Abstract
US presidents have routinely ordered the use of force without seeking prior authorisation from Congress. However, this practice does not mean that the legislature is irrelevant, as Congress often influences decisions by exercising informal political levers. One of the most important is through Congress’ ability to affect popular support for the commander-in-chief. Through a pair of experiments embedded on nationally representative opinion surveys, this article evaluates whether Congress’ constitutional prerogatives in war powers remain relevant when battling the president in the public sphere. Policy criticism significantly decreased support for the use of force, as did challenges to administration actions on constitutional grounds. Although Congress routinely fails to use the constitutional tools at its disposal to check the commander-in-chief, these powers bolster Congress’ capacity to influence public opinion. Hence, while presidents enjoy considerable leeway in the military arena, Congress’ capacity to erode public support can serve as a check on presidential power.
An emerging literature challenges the conventional portrayal of executive-driven foreign policy-making that leaves little room for parliamentary influence or constraint (see Mello and Peters, 2018, the introduction to this Special Issue, as well as Mello, 2014; Raunio and Wagner, 2017). However, much of this literature has focused strictly on legislatures’ assertions of their formal powers to constrain executive initiative in security policy (Dieterich et al., 2015; Mello, 2012; Raube, 2014), and perhaps most importantly, their ability to exercise an ex ante veto on the use of force (Peters and Wagner, 2011, 2014; Wagner, 2017). For example, recent research argues that the Parliament of the United Kingdom, by virtue of securing votes on authorising the use of force in Iraq, Libya, and Syria—and in the latter case blocking the Prime Minister’s desired intervention—has established a parliamentary prerogative to vote before the use of force (Strong, 2015). This parliamentary assertion is particularly notable given that the Parliament of the United Kingdom is traditionally classified as being relatively ‘weak’ in terms of its capacity to influence security policy compared to other European legislatures (Kaarbo and Kenealy, 2015; Mello, 2017).
While these studies offer an important corrective to conventional models that all but ignore the legislature’s role in shaping security policy, they largely omit many important informal mechanisms through which legislatures constrain executive initiative in the international arena. In the American context, these informal levers routinely afford Congress greater influence over policy outcomes than do its formal constitutional powers.
With respect to its share of constitutional war powers, the United States Congress is perhaps the most powerful legislature in the world. Congress alone can declare war. Article I also grants Congress authority over the initiation of limited military actions, through its sole power to grant Letters of Marque and Reprisal (Lofgren, 1972), and it empowers Congress to raise and regulate the Armed Forces. Ultimately, Congress also controls the federal purse strings.
Yet throughout American history, Congress has struggled mightily to employ any of these tools legally to constrain the commander-in-chief. Presidents have routinely deployed American troops abroad without first seeking congressional authorisation (Torreon, 2016). The power of the purse has proved a blunt instrument; the prospect of cutting off funding for American forces committed to the field is so politically toxic that it all but precludes the power’s successful use. 1 Similarly, the War Powers Resolution, which was intended to create a mechanism for terminating military actions of which Congress disapproved without having to cut off funding, has similarly failed to achieve the lofty goals of its architects (Auerswald and Cowhey, 1997; Fisher and Adler, 1998; Glennon, 1984). Such failures have led many analysts to write off Congress as all but irrelevant when pitted against an ‘imperial presidency’ (Griffin, 2013; Rudalevige, 2005; Savage, 2008; Schlesinger, 1973).
The formal interactions of legislatures and executives are, of course, important. Moreover, there is an essential truth underlying the conventional view of inter-branch power in American military affairs. Contemporary presidents are undoubtedly the focal point of American military policy-making. However, scholarship examining the legislature’s influence in security policy both in the American and comparative contexts would do well to consider more explicitly the legislature’s capacity to influence both the course and conduct of security policy through more informal means (Howell and Pevehouse, 2007; Kriner, 2010). Perhaps most importantly, it should examine the capacity of legislatures to influence public support for the use of force.
