The Politics of Presidential Press Conferences

Abstract

What explains the president’s tendency to deliver press conferences? Unlike other forms of going public, the press conference presents a unique venue that allows presidents to grapple with their policy and political decisions in a highly public and interactive format with journalists. This should influence how the political environment affects not only the number of press conferences presidents deliver in a month but also the time between press conferences. Indeed, lower approval ratings and unified government lead to more solo press conferences and less time between them. Whereas reelection years and political scandal decrease the number of solo press conferences, only reelection years increase the time between them. This article also reminds us of the relevance of individual presidents to explaining press conference activity and has implications for a broader understanding of the president’s public appeals.

Keywords

president leadership media press conference

On October 6, 2011, President Barack Obama held a late morning press conference in the East Room of the White House.¹ In addition to addressing a joint session of Congress on national television and delivering speeches at several locations around the country, the president used his opening statement of this press conference to discuss his jobs bill. Clearly, this press conference was part of the president’s larger effort to advertise his jobs plan to the news media and American people. Yet it failed to generate substantial news coverage² and did not move public support for his jobs plan.³ Observers may not have expected otherwise, given the negative impact that press conferences appear to have on public opinion (Rottinghaus, 2010) and their limited effect on news coverage (Cohen, 2010, p. 116). Prospects may have been even bleaker for Obama since he faced an unfavorable political environment measured by a job approval rating at 39%⁴ and the unemployment rate stuck at 9.1%. Holding a press conference under these conditions surely contradicts the classical expectation, that presidents avoid press conferences when the political environment worsens (Lammers, 1981). Still, President Obama delivered this press conference, along with 65 others through his 41st month in office.

In the age of the permanent campaign, presidents have myriad options to reach out to the news media and the American people. Indeed, the president could have held a short question and answer session with reporters or targeted local media, a presidential communications strategy that is virtually guaranteed to generate news coverage and, as some conclude, increase public support for the president (Cohen, 2008, 2010). So why would President Obama hold a press conference when neither conventional wisdom nor political conditions foretold favorable treatment by the Washington Press Corps or an enthusiastic response by the American people? What explains the president’s decision to deliver a press conference?

There are two likely answers to these questions. First, the White House views the press conference as an opportunity to explain policy decisions before reporters and lead news coverage (Kumar, 2007, pp. 256-257). Although the White House also sees the press conference as an opportunity to lead the public, the unique interactive format of the press conference emphasizes the president’s effectiveness in justifying his actions, not leading public opinion. Second, the press expects the president to hold press conferences, at least occasionally. Thus, the time between press conferences should weigh on the president’s decision to hold them. All in all, the impact of the political environment on press conference activity should be the result of the tendency (and frequency) for presidents to want to explain their policy decisions as they grapple with reporters’ questions and how often reporters expect presidents to do so.

The purpose of this article, then, is to answer the following two questions. What explains the number of presidential press conferences by month? And what explains the time between presidential press conferences? To answer these questions, I examine solo, joint, and all presidential press conferences held by month between January 1989 and May 2012. I build upon the existing going public literature but argue that press conferences are qualitatively different from other public activities. Because press conferences are used to achieve a different set of goals—to help presidents explain their policy decisions before reporters and meet reporters’ expectations—than other means of going public, the political environment will affect presidential press conferences in different ways, too. I find that presidents deliver more, but primarily solo, press conferences during unified government and when their approval ratings are low. These conditions also decrease the time between press conferences, which help presidents to meet the press’ expectations of relatively frequent press conferences.

Exploring the politics of presidential press conferences is important to the study of American politics for several reasons. First, the presidential press conference is a part of our democratic tradition and is important to having a free press (Kumar, 2007, p. 255). The press conference, unlike other lower-profile avenues of president–press interactions, provides the press with an opportunity to check presidential power, perhaps to act as a watchdog, an important but underplayed feature of news media in American politics (see Bennett, Lawrence, & Livingston, 2007). By requiring presidents to deviate from a prepared script and think through questions posed by the “fourth branch” of government (Cook, 1998), reporters’ questions may force presidents to offer a more honest assessment of politics and public policy unlikely to be reflected in a president’s prepared remarks. Second, knowing the conditions under which presidents are most likely to deliver press conferences is important to understanding when the president is most likely to subject himself to rigorous questioning by the press. In turn, the duration analysis allows us to assess when we are likely to see heightened expectations for presidents to hold more press conferences. Third, this article speaks to the changing nature of president–press relations, which some deem have become less adversarial in recent years (Iyengar & McGrady, 2007, Chapter 4). Understanding variation in presidential press conference activity has important implications for assessing the overall tenor of president–press relations.

The State of the Presidential Press Conference

The presidential press conference is a significant public relations event. In one, a president typically issues a brief statement and then calls on reporters to answer their questions. Presidents can hold formal, primetime press conferences in the East Room or informal afternoon conferences in the press briefing room. They can hold them individually or jointly with foreign dignitaries. They can hold them in Washington, D.C., anywhere throughout the United States, or overseas. Although presidents can engage reporters in other formats, such as interviews or informal question and answer sessions, the press conference is unique and restricted to these types of interactions.⁵

