An earlier version of this work was presented at the 2010 meetings of the Midwest Political Science Association.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Bios

Edward B. Hasecke is an Associate Professor of Political Science, Wittenberg University. His research interests focus on congressional lawmaking, with recent publications appearing in Political Research Quarterly and this journal.

Scott R. Meinke is an Associate Professor of Political Science at Bucknell University. His research addresses legislative institutions and representation. His current work examines party leadership organizations in the House of Representatives, and his recent publications have appeared in Political Research Quarterly, Legislative Studies Quarterly, and this journal.

Kevin M. Scott is a Social Science Analyst for the Statistics Division, Administrative Office of the U.S. Courts. His research interests focus on judicial selection and judicial decisionmaking, with recent publications appearing in Presidential Studies Quarterly, Political Research Quarterly, and this journal.

References

Anderson

C. L.

(in press). Which party elites choose to lead the nomination process? Political Research Quarterly.

Arnold

R. D.

(1990). The logic of congressional action. New Haven, CT: Yale University Press.

Berg

J. E.

Nelson

F. D.

Rietz

T. A.

(2008). Prediction market accuracy in the long run. International Journal of Forecasting, 24, 285-300.

Bishin

B. G.

(2000). Constituency influence in Congress: Does subconstituency matter? Legislative Studies Quarterly, 25, 389-415.

Box-Steffensmeier

J. M.

Arnold

L. W.

Zorn

C. J. W.

(1997). The strategic timing of position taking in Congress: A study of the North American Free Trade Agreement. American Political Science Review, 91, 324-338.

Box-Steffensmeier

J. M.

Jones

B. S.

(2004). Event history modeling: A guide for social scientists. Cambridge, UK: Cambridge University Press.

Caldeira

G. A.

Zorn

C. J. W.

(2004). Strategic timing, position-taking, and impeachment in the House of Representatives. Political Research Quarterly, 57, 517-527.

Carroll

Lewis

McCarty

Poole

Rosenthal

(2008). Who is more liberal, Senator Obama or Senator Clinton? Retrieved from http://voteview.com/Clinton_and_Obama.htm

Cohen

Karol

Noel

Zaller

(2008). The party decides: Presidential nominations before and after reform. Chicago, IL: University of Chicago Press.

10.

Dow

J. K.

Endersby

J. W.

(2004). Multinomial probit and multinomial logit: A comparison of choice models for voting research. Electoral Studies, 23, 107-122.

11.

Dowdle

A. J.

Adkins

R. E.

Steger

W. P.

(2009). The viability primary: Modeling candidate support before the primaries. Political Research Quarterly, 62, 77-91.

12.

Fenno

R. F.

Jr. (1978). Home style: House members in their districts. Glenview, IL: Scott, Foresman and Company.

13.

Fenno

R. F.

Jr. (1986). Observation, context, and sequence in the study of politics. American Political Science Review, 80, 3-15.

14.

Forsythe

Nelson

Neumann

Wright

(1992). Anatomy of an experimental political stock market. American Economic Review, 82, 1142-1161.

15.

Green

M. N.

Harris

D. B.

(2007). Goal salience and the 2006 race for House Majority Leader. Political Research Quarterly, 60, 618-630.

16.

Hearn

(2008a, February 7). Super Tuesday over, superdelegates enlisted. Politico. Retrieved from http://www.politico.com

17.

Hearn

(2008b, April 4). Superdelegates fret over blowback. Politico. Retrieved from http://www.politico.com

18.

Herrera

(1994). Are “Superdelegates” super? Political Behavior, 16, 79-92.

19.

Highton

Rocca

M. S.

(2005). Beyond the roll-call arena: The determinants of position-taking in Congress. Political Research Quarterly, 58, 303-316.

20.

Hook

(2008, April 7). Some delegates consider their own fate. Los Angeles Times. Retrieved from http://www.latimes.com

21.

Jacobson

G. C.

(2010). George W. Bush, the Iraq War, and the election of Barack Obama. Presidential Studies Quarterly, 40, 207-224.

22.

Judd

K. L.

(1998). Numerical methods in economics. Cambridge, MA: MIT Press.

23.

Kingdon

J. W.

(1989). Congressmen’s voting decisions (3rd ed.). Ann Arbor: University of Michigan Press.

24.

Mayer

W. G.

(2009). Superdelegates: Reforming the reforms revisited. In Smith

S. S.

Springer

M. J.

(Eds.), Reforming the presidential nomination process (pp. 85-108). Washington, DC: Brookings Institute.

25.

Mayhew

D. R.

(1974). Congress: The electoral connection. New Haven, CT: Yale University Press.

26.

Mayhew

D. R.

(2000). America’s Congress: Actions in the public sphere, James Madison through Newt Gingrich. New Haven, CT: Yale University Press.

27.

Nagourney

Zeleny

(2008, March 16). For Democrats, increased fears of a long fight. New York Times. Retrieved from http://www.nytimes.com

28.

Parnes

Hearn

(2008, April 30). Obama backer predicts victory in Hill war. Politico. Retrieved from http://www.politico.com

29.

Presidential endorsement race. (2007, February 13). The Hill. Retrieved from http://www.thehill.com

30.

Southwell

P. L.

(2012). A backroom without the smoke? Superdelegates and the 2008 Democratic nomination process. Party Politics, 18, 267-283.

31.

Steger

W. P.

(2008). Interparty differences in elite support for presidential nomination candidates. American Politics Research, 36, 724-749.

32.

Welna

(2008, April 7). In Senate, a dilemma for 19 superdelegates. NPR.org. Retrieved from http://www.npr.org/templates/story/story.php?storyId=89424926