Rules and Representation: The Dual Nature of Legitimacy and the Presidency

People often equate authority and legitimacy when talking about government power (Reference Gibson and CaldeiraGibson and Caldeira 2009: 38–39). In fact, the terms refer to distinct aspects of state capacity. At its most general level, authority involves the power to make rules and compel people to follow them. Legitimacy, conversely involves a sense of “rightfulness” in the exercise of government authority (Freidman 1998; Reference TylerTyler 1990, Reference Tyler2006). Basically, citizens want to be confident that the people making the rules are the appropriate ones to be doing so, and that they are following required procedures in the implementation and enforcement of rules promulgated pursuant to that authority.

Tyler has written at length about how conceptions of “procedural justice,” or the idea that officials should follow prescribed rules, influences citizens’ views of government institutions and their outputs. He writes, “people are found to believe authorities are more legitimate when they view their actions as consistent with fair procedures” (2006: 381). Indeed, there are a number of studies by Tyler and others, that focus on procedural considerations in the judicial and legislative contexts, such as whether interested parties are treated fairly (Reference TylerTyler 1990), whether affected constituencies have a “voice” in debate (Reference TylerTyler 1994), and the deliberative nature of decision making (Reference Doherty and WolackDoherty and Wolack 2012; Reference GanglGangl 2003; Reference TylerTyler 1994).

While these studies do much to highlight the importance of procedure in assessments of government outputs and institutions, they fail to consider how citizens think about state action in the context of specific constitutional provisions that actually empower and limit the authority of our federal institutions. Moreover, they fail to acknowledge that some decisions made by government officials are purposefully removed from these kinds of democratic controls. The founders, for instance, put ultimate military and executive authority in the hands of the President, a single individual, with superior access to information about foreign and domestic affairs, so that certain decisions could be made quickly and with resolution.

I begin with the premise that there are two primary sources of governmental authority in the United States, institutional rules and democratic support. Both are uniquely reflected in our constitutional structure and can be important in how citizens think about the appropriate exercise of state power. In our federal government authority is conferred by constitutional provisions, which bestow powers to institutions (like Congress) and individuals holding particular office (like the President). Government officials also derive authority from democratic elections where specific individuals are chosen to fill institutional roles. Article II sets forth broad powers for the office of the Presidency and individuals who are elected to that office are thought to have a democratic mandate to pursue policy goals as they see fit within the confines of that institutional role (Reference KellyKelly 1983).

Constitutional provisions pertaining to executive action grant and limit presidential authority to powers that are in the Constitution and those can be reasonably inferred from its provisions. Constitutional rules, however, vary a great deal in terms of specificity. Sometimes it is clear the President has the power and is the appropriate government official to take action; other times, the President's authority to do things is less clear because his constitutional powers are not clearly delineated. In such instances, executive authority is often defined through the interplay of executive action and Supreme Court decisions ruling on the constitutionality of particular conduct (Reference WhittingtonWhittington 2007). Sometimes, however, the Chief Executive may take action that is constitutionally questionable, but never formally litigated in a case that makes its way to the Supreme Court. Such unchallenged executive action may take on an “air of validity,” that has real consequences for state power (Reference MarshallMarshall 2008). For instance, if the President does something of dubious constitutional validity that is immediately challenged and struck down by the Court, it may be widely perceived as illegitimate. But if he does something that is not challenged, the action can take on an appearance of validity that augments his authority over time.Footnote ¹

Hypotheses Regarding Assessments of Executive Authority

Acknowledging that the sort of representation citizens seek from the President is likely to be different from what they expect from legislators, Barker and Carman provide evidence that it is the nature of the issue that determines the type of leadership citizens seek from the Chief Executive (2012: 36–7).Footnote ² Consistent with classic research by Miller and Stokes (1963) they find that when surveyed, people prefer trustee style representation from the President in the domain of foreign relations. With regard to domestic issues they find that Carmines and Stimson's (1980) well-worn distinction between “hard” and “easy” issues helps to explain the latitude people are willing to extend to the President. Here, I use two distinct scenarios involving assertions of authority in domestic and foreign policy domains to allow for the possibility that people are more likely to defer to the President in judging actions relating to foreign affairs.

Issue Hypothesis: Assertions of executive authority involving foreign relations will receive higher legitimacy ratings.

Tyler argues that one of the most important aspects of legitimacy is that people will come to respect and obey the law even if they do not agree with it (Reference TylerTyler 1990, Reference Tyler1994). Thus, the idea that a particular action is legitimate or appropriate should be distinct from citizens’ assessment about whether it is a desirable course of action. Citizens who do not agree with an executive action may yet acknowledge the President has the legitimate authority to take that action. As such, I expect different considerations will be significant in assessments of the desirability and legitimacy of presidential action.

Differential Reasoning Hypothesis: Participants will think differently about the desirability and legitimacy of proposed action. Specifically, participants’ views about the policy problem the President seeks to address should significantly predict their views about the desirability of the action but have less of an impact on assessments of legitimacy. Conversely, expert consensus about the constitutionality of proposed action should have a strong effect on legitimacy assessments, but less of an impact on views of the desirability of proposed action.

Hypotheses Pertaining to Rules and Politics in Legitimacy Assessments

Citizens are not constitutional experts. Yet, because of the dual nature of legitimacy set forth above, I posit constitutional considerations will weigh heavily in their judgments of appropriate executive authority. In short, people are aware that there are institutional rules that the President must follow and they care that he respects the constitutional boundaries of executive power.Footnote ³ Very little is known, however, about how people understand institutional power and constraints in the context of particular problems and what the implication of that understanding is for thinking about the exercise of government authority (Boudreau and Lupia 2011). Indeed, citizens can often feel ill equipped to understand esoteric constitutional provisions and case decisions pertaining to executive authority. Yet, if Tyler is correct, this sort of information should be particularly relevant to citizens making decisions about whether a proposed action is a legitimate, or appropriate, exercise of presidential power.