The press conference was born at the dawn of the modern presidency, first held in an informal setting and characterized by scripted questions asked of Woodrow Wilson and then Franklin Roosevelt (Maltese, 1994, p. 4). Although Truman formalized the press conference, mimicking the press conference format we see today, it was with the advent of television during the Eisenhower administration that the press conference evolved from an “off-the-record” exchange to a public recording, later to be broadcast as a live exchange during the Kennedy years (Kumar, 2007, pp. xxviii-xxix). Presidents after Kennedy engaged in more avoidance than interaction through the press conference, prompting Grossman and Kumar (1981, p. 246) to conclude that presidents ceased giving press conferences when their political environments became more difficult. This coincided with an increase in the adversarial nature of reporters’ questions (Clayman, Elliot, Heritage, & Beckett, 2010). Just as the yearly number of press conferences declined through the Reagan years, however, the Bush and Clinton Administrations reversed this downward trend (Eshbaugh-Soha, 2003, p. 352), with recent presidents having held one press conference in, at minimum, 75% of their months in a term (Kumar, 2007, p. 267). These numbers reflect fewer primetime press conferences but an increase in the number of afternoon press conferences, press conferences held abroad, and those held jointly with foreign dignitaries. Perhaps because of these changes to the press conference format, the relative accountability of questions has declined steadily since Reagan (Clayman et al., 2010, p. 242) suggesting, at least, that press conferences are less adversarial than they used to be. These changes imply that the president’s decision as to when to hold a press conference may have also evolved based on a press conference’s perceived cost or benefit to the president’s communications strategy.⁶

Still, the rationale for the press conference has remained strikingly similar over the years: “to invoke a president’s unrehearsed response to questions, probe the justifications a president offers for his policies” and satisfy reporters’ desires that presidents hold them (Grossman & Kumar, 1981, pp. 241, 244). More recent scholarship still reflects this classical view of the president–press conference interaction. According to Martha Kumar (2007, Chapter 7), presidents use the press conference to work through difficult policy problems in an engaging and interactive format with the press and perhaps even the American people. Second, the press conference is designed to generate news coverage and communicate the president’s policy goals to the media and American people. Smith (1990, p. 66) notes that the press conference is as an opportunity for “mediated presidential persuasion” and provides the president with an opportunity to lead the policy agenda, especially on single-issue press conferences, often relegated to foreign crises (p. 83). Although reporters expect presidents to hold them regularly as they see press conferences as fundamental to a democratic free press, the White House has the authority to decide when to schedule a press conference and with whom. Presumably, the White House would schedule a press conference to benefit the president, but we know little about when presidents are likely to hold them. Even Kumar’s formative treatise on White House communications strategies, which devotes an entire chapter to the press conference, offers no systematic predictions as to when and under what conditions presidents are likely to hold one.

Despite the White House’s goals in holding them, their sheer volume, and the press’ expectations that presidents hold them, some scholars believe that the press conference is not akin to going public as Kernell (1997) conceived of it. Barrett (2004), for example, excludes press conferences in his measure of going public. Hager and Sullivan (1994, p. 1094) go so far as to remark that “press conferences offer few attractive opportunities for presenting the president’s position,” likely due to their adversarial nature (Clayman, Elliot, Heritage, & McDonald, 2007; Smith, 1990). Indeed, very little research shows much tangible benefit to the president in the classic going public sense. Rottinghaus (2010), for example, illustrates a decline in public support for the president’s policy positions uttered during press conferences. Moreover, unlike other public events that may increase news coverage of the president’s policies (Peake & Eshbaugh-Soha, 2008) the data belie the expectation that presidential press conferences will generate local news coverage. They do not (Cohen, 2010).

In short, the presidential press conference is an important and relevant public relations event to the White House. Although Kumar (2007) is clear that a primary goal of the press conference is to provide presidents with a public opportunity to answer difficult questions, she does not predict when presidents may hold them. Indeed, most of the literature’s explanations for presidential public activities is longitudinal, highly aggregated, and centers on major changes in technology (Hager & Sullivan, 1994) and personal preferences (Eshbaugh-Soha, 2003). Given that the press conference does not seem to be effective in generating news coverage (Cohen, 2010) or moving public opinion (Rottinghaus, 2010), those factors that explain presidential efforts to go public as Kernell (1997) observes may not motivate the president’s press conference activity in the same way. In the next section, I elaborate upon the distinct goal of press conferences to develop testable hypotheses about how political conditions affect the number of and time between solo and joint press conferences.

Theory

What explains the propensity of presidents to deliver press conferences? And what predicts time between press conferences? In the tradition of Canes-Wrone (2006) and others, I assume that presidents are rational and act as if to maximize benefits and minimize costs. The president’s decision calculus is driven by his desire to achieve three broad goals: good public policy, reelection, and historical achievement (Light, 1999). Presidents’ most high profile and a potentially beneficial means through which they may achieve these goals is by speaking publicly, what some scholars have termed “going public” (Kernell, 1997). The idea is that a president will assess his current political context and make a decision as to whether to issue a public appeal when it will increase the chances that he will achieve his goals.

For much of the going public literature, the decision to deliver a speech is in response to an unfavorable political environment, particularly the state of the congressional environment. That is, when conditions of divided government and congressional gridlock otherwise undermine the president’s legislative goals, presidents go over the heads of Congress and enlist the American people as soldiers in the president’s battle for legislative victory (Kernell, 1997). By speaking publicly, Kernell argues, presidents can move the American people to support their policy positions, thus making it more difficult for legislators to oppose the president. The goal here is to move the American people by speaking more to increase the president’s legislative success rate.

This paints an incomplete picture of the president’s decision to speak, however. Others demonstrate just the opposite, that a favorable political environment encourages public appeals because it most benefits the president’s goal achievement. For example, Canes-Wrone (2006, p. 75) shows that presidents are most likely to parlay public appeals into legislative success on policies that are already popular with the American people. Page and Shapiro (1985) illustrate that popular presidents are more likely to influence public support, a condition that Rottinghaus (2010, pp. 66-67) has replicated. Moreover, both Eshbaugh-Soha (2003) and Hager and Sullivan (1994) show that unified government increases the number of yearly press conferences. Thus, presidents may benefit from going public under some conditions but not others. Moreover, since different public activities have different goals, the intent of holding one type of public event matters to what might explain their frequency. The question then becomes: How do political conditions affect the number of presidential press conferences and the time elapsed between them?⁷

To answer, I maintain that presidential press conferences are focused not on moving public opinion (as with Kernell, 1997) but on meeting other goals more specific to press conferences. First, press conferences are designed to assist the president achieve two primary goals identified by Kumar (2007, p. 256): explaining policy decisions, and influencing the news agenda. This article focuses on the former of these goals to explain the number of monthly press conferences. Second, press conferences are held to satisfy reporters’ expectations that presidents hold them. This expectation should influence the time between press conferences. Both should be affected by several conditions in the political environment, including presidential approval, divided government, and the economy.

