MSU LIBRARIES “ RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. PRESIDENTIAL MANAGEMENT OF PUBLIC SUPPORT: A THEORETICAL AND EMPIRICAL INQUIRY by Dennis Michael Simon A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Political Science 1981 ABSTRACT PRESIDENTIAL MANAGEMENT OF PUBLIC SUPPORT: A THEORETICAL AND EMPIRICAL INQUIRY BY Dennis Michael Simon Recent scholarship on the American presidency and particularly the research which focuses upon the determinants of public support for American presidents has emphasized the weakness of the office and the inability of presidents to influence how the public evaluates their performance. The expectations created by this literature are that public support will invariably decline over the course of the term and that presidents can neither halt nor reverse this pattern. This thesis challenges such research and offers an alternative explanation of public support which, unlike prior studies, includes presidential behavior as an integral component and asserts that support is subject to presidential control. This explanation is formulated and assessed in three phases. Chapters 1 and 2 develop the arguments that public support operates as a presidential resource and that, because of its value, presidents will be motivated to manage this commodity. To substantiate these assertions, it is shown that public support is a determinant of both electoral and legislative outcomes. Having established that there exists a rationale.for.presidential management of public support, the next phase of the thesis constructs a mathematical model of citizen and presidential decision making. A formal argument then illustrates how the value of public support impinges upon the pursuit of presidential goals and the content of presidential actions. The objective in the third portion of the thesis is to estimate and evaluate an econometric model based upon several prepositions derived from the formal model of citizen and president interaction. The model is then applied to the case of the Vietnam.war during the administrations of Lyndon Johnson and Richard Nixon. The econometric specification is grounded upon the argument that public disapproval with the conduct of the war by these presidents is a function of (1) war- related outcomes, (2) personal and policy-based presidential actions, and (3) exogenously-induced, war-related events. The model is estimated using data from the Harris and Gallup Polls and its performance is compared to the results generated by three alternative models derived from prominent arguments in the public support literature. In contrast to the denial of presidential control and mechanistic explanations of public support, the results of the analysis strongly support the conclusion that performance assessments are subject to presidential control. Copyright by Dennis Michael Simon 1981 For Debbie A Partner and Companion ii ACKNOWLEDGEMENTS I would like to express my gratitude to the members of my diss- ertation committee: Professors Charles W. Ostrom, Jr. and John H. Aldrich, who served as co-chairpersons of the committee; Professor David W. Rohde; and Professor Charles F. Cnudde. These individuals made graduate study at Michigan State a worthwhile and exciting experience. I am most appreciative for their assistance, encouragement, and advice. A special debt is owed to the co-chairs of my committee. Charles Ostrom has not only served as an advisor, but has been both a colla- borator and a friend. His creativity, criticism, and concern have been and will remain highly valued. The help and friendship of John Aldrich, adviser and now colleague, have been substantial. His thoroughness and precision continue to provide a worthy example. Thanks are also due to Professor Paul Abramson, who cheerfully provided me with information useful in writing this thesis and to Harriet Dhanak, head of the Politometrics Center at Michigan State, who gave much aid and comfort in compiling the data employed in this work. An incalculable debt is owed to Debra Simon. Her assistance in preparing this thesis was substantial; her confidence and enthusiasm provided constant encouragement. iii TABLE OF CONTENTS LIST OF TABLES vi LIST OF FIGURES viii CHAPTER Oh -- THE VALUE OF PUBLIC SUPPORT . 