Perhaps as result of the practical importance of this kind of information, there has developed a norm in media accounts of novel assertions of executive power to cite constitutional law “experts” on such matters (Reference BrodyBrody 1991; Reference GrecoGreco Larson 1988). Here, I take advantage of this norm by providing information about executive authority in the context of a realistic article discussing proposed action. Experimental conditions provide information to participants about (1) constitutional provisions that might provide authority for the President's conduct and (2) what consensus there is among legal scholars about the executive claim to authority under those provisions. The nature of consensus (clear authority to act v. unclear authority v. clearly no authority to act) is specifically manipulated to test the influence constitutional norms on participants’ assessments of legitimacy.

Institutional Rule Hypotheses: Expert consensus will have a strong and significant effect on participants’ assessment of the legitimacy of proposed action. Assessments will be highest when experts agree he has the authority to take action and lowest when they agree he does not.

Citizens do not make assessments about the legitimacy of executive action in a vacuum. There is a larger context where the democratic environment and political factors can influence judgments about what is legitimate in particular circumstances. Of course, just saying that the democratic environment is important does not tell you what aspects of the larger political context matter or when they will matter to citizens in thinking about presidential power. Here, I identify three political variables that should be of particular relevance to citizens in considering the appropriate use of executive authority: (1) the level of public support for the proposed action, (2) participants’ views about the problem being addressed, and (3) their feelings about the President. I also specify alternative hypotheses regarding the “conditions under which” these variables might play a significant role in the assessment of appropriate executive authority.

Majoritarian conceptions of democracy suggest that government acts appropriately where it is doing what a substantial majority of citizens want it to do (Reference LipsetLipset 1983; Reference CharltonCharlton 1986); state action appears most legitimate when it enjoys widespread support of the citizenry. On average, I expect that public support for executive action will boost participants’ assessments of the legitimacy of the proposed action.

Majority Support Hypothesis: Substantial public support for the executive action will increase legitimacy evaluations.

The other political variables I consider as part of the democratic context recognize that people bring their own political values and preferences into assessments of appropriate executive authority. Research from psychology and political science illustrates that personal preferences often make a difference in how individuals assess evidence (Reference KundaKunda 1990; Reference MacCounMacCoun 1998) and form opinions about policy and candidates (Reference Lodge and TaberLodge and Taber 2006, Reference Lodge and Taber2013). Citizens for instance, tend to make more positive assessments about the performance of candidates that they admire (Reference Lau and RedlawskLau and Redlawsk 2006) and assess research that supports their opinions about the death penalty as more skillfully executed than research that does not (Reference Lord, Ross and LepperLord, Ross, and Lepper 1979). This is often referred to as “confirmation bias” (Plous 1993) or “motivated reasoning” because psychologists have speculated that these tendencies are driven by a motivation to believe we are correct. Making assessments that support our prior beliefs is a directional goal that serves esteem in complex ways. Indeed, this sort of motivated decision-making can happen even where people have no obvious reason for doing so, or where they are sincerely trying to achieve accuracy goals (Reference KundaKunda 1990).

It makes sense that these tendencies could also come into play in making judgments about the legitimacy of government authority. Put simply, people may be more likely to believe actions that they agree with are the result of legitimate executive authority. Thus, how citizens feel about the particular problem being addressed is likely to factor into whether or not they see presidential efforts as appropriate.

Policy Preferences Hypothesis: Participants who believe the problem the President seeks to address is particularly pressing will be more likely to see the proposed action as legitimate.

Moreover, how people feel about the particular government official who is taking action may influence their judgments about the legitimacy of proposed action. This should be especially true for assessments of executive authority where actions are tied to a single identifiable official who citizens often feel quite strongly about. Individuals who are dissatisfied with the President may judge the legitimacy of his deeds less favorably than those who are satisfied with his performance in office. This can be seen as alternative conceptualization of the how presidential approval can influence individual judgments about legitimacy where personal support for the individual holding office influences assessments of appropriateness.Footnote ⁴

Presidential Satisfaction Hypotheses

Dissatisfaction with the President will decrease legitimacy assessments.

How Might Rules and Context Interact?

Critical to my theorizing and empirical inquiry, I am interested, not only in which of the identified variables are significant in citizen's thinking about presidential authority, but when they matter as well. Abstract questions about what we want from our chief executive may be different from how we think about specific assertions of authority in the context of real world problems and presidential personalities we care deeply about.

Here, Reference Hibbing and Theiss-MorseHibbing and Theiss-Morse's (2002) insights may be instructive. Although they corral evidence that has more to do with feeling about legislative actors than the President, these authors contend that people want elected officials who will make hard decisions for them, keeping public interest in mind, without a lot of citizen input.Footnote ⁵ They argue that most citizens want a government that runs on “autopilot” where representatives can be trusted to pursue the public good, without a lot of citizen involvement. The problem, according Hibbing and Theiss-Morse, is most citizens believe elected officials cannot be trusted—that self-interest will inevitably get in the way of enlightened officials acting on the public's behalf. They write, “people do not want responsiveness and accountability in government; they want responsiveness and accountability to be unnecessary,” contending that citizens have an “intense desire to give decision-making authority to someone else and to give those decision makers wide berth, as long as they are barred from taking advantage of their position for personal gain” (2002: 4,9).