First, approval ratings should affect the number of and time between presidential press conferences. Consistent with both Ragsdale (1984) and Eshbaugh-Soha (2010), who find that lower approval ratings lead to more national addresses and policy-based minor speeches, respectively, lower approval ratings should also increase the number of press conferences. The American people (and reporters) are more likely to require explanations of policy decisions when presidents are unpopular. (If presidents are popular, presumably, Americans agree with them, and this, in turn, requires less explanation.) Because presidents hold press conferences to explain policy decisions to reporters and the American people,

Hypothesis 1: Lower approval ratings will increase the number of press conferences delivered in a month.⁸

Given that reporters should be more likely to wish to hold unpopular presidents accountable and ask them to explain their decisions, reporters should be more likely to expect presidents to hold press conferences when they are unpopular. This leads to a second hypothesis concerning approval ratings:

Hypothesis 2: Low job approval ratings will decrease the time between press conferences.

Second, divided government presents an important condition to the public presidency. Although Kernell (1997) argues that these conditions should increase the number of public appeals, others show no relationship between divided government and speechmaking (Eshbaugh-Soha, 2010; Powell, 1999). To the contrary, research on press conferences shows that divided government actually suppresses annual press conference activity (Eshbaugh-Soha, 2003; Hager &Sullivan, 1994). I hypothesize this, as well:

Hypothesis 3: Instances of divided government will decrease the number of press conferences per month.

The rationale, like with presidential approval ratings, is premised on the president’s goal of explaining policy decisions before reporters and the American people through the press conference. That is, under unified government, presidents are of the governing, majority party, which is responsible for the policy actions of government.⁹ This should encourage reporters to hold presidents accountable for the government’s policy decisions. And although the opposition can counter the president’s perspective through its own rhetoric and actions, it is the president—as leader of the majority—who must take responsibility for the actions of his party. Thus, the press conference provides an opportunity for the president to justify the government’s policy decisions and is more likely to occur under unified government. If this reasoning proves true, moreover,

Hypothesis 4: Unified government will also decrease the time between press conferences.

Third, the state of the economy matters to the public presidency. A favorable economy is an obvious boon to the president and his reelection chances (Campbell, 2000; Holbrook, 1994), but an unfavorable one has been known to depress the number of national addresses (Ragsdale, 1984) and policy speeches (Eshbaugh-Soha, 2010) presidents deliver. Still, the rationale for holding a press conference is distinct from these speeches, and, therefore, the impact of the economy on press conference frequency should also differ. Accordingly, Americans expect presidents to deliver the good life. If they do not, then presidents are more likely to need to explain their policy decisions to the American people and the news media. The expectation for more and more frequent press conferences is likely to increase when the economy sours as reporters seek answers to questions concerning the president’s policy actions. Although previous research shows no relationship between yearly change in economic conditions and the frequency of yearly press conferences (Eshbaugh-Soha, 2003; Hager & Sullivan, 1994), I hypothesize that

Hypothesis 5: An increase in the unemployment rate will increase the number of press conferences presidents deliver in a month. An increase in the unemployment rate will also decrease the time between press conferences.

Type of Press Conference

Prior to the George H. W. Bush Administration, presidents gave primarily solo, primetime televised press conferences. Since 1989, however, each president except George H. W. Bush has held more joint than solo press conferences (Kumar, 2007, p. 266). The number of joint press conferences as a percentage of all press conferences is also substantial, ranging from 41.3% (for George H. W. Bush) to a high of 80.9% (for George W. Bush’s first term). Presidents see joint press conferences as less risky (Kumar, 2007), even under difficult conditions. For example, President Clinton held 13 press conferences in 1998, the year of his impeachment. Of these, 11 were joint sessions held with foreign dignitaries and leaders. Unlike solo press conferences, which presidents use to explain their policy decisions before reporters, joint press conferences are designed more to engage in diplomacy with leaders from other nations. Thus, the factors identified above should be most likely to hold for solo press conferences. In other words, since joint press conferences are geared more toward satisfying diplomatic goals rather than explaining policy decisions or generating news coverage, then they should be more impervious to political conditions than solo press conferences, an expectation controlled for in the quantitative analysis.¹⁰

Political Scandal

There is one condition that should deviate from the theoretical expectations outlined above. During political scandal, presidents should be less likely to want to hold press conferences. When a political scandal afflicts the White House, the press conference format still provides the president with the opportunity to explain his decisions, albeit not on policy, but on the scandal saturating the news. The president could also influence the news agenda, as it relates to the scandal, with his comments on the scandal. Yet it is likely that presidents do not wish to hold press conferences when they know that the nature of the questioning is going to be driven by the scandal, thereby distracting the nation from presidents’ policy agendas and their administrations’ successes. Therefore,

Hypothesis 6: Presidents will hold fewer press conferences when a scandal is in the news.

It is possible, too, that presidents will deliver more joint press conferences as a way to meet the press expectations for presidential press conferences. Without question, the press’ expectations for more access to the president, preferably through a press conference, increase during a political scandal. It is therefore possible, however unlikely, that these expectations will drive presidents to reduce the time between press conferences during a scandal.