While Congress rarely imposes formal legal constraints on the commander-in-chief, congressional opponents of administration policy can raise the political costs of pursuing a policy course that strays too far from congressional preferences. One of the most important mechanisms through which Congress can impose political costs on the president is by shaping public opinion and eroding popular support for the president and his policies. Public support is one of the most salient and significant metrics on which other politicians assess the president’s political capital. Presidents who enjoy strong popular support are much more successful in getting their policy proposals through Congress (Beckmann, 2010; Canes-Wrone and De Marchi, 2002; Rivers and Rose, 1985) and, ultimately, in securing reelection for themselves or for their partisan successors than are presidents who do not enjoy strong support among the public (Abramowitz, 2008; Sigelman and Brody, 1979; Wlezien and Erikson, 2004). Indeed, precisely because presidents are sensitive to the political costs of military policy, defense policy-makers have long-listed strong public support as a precondition for the successful use of force (Klarevas, 2002).
Prior scholarship has amassed considerable evidence that congressional opposition erodes public support for the president and his military policies. 2 However, important questions remain about what types of congressional challenges to presidential actions as commander-in-chief resonate with the public.
This analysis addresses this gap by examining the extent to which Congress’ constitutional war powers remain relevant in shaping its capacity to influence public opinion. Specifically, it examines whether congressional challenges asserting that the president has overstepped his constitutional authority and transgressed on congressional prerogatives are influential with the public. Results from two experiments embedded in nationally representative opinion surveys suggest that Congress’ formal constitutional powers remain more relevant than often supposed: they open another line of attack for opponents of the president’s military policies that can impose significant political costs on the commander-in-chief.
Congress and public support for military action
Public opinion scholars have long argued that Members of Congress have the potential to shape popular attitudes toward war. Most Americans lack relevant political knowledge on a host of policy issues (Delli Carpini and Keeter, 1997). However, this deficit is perhaps most acute on questions of foreign affairs (Almond, 1950; Lippmann, 1922). As a result, citizens rationally rely on elite cues when evaluating military policy. When other political elites rally around the White House, the public also falls in line behind the president and his policies. 3 By contrast, when elites are divided on the wisdom of a use of force and offer competing narratives to that advanced by the administration, then public support for the president and his military policies declines (Zaller, 1992).
Scholars using observational polling data have found evidence of congressional leadership of wartime American public opinion in a range of settings. Brody (1991) found that the rally around the flag effect in the wake of international crises does not arise automatically; rather, the size of the rally, and even its existence, is largely a function of the congressional response (see also Groeling and Baum, 2008; Lian and Oneal, 1993). Zaller’s (1992) study of popular support for the Vietnam War showed that public support for the war remained resilient, even in the face of mounting casualties for most of the 1960s. However, once prominent cue-givers, primarily in Congress, began to break with the president, support for the war fell and polarised along partisan lines. Similarly, Berinsky (2009) demonstrated the importance of elite cues in explaining patterns in support for war throughout World War II, and recent research has also found varied evidence of elite opinion leadership throughout the Iraq War (Berinsky, 2007, 2009; Howell and Pevehouse, 2007; Kriner and Shen, 2014).
While these studies marshaled a range of opinion data across many decades at both the aggregate and individual level, observational analyses always raise questions of causality. Are political elites leading or following public opinion? Moreover, to what extent are elites simply responding to events, which are driving both shifts in elite cues and public attitudes? To address questions of causality, recent research has employed a range of survey experiments examining both hypothetical scenarios (Berinsky, 2009; Gelpi et al., 2009; Grieco et al., 2011; Howell and Pevehouse, 2007) and ongoing conflicts (Baum and Groeling, 2009a; Kriner, 2017; Kriner and Howell, 2012) to establish the capacity of congressional elites to influence popular support for war. Consistent with observational analyses, multiple experiments have shown the power of congressional support for or opposition to presidential military policies to shape public support for the use of force.