Hibbing and Theiss-Morse's theory points to the type of information that might be relevant in assessing citizens views of the appropriateness of executive authority: whether or not the president is acting on the public's behalf or in a self-interested manner. Reference Barker and CarmanHeeding Barker and Carman's (2012) assertion that the constitutional distinctions are particularly important in thinking about the office, one way the President could be perceived as acting in a self-interested manner is if he is viewed as seeking to augment his own power by overstepping the bounds of constitutional authority.Footnote ⁶ This suggests the political factors examined here could be particularly important when there is evidence the President is violating institutional rules. It implies that in the usual course of events people are willing to extend the President significant latitude in assessing the legitimacy of executive action, but where circumstances suggest he is acting in accordance with his own self-interest they will call on political context to judge the appropriateness of his actions.

Research on motivated reasoning suggests a different way rules may moderate the role of the political variables. Specifically, psychological studies tell us motivated biases are most powerful where criteria for assessment are ambiguous (Reference Hodson, Dovidio and GaertnerHodson et al. 2002, Reference Braman and NelsonBraman and Nelson 2007). Research demonstrates that where people have a desire to be accurate they will use the most appropriate criteria available to make a decision, but where that criteria is indeterminate, directional goals and context can affect judgments in systematic ways (Reference KundaKunda 1990). Here, I believe participants will try to be accurate in their assessments of executive authority. The most appropriate criteria to judge the legitimacy of executive authority is information participants are provided about the President's compliance with constitutional rules. Thus, where participants are told experts agree that executive action is consistent or inconsistent with constitutional requirements I do not expect political factors to influence assessments about the legitimacy of proposed action. Instead, I expect the political variables identified will play more of a role where constitutional cues regarding executive authority are unclear; where rules are not determinative participants will use political context and their own preferences as information to make judgments about the appropriateness of executive action. In a similar vein, Doherty and Wolak (2012: 305) provide evidence that people use their own beliefs as “tool(s) of inference” to interpret ambiguous situations when assessing the fairness of legislative procedures. I believe the same dynamic will apply to assessments of executive authority.

Finally, it is entirely possible that constitutional considerations will not influence the role of the political variables I investigate in a manner consistent with Hibbing and Theiss-Morse's thinking or motivated reasoning hypotheses. Political factors might be important in assessments of legitimacy regardless of rules pertaining to executive authority. This would be the case if one or more of the political variables was significant in judgments of legitimacy across all expert consensus conditions.

As theory points to somewhat different ways political factors might operate in light of what participants are told about compliance with constitutional rules, I set forth three alternative hypotheses regarding how rules might moderate the influence of political factors on legitimacy assessments in this study.

Alternative Moderation Hypotheses

1. 1. No Moderation Effect: One or more political variables will be significant across all three expert consensus conditions.

2. 2. Motivated Reasoning Hypothesis: One or more of the political variables will be significant where presidential authority is unclear. This is consistent with the idea that political context will be used as additional information that influences judgments about legitimacy where rules are not determinative.

3. 3. Trigger Hypothesis: Political factors will be especially important when there is evidence the chief executive is not complying with institutional rules. This would be consistent with the idea that citizens prefer that government run on autopilot unless and until there is reason to believe officials are acting against the public interest. Under these circumstances a violation of institutional rules could serve as a “trigger” for political factors to kick-in in assessments of the appropriate use of executive authority.

Justification of Experimental Approach

Reference Druckman and HolmesDruckman and Holmes (2004) and Reference Gronke and NewmannGronke and Newman (2003) have commented on the need to investigate individual level thinking about phenomena related to the presidency. This study represents an attempt to do so, taking advantage of extant research involving executive authority and political psychology. Experiments provide unparalleled leverage to discover the cognitive processes underlying the assessments that are the subject of this inquiry. Moreover, because the approach allows for the manipulation of aspects of the factual scenario relevant to theorizing about how citizens think about executive power, we can observe how participants assess the legitimacy of executive action under relevant counterfactual conditions. For instance, by altering the level of democratic support for intended action, we can see how majority support influences people's thinking about the legitimacy of executive action where there is substantial support for the intended action and where there is substantial opposition. This would be exceedingly difficult to do using other techniques because public opinion involving the exercise of presidential power, if it is measured at all, tends to go in one direction. The effective isolation of such factors will allow us to discover, not only what variables are important in assessments of executive authority, but also, the conditions under which those variables will be influential in citizens’ evaluations in line with our theory.

This experiment was conducted with and undergraduate sample and then replicated with a nonstudent sample using Amazon's Mechanical Turk (M-Turk). Here, I present findings from both administrations; the logic is two-fold. First, I note there are those who question the generalizability of findings from experiments using undergraduate samples; college students are unquestionably younger and more educated, on average, than the general population (Reference SearsSears 1986). Still, findings from the student administration are no less “real” because they pertain to undergraduate assessments of authority (Reference Braman and NelsonBraman and Nelson 2007).

Concerns about external validity are strongest where there is some reason to think that differences between experimental populations and the general public would cause the judgments and cognitive processes investigators are looking at to be different (Reference KeppelKeppel 1982). Here, one of the differences that could be relevant is that students may be more likely to defer to “constitutional law experts” in the context of this inquiry. Admittedly, this difference might detract from the ability to generalize findings to the larger citizenry—but it also aids in the testing of my theoretical hypotheses (Reference Druckman, Kam, James, Green, Klukinski and LupiaDruckman and Kam 2011). Because, I am interested in seeing how institutional and political factors influence assessments of legitimacy where individuals sincerely believe the President is acting in ways that are consistent (or inconsistent) with his constitutional authority, this particular difference makes students uniquely sensitive to the theoretical states of mind the treatments are designed to create. Thus, the undergraduate sample is especially well suited to assess the internal validity of hypotheses regarding the influence of political factors under various types of constitutional constraint.