Controls

As a final matter, I control for the timing of press conferences in three ways. First, Grossman and Kumar (1981) explain how president–press relations operate according to three phases of which the honeymoon period is most favorable to the president. Thus, the president’s honeymoon period should witness a larger number of press conferences in comparison with other times in the president’s tenure (see Manheim, 1979). Second, the president’s second term in office is qualitatively different than his first. It tends to be less effective in terms of overall communications successes (Kumar, 2007) and is the term in which presidents have typically experienced political scandals (Shogan, 2006). In addition, Clayman et al. (2007) find that presidents face more aggressive questions during their second term in office. If the White House is aware of greater exposure to negative questions in the president’s second term, presidents may be less likely to deliver press conferences during their second terms (see Lammers, 1981). Finally, reelection years severely constrain the president’s time. Presidents alter their focus from going public on policy to campaigning for reelection during their reelection years. They are likely to deliver more fundraising or political speeches during reelection years (Hager & Sullivan, 1994) and often refocus their communications office to focus on reelection and not on policy (Maltese, 1994). In short, presidents should hold fewer press conferences during reelection than other years in office.

Individual Presidents

Unlike with other speeches, individual presidents explain a fair share of the variation in the number of annual press conferences delivered between 1989 and 1998 (Eshbaugh-Soha, 2003), indicating that individual presidents are necessary controls in any model that attempts to explain the number of presidential press conferences. It also makes sense for the theoretical frame of the article. Since the president’s goal in holding a press conference is to justify policy decisions before an audience of reporters, some presidents may be more or less comfortable in this format and adjust the number of their press conference activity relative to other presidents. Because of this, I account for variation in presidential administrations and how this might affect both the number of and time between press conferences.

Data

I tallied the universe of press conferences, whether solo or joint, held in the United States (domestic) or abroad (foreign), that presidents held monthly between January 1989 and May 2012 from the Public Papers of the Presidents, as catalogued at the American Presidency Project, http://www.presidency.ucsb.edu. I searched only “Oral: News Conferences” and “Oral: News Conferences—Joint” and counted the number of each type of press conference per month. This count is what the Public Papers identifies to be a “press conference” whether held during primetime in the East Room, afternoon conferences held in the briefing room, or morning sessions held in the Rose Garden. It does not include question and answer sessions or interviews.¹¹ The dependent variables in the first sets of models are therefore the number of press conferences that the president delivered in a month, in total, solo, or joint.¹²

I examine press conferences between 1989 and 2012 for a number of reasons. First, it is with the George H. W. Bush administration that we witness a trend reversal, from a long-term decline in press conferences through Reagan toward an increased tendency for presidents to deliver more press conferences, generally, and more joint press conferences, in particular. Indeed, of Reagan’s 96 press conferences 30 were held during primetime, 40 were televised, and none were joint press conferences. The post-1989 era of presidential press conferences is thus substantially different from the previous era. I also restrict the time series to avoid changes associated with pre- and post-Watergate differences in press conferences (Clayman et al., 2010). This still ensures variation across numerous presidents as a way to address potential differences in presidential style in dealing with the press directly (Eshbaugh-Soha, 2003). In addition, I limit the analysis to press conferences and do not analyze interviews, short question and answer sessions with reporters, or other interactions with the press. Press conferences are rarer events, suggesting that they are not only qualitatively different from other president–press interactions but they should also be affected by a different set of variables. Finally, I examine solo and joint press conferences, rather than some combination of domestic or foreign press conferences. The goal of a joint press conference is the same: to engage in diplomacy with other leaders, regardless of location. The vast majority of solo press conferences are held domestically, anyway, such that a model of domestic-solo press conferences produces results similar to those for solo conferences presented in Table 2.

Independent variables are coded in the following ways. First, the president’s approval rating is the percentage of the public who support the president’s job performance according to the average Gallup poll approval rating for each month in the analysis, lagged 1 month. Second, divided government is coded 1 when either house of Congress is controlled by the opposition party or zero otherwise. Third, the state of the economy is coded as the unemployment rate, which is provided by the Bureau of Labor Statistics.¹³ The controls for timing include the presidential honeymoon, the first six months of the president’s first term in office, which is coded as zero;¹⁴ the president’s second term, which is coded 1 for each month of Presidents W. Bush’s and Clinton’s second terms in office; and reelection years, which are coded as 1 for the months of January through November for 1992, 1996, and 2004. Fourth, scandals are coded, from Basinger and Rottinghaus (2012), as 1 in the month that the scandal occurred.¹⁵ Finally, presidential dummies are coded 1 for each month an individual was president, with George H. W. Bush being the reference category in all models.¹⁶

Findings

I begin by presenting basic descriptive data of the president’s press conferences by administration and type. Figure 1 displays the number of all press conferences delivered by presidents in each month from January 1989 through May 2012, overlaid with a moving average trend line of 6 months for better visualization of dynamic patterns of these press conferences. The figure reveals substantial variability in press conference activity. George H. W. Bush delivered consistently more press conferences per month than any other president in our sample, followed in number by Clinton’s first term and George W. Bush’s second term in office. Clinton’s second term and, to a slightly lesser extent the second and third years of Obama’s first term, represents the nadir of presidential press conferences for this time period. Obama’s press conference activity peaked early in his first year—when he delivered numerous foreign press conferences during his European and Middle Eastern international tour—only to decline to numbers reminiscent of Clinton’s second-term activity. All in all, these presidents averaged 2.15 press conferences per month, with an overall decline in the number of press conferences per month since Clinton’s first term.

Figure 1.

The Number of Monthly Presidential Press Conferences, January 1989-May 2012.

Admittedly, there is much more variation to press conference activity than what is presented in Figure 1. Presidents give solo and joint press conferences, both on domestic and foreign soil. I provide the counts for these press conferences in Table 1.¹⁷ One observation is immediately noteworthy: The number of solo press conferences has decreased sharply since the George H. W. Bush Administration, being replaced by more joint press conferences by president and over the entire sample of administrations. These joint press conferences occur almost equally at home or abroad. Solo press conferences are unlikely to be held on foreign soil, with President Obama holding the only seven solo foreign press conferences since the end of the Clinton administration. Moreover, although George H. W. Bush preferred solo to joint press conferences at an approximate 2 to 1 ratio, Presidents Clinton and W. Bush preferred joint to solo, at a ratio of over 2 to 1, with Obama favoring joint press conferences at a lower ratio.