Taken together, recent scholarship significantly bolsters arguments that Congress may retain an informal check on presidential war-making through its ability to shape public opinion and bring popular pressures to bear on the White House. However, existing research offers less insight into what types of congressional criticisms resonate with voters. Specifically, do appeals to Congress’ constitutional prerogatives in war powers remain relevant when battling the White House in the court of public opinion?
When presidents assert their constitutional role as commander-in-chief to order American military forces abroad unilaterally, they open themselves up to two distinct lines of attack. Congressional critics can oppose the president on both policy and constitutional grounds. By focusing on policy, congressional opponents can challenge the dominant frame offered by the administration. They can contest administration claims about the importance of a venture to vital national interests. They can emphasise the costs of military action, both in terms of casualties incurred or risked and dollars spent. Ultimately, they can also raise doubts about the mission’s prospects for success. Each of these dimensions has been shown by past research to influence popular support for war. 4
However, because the Constitution entrusts the bulk of the enumerated war powers to Congress, congressional opponents can also challenge the administration on constitutional grounds. When presidents unilaterally dispatch American forces abroad, Members of Congress can and frequently do argue that presidents are overstepping the bounds on their authority and transgressing constitutional limits on executive power. Will such critiques based on constitutional grounds alone erode public support for military action?
This question is more than of just academic interest. Past research has demonstrated that ‘costly’ criticism from the president’s co-partisans attracts significantly more attention from media outlets (Groeling and Baum, 2008). In many recent cases, members of the president’s party have been reticent to criticise presidential uses of force on policy grounds, even as they have criticised the president’s decision to act unilaterally and tread on congressional prerogatives. In short, they have focused their criticism on the means through which the president has acted, while remaining largely silent on the ends for which force was employed. As a result, such challenges may be the most likely to be reported extensively to the public through the mass media.
For example, President Barack Obama’s unilateral air strikes in Libya in 2011 attracted bipartisan criticism, including from many Democrats, that he had overstepped his constitutional authority. Instead, critics charged, military action in Libya required congressional authorisation (Savage, 2011). When the administration later argued that the mission in Libya did not meet the threshold of ‘hostilities’ as defined in the War Powers Resolution to trigger automatically its sixty day withdrawal clock, the decision was roundly criticised by both Democrats and Republicans (Fahrenthold and Sonmez, 2011). Similarly, when President Obama escalated American military actions against Islamic State in Iraq and Syria (ISIS) by dispatching a small contingent of special forces troops to Syria in 2015, he again aroused criticism from a number of House and Senate Democrats who argued that such actions required new and explicit authorisation from Congress (Carney, 2015). Finally, while most Republicans cheered President Trump’s decision to fire missiles at Syrian targets in response to President Assad’s renewed use of chemical weapons, several Republicans along with a number of Democrats openly criticised the president’s actions on constitutional grounds. While they largely resisted challenging the administration on policy, these critics nonetheless maintained that such strikes require congressional authorisation (Prokop, 2017).
In each of these cases, co-partisan constitutional criticism of the president’s unilateral action attracted considerable media attention. Such intra-party conflict is inherently newsworthy and is often magnified by the media in importance beyond the sometimes small number of co-partisans leveling such charges (Groeling, 2010). However, past research has not explored whether such constitutionally based criticisms, devoid of any challenges to the administration on policy grounds, are able to erode public support for war.
To address this gap, I embedded a pair of experiments on nationally representative opinion surveys that explore the influence of congressional criticism of presidential actions on both policy and constitutional grounds on public support for the use of force. While experiments have important advantages in establishing causality in cases where the causal arrow could well be reversed, this benefit comes at the cost of external validity. To minimise such concerns, both experiments examine ongoing policy crises and employ treatments based on actual arguments made by congressional actors. It is exceedingly difficult to determine how well any observed treatment effect sizes correspond to the effects that congressional opposition produces in the real-world. On one hand, in the artificial experimental setting subjects receive the congressional treatment and then immediately respond whether or not they support or oppose the president’s policies. On the other, the treatments themselves are fairly weak—each comprises a single statement in which Congress challenges the president’s action. In real-world politics, congressional critics enjoy multiple fora, including investigative hearings, public speeches, and media interviews, to challenge repeatedly and consistently the president’s policies in the public sphere.