1 The Salience of Public Support for the President The Value of Public Support: Some Impressionistic Evidence 4 Public Support as a Resource: Some Implications 10 The Management of Public Support: An Outline 13 Notes to Chapter 1 19 Tables 20 Figures 21 CHAPTER TWO - PUBLIC SUPPORT AS A PRESIDENTIAL RESOURCE: AN EMPIRICAL ASSESSMENT 22 Presidential Support and Electoral Outcomes 23 Presidential Support and Congressional Outcomes 27 Presidential Performance and Other Citizen Evaluations 47 Conclusion 49 Notes to Chapter 2 52 Tables 55 Figures 67 CHAPTER THREE - THE CALCULUS OF PUBLIC SUPPORT: CITIZEN AND PRESIDENTIAL DECISION MAKING 68 The Criteria of Citizen Choice 70 Institutional Demands 73 Electoral Demands 76 Situational Demands 79 The Criteria of Citizen Choice: A Spatial Representation 81 The Formation of the Initial Support Coalition 88 iv CHAPTER THREE (cont'd) A Model of the Citizen Support Decision 92 Presidential Decision Making 106 Presidential Goals 109 A Calculus of Presidential Action 111 The Choice of Presidential Actions: An Illustration 123 Conclusion 133 Notes to Chapter 3 135 Tables 137 Figures 140 CHAPTER FOUR -- THE MANAGEMENT OF PUBLIC SUPPORT 146 Toward An Alternative Model of Public Support 146 Presidential Actions As Levers of Control 150 The Pursuit of a Management Strategy: Richard Nixon and Vietnam. 171 A Strategy for Testing the Alternative Explanation 175 Notes to Chapter 4 178 Tables 181 Figures 186 CHAPTER FIVE - MANAGING PUBLIC SUPPORT: AN EMPIRICAL ASSESSMENT 189 we: and Public Support: The Conclusions of Prior Research 191 A Model of Specific Performance Assessments: The Case of Vietnam. 196 Operationalization 208 Model Estimation and Evaluation 216 Conclusion 222 Notes to Chapter 5 225 Tables 227 Figures 236 CONCLUSION A Summary of Results Declines in Support: Inevitable Decay or Presidentially Induced? BIBLIOGRAPHY . vi 239 239 240 244 Table Table Table Table Table Table Table Table Table Table Table 1.1- 2.1—- 2.2-- 2.3-- 2.4 2.5-- 2.6-- 2.7-- 2.8-- 209-- LIST OF TABLES Citizen Evaluations of Presidential Performance: The Frequency of Measurement by the Gallup Opinion Poll, 1938-1980 Citizen Evaluations of Presidential Performance and Candidate Choice in House Midterm Elections Citizen Evaluations of Presidential Performance and Turnout in Midterm Congressional Elections Citizen Evaluations of Lyndon B. Johnson and Voting Behavior in the 1968 National Election, By Partisan Affiliation Citizen Evaluations of Richard M. Nixon and Voting Behavior in the 1972 National Election, By Partisan Affiliation Citizen Evaluations of Gerald R. Ford and Voting Behavior in the 1976 National Election, By Partisan Affiliation Public and Congressional Support for the President, 1953-1978 The Impact of Presidential Support among Partisan Identifiers upon Roll Call Support for Presiden- tially Endorsed Legislation The Impact of Presidential Support Among Republican Identifiers upon Roll Call Support for the Pres- ident by Republican Members of Congress (Estimates Obtained from Cochrane-Orcutt Procedure) The Relationship Between Public Support and Presidential Roll Call Victories in the House and Senate, 1953(I)-1978(IV) 2.10--A Comparison of Presidential Success on Domestic and Foreign Roll Call Votes in the House of Representatives vii 20 55 56 57 59 6O 61 62 63 64 65 Table .Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table 2.11-4Presidential Roll Call Victories as a Function 3.1-- 3.2- 3.3-- 5.1- 5.2-- 5.3-- 5.4- 5.5-- 5.6-- 5.7-- 5.8-- 5.9-- of Public Support, Legislative Activity, and Legislative Positions, 1959(I)-1978(IV) The Demographic Composition of the Initial Support Coalition A Comparison of Electoral and Initial Support Coalitions: Republican and Democrat Presidents The Impact of Differential Public and Congression- al Salience Weights on the Optimal Presidential Locations: An Example . Econometric Models of Public Support: A Summary Presidential Decision Making and Issue Activation: An Example of the Impact on Public Support Presidentially and Exogenously Induced Rally Points: A Preliminary Classification Televised Reports to the Nation on Vietnam by Richard Nixon, 1969-1972 The Reduction of U.S. Forces in Vietnam,1969-l972: A Summary The Impact of War on Public Support for American Presidents: A Summary The Set of Policy-Based Presidential War Actions Public Opinion Toward Nixon's Troop Withdrawal Policy Major Domestic Anti-War Demonstrations, 1965-1972 Major North Vietnamese-Viet Cong Offensives, 1965-1972 Evaluations of Presidential war Performance: The Presidential Action Model Evaluations of Presidential war Performance: MMd.l Evaluations of Presidential war Performance: Model 2 Evaluations of Presidential War Performance: Model 3 viii 66 137 138 139 181 182 183 184 185 227 228 229 230 231 232 233 234 235 Figure 2.1- Figure 3 .1-- Figure 3.2-- Figure 3.3- Figure 3.4-- Figure 3.5-- Figure 4 .1-- Figure 4.2-- Figure 4.3- LIST OF FIGURES A Model of Citizen-President Decision Making The Relationship Between Presidential and Cong- ressional Dissatisfaction, 1943-1978 The Representation of Expected and Actual Pres- idential Positions in the Action-Outcome Space: An Example A Comparison of Public Support During the Initial Six Months of the Presidential