Table 1.

Counts of Press Conferences by Presidential Term and Type.

	Domestic	Foreign	Total
GHWB
Solo	81	13	94
Joint	24	21	45
	105	34	139
WJC (1)
Solo	34	9	43
Joint	57	32	89
	91	41	132
WJC (2)
Solo	16	2	18
Joint	19	23	42
	35	25	60
GWB (1)
Solo	18	0	18
Joint	35	36	71
	53	36	89
GWB (2)
Solo	33	0	33
Joint	43	43	86
	76	43	119
BHO^a
Solo	20	7	27
Joint	17	21	38
	37	28	65
All
Solo	202	31	233
Joint	195	176	371
Total	397	207	604

Note: GHWB = George H. W. Bush, WJC = Williams Jefferson Clinton; GWB = George W. Bush; BHO = Barack Hussein Obama.

Through May 2012.

Table 1 also leads us to consider the following two points. First, it is likely that President George W. Bush held more joint press conferences on foreign soil than any other president given his more foreign policy–oriented presidency. Although any definitive conclusions about Barack Obama’s press conference activity must wait, he has not moved to deliver more solo press conferences, as George H. W. Bush did, but has mirrored Clinton and George W. Bush’s preference for jointly held press conferences. Second, as 2012 is a reelection year—an occurrence that I have hypothesized will suppress press conferences—it is likely that Barack Obama will have held the fewest number of first-term press conferences since Ronald Reagan. If his press conference availability proves particularly stingy, Obama may signal a new nadir in presidential press conferences and, perhaps, a shift away from holding fewer press conferences all together, a fear shared by journalists today.

Table 2 presents the systematic tests of hypotheses that predict the number of monthly press conferences. Since the dependent variables are counts with a relatively low average, I use either Poisson or Negative Binominal count models, depending on the results of a test for overdispersion. Although one can simply compare whether the conditional variance exceeds the conditional mean to conclude overdispersion (Cameron & Trivedi, 1998, p. 77), for ease of presentation, I simply report Stata’s goodness-of-fit statistics for determining overdispersion. When overdispersion is a problem, the goodness-of-fit measure is statistically significant, indicating that a negative binomial model is more appropriate, as is the case for the total and joint press conference models. Other than this distinction, interpretation of the individual coefficients is the same for both sets of count models, with each statistically significant relationship interpreted as a percentage change in the dependent variable being the result of a one-unit change in the independent variable. In addition to examining all press conferences together, I focus on two particular subsets of press conferences: the solo press conference, which should be most susceptible to changes in the political environment, and joint press conferences.¹⁸

Table 2.

Determinants of the President’s Press Conferences, January 1989 to May 2012.

Variable	All conferences	Solo	Joint
Job approval, _t-1	−0.01 (0.001)	−0.02 (0.01) * [−1.8]	0.002 (0.01)
Unemployment	0.06 (0.07)	−0.14 (0.10)	0.28 (0.11) * [+31.7]
Divided government	−0.39 (0.12) * [−32.6]	−0.60 (0.21) * [−45.1]	−0.32 (0.17) [−27.1]
Scandal	−0.29 (0.13) * [−25.2]	−0.34 (0.20) * [−29.1]	−0.23 (0.18)
Honeymoon	0.09 (0.17)	0.11 (0.22)	0.001 (0.26)
Second term	−0.10 (0.16)	−0.31 (0.25)	0.19 (0.23)
Reelection year	−0.31 (0.15) * [−26.3]	−0.67 (0.25) * [−48.8]	−0.02 (0.21)
Clinton	−0.41 (0.16) * [−33.5]	−1.45 (0.23) * [−76.6]	0.59 (0.25) * [+79.7.0]
G. H. W. Bush	−0.52 (0.18) * [−40.3]	−1.97 (0.30) * [−86.1]	0.65 (0.27) * [+91.2]
Obama	−1.01 (0.26) * [−63.8]	−1.18 (0.37) * [−69.1]	−0.82 (0.37) * [−56.1]
Constant	1.55 (0.65)*	3.43 (0.94)*	−1.63 (0.97)*
Pseudo R²	.04	.15	.03
LRχ²	42.09*	103.06*	28.75*
Goodness of fit χ²	378.24*	228.53	426.99*
N	280	280	280

Note: Standard errors in parentheses, with percentage change in brackets. George H. W. Bush is the reference category. The Solo Press Conferences model uses a Poisson distribution.

p < .05 (one-tailed).

To make the most organized sense of the findings in Table 2, I discuss the statistically significant relationships, noting as well any variation by press conference type. First, lower approval ratings lead to an increase in the number of solo monthly press conferences. As hypothesized, presidents are in higher demand to explain their policy decisions through a press conference when they are less popular. Specifically, a 1 percentage point decrease in presidential approval ratings leading to a 1.8% increase in the number of monthly press conferences. Alternatively, a 10% decrease in approval ratings leads to an 18% increase in the number of monthly solo press conferences, which increases the monthly average from 0.83 to nearly 1.0.¹⁹ Second, conditions of divided government depress presidential press conference activity, a finding that holds for all three models. The strongest finding, which is for solo press conferences, reveals that unified government leads to a 45% increase in press conferences, from an average of 0.83 per month to nearly 1.25 press conferences per month. Consistent with previous annual analyses of press conferences, this finding also flows from the president’s goals in holding press conferences. As head of the governing party, the president is in a stronger position to justify his policy decisions under unified government, with the press conference approximating question time in the English parliamentary system. Third, the unemployment rate matters in the direction hypothesized, but only for joint press conferences. Admittedly, this does not follow tightly from the theoretical rationale put forth, as I had expected that the state of the economy would be more influential over solo than joint press conferences.²⁰