That said, I fully acknowledge that this tendency to defer to experts could undermine the external validity of results from the undergraduate sample for making generalizations to the general public. This is why I replicate the experiment with the nonstudent sample on M-Turk.Footnote ⁷ Specifically, the replication serves several distinct purposes. Besides the main function of all replications for seeing how findings “hold up” over distinct administrations, looking at results across samples should reveal whether undergraduates are, indeed, more likely to defer to expert opinion in this context. Finally, if undergraduates are, in fact, different in this important respect, results from the M-Turk sample should be useful for making the sort of generalizations many political scientists are most interested in.

Design Specifics

The experiment involves a 2 × 3 × 2 design. I present participants with a mock newspaper article that raises the question of the President's authority to take particular action. In the article I manipulate the issue context (debt ceiling vs. humanitarian aid to foreign states), nature of legal consensus regarding the assertion of executive authority (clear authority vs. unclear authority vs. no authority), and the level of democratic support for the proposed action (majority support vs. majority opposition). In a questionnaire administered prior to the presentation of the mock article I also measured participants’ satisfaction with President Obama, and their opinions on the issues that are the subject of each scenario.

Both articles involved constitutional issues that were in the headlines in the summer of 2011. The debt ceiling scenario takes advantage of the debate that occurred concerning what options might be available to President Obama if Congress failed to raise the debt ceiling under a significant threat of default. Some pundits argued that the President would be within his authority to unilaterally raise the debt ceiling, citing his authority as Chief Executive and a provision in the 14th Amendment stating that the debt of the United States “shall not be questioned.” The second scenario involved the commitment of troops to Somalia amid a significant humanitarian crisis. There was a substantial drought and famine in that region where there was talk of the need for President Obama to take such action; some pundits and politicians were concerned such a commitment would spread United States forces too thin. The mock article stated that, although prior Presidents had taken similar steps to commit troops under their authority as Commander in Chief, the Supreme Court had never definitively ruled on the President's constitutional authority to do so without congressional approval.

In each instance participants read an article describing the conflict and stating that the President was considering taking action pursuant to his authority as Chief Executive bolstered by language in the 14th Amendment (for debt ceiling scenario) or in his capacity as Commander in Chief (humanitarian aid scenario). Participants were explicitly told that the Supreme Court had not yet considered the issue, but that constitutional experts were weighing in on the matter. One third of participants were told legal experts agreed that it was clear the President had the constitutional authority to take the proposed action, one third were informed that opinion among legal experts was divided and that it was unclear how the Supreme Court would rule on the issue, and one third were informed that legal experts agreed he lacked the authority to take action. I also manipulated the level of democratic support for the proposed action. Half of the participants were told a recent poll indicated 85 percent of Americans agreed the President should take measures described, and half were told that that only 15 percent of Americans agreed with the proposed action.

Following a series of manipulation checks, participants in each scenario were asked (1) whether they agreed the President should take the proposed action (measured on a six-point scale) and (2) regardless of whether or not they thought it was a good idea, whether in their view, the action represented a legitimate (or appropriate) exercise of presidential authority (measured on a seven-point scale). These questions were used as the dependent variables in the analysis. Precise wordings for mock articles and all measures are available in the Supporting Information Appendix.

Procedure

Two hundred and twenty three undergraduates participated in the undergraduate administration of this study in January of 2012.Footnote ⁸ Students at a large midwestern university who agreed to take part in political science experiments for course credit were given an internet link to a Qualtrics website where the experiment was programmed. Participants were told the study was about how internet media report about government officials to avoid experimental effects that might be caused by their being sensitized to experimental hypotheses. The replication study was administered via Amazon's Mechanical Turk in February of 2013. Those in the nonstudent sample were paid for their participation in accordance with Mechanical Turk practice and procedures. A total of two hundred and sixty-three participants took part in the M-Turk administration.Footnote ⁹ On agreeing to take part they were directed to a similar Qualtics website to participate in the study.

In each administration participants were first asked to answer a number of policy questions relating to their views on political issues of the day. Among these questions were the two policy measures that are particularly relevant for our analysis. Participants were asked how serious they thought the threat financial default was to the health of our economy. They were also asked whether they agreed the United States was obligated to help foreign states in times of humanitarian crisis. The initial battery of questions also asked participants how satisfied they were with President Obama's job performance.Footnote ¹⁰

After answering these questions each participant read one of the mock articles containing the experimental manipulations involving either the debt ceiling or the commitment of troops to Somalia. Participants then answered a series of manipulation checks and questions about the desirability and legitimacy of the proposed executive action.Footnote ¹¹ Finally, at the end of the experiment participants were asked a number of demographic questions and debriefed as to the nature of the study.

Analysis

To see how these variables influenced participants thinking about the executive authority that was the subject of the experimental scenarios and test the influence of manipulated variables I conduct a series of Analysis of Covariance (ANCOVA) analyses particularly well suited to interpret experimental data (Reference KeppelKeppel 1982). I also present disaggregated regression analyses to test the influence of political variables in each constitutional consensus condition to see how their influence is moderated by information about institutional rules.