Term The Optimal Trajectory of Presidential Actions: An Example The Impact of Varying Salience Assumptions on the Optimal Trajectory of Presidential Actions: An Example The Impact of Differential Public and Cong- ressional Salience Weights on the Optimal Presidential Trajectory: An Example The Relationship Between Public Concern About the Economy and Economic Outcomes: An Illustration Public Evaluations of Presidential war Management: Johnson, Nixon, and Vietnam, 1965-1972 Monthly American Battle Casualties Incurred in South Vietnam, 1965-1972 A Comparison of Predicted and Actual war Disapproval: Lyndon Johnson, 1965-1968 A Comparison of Predicted and Actual War Disapproval: Richard Nixon, 1969-1972 The Impact of Outcomes, Actions, and Events on War Disapproval 21 67 140 141 143 144 145 186 187 188 236 237 238 CHAPTER ONE THE VALUE OF PRESIDENTIAL SUPPORT The prince can win their favor in many ways, which vary according to circumstances, for which no certain rule can be given, and will therefore be passed over. I will only say, in conclusion, that it is necessary for a prince to possess the friendship of the people; otherwise he has no resource in time of adversity. Machiavelli, The Prince Chapter 9: Of the Civic Principality The Salience Of Public Support For the President The growth of public opinion measurement in the last forty years has added a new dimension to the study of the American presidency. Un- like their predecessors, occupants of the Oval Office since Franklin Roosevelt have been subject to practically continuous evaluation via the public opinion poll. If one examines the Gallup Opinion Index or similar compilations of poll results, it is seen that the presi- dential activities submitted for public evaluation have ranged from inquiries about specific policy decisions,1 to the chief evecutive's reactions in the face of crises,2 to his social or recreational habits.3 Despite such volume and variety, no question has proved more important or intriguing than the general performance item first introduced by the Gallup P011 in 1938. With a frequency that is unmatched in pol- itical surveys, the Gallup organization has posed the following ques- tion to cross sections of the American public: "Do you approve or disapprove of the job (incumbent) is doing as President?" Table 1.1 2 summarizes the regularity with which this item has appeared in the Gallup surveys. Since the wording of the question has remained constant over this period, the aggregate responses to this item may be combined to create a time series which spans forty-two years and includes 538 data points. Such information provides a novel oppor- tunity for examining the ability of men who occupy the "Modern Presidency" (Greenstein, 1978) to cultivate the friendship of the people. The performance question itself seems quite straightforward. It simply seeks to elicit from respondents a general evaluation of the president's conduct of office: approve, disapprove, no opinion. Most typically, the proportion of citizens who select the "approve" Option is taken as a measure of the public support, approval, or popularity enjoyed by the chief executive. Yet the simplicity of this measure is more apparent than real. What remains particularly ambiguous are the specific attitudes or sentiments tapped by this performance item. From the social scientist's perspective, this imprecision is quite troublesome. Little is known about the extent to which the performance item measures a general sense of trust in government, respect for the Presidency as an institution of democratic government (generally termed diffuse support), or satisfaction with the tenure of a particular incumbent (generally referred to as specific support).4 Furthermore, there is no information concerning the criteria employed to arrive at the assessments. As Neustadt (1976:163) warns, responses to this question may convey "anything or nothing" about citizen perceptions of presidential performance or the standards upon which these evaluations rest. It is for such reasons 3 that Roll and Cantrol (1972:129) offer the skeptical conclusion that O popularity ratings represent little.more than an artifact of the polling technique created in response to journalistic in- terests, and they certainly are not mean- ingful for guidance at the presidential level. At first glance then, presidential support ratings appear to be little more than imprecise summaries of satisfaction and grievances molded by some mixture of individual predispositions with vague perceptions about the president's conduct of office. Beyond this description, the underlying content or substance of presidential support is a matter left largely to speculation and debate. The speculation and debate have become more frequent