Timing and scandal matter, but variably and by press conference type. Reelection years depress all press conferences (by 26%) and solo press conferences (by 49%) as presidents turn to campaigning and fundraisers. Scandal has a similar, although weaker, effect for solo press conferences. This impact does not translate into an increase in joint press conferences, however. A lack of a honeymoon effect may be unsurprisingly simply because presidents are using this time period to deliver national addresses and speeches to local audiences. Even so, this suggests once more that presidents do not prioritize the press conference format, even when their relationship with the Washington Press Corps may be at its best. Despite previous research that finds second terms to produce more adversarial questioning, this does not translate into fewer second-term presidential press conferences, at least at conventional levels of statistical significance.²¹

Finally, individual presidents have sizeable effects on the expected number of solo, joint, and all press conferences delivered each month. Each president is much less likely to deliver solo and all press conferences than George H. W. Bush. Moreover, each president except Obama has given a larger number of joint press conferences than George H. W. Bush, with Clinton holding 102% more domestic joint press conferences per month than Bush had held. Although not unexpected given the numbers presented in Table 1—and, indeed, these coefficients reinforce the trends revealed there and in Figure 1—these results usefully buttress the importance of individual presidents to explaining press conferences (Eshbaugh-Soha, 2003), even at this disaggregated level of analysis. Of course, the statistically significant presidential dummies only tell us so much about the impact of individual presidents, that Clinton, Bush, and Obama differ from George H. W. Bush in their press conference activity. Separate analyses show some variation by the independent variables, with scandal significantly decreasing President Clinton’s solo press conference activity and reelection years depressing President George H. W. Bush’s monthly solo press conferences. President Bush learned in June 1992, for example, that the press would not give him television airtime for a press conference, a factor that appears to have had an impact on his decision to hold no other press conference during his reelection campaign and, perhaps, future presidents’ decisions, too.

An event history count model reveals the relationship between the set of independent variables and the time between press conferences. The dependent variable for this analysis is a press conference occurring during a month, with duration being the number of months between press conferences. Because the month is the unit of time, I do not differentiate, for example, between press conferences that occurred within 1 week of another and those that occurred within 3 weeks of another. The statistical technique is a Cox proportional hazards model with time-varying covariates (see Box-Steffensmeier & Jones, 2004, Chapter 7).²² Whether or not a variable is determined to be a time-varying covariate is a function of its variation between months for each press conference in the data set (Cleves, Gould, Gutierrez, & Marchenko, 2008). Since approval ratings, the unemployment rate, instances of scandal, and the honeymoon vary regularly between press conferences, I model these as time-varying covariates. In Table 3, I report the hazard ratios, which reveal an increase or decrease in the risk that the press conference is likely to happen in terms of the magnitude by which the ratio is above or below 1. The average time between press conferences ranges from 1.24 months between all press conferences and 1.88 months between solo press conferences.

Table 3.

A Duration Model of Presidential Press Conferences, January 1989 to May 2012.

Variable	All conferences	Solo	Joint
Job approval, _t-1	0.99 (0.01)	0.99 (0.004)*	0.99 (0.004)
Unemployment	1.04 (0.07)	1.00 (0.05)	0.99 (0.04)
Divided government	0.73 (0.13)*	0.53 (0.12)*	0.72 (0.14)
Scandal	0.99 (0.14)	0.85 (0.09)	1.01 (0.11)
Honeymoon	0.73 (0.16)	0.91 (0.11)	0.78 (0.11)*
Second term	1.01 (0.22)	0.96 (0.28)	0.83 (0.19)
Reelection year	0.66 (0.15)*	0.48 (0.13)*	0.93 (0.23)
Clinton	0.82 (0.19)	0.26 (0.08)*	2.06 (0.61)*
George H. W. Bush	0.68 (0.17)	0.18 (0.06)*	1.87 (0.59)*
Obama	0.51 (0.16)*	0.25 (0.09)*	0.98 (0.37)
Likelihood ratio χ²	11.61	49.99*	24.07*
Log likelihood	−1227.01	−755.81	−894.70
Average time	1.24	1.88	1.63
Time at risk	348	527	457
N	280	280	280

Note: Hazard ratios reported, with standard errors in parentheses. Approval, unemployment, scandal, and honeymoon are time-varying covariates. George H. W. Bush is the reference category.

p < .05 (one-tailed).

The results are similar in specific ways to those produced by the count models, with the strongest effects occurring for solo press conferences. This makes sense because solo press conferences are what journalists want most from presidents. Approval, reelection years, and divided government ratios are negatively related to time to another press conference. That is, a 1 percentage point decrease in approval ratings increases the risk of a solo press conference occurring by about 1%, consistent with my rationalization that the press expects the president to hold press conferences more frequently when he is unpopular. In addition, divided government decrease the risk of a solo press conference occurring by 47%, meaning that the time between solo press conferences is about 47% shorter under unified government. Reelection years predictably decrease the risk of (or increase the time to) another solo press conference, too, by about 52%, as presidents engage in campaign activities during reelection years. Scandal reveals a similar relationship, albeit not a conventional levels of significance. Finally, all three presidents listed in the second column of Table 3 waited longer between delivering solo press conferences than President George H. W. Bush had.

The total and joint categories of press conferences expose fewer relationships. As with the solo press conference model, reelection years also increase the time to any press conference. In addition, conditions of divided government matter to both total and joint press conferences, although at less than p = .05 (one-tailed test) for joint press conferences. The impact is similar, at 28%. The honeymoon period also reveals similar relationships for total and joint press conferences. But this time, the ratio is not quite statistically significant for the total press conference category. Despite these important effects on the number of months between press conferences, the solo press conference model, as with the count models presented in Table 2, reveals the strongest relationships most likely because the press values solo press conferences over joint ones.