Despite these limitations, experiments provide an important complement to observational studies strongly suggestive of congressional influence on foreign affairs. Moreover, in the current context, they are absolutely essential as experiments allow us to manipulate the content of congressional rhetoric to determine the relative influence of different types of congressional challenges on public support for the use of force.
Experiment 1: Obama’s strikes against ISIS
The first experiment examined the influence of a range of congressional cues on support for President Obama’s decision to launch air strikes against ISIS targets in Iraq and Syria. 5 This experiment was embedded in a nationally representative online survey conducted by YouGov from 10–13 December 2016. All subjects first read the following prompt: ‘As you may know, President Barack Obama has unilaterally launched a series of air strikes against ISIS militants in Iraq and Syria’. Subjects were then randomly assigned to one of four experimental groups. Those assigned to the control group received no further information. Subjects assigned to the first treatment group read an additional statement expressing bipartisan congressional support for President Obama’s actions on policy grounds. 6 These subjects were told, ‘Many members of Congress from both parties support the President’s decision. They argue that air strikes are necessary to target ISIS terrorists who are threatening American interests and to defend hard-won gains in Iraq’. Based on past scholarship, we would expect this cue emphasising bipartisan agreement with the mission’s objectives to increase support for Obama’s actions.
Those in the second treatment group instead received a cue informing them of bipartisan congressional criticism of Obama’s air strikes against ISIS targets in Iraq and Syria on the grounds that it risked miring the military in another intractable situation in the Middle East. These subjects were told, ‘Many members of Congress from both parties, however, oppose the President’s decision. They argue that the air strikes could deepen American’s military involvement in the region’. To maximise external validity and reflect the tenor of media coverage, this treatment concluded with a strong rebuttal of the policy critique from the White House: ‘President Obama rejects this criticism and maintains that his actions will not deepen the American military’s involvement in the region’.
Subjects assigned to the final treatment group were told that Members of Congress from both parties opposed President Obama’s air strikes on constitutional grounds. These subjects were told, ‘Many members of Congress from both parties, however, oppose the President’s decision. They argue that President Obama has overstepped his constitutional authority and that military action requires authorisation from Congress’. After receiving this congressional opposition cue, all subjects also received a rebuttal of the constitutional argument: ‘President Obama rejects this criticism and maintains that his actions are consistent with his constitutional authority as commander-in-chief’.
Finally, all subjects were asked the following question: ‘Do you support or oppose President Obama’s decision to unilaterally launch air strikes against ISIS in Iraq and Syria?’ Subjects indicated their response on a 4-point Likert scale ranging from strongly support to strongly oppose. The strongly support and somewhat support categories were then collapsed to determine the percentage of subjects that supported the strikes against ISIS targets in the Middle East.
Two features of the experiment merit notice. First, in each experimental group, the president’s position is presented first, reflecting the White House’s privileged position in shaping most media coverage (Entman, 2004). Second, in both congressional criticism treatments, the congressional challenge is immediately followed by a strong presidential rebuttal. In the policy treatment, the president refutes congressional assertions that the intervention risks deepening the American military presence in the Middle East. In the constitutional treatment, the rebuttal is perhaps even stronger as President Obama not only rejects the constitutional arguments of his congressional opponents but also refers explicitly to his constitutional powers as commander-in-chief. These features were designed, if anything, to bias the experiment against finding evidence of congressional treatment effects. Indeed, the framing literature suggests that competing frames from equally credible sources should simply cancel each other out, resulting in no opinion change (Chong and Druckman, 2007). However, if congressional criticism is judged to be particularly credible (Kriner and Schickler, 2014), then both treatments will erode support for the unilateral use of force.