in recent times. The growth of interest in the relationship between the president and the public is evidenced not only by increases in the number of popularity polls conducted by the Gallup organization (See Table 1.1), but also by the fact that each of the commercial networks (CBS/New York Times, NOB/Associated Press, ABC/Louis Harris) has in- corporated the presidential performance question as well as more specific variants into its public opinion surveys. The public standing of the president is now treated as a regular feature in both news- paper headlines and network news broadcasts. Regardless of the under- lying attitudes measured by the question, the public standing of the president has grown increasingly newsworthy. If one sifts through the relevant editorial statements, scholarly discussions, memoirs and histories, it is quickly distovered that despite the uncertainty about what is being measured, the public 4 standing of the president in an indicator or signal to which decision makers and opinion leaders, particularly those in the Washignton establishment,5 pay attention (works which contain this theme in- clude Evans and Novak, 1966,1971; Kearns, 1976; Safire, 1975; and White, 1973,1975). The measurements of presidential support are treated like readings from a political barometer and provide a basis or justification for a variety of judgments offered by journalists and politicians about the competence of a given administration. At times, the ambiguity surrounding the performance question and its responses even seems useful. The release of a Gallup or other pre- sidential performance rating is often accompanied by a host of com- mentaries, editorials, and by-line columns which attempt to decipher "the" meaning hidden within this latest reading of public sentiment. Thus, the ratings have been employed to assess the performance of the president relative to his predecessors (for a comparison of the Carter first year with those of presidents Truman through Ford, see the Gallup Opinion Index, Report Number 152,p.4), the public's re- action to a specific domestic and international events (for example, Polsby, 1976:66; Lee, 1977:254-255), the electoral prospects of the president amihis party (e.g., Gallup Opinion Index, Report Number 125, p.5), and both the image and credibility of the chief executive (Neustadt, 1976:154-175; Cornwell, 1965). The Value of Public Support: Some Impressionistic Evidence Given the skepticism of social scientists and the multiple inter- pretations of journalists, what accounts for the fact that citizen 5 evaluations of presidential performance are accorded such significance? Stated most directly, measurements of public support are quite useful to those who engage in and write about politics in the nation's capital. As an element of public Opinion, the level of support is, first of all, an ingredient of the Washington political context. Whether trans- lated in terms of the chief executive's image, credibility or com- petence, public support stands as an intrinsic component of what Barber (1977:8) terms the power situation confronting the president. A consideration of the post-war presidencies illustrates that power situations vary both across and within administrations. Consider the contrast between Harry Truman, his public support at 33%, confronted by the recalcitrant 80th Congress and Dwight Eisenhower parlaying widespread public support to establish a workable legislative relation- ship with the Democrat leaders of the House and Senate, Sam Rayburn and Lyndon Johnson (Evans and Novak, 1966:1968-194). That the power situation seldom remains constant within a single term is clearly documented by the examples of Lyndon Johnson and Richard Nixon. These instances suggest that the public standing of the president helps to define a prevailing atmosphere within which competition among president, congress and bureaucracy takes place. If we consider the character and style of this competition, the public standing of the president assumes an even more significant role. The primary characteristic of washington politics is that formal decision making authority is divided and distributed among multiple pockets of power and rival elites - the president and his White House organization, the congressional leadership, congressional committees and subcommittees, the various executive branch deparUments, 6 independent commissions and agencies - each pursuing their own objectives and each responsible to different constituencies. In this setting shared partisanship and formal lines of organizational authority offer little guarantee of cooperation in the formulation, implementation and evaluation of public policies. Policy initiatives in different substantive domins typically