Conclusion

The presidential press conference is a high-profile, democratically important political event. It not only allows presidents to justify their policy decisions before the press and American people but also provides the press with the chance to scrutinize the president and his policies. Even though the adversarial nature of a press conference contributed to their decline throughout the 1970s and 1980s, the continuation of modest press conference activity in recent years reconfirms the press conference as an important part of the president’s public relations resources. Despite the importance of the press conference to the White House, the news media, and American government, very little research had explored the conditions under which presidents are likely to hold one. This article makes an important contribution to the literatures of the public presidency and president–press relations by noting that the president’s desire to work through policy problems and the press’ expectations that presidents hold press conferences drive those variables that influence the number of and time between presidential press conferences.

To this end, the results in this article illustrate that presidents will give more solo press conferences with less time between them when they are unpopular and during unified government. Reelection years and political scandal tend to depress press conference activity, especially solo press conferences. Theoretically, this should be so given a press conference’s primary goal of allowing the president to grapple with policy decisions before an interactive audience. It may also be because presidents do, indeed, hold fewer high-risk, primetime press conferences but instead now hold a range of other kinds of press conferences, such as solo, afternoon sessions, or joint conferences with foreign leaders. This, in itself, may have spawned this reaction to the political environment, also supported by the duration models.

This article also provides an answer to the question raised at the beginning of the article: Why did Obama hold a press conference when it was unlikely that it would have any of the classic benefits of going public? The unfavorable political conditions and lack of influence over the public and news media suggest that the president may have been unwise to hold a press conference if popular influence was his primary goal. Instead, I have argued that the press conference is geared toward providing presidents with an opportunity to work through policy problems and justify their actions, as well as meeting the press’ expectations for holding them. And, given the results of this article, this goal is much more plausible than if the president were holding a press conference to lead public opinion, which is more likely to occur through other public activities (Rottinghaus, 2010). Measuring whether President Obama successfully explained his policy decisions is another matter beyond the scope of this article. But the anecdote reinforces why the president would deliver a press conference when he had other, potentially more favorable public relations options in the face of a relatively unfavorable political environment.

In addition, the results substantiate the significance of individual presidents to explaining press conference activity. Even though we are confident about the limits of the president’s public appeals on public preferences (Edwards, 2003), news coverage (Edwards & Wood, 1999), or legislation, that national addresses have remained constant over time (Ragsdale, 2009) and that all presidential speeches have increased significantly since the Nixon Administration (Kernell, 1997; Powell, 1999), I find substantial variation in the number and type of press conference by individual president. This speaks to the role that individual presidents play in their public activities and suggests that individual presidents can color their presidency by deciding whether they prefer to deliver press conferences and what types of press conferences best suit their rhetorical style. This may be a function of personality. As Barber (1992) notes, presidential personality may affect many aspects of the presidency, not the least of which is the president’s rhetoric. Still, this article falls short of confirming that my results illustrate the impact of presidential style—and not simply time breaks measured by presidential dummies—on press conferences. Although Lammers (1981) offers an early test of presidential personality on press conferences as a function of Barber’s presidential character typology, the typology is no longer a valid measure with which to test variation in presidential personality and its impact on presidential activities. Future research should look to uncover more evidence of the importance of individual presidents to press conference activity, perhaps by scouring the libraries of these presidents when documents become more readily available.

A final implication of this article concerns the nature of president–press relations. I stated at the outset that some have observed a less adversarial relationship between the Washington Press Corps and the president of the United States in recent years. Recall, indeed, that press conference questions have become less adversarial and that solo press conferences have been held less frequently since the George H. W. Bush Administration. Indirectly, I suspect that the rarity of solo press conferences ultimately depresses the cost to holding press conferences if only because, as a rare event, the press want to encourage presidents to hold more of these direct exchanges to benefit their own reputations and media’s prestige. The press will be less adversarial in these rare meetings to encourage more of them. My findings cannot speak to this implication directly, nor can this article address whether press conference activity is a response to the larger media environment. But they suggest that exploring further whether the president varies his press conference activity in response to the tone of news coverage may be a good barometer for measuring the negativity surrounding the president–press relationship overall.

Footnotes

Acknowledgements

Thanks to Jun-Deh Wu and Lilly Lu for assistance with data collection. Thanks to the University of North Texas and Marijke Breuning for providing research support, and to Steve Liebel and several anonymous reviewers for helpful comments. Statistical analysis conducted in Stata 12.1.

Declaration of Conflicting Interests

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

Funding

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

Notes

Author Biography

Matthew Eshbaugh-Soha is an associate professor of political science at the University of North Texas. He studies the presidency, media, and public policy and is coauthor of Breaking through the Noise: Presidential Leadership, Public Opinion, and News Media (Stanford University Press).

References

Aldrich

J. H.

Rohde

D. W.

(2000). The consequences of party organization in the House: The role of the majority and minority parties in conditional party government. In Bond

J. R.

Fleisher

(Eds.), Polarized politics: Congress and the president in a partisan era (pp. 31-72). Washington, DC: CQ Press.

Barber

J. D.

(1992). The presidential character: Predicting performance in the White House (4th ed.). Upper Saddle River, NJ: Prentice Hall.

Barrett

A. W.

(2004). Gone public: The impact of going public on presidential legislative success. American Politics Research, 32, 338-370.

Basinger

S. J.

Rottinghaus

(2012). Skeletons in the White House closets: An empirical investigation into modern presidential scandals. Political Science Quarterly, 127, 213-239.

Beckmann

M. N.

Godfrey

(2007). The policy opportunities in presidential honeymoons. Political Research Quarterly, 60, 250-262.

Bennett

W. L.

Lawrence

R. G.

Livingston

(2007). When the press fails: Political power and the news media from Iraq to Katrina. Chicago, IL: University of Chicago Press.