Before I discuss the results from those analyses it is important to note there is no support for the Issue Hypotheses in either sample. Participants’ responses to questions about the desirability and legitimacy of the proposed executive action did not significantly differ across issue area. In the undergraduate administration the mean on our six-point scale asking participants if they thought the president should take the proposed action was 3.37 for the debt ceiling scenario and 3.35 for those given the foreign policy scenario (t[219] = 0.85; sig = 0.93). For the legitimacy measure (seven-point scale) means were 4.37 and 4.63, respectively (t[219] = 1.03; sig = 0.31). In the M-Turk administration means on the desirability questions were also quite similar, 3.29 for the debt ceiling and 3.24 for the foreign policy scenario (t[244] = 0.26; sig = 0.78). Means on the legitimacy measures were 4.27 and 4.24 (t[244] = 0.14; sig = 0.89), respectively. Thus, it appears participants found both scenarios about equally plausible. As such, I pool the data for both issue scenarios to investigate the influence of variables on assessments of the desirability and legitimacy of proposed executive action in each sample.Footnote ¹²

ANCOVA Analyses: Comparing Desirability and Legitimacy Assessments

The Differential Reasoning Hypothesis posits that people think differently about the desirability and legitimacy of proposed executive action. As one might expect, these measures are significantly correlated in each sample, but the correlations are moderate, r = 0.56 in the undergraduate sample, r = 0.62 in the M-Turk sample. This suggests, consistent with the literature, one does not have to see an action as desirable to acknowledge its appropriateness.

I also hypothesized that different factors would be relevant in participants assessments about the desirability and legitimacy of proposed action. To test this hypothesis I conducted four-way ANCOVAs of each measure in both samples. I include three factors representing the manipulated variables: issue area, expert consensus about the use of executive authority, and democratic support. I also created a factor based on respondents’ satisfaction with the president's job performance. Specifically, I divided participants who expressed dissatisfaction with President Obama and those who were satisfied or felt neutral about his job performance. The “fully saturated” models include tests for these four factors and all possible (two-, three-, and four-way) interactions between them (Reference Akien, West and RenoAkien, West, and Reno 1991: 174). Participants’ response to the relevant policy question for the treatment to which they were assigned was included as a covariate in each pooled model to test and control for the influence of policy views. Table 1 sets forth the results for questions about the desirability of proposed actions in the undergraduate and M-Turk samples.

Table 1. Results for ANCOVA Analyses on the Desirability of Proposed Action for Student and Nonstudent Samples

^* p < 0.05; ^**p < 0.01; ^***p < 0.00.

Results demonstrate that, as hypothesized, participants’ policy preferences about the issue had a significant impact on assessments of the desirability of proposed action in both samples (F[1, 220] = 13.23; sig < 0.001 undergraduate; F[1, 145] = 22.26, sig < 0.001 M-Turk). Indeed, it makes sense that people who see the possibility of default as a serious threat to the economy and/or think the United State has an obligation to help foreign nations in times of humanitarian crisis would be more likely to think the President should act to address these issues. In the nonstudent sample, the issue variable also has a significant effect on desirability that must be interpreted in light of its significant interaction with presidential satisfaction (Reference KeppelKeppel 1982—see discussion below).

Results also indicate that the level of expert consensus about the President's constitutional authority had a significant, though modest effect (effect size = 0.05) on participants’ views of the desirability of proposed action in the undergraduate sample (F[2, 219] = 4.16; sig. < 0.01). Specifically, undergraduates who were told that experts believed the President had clear authority to act were significantly more likely to say the President should take the proposed action than those who were told experts agreed he lacked authority to do so. The mean for desirability where the expert consensus about authority was unclear fell between these two extremes. In the M-Turk sample the pattern of means was identical although the effect of the expert consensus variable was not significant in assessments of desirability of proposed action. This could be one indication that undergraduates were, indeed, more likely to be influenced by expert opinion than those in the nonstudent sample; it shows that the expert consensus conditions had a persuasive influence on undergraduates opinions about the desirability of proposed action that was not evident in the M-Turk sample.Footnote ¹³

Finally, results demonstrate that in both samples participants’ level of satisfaction with the President had the most dramatic influence on desirability assessments (F[1, 220] = 31.66; sig. < 0.001 undergraduate; F[1, 245] = 36.28, sig. < 0.001 M-Turk). Those who were dissatisfied with the President were less likely to think he should act than those who were satisfied or felt neutral about his job performance; this variable had the largest effect size in assessments of the desirability of proposed action, accounting for approximately 0.13 of the variance observed in the undergraduate sample and 0.15 in the nonstudent sample. Moreover, in the nonstudent sample the satisfaction variable was embedded a significant interaction with the nature of the issue (F[1, 250] = 12.77; sig. < 0.01). The pattern of marginal means indicates that in that in the M-Turk administration, how participants felt about Obama was especially important in shaping participants’ views of the desirability of domestic action. Specifically, those who were dissatisfied with Obama were much less likely to see action with respect to the debt ceiling as desirable (mean of 1.86) compared to those who were satisfied or felt neutral about his job performance (mean 3.79). The means of those who were satisfied and dissatisfied with Obama were closer with respect to the desirability of aid to foreign states in times of humanitarian crisis (3.54 vs. 3.01, respectively).

The level of public support for the proposed action did not significantly influence participants’ views of whether or not the president should take the action, although values were slightly higher in conditions indicating significant support. This could reflect a slight influence for majority sentiment on participants’ expressed opinions about the desirability of proposed action in this experiment.

Consistent with the Differential Reasoning Hypothesis, analysis of assessments of the legitimacy of executive action, set forth in Table 2, reveal a somewhat different pattern of results. First, policy views do not have the significant effect they had on desirability assessments in judgments of legitimacy in either sample. Moreover, the level of expert consensus about the President's constitutional authority to act, has a strong and significant influence on participants’ assessments of the legitimacy of proposed action in both samples (F[2, 219] = 25.05; sig < 0.001 undergraduate; F[2, 244] = 16.34, sig < 0.001 M-Turk). Relative means are as predicted; assessments of legitimacy are at their highest when participants are told there is a clear consensus among legal scholars that the President has constitutional authority to take the proposed action and lowest when they are told they agree he does not. The effect size for this variable is the largest of those in the analysis of legitimacy assessments in both samples accounting for 0.20 of variance for undergraduates and 0.13 in the M-Turk sample. Thus, it seems people pay particular attention to constitutional factors and institutional rules in assessing the legitimacy of executive action.