Results
Because subjects were randomly assigned to one of the four experimental groups, the resulting differences in means are unbiased. As an initial presentation of the results, Table 1 summarises the percentage of subjects supporting President Obama’s unilateral decision to launch a series of air strikes against ISIS targets in Iraq and Syria across the experimental and control groups. President Obama’s actions enjoyed overwhelming support in the control group. A full 80% of subjects in the control group backed Obama’s unilateral decision to launch air strikes against ISIS targets. Because of this very high baseline level of support, it is perhaps unsurprising that the congressional support treatment had little influence on public attitudes toward the ISIS strikes; support simply had little room to increase further. Just under 80% of subjects backed Obama’s actions in the congressional support treatment, a figure that is statistically indistinguishable from that observed in the control group.
Support for Obama’s ISIS strikes across treatments.
Differences in support from that observed in the control group that are statistically significant, p < 0.05, are indicated in italics.
Congressional criticism, however, significantly eroded support for Obama’s unilateral air strikes. The policy criticism treatment—in which Members of Congress warned that the escalation risked deepening the United States’ military commitments in the Middle East—had the greatest effect. Support for the strikes fell to 62% in this treatment, which represents an 18% decrease from that observed in the control group. Raising the specter of again miring the country in a Middle Eastern war clearly resonated with a significant share of subjects, despite President Obama’s strong rebuttal that the ordered strikes would not risk further entangling US forces in the region.
Congressional challenges to the president’s decision to use force on constitutional grounds also significantly decreased support for the use of force. In this treatment, congressional opponents offered no policy critique of Obama’s decision to launch strikes against ISIS; rather, they solely criticised the unilateral nature of these strikes, arguing that Obama had overstepped his constitutional authority and that the use of force required congressional authorisation under the Constitution. Support for the use of force in this group was a still robust 71%, but it was 9% lower than that observed in the control group.
Table 2 presents a pair of logistic regression models to assess further the effects of the congressional cue treatments. In each regression, the independent variables of interest are a series of indicator variables identifying assignment to each of the congressional treatment groups. The first model in Table 2 includes only the relevant treatment variables. The second model adds a number of control variables that might also predict support for the strikes against ISIS. Although subjects were randomly assigned across the experimental groups, randomisation can still result in imperfect balance across likely confounders. Multivariate logit models can therefore account for any remaining imbalances and improve the precision of the estimates.
Effect of Congressional cues on support for Obama’s ISIS strikes.
Standard errors in parentheses. All significance tests are two-tailed.
p <0.01, **p <0.05, *p <0.10.
The first model in Table 2 mirrors the results from the difference in means analysis. The coefficients for the two congressional criticism treatments are both negative and statistically significant. The coefficient for the congressional support treatment, by contrast, is substantively small and statistically insignificant. The multivariate logit model also yields similar results. The coefficients for both congressional opposition treatments remain negative and statistically significant.
Figure 1 illustrates the estimated effect of each treatment, as well as the estimated effects of the control variables, on the probability of the median-independent subject supporting President Obama’s decision to launch air strikes against ISIS targets. When congressional critics raised the specter of Obama’s actions further entangling the American military in the Middle East, it significantly reduced the predicted probability of the median-independent backing the use of force by about 0.25. When Members of Congress challenged the action on constitutional grounds, the effect was smaller, reducing the predicted probability of support by about 0.12. However, this effect is both statistically and substantively significant, and it occurred despite President Obama’s rebuttal of this criticism through an appeal to his powers as commander-in-chief.

Effects of treatments and demographic factors on support for ISIS strikes.
Finally, several of the control variables also affected the probability of supporting the ISIS strikes in the expected direction. Most importantly, partisanship was an important predictor of support for the strikes. Both Democrats and Republicans were more supportive of the use of force, on average, than were independents, with the president’s co-partisans being the most supportive of his actions.