require different supporting coalitions and success on one policy front ensures neither future success nor its extension to other areas. Coalitions are thus fragile and transient. The ability to construct and maintain such alliances is regarded as an acid test of successful political leader- ship. Most importantly, it is commonly acknowledged that this ability is less dependent upon formal authority than upon informal levers of influence and each participant's assessment of the strengths and weaknesses of the competitors. According to this description, the president is envisioned not as an impartial arbiter presiding over this decentralized system of competition, but rather as an active participant pursuing his own set of goals and dependent upon his own powers and resources. If there is one lesson to be drawn from recent literature on the American presidency, it is that the power of the president is comprised of more than the textbook list of constitutional perogatives. For example, Neustadt's (1976) study of the modern presidency advanced the agrument that presidential power or influence is embodied primarily 'hathe chief exeuctive's ability to persuade. What must be emphazised is that presidential persuasion or, more appropriately, the suscep- tibility of others to such persuasion rests largely upon the perceptions of those who are the potential targets of the chief exeuctive's efforts. 7 Herein lies the import of the president's public standing and the rationale for the attention which its measurement receives. Aside from its role in defining the context of political competition, the popular support enjoyed by a president is one piece of information which operates to shape the perceptions of Washington decision makers. As Neustadt (1976:165) argues, Most members of the washington community de- pend upon outsiders to support them or their interests. The dependency may be as direct as votes, or it may be as indirect as passive toleration. Dependent men must take account of popular reaction to their actions. What their publics may think of them becomes a factor, therefore, in deciding how to deal with the desires of a president. His prestige enters into that decision; their publics are part of his. Their view from inside Washing- ton on how outsiders view him thus affects his influence with them. For example, the situation which confronted Lyndon Johnson as Senate Majority Leader after the Democrats had recaptured control of Congress in the 1954 midterm elections offers an illustration of the signifi- cance of mass support upon the assessments of those who are subject to presidential persuasion (Evans and Novak, 1966:193): ' I've got the committee chairmen and I could make his sweat,‘ Johnson would tell his closest friends in the Senate. He was thinking, among others, of begign Senator Carl Hayden of Arizona, Chairman of the Appropriations Committee, who was strictly nonpartisan in dealing with Eisen- hower's spending plans. Hayden would have twisted the President's budget out of shape at a word from Johnson ... The fact that Johnson did not give that word was sometimes self- torture for Johnson himself. Indeed, the fury of Johnson's private discussion about Eisenhower has only one course -- Johnson's deep frustration over his unused power. 8 ... It is doubtful that Johnson would have broken openly with Eisenhower. Johnson's in- stinct for domestic politics would not permit it. Unlike many Democratic leaders, he had measured the depth and permanence of Eisehower's — popularity beyond the Potomac and foresaw political disaster for any man who broke openly with the President. Two points are underscored by this illustration. First, despite the ambiguity and multiple interpretations, popularity ratings are treated by members of the Washington community as current and readily available measures of presidential power of influence. Second, it is seen that substantial public support for a president may condition or restrict the actions of congressional or bureaucratic decision makers. As Page and Brody (1975:136) have observed, the most recent opinion poll provides such actors with an indication of whether "it is wise or foolish to resist presidential persuasion." For this reason the public standing of the president is likely to be significant as a consideration or element which enters into the formulation of presidential, congressional and bureaucratic game plans. Although it would be unrealistic to expect a one-to-one correspondence between mass opinion and the outcomes generated by the competition between the president and other decision makers, public evaluations do shape the structure of choices confronting these in- dividuals. Variations in presidential support may not only expand or contract the set