Table 2. Results for ANCOVA Analyses on the Legitimacy of Proposed Executive Action for Student and Nonstudent Sample

^* p < 0.05; **p < 0.01; ^***p < 0.001.

Majority support does not influence participants’ assessments of legitimacy in either administration. Thus, the only political variable that is significant in legitimacy assessments is participants’ level of satisfaction with President Obama. The main effect for the variable is highly significant in both samples (F[1, 220] = 26.10; p < 0.001 for the undergraduate sample and F[1, 245] = 24.69, p < 0.001 in the M-Turk Administration). As predicted, those who are dissatisfied with the President's job performance are much less likely, on average, to see the proposed action as a legitimate exercise of executive power.

Turning to the hypotheses about when political factors are likely to matter, there is a significant interaction (F[2, 219] = 3.13; p < 0.05) between expert consensus and participants’ feelings about the President in the analysis of legitimacy assessments in the undergraduate sample, but not in the nonstudent sample (F[2, 245] = 1.09, n.s.). To get a better idea of what is going on, I present the marginal means for the presidential satisfaction factor for each of the consensus conditions in Figure 1. Panels A and B demonstrate a main effect for constitutional consensus conditions that is consistent with hypotheses. In each sample those dissatisfied with Obama were less likely to view actions as legitimate across all consensus conditions. For undergraduates the differences appear quite substantial where presidential authority is unclear and when he lacks authority. Where experts agree he has the authority to take action, however, there is not a large difference; this is why the interaction between presidential satisfaction and constitutional consensus conditions is significant; it indicates presidential satisfaction has a somewhat different effect across consensus conditions. In the M-Turk sample differences appear quite substantial across all three consensus conditions; thus the interaction is not significant in the main ANCOVA of legitimacy assessments for that administration.

Figure 1. Marginal Means for Legitimacy Assessments Across Expert Consensus Conditions. Panel A: Undergraduate Sample. Panel B: M-Turk Sample.

To confirm this intuition, main effects from disaggregated ANCOVAs for each expert consensus condition (using the same factors and policy covariate) are presented in Table 3. In the undergraduate sample, dissatisfaction is not significant where participants are told experts agree that the President has clear authority to act, but the factor is significant where they are told his authority is unclear and when participants are told experts agree he does not have authority to take the proposed action. In the nonstudent administration, the factor gauging presidential satisfaction is significant across all three expert consensus conditions.

Table 3. F-Test Results for Disaggregated ANCOVA Analyses on the Legitimacy of Proposed Executive Action for Student and Nonstudent Samples

⁺ p < 0.10; ^*p < 0.05; ^**p < 0.01; ^***p < 0.001.

Disaggregated Regression Analyses

To further probe the influence of presidential satisfaction under different levels of constitutional constraint I also conducted disaggregated regressions of legitimacy assessments for each sample across the consensus conditions. Specifically, the disaggregated regressions allow for the exploration of the role of satisfaction across the variable's entire seven-point range. It was necessary to dichotomize the variable to create a factor allowing us to see how presidential satisfaction interacted with manipulations for the ANCOVAs set forth above, but given the skew in the variable (see note 10) some readers will, no doubt, want to see this more nuanced approach.Footnote ¹⁴ The regression analyses also allow us to observe the influence of additional variables like gender, race, and party affiliation that have been theorized to influence perceptions of government action.Footnote ¹⁵

Each model includes variables for the issue type (coded 1 for debt ceiling, 0 for humanitarian aid), majority support (1 for majority support, 0 for opposition), presidential satisfaction (1–7 scale with higher values indicating greater satisfaction with Obama), policy opinion on relevant measure (1–6 scale with higher numbers indicating participants think debt ceiling serious problem or United State has strong obligation to aid foreign states in times of humanitarian crisis), party (1–7 scale with higher numbers coded as Democratic affiliation), gender (1 male, 0 female), and race (1 African American, 0 other). I also include an interaction for the manipulated factors of issue area and majority support to ensure all experimental effects are accounted for in the regression analysis (Reference Auspurg and HinzAuspurg and Hinz 2014).Footnote ¹⁶

Results of these analyses appear in Tables 4 and 5. For participants in the undergraduate sample, again satisfaction with Obama is only political variable that significantly influences legitimacy assessments (see Table 4). Neither majority support, nor participants’ expressed policy views, influenced their feelings about whether the President is acting approptiately. The direction of the satisfaction coefficient indicates, as we observed in the ANCOVA analyses, those who approve of Obama are more likely to judge his actions as legitimate. Moreover, the variable only obtains statistical significance, where his authority to act is unclear. This is consistent with the motivated reasoning hypothesis suggesting feelings about the President can act as information where constitutional rules regarding executive authority are ambiguous. Satisfaction is not significant where there is a consensus that the President has authority to take action; it seems constitutional considerations are driving participants relatively high assessments of the legitimacy of proposed action in this category.Footnote ¹⁷

Table 4. Ordinary Least Squares Regression for Legitimacy Assessments in Each Expert Consensus Category—Undergraduate Sample

⁺ p < 0.10; ^*p < 0.05; ^**p < 0.01; ^***p < 0.001 (two tailed test).

Table 5. Ordinary Least Squares Regression for Legitimacy Assessments in Each Expert Consensus Category—Mechanical Turk Sample

⁺ p < 0.10; ^*p < 0.05; ^**p < 0.01; ^***p < 0.001 (two tailed test).