Experiment 2: Trump’s strikes against Syria
The first major use of force of the Trump presidency offered an interesting opportunity to examine whether congressional criticism retains the capacity to erode support for unilateral presidential military action in a dramatically different political environment. On 6 April 2017, President Trump ordered a series of retaliatory strikes against Syrian regime targets to punish President Bashar al-Assad for the renewed use of chemical weapons in the country’s long-standing civil war.
In terms of policy substance, Trump’s strikes were remarkably similar to those ordered by Obama against ISIS examined in the preceding experiment. In both cases, the president unilaterally ordered air strikes against targets in Syria. However, several features of the 2017 strikes stack the deck against finding evidence of congressional treatment effects. First, Trump’s missile strikes took place less than 3 months after the new president assumed office in the middle of the traditional ‘honeymoon’ period. It is possible that any president may be more insulated from congressional criticisms during this period. Second, while it has almost become cliché to describe contemporary presidential approval ratings as intensely polarised, public attitudes toward Donald Trump have polarised more quickly than toward any of his predecessors. 7 If public opinion concerning Trump is already intensely polarised, then modest cues about congressional opposition to his policies may be unlikely to sway popular assessments of Trump and his actions.
To examine whether congressional challenges to unilateral military action on both constitutional and policy grounds continue to influence public support for the use of force in the Trump era, I embedded a second experiment on a nationally representative survey fielded by YouGov from 7–10 April 2017, immediately following the announcement of Trump’s missile strikes. All subjects received the following prompt: ‘As you may know, President Donald Trump has unilaterally launched a series of missile strikes against Syrian forces in retaliation for the Syrian government’s use of chemical weapons against civilians’.
Subjects were then randomly assigned to one of three experimental groups. 8 Those in the control group received no further information. Subjects assigned to the first treatment group received a congressional cue criticising President Trump’s decision on policy grounds that was virtually identical to that used in the preceding experiment: ‘Many members of Congress from both parties, however, oppose the President’s decision. They argue that the missile strikes could deepen American’s military involvement in the region’. Subjects in the second treatment group learned of congressional objections to Trump’s unilateral decision to order the strikes on constitutional grounds: ‘Many members of Congress from both parties, however, oppose the President’s decision. They argue that President Trump has overstepped his constitutional authority and that military action requires authorisation from Congress’. The wording of this congressional cue was also virtually identical to that used in the corresponding treatment of the Obama experiment. Because recent research on support for unilateral action in other contexts found little evidence that presidential rebuttals mute the effects of congressional criticism (Christenson and Kriner, 2017)—indeed, in the preceding experiment both criticism treatments significantly eroded support for Obama’s strikes against Syria despite a strong rebuttal—presidential rebuttals were omitted from the second experiment.
All subjects then received the same question: ‘Do you support or oppose President Trump’s decision to unilaterally launch missile strikes against Syria?’ Responses were coded on a 4-point Likert scale ranging from strongly support to strongly oppose. The strongly support and somewhat support categories were then collapsed to determine the percentage of subjects that supported the strikes against Syria.
Results
Table 3 presents the percentage of subjects supporting President Trump’s unilateral decision to launch a series of air strikes against Syrian regime targets across the two treatment and control groups. A significant majority of Americans backed Trump’s action in the control group: 64%. However, this percentage is significantly lower than the control group percentage supporting Obama’s strikes against ISIS the preceding year. The main reason for this gap is that support for Trump’s actions was significantly more polarised along partisan lines. Whereas 90% of Republicans in the control group backed Trump’s unilateral decision to launch retaliatory missile strikes, just 45% of Democrats did so. By contrast, in the Obama experiment, the percentage of Democrats and Republicans supporting the air strikes in the control group was virtually identical: 86% and 85%, respectively.
Support for Trump’s Syria strikes across treatments.
Differences in support from that observed in the control group that are statistically significant, p <0.05, are indicated in italics.