of feasible alternatives for a particular decision, but may also enhance or reduce the value of those actions which are feasible. From the President's perspective (Neustadt, 1976:158): 9 (T)he weaker his apparent popular support, the more his cause in Congress may depend on negatives at his disposal like the veto or "impounding". He may not be left help- less, but his options are reduced, his opportunities diminished, his freedom for maneuver checked in the degree that washing- ton conceives him unimpressive to the public. Because it is an indicator to which decision makers react and there- fore because it shapes the context of choice, public evaluations be- come a matter of concern, whether acknowledged or not, for a sitting president. Indeed, the proposition that presidential influence varies with public support appears to underlie the strategies devised by administration tacticians. At the outset of the Carter Administration, for instance, one news organization characterized such calculations by reporting that (Newsweek, May 2, 1977, p.36): There is an adversary undertone to the preparation -- that us-against-them mode in which Carter's Georgians tend to view Capitol Hill and other potential seats of organized resistance. 'The only way to keep these guys honest,‘ says one Carter operative, 'is to keep our pop- ularity high -- and convert it into the sort of mass support a senator or con- gressman can ignore only at his peril.‘ Such descriptions suggest that presidents themselves look upon public support as an implicit source of leverage in their relationships with other decision makers. As such, public support appears to be useful not only to Washington decision makers but to the president himself. What the multiple uses of presidential support underscore is not simply that many contradictory judgments are drawn from and read into 10 such poll results but, more importantly, that the current state of the relationship between president and public is regarded by politicians and journalists as a significant and useful piece of information. Indeed, because of its salience and, as the above examples suggest, its import as a shaper of perceptions, presidential support can be expected to exercise an impact on the behavior of decision makers. If decision makers are influenced by the public standing of the president, then it can be argued that public support operates as a presidential re- source; that is to say, public support will provide the president with a form of political currenty which can be used as an instrument in the pursuit of his goals and objectives. Public Support as a Resource: Some Implications This theme of public support as a presidential resource serves as a point of departure in this thesis for two reasons. First of all, if public support is found in influence political outcomes deemed important to the president, it becomes a valuable commodity, a re- source which enhances the president's power to persuade. Further, this value creates an incentive for the president to manage, manipulate, or otherwise influence the level or pattern of citizen evaluations. Perhaps the most clearly articulated recognition. of such an incen- tive was provided in the observations of Theodore Roosevelt (Burns, 1972:52): 11 I have felt a slightly contemptuous amusement over the discussion that has been going on for several days about my popularity or waning popularity or absence of popularity. I am not a college freshman ... and therefore I am not concerned about my popularity save in exactly so far as it is an instrument which will help me achieve my purposes ... The notions of incentive and value embedded within this argument are critical. If public support proves to be unimportant in accounting for political outcomes, then as Roll and Cantril suggest, presidents could be advised to ignore the polls and dismiss the analyses Of poll watchers and opinion experts as irrelevant to the fortunes of an administration. However, if politically relevant outcomes are in- fluenced by how the public evaluates the president, then it can be expected that presidents will formulate strategies and take actions designed to protect or enhance their public standing. Support will Operate, in this circumstance, as a determinant of presidential decision making. As a result, the study of public support is intrin- sically connected to the study of presidential behavior. The enter- prise Of explaining public support must not only acknowledge that presidents are motivated to influence public support but also specify how presidential actions impinge upon public assessments of presidential performance. This is the second reason for focusing upon the theme of public support as a presidential resource. It is my contention that the