The president's representational responsibilities are unique and varied. Because a national constituency elects the president, the centrist model of representation contends that presidents must respond to and lead the entire nation. Woodrow Wilson (1961, 67-68) observed this when he wrote that as "political leader of the nation," the president is "representative of no constituency but of the whole people." James MacGregor Burns (1973, 106) echoed this perception: "the President is custodian of popular safety, national destiny, and the conscience of the people." Consistent with the centrist view of representation, numerous scholars have found that presidents are highly responsive to changes in national public mood (Erikson, MacKuen, and Stimson 2002; Jacobs 1992; Stimson, Erikson, and MacKuen 1995), respond to the national public concerns about foreign and economic issues (Cohen 1999), and respond to national public preferences conditionally, by issue area, popularity, and the president's electoral cycle (Canes-Wrone and Shorts 2004; Rottinghaus 2006).
Being a partisan political figure, the president also represents party interests (Skinner 2008). This partisan model of representation holds that presidents must consider their partisan's policy preferences to win their party's nomination (see Key 1964), and the president's success in Congress is predicated on party control (Bond and Fleisher 1990; Edwards 1989). Unsurprisingly, party predicts presidential liberalism and contributes to presidential representation of public opinion (Stimson, MacKuen, and Erikson 1995). Wlezien (1996), in particular, finds that the president's party affiliation is a strong indicator of representation on defense spending in the 1970s and 1980s. Wood (2009) not only shows that the partisan model best explains the relationship between presidential liberalism and public mood, but, consistent with Wlezien's (1995) thermostatic model of representation, he also demonstrates that the national public does not follow, but rather tends to move away from the president's policy liberalism over time.
Yet, debate persists on two levels regarding the president's responsiveness to public opinion. First, despite the plausibility that presidents may prefer taking positions on partisan issues and recent evidence that confirms this tendency (Wood 2009), most research that considers presidential-public relationships builds upon the centrist model of representation, that presidents represent the nation. Canes-Wrone's (2006) strategic model of going public, for example, hinges on presidential leadership of issues that are popular with the mass public. Canes-Wrone and Shotts (2004) also develop a model of centrist presidential representation. Still others argue that presidents may try to satisfy both constituencies by switching their support between mass and partisan public opinion (Druckman and Jacobs 2006; Pious 1996, 184), but this is not without costs, as the public shifts its support away from the president when he takes partisan positions (Wood 2009, 158). Second, public opinion is measured in both aggregated, ideological ways (Simson, MacKuen, and Erikson 1995; Wood 2009; Wood and Lee 2009) and directly through public opinion polling (Canes-Wrone and Shorts 2004; Rottinghaus 2006). Despite having certain advantages, these alternative measures may produce different conclusions about presidential representation.
Since presidents have myriad reasons to represent partisans and there are multiple ways to measure representation, reexamining presidential representation may clarify the role presidential partisans play in the president's policy positions. The purpose of this article is to build on the existing literature by analyzing the impact of mass and partisan opinion on the representational position taking of several presidents. We ask, do presidents tend to be more representative of mass or partisan publics on their specific policy positions? And what factors explain this behavior? We first catalog presidential partisan and national representation according to the president's publicly stated positions on specific policies from 1989 to 2008. We then match the president's positions with national and partisan survey responses to those issues and indicate whether a majority of the national or partisan public supported the president. Our theoretical framework, adapted from Canes-Wrone and Shorts (2004) and Wood (2009), illustrates that presidents are more representative of partisan than national preferences and are most representative of partisan preferences when they are popular during presidential reelection years.
This topic is important for several reasons. First, the type and quality of representation is fundamental to exploring the effectiveness of a democracy. A tradition of democratic theory assumes that presidents represent the nation. But if they do not, we may wish to revisit what a partisan representational presidency means for presidential responsiveness, presidential speechmaking, and legislative success. Second, the tendency for presidents to represent the masses or their partisans is important for how presidents lead as the only nationally elected official in the United States. If we find more evidence of partisan than centrist leadership, this raises questions about the conditions under which presidents may actually represent the nation and not their partisans. Third, the relative balance between partisan and centrist presidential representation may encourage researchers to ask whether presidents can successfully go public on partisan issues, or whether they are bound to speak on issues that have majority support of the mass public.
Centrist and Partisan Presidential Representation
Presidential representation may be conceived of in two forms. On the one hand, according to the centrist model of representation, presidents take issue positions that are congruent with a majority of the mass public to build favorable approval ratings, which assist in the president's policy and reelection goal achievement. Presidents who are more popular with the American people stand a better chance of influencing Congress (Neustadt 1990). If presidents hope to win reelection, having a majority of Americans approve of the president's job performance predicts reelection victory (Holbrook 1994), especially as reelection day nears. Indeed, one constant of political representation in the United States is that it varies by electoral cycle. As scholars have repeatedly shown that senators become more representative of their constituents as their reelection campaign nears (Elling 1982; Kuklinski 1978), so too has scholarship begun to test the variation in presidential representation by proximity to Election Day.
In a seminal article, Canes-Wrone and Shorts (2004) (or CWS, hereafter) held that presidents are more likely to take positions popular with a majority of the public during the second half of their first term in office. Simply, presidents appeal (or even pander) to the public when Election Day nears so as to maximize voter support. Yet proximity to reelection year is only one factor in the president's decision calculus. The president's likelihood of advocating policies that are congruent with a majority of public opinion tends to be conditional on his current popularity. CWS demonstrated that presidents with below (above) average approval are more (less) likely to be congruent with national public opinion as their reelection nears. In other words, presidents take positions popular with the majority when they are unpopular to improve their popularity and chances for reelection.
On the other hand, presidents are responsive to partisan publics. Wood (2009) illustrates that presidents are more likely to respond to trends in partisan, not mass public opinion. Bailey, Sigelman, and Wilcox (2003) find that Democrats were more likely to support President Bill Clinton's policy on gays in the military. Murray (2006) notes that President Ronald Reagan was more responsive to Republican Party activists (see also Druckman and Jacobs 2011). In addition, Miller and Sigelman (1978) report that President Lyndon Johnson tailored his public statements on Vietnam to the hawkish or dovish partisan Democratic audience to whom he was speaking. Signals sent to core political consistencies are often as important as leading public opinion in the aggregate sense and is often easier for the White House to accomplish.
There are several reasons why presidents should be representative of partisan opinion. If presidents have to bargain with Congress to achieve their legislative goals while also desiring reelection, they will be more interested in responding to preference of active segments of their party. These individuals should be more proactive in the reelection process than the masses who may be placated with general ideological representation through rhetoric or symbolic actions (Druckman and Jacobs 2006). Especially when the president is in the majority, he has less incentive to respond to the opinions of the median voter but prefers to maintain his majority by responding to partisan preferences (Wood 2009). And when presidents are confronted with a reluctant and inattentive public (Edwards 2003), there are still critical issue publics that pay attention to the president. Presidents find partisan publics to be receptive audiences when attempting to build support for their policies (Brace and Hinckley 1992; Heith 2004; Rottinghaus 2006). It may also follow that if presidential rhetoric does not persuade the nation to support the president's issue positions (Edwards 2003), then a more productive strategy may be for presidents to propose policies that are already likely to be supported by their copartisans.
The partisan nature of primaries, ideology, and the president's job approval ratings also foretell presidential partisan representation. First, presidents may be more...