Even in the contemporary political environment with a newly minted and intensely polarising president, both congressional cues again decreased support for the use of force. Support for the missile strikes fell from 64% in the control group to 56% in the congressional policy criticism treatment. This 8% drop is much smaller than that observed for the same policy critique in the Obama experiment. However, it is still both substantively and statistically significant. In the Trump experiment, constitutional challenges to the president’s unilateral military action also continued significantly to depress support for the use of force. Just 53% of subjects in this treatment supported Trump’s decision to use force, a difference of 11% from the level of support observed in the control group.
As a robustness check on the difference in means results, Table 4 estimates a pair of logistic regressions. The first model includes only two variables identifying assignment to each of the congressional opposition treatments. Consistent with the difference in means results presented in Table 1, the coefficients for both treatment variables are negative and statistically significant. The second model in Table 4 adds an identical set of demographic control variables to those included in the multivariate logit in Table 2. When including these additional controls, the coefficients for both treatment variables remain negative, although the coefficient for the policy criticism treatment narrowly misses conventional thresholds of statistical significance.
Effect of Congressional cues on support for Trump’s Syria strikes.
Standard errors in parentheses. All significance tests are two-tailed.
p <0.01, **p <0.05, *p <0.10.
Figure 2 illustrates the estimated effect of each treatment, as well as the estimated effects of the control variables, on the probability of the median-independent subject supporting President Trump’s decision unilaterally to launch a series of missile strikes against Syrian regime targets. Congressional challenges to President Trump’s actions on constitutional grounds—that by acting unilaterally, he had transgressed constitutional limits on his power and that military action requires congressional authorisation—resonated with many Americans and lowered the predicted probability of the median-independent supporting Trump’s action by 0.11. Congressional policy criticisms that Trump’s action risked further embroiling the United States in the Middle East also decreased the probability of supporting Trump’s action. However, the estimated effect is considerably smaller than in the Obama experiment, and the confidence interval for this estimated effect includes zero.

Effects of treatments and demographic factors on support for Syria strikes.
Finally, many of the control variables influenced support in the expected direction. Support for Trump’s action was intensely polarised along partisan lines. Republicans were significantly more likely to back Trump than were independents, all else being equal. By contrast, the median Democrat was significantly less likely to support Trump’s action than was the median independent. Consistent with the literature on gender gaps in support for the use of force (Conover and Sapiro, 1993; Eichenberg, 2016), men were more likely to back Trump’s decision than were women. Reflecting patterns in electoral support for Trump, older Americans were more likely to back Trump’s action, while the probability of supporting Trump decreased with educational attainment.
Treatment effects by partisan affiliation
Both experiments offered considerable evidence that congressional challenges to unilateral presidential decisions to use force can significantly erode popular support for the commander-in-chief and his conduct in military affairs. Across both experiments, when Members of Congress argued that, by acting unilaterally, the presidents had trampled on their constitutional prerogatives in war powers, support for the use of force decreased by about 10%. Interestingly, however, the efficacy of the policy critique that air strikes risked deepening American military involvement in the Middle East varied significantly across the two experiments. This critique was much more effective in the Obama experiment than in the Trump experiment.
To explore this difference further, Table 5 estimates separate regressions for both Democratic and Republican respondents in the Obama and Trump experiments. 9 Unsurprisingly, both the policy and constitutional congressional criticisms significantly reduced support for Obama’s decision to order strikes against ISIS targets among opposition Republicans. The policy criticism treatment also significantly reduced support for Obama’s action among Democrats. Perhaps because it prompted many Democrats to consider their long-standing opposition to the Iraq War, this treatment, which raised the prospects of again miring the United States in Middle Eastern conflicts, encouraged some Democrats to break with their party leader in the White House.
Effect of Congressional treatments by party.
Standard errors in parentheses. All significance tests are two-tailed.
p <0.01, **p <0.05, *p <0.10.
In the Trump experiment, both the policy and constitutional treatments again significantly reduced support for the president’s actions among the partisan opposition (for instance, among Democrats). However, neither congressional treatment reduced support for Trump among co-partisan Republicans. Indeed, the share of Republicans backing Trump’s actions remained steady between 88% and 92% across all three experimental groups. Thus, the greater resonance of the policy critique with co-partisan Democrats in the Obama experiment explains the dramatic difference in the estimated effect size for this treatment in the aggregate across the two experiments.