Box-Steffensmeier

J. M.

Jones

B. S.

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

Brandt

P. T.

Sandler

(2012). A Bayesian Poisson vector autoregression model. Political Analysis, 1-24.

Burden

Mughan

(2003). The international economy and presidential popularity. Public Opinion Quarterly, 67, 555-78.

10.

Cameron

A. C.

Trivedi

P. K.

(1998). Regression analysis of count data. New York, NY: Cambridge University Press.

11.

Campbell

J. E.

(2000). The American campaign: U.S. presidential campaigns and the national vote. College Station: Texas A&M University Press.

12.

Canes-Wrone

(2006). Who leads whom? Presidents, policy, and the public. Chicago, IL: University of Chicago Press.

13.

Clayman

S. E.

Elliot

M. N.

Heritage

Beckett

M. K.

(2010). A watershed in White House journalism: Explaining the post-1968 rise of aggressive presidential news. Political Communication, 27, 229-247.

14.

Clayman

S. E.

Elliot

M. N.

Heritage

McDonald

L. L.

(2006). Historical trends in questioning presidents: 1953-2000. Presidential Studies Quarterly, 36, 561-583.

15.

Clayman

S. E.

Elliot

M. N.

Heritage

McDonald

L. L.

(2007). When does the watchdog bark? Conditions of aggressive questioning in presidential news conferences. American Sociological Review, 72, 23-41.

16.

Cleves

Gould

W. W.

Gutierrez

R. G.

Marchenko

(2008). An introduction to survival analysis using Stata. Stata Press.

17.

Cohen

J. E.

(2008). The presidency in an era of 24-hour news. Princeton, NJ: Princeton University Press.

18.

Cohen

J. E.

(2010). Going local: Presidential leadership in the post-broadcast age. Cambridge, UK: Cambridge University Press.

19.

Cook

T. E.

(1998). Governing with the news: The news media as a political institution. Chicago, IL: University of Chicago Press.

20.

Edwards

G. C.

(2003). On deaf ears: The limits of the bully pulpit. New Haven, CT: Yale University Press.

21.

Edwards

G. C.

Wood

B. D.

(1999). Who influences whom? The president, Congress, and the media. American Political Science Review, 93, 327-344.

22.

Eshbaugh-Soha

(2003). Presidential press conferences over time. American Journal of Political Science, 47, 348-353.

23.

Eshbaugh-Soha

(2010). The politics of presidential speeches. Congress & the Presidency, 37, 1-21.

24.

Grossman

M. B.

Kumar

M. J.

(1981). Portraying the president: The White House and the news media. Baltimore, MD: Johns Hopkins University Press.

25.

Hager

G. L.

Sullivan

(1994). President-centered and presidency-centered explanations of presidential public activity. American Journal of Political Science, 38, 1079-103.

26.

Holbrook

T. M.

(1994). Campaigns, national conditions, and U.S. presidential elections. American Journal of Political, 38, 973-98.

27.

Iyengar

McGrady

J. S.

( 2007). Media politics: A citizen’s guide. New York, NY: Norton.

28.

Kernell

(1997). Going public: New strategies of presidential leadership (3rd ed.). Washington, DC: CQ Press.

29.

King

Tomz

Wittenberg

(2000). Making the most of statistical analyses: Improving interpretation and presentation. American Journal of Political Science, 44, 347-361.

30.

Kumar

M. J.

(2003a). Source material: Does this constitute a press conference? Presidential Studies Quarterly, 33, 221-237.

31.

Kumar

M. J.

(2003b). Communications operations in the White House of President George W. Bush: Making news on his terms. Presidential Studies Quarterly, 33, 366-80.

32.

Kumar

M. J.

(2007). Managing the president’s message: The White House communications operation. Baltimore, MD: Johns Hopkins University Press.

33.

Lammers

W. W.

(1981). Presidential press-conference schedules: Who hides, and when? Political Science Quarterly, 96, 261-278.

34.

Light

(1999). The president’s agenda (3rd ed.). Baltimore, MD: Johns Hopkins University Press.

35.

Maltese

J. A.

(1994). Spin control: The White House Office of Communications and management of presidential news. Chapel Hill: University of North Carolina Press.

36.

Manheim

J. B.

(1979). The honeymoon’s over: The news conference and the development of presidential style. The Journal of Politics, 41, 55-74.

37.

Mayhew

D. R.

(1991). Divided we govern: Party control, lawmaking, and investigations 1946-1990. New Haven, CT: Yale University Press.

38.

Page

Shapiro

(1985). Presidential leadership through public opinion. In Edwards

G. C.

III Shull

Thomas

(Eds.), The presidency and public policy making (pp. 22-36). Pittsburgh: University of Pittsburgh Press.

39.

Peake

J. S.

Eshbaugh-Soha

(2008). The agenda-setting impact of major presidential TV addresses. Political Communication, 25, 113-37.

40.

Peterson

D. A. M.

Grossback

L. J.

Stimson

J. A.

Gangl

(2003). Congressional response to mandate elections. American Journal of Political Science, 47, 411-26.

41.

Powell

R. J.

(1999). “Going public” revisited: Presidential speechmaking and the bargaining setting in Congress. Congress & The Presidency, 26, 153-70.

42.

Ragsdale

(1984). The politics of presidential speechmaking, 1949-1980. American Political Science Review, 78, 971-984.

43.

Ragsdale

(2009). Vital statistics on the presidency: George Washington to George W. Bush (3rd ed.). Washington, D.C.: CQ Press.

44.

Rottinghaus

(2010). The provisional pulpit: Modern presidential leadership of public opinion. College Station: Texas A&M University Press.

45.

Shogan

C. J.

(2006). The contemporary presidency: The sixth year curse. Presidential Studies Quarterly, 36, 89-101.

46.

Smith

(1990). Presidential press conferences: A critical approach. Westport, CT: Praeger.

47.

Yamaguchi

(1991). Event history analysis. Newbury Park, CA: SAGE.