Interestingly, where undergraduates are told the President lacks authority to take action, the regression analysis indicates presidential satisfaction has a marginal effect on judgments of legitimacy—this hints at the possibility, derived from Hibbing and Theiss-Morse's thinking, that deviation from constitutional rules can act as a trigger for political considerations to factor into assessments of legitimacy as citizens try to assess appropriateness of executive motivations. Those who are dissatisfied with Obama may be more likely to impute self-interested motives in questionable executive action.Footnote ¹⁸ This is similar to the finding from the disaggregated ANCOVA (Table 3) where those who were dissatisfied with the President had significantly lower assessments of proposed action in this category. The fact that it is only marginally significant in the disaggregated regression analysis is probably because there is less pronounced variation in legitimacy ratings across the variables’ entire range than for the dichotomized version of the variable.Footnote ¹⁹ Results from both the disaggregated analyses are similar in that they suggest that participants who are generally satisfied with Obama are more likely to extend the benefit of doubt in judging his actions where experts calls his constitutional authority into question.

Results in the nonstudent sample, set forth in Table 5, provide even stronger support for this trigger hypothesis. Again, satisfaction with Obama is the only political variable that influences participants’ legitimacy assessments, but the pattern of influence is distinct from that in the undergraduate sample. Here, feelings about Obama have at least a marginal effect across all three categories of constitutional consensus. Again, those who are satisfied with Obama are more likely to view his actions as legitimate. This is likely because, as we observed in the ANCOVA analyses, the constitutional consensus conditions did less to constrain the role of satisfaction in legitimacy assessments; M-Turk participants were simply less likely than undergraduates to defer to the “constitutional experts” in making their assessments about legitimacy to the exclusion of other considerations. Here, presidential satisfaction has its strongest effect where M-Turk participants are told that experts agree the Chief Executive lacks authority to take proposed action. This echoes the finding in the undergraduate sample that a violation of constitutional rules may serve as a trigger for feelings about Obama to influence assessments of the legitimacy of proposed action. I consider the implications of these findings in the discussion section that follows.

Discussion

A summary of results with respect to each hypothesis is presented in Table 6. This study reveals several important things about how people think about executive authority. First, there was no support for the Issue Hypothesis; although legitimacy assessments were slightly higher for the foreign policy scenario, differences were not significant. Participants in both samples made similar judgments about the authority of the president to act with regard to the foreign and domestic issues that were the subject of experimental manipulations.

Table 6. Summary of Results

Consistent with prior research, evidence supports the Differential Reasoning Hypothesis. It seems people do not have to agree with proposed measures to recognize the President's authority to act; desirability and legitimacy judgments were only moderately correlated. Moreover, different variables were significant in each assessment. As predicted, policy preferences mattered more in assessments of whether or not the President should take particular action than in judgments about whether he had the legitimate power to do so. Moreover, consistent with the Institutional Rule Hypothesis, constitutional considerations weighed heavily in judgments of legitimacy; expert assessment of Presidential compliance with institutional rules was highly significant in both samples.

Of the three political variables tested, only one, satisfaction with the President, significantly influenced assessments of about the legitimacy of proposed executive conduct. But even where satisfaction factored into evaluations about the appropriateness of executive conduct, as Figure 1 clearly illustrates, it operated in a range that was determined by compliance with constitutional rules.

Surprisingly, there was no support for the Majority Support Hypothesis; the level of public support for proposed action did not matter for legitimacy assessments in either sample. Thus, at least for the issues that were the subject of this study, participants’ seemed to treat the President as a “trustee” representative, making legitimacy assessments regardless of the level of popular support for proposed conduct. This could be a reflection of citizens’ general faith in the President's discretion as a policy expert—but it could also be an artifact of the particular issues employed in this study. The debt crisis and the determination of circumstances that merit humanitarian intervention could both represent complex issues where people are likely to give politicians significant latitude to act against what a majority of citizens want. Further testing should be done to test the durability of this finding before we conclude broad based public support cannot augment the perceived legitimacy of proposed action for different types of issues.

With regard to the preference variables tested in this study, it is clear that feelings about the person acting as President mattered more than preferences about the issue that proposed action sought to address. Although evidence did not support the Policy Preferences Hypothesis, there was substantial support for the Presidential Satisfaction Hypothesis. Results indicate that those who were dissatisfied with President's job performance were less likely to deem proposed conduct a legitimate exercise of executive authority than those who were satisfied or felt neutral about his performance. Significantly, however, findings also demonstrate that the influence of presidential satisfaction was conditioned on the level of compliance with constitutional rules, albeit in different ways across samples.

Regression results from the undergraduate administration were most consistent with the hypothesis that rules can constrain the role of political factors in participants’ judgments; satisfaction with Obama had its strongest influence on judgments of legitimacy where the President's authority to act was unclear. This is consistent with the research on motivated reasoning showing that citizens often use “feelings as information” when other criteria for evaluation are ambiguous. This likely occurred in the undergraduate sample because student participants were willing to defer to the “constitutional experts” cited in our scenarios about the constitutionality of proposed action to the exclusion of other considerations. In other words, they were particularly sensitive to the influence of our manipulations of constitutional constraint; they took expert information at face value in making their evaluations. As such, the results from the undergraduate sample are particularly useful for evaluating the internal validity of experimental hypotheses; they demonstrate the potential for institutional rules to constrain political factors where citizens accept as true evidence that the president is complying with (or violating) constitutional dictates.

Of course, ordinary citizens may be less likely to defer to experts. This is why the experiment was replicated using the nonstudent sample. Results from the M-Turk replication, while consistent with the undergraduate administration in many important respects, do seem to demonstrate that nonstudent participants were less likely to defer to experts in legitimacy evaluations to the exclusion of other factors. Satisfaction with Obama was at least marginally significant across all three levels of constitutional constraint. Thus, in the M-Turk sample, rules played less of a role in dulling the role of presidential satisfaction in legitimacy assessments across all levels of constraint.