Discussion
Recent scholarship has rightly challenged the conventional view of executive dominance in security policy by showing how legislatures have increasingly used their formal constitutional powers to exercise a check on executive initiative in foreign affairs (Mello, 2014, 2017; Peters et al., 2010; Peters and Wagner, 2014; Raunio and Wagner, 2017; Strong, 2015). This renewed scholarly attention to legislative constraints is long overdue and provides an important corrective. However, scholarship in this vein must not overlook the informal checks that legislatures can also exercise on executive authority. Legislatures need not necessarily exercise their constitutional war powers to influence the conduct of military affairs. Rather, these powers’ very existence—and arguments that they are being usurped—may give legislative critics of administration policies an important weapon when seeking to influence public opinion and raise the political costs of war for the executive.
Although possessing considerable constitutional authority in foreign affairs, the US Congress has a dismal track record in successfully using its formal tools to check presidential initiative in military affairs. Instead, congressional critics have sought to influence the decision calculus of the commander-in-chief through more informal means, primarily by shaping anticipations about the political costs the White House stands to incur from choosing a particular policy course. One of the most important mechanisms through which congressional opponents of administration policy impose political costs is by mobilising public opinion against the president.
Throughout American history, wartime presidents have struggled when they are unable to maintain strong levels of popular support for their military actions. With a depleted reservoir of political capital, they have failed to gain traction on other elements of their policy agenda (Kriner, 2010; Neustadt, 1960). And ultimately, they and members of their party have faced electoral retribution at the ballot box (Cotton, 1986; Karol and Miguel, 2007; Kriner and Shen, 2007). In the early 20th century, the great legal and presidential 10 scholar Edward Corwin (1917) wrote, ‘For the president, even in the exercise of his most unquestioned powers, cannot act in a vacuum. He must ultimately have the support of public sentiment’. During the Vietnam Era, Leslie Gelb (1972) reached the same conclusion, calling public support the ‘essential domino’ of military policy-making, a view reflected in later scholarship (Klarevas, 2002).
The experimental results suggest that, somewhat paradoxically, Congress’ rarely used formal constitutional powers afford a significant source of leverage when seeking to battle the president in the court of public opinion. When Members of Congress object that presidents have overstepped the constitutional limits on their power by ordering the use of force unilaterally, the public responds and support for the president’s actions wanes. Moreover, as demonstrated in the first experiment, congressional constitutional arguments resonate with the public, even when presidents counter them by appealing to their own constitutional authority as commander-in-chief. Despite Congress’ abysmally low institutional approval rating, most Americans, like the Framers, still desire collective judgment in questions of war and peace. And when congressional opponents accuse presidents of ignoring this constitutional directive, public support for the president and his policies attenuates.
Congressional opponents also routinely challenge the president’s actions on policy grounds. 10 By questioning the administration’s casus belli, highlighting a mission’s human and monetary costs, and raising concerns about its long-term consequences, Members of Congress can challenge the dominant frame emanating from the White House and undercut its efforts to manage the news cycle. The experimental results confirm that such criticisms can also significantly lower public support for war. However, even when they are not explicitly evoked, Congress’ constitutional prerogatives in war powers may nonetheless remain relevant by lending its policy critiques with institutional credibility. Given the informational asymmetries across the executive and legislative branches in military matters (Wildavsky, 1966), one might rationally discount legislative criticisms on the intricacies of policy. However, congressional policy challenges may resonate with the public, despite most members’ limited access to information concerning and engagement with military matters, precisely because the Constitution so plainly entrusts to Congress an important role in war powers. Even if not exercised, the existence of these formal powers may bolster the credibility of congressional challenges to the president with the public and thereby afford a tangible, if conditional constraint on the actions of the commander-in-chief.
Footnotes
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