Regression results in the nonstudent sample also differed from those in our student sample, in that, attitudes toward the President had their strongest influence when participants were told that experts agreed Obama was violating his constitutional authority. This is consistent with the idea that political factors are especially important where there is evidence the President is acting outside the bounds of his authority. In line with Reference Hibbing and Theiss-MorseHibbing and Theiss-Morse's (2002) thinking, we saw that an alleged violation of constitutional authority can act as a “trigger” for citizens to use their feelings about the Chief Executive to judge the legitimacy of his actions.

This particular result is also consistent with findings from political psychology suggesting that citizens are not passive, disinterested observers of politics. Reference Lodge and TaberLodge and Taber (2006) argue that people tend to treat politics like sports fans—supporting political parties and politicians they like and assessing those they do not like more critically. In essence, it could be that the expert consensus against constitutional authority signaled to participants that a debate about constitutional powers was imminent; thus, it may have served as a cue for participants to “choose a side” and rally behind the President if they liked him or against the Chief Executive if they were dissatisfied with his performance. This raises the intriguing possibility that popular presidents may have an easier time than those who are unpopular in expanding their authority in response to current issues without widespread public opposition. If people like the President they will be more likely to accept even questionable assertions of authority that might be interpreted as “power grabs” coming from less popular Chief Executives. It also suggests that Chief Executives will have the most success with novel assertions of power at particular points in their terms when they enjoy a good deal of public support; this is more likely to be the case, for instance, at the beginning, rather than at the end of their tenure.

Of course, further testing should be done to test the durability of these findings and probe the cognitive mechanisms underlying these results. Some argue that President Obama is a particularly polarizing figure, so it would be useful to see how similar variables apply to citizens’ thinking about assertions of authority made by our next chief executive. Moreover, future studies might investigate how citizens think about Presidential authority with regard to other important issues. One might hypothesize, for instance, that how people think about the commitment of troops to active hostilities, like the fight against Islamic State in the Middle East, is distinct from how they think about the President sending troops to aid in the sort of humanitarian crisis investigated here; or that public sentiment could play an more important role with regard to domestic issues that are less complex and more emotionally charged than the debt crisis (such as the recent executive order on immigration reform). Future studies can (and should) be designed to look at these interesting questions; this inquiry gives us a good start in investigating such phenomena.

Conclusion

This paper set forth a theory of how citizens think about the appropriate exercise of executive power that accounted for institutional rules, political factors and the interaction between them. It revealed not only which political factors were important among a number of theoretically based variables, but when they mattered in assessments of executive action as well. Employing an experimental design, we saw that compliance with constitutional rules and satisfaction with the President weighed most heavily in the minds of student and nonstudent participants when judging the legitimacy of executive action. Findings suggest that institutional rules can constrain the role of political factors in assessments of the legitimacy, but this is not inevitably the case. Strong feelings about the individual acting as Chief Executive and an added measure of skepticism about the wisdom of “experts” may be prevalent where ordinary citizens think about government authority in the context of real world issues.

These findings complement recent studies investigating how similar factors shape evaluations of judicial and legislative authority shedding light on how government action is assessed across different branches of government. For instance, Reference Braman and EasterBraman and Easter (2014) find that judges’ compliance with decision-making rules is by far the most important factor in assessments of the appropriate exercise of judicial authority. Judges are rated as acting significantly more appropriately when their decisions are characterized as being motivated by legal considerations compared to when they are portrayed as being driven by political ideology, contributions, or bribes. Interestingly, Braman and Easter find that nothing can bolster assessments of decisions based on grounds experimental participants view as particularly suspect (bribes and political contributions), but political factors (like significant public support for the outcome of a case involving tort liability, and participants’ own agreement with the policy implications of case a involving gay marriage) can enhance the legitimacy ratings of decisions characterized as being motivated by legal and ideological considerations.

In the congressional context, Reference BramanBraman (2015) reports compliance with constitutional rules pertaining to federalism is significant in assessments of exercise of national legislative authority. Where experimental participants are told that constitutional experts agree it should be the national government acting on immigration and gun control issues, their assessments are significantly higher than when they are told experts believe that Congress would be infringing on state prerogatives in taking proposed action. Indeed, the pattern of results for the impact of constitutional rules on legitimacy assessments looks quite similar to that observed in the current study. In the legislative context, however, it is participants’ agreement with the proposed policy action that interacts with rule compliance most prominently. As we observe here with feelings about Obama, where they are significant, policy views can have a strong influence in evaluations of legislative authority. This is quite distinct from the judicial context where compliance with decision-making rules predominate and political factors seem to matter at the margins.

The picture emerging from this nascent line of research indicates that while individuals may be willing to extend latitude to presidents they like and/or legislative actions they agree with in judging the appropriateness of government deeds, that latitude is not unlimited. It is bound by conceptions of appropriate behavior across the different branches of government. Clearly, citizens pay very close attention to whether government actors are following prescribed rules in evaluating the legitimacy of state action. Moreover, evaluations of the appropriate exercise of government power are highly contextual, depending on the political environment at the time of the assertion of contested authority.

Still, it is possible for political actors to take advantage of hospitable conditions when they arise to expand the authority of particular offices like the Presidency in concrete ways. This line of research may help us to understand why such expansions of power tend to be incremental rather than revolutionary. Perhaps most importantly, it can illuminate the “conditions under which” these shifts in authority are more (and less) likely to occur. Clearly, both rules and political context play an important role in how citizens think about the appropriate exercise of state authority and they can interact in complex ways that are worthy of additional theoretical and empirical attention.