existence Of a presidential incentive to manage fundamentally alters the nature of the enterprise Of theorizing about and analyzing public support for American presidents. Because it represents a valuable commodity, public support can no longer be treated, in vacuo, as simply 12 an interesting component Of public Opinion. Indeed, a recognition that presidents have an incentive to manage public support redefines both the "normal science" and research agenda associated with the study of how citizens evaluate the president. This normal science is illustrated by a rather extensive body of literature which is primarily concerned with identifying aggregate-level determinants Of support and specifying econometric models tO account for the movement of support over time. It has been hypothesized, for example, that the state Of the economy has an effect on presidential support. This body Of research has attempted to explain variations in presidential support in terms of aggregate economic variables such an unemployment and inflation (Mueller, 1970; Hibbs, 1974,1980; Kenski, 1977a,l977b; MOnroe, 1977; Schneider, 1977; Kernell, 1978; Frey and Schneider, 1978; Golden and Poterba, 1980). There is another body Of literature which seeks to explain support in terms Of more episodic influences such as wars, domestic and international crises, scandal (Mueller, 1973; Stimson and LeGette, 1975; Stimson, 1976; Kernell, 1978), and the monthly balance between good and bad news (Brody and Page, 1975). Finally, it has been argued that there are factors which are responsible for certain regularities within each presidential term. These explanations have formulated hypotheses such as the coalition-of-minorities (Muller, 1970), expectation-disillusionment-forgiveness (Stimson, 1976), and incrementalism (Kernell, 1978) to explain the basic shape of the presidential support time path.during each term. Now it must be emphasized that, whether acknowledged.or not, these dynamic models of presidential support invariably touch upon questions 13 about the capacity of presidents to influence the path Of support. For example, the model formulated by Stimson (1976) implies that because of public inattention and exaggerated expectations, support for presidents will invariably decline regardless Of the actions taken by a particular incumbent. In this instance the president emerges as simply a passive Observer. The more general point however is that assuming public support Operates as a resource, any study Of its determinants is simultaneously a study Of the methods Of control and therefore a study Of presidential decision making. In other words, if it is the case that presidents -- motivated by the payoffs which accompany widespread public support -- do attempt to influence citizen assessments, then public support must be regarded as a joint function of citizen and presidential decision making. This, in turn, implies that it is the interaction of choices made by citizens and presidents which acts as the driving force or dynamic underlying the time pattern Of this commodity. It is easily seen then that the Operation of support as a resource stretches the prevailing research agenda beyond its typical concern with the specification of econometric models to include questions pertaining to whether and how presidents attempt to manage public support. The Management of Public Support: An Outline The purpose of this thesis is to conduct such an inquiry. That is to say, the research and discussion to be presented is designed to investigate the management Of public support -- the question of whether and how presidents, via their actions and decisions, can influence the manner by which their performance will be evaluated. As suggested in 14 the previous section, it can be expected that public support and its pattern over time are a joint function Of citizen and presidential behavior. Indeed it is the interaction of choices made by citizens and the president which defines the fundamental character Of the manage- ment question. Yet the normal science associates with the study of public support has, as will be shown, relied upon vaguely articulated descriptions of citizen behavior and, more importantly, excluded the behavior Of the president from its explanations. It is my contention that a richer and more thorough understanding Of public support demands that both of these shortcomings be rectified. More specifically, any systematic treatment Of public support and its potential manipulability by the president requires that attention be directed to the following agenda of concerns: (1) Rationale for Management Does there exist an incentive for presi- dents to pursue a support management strategy? (2) Citizen Behavior What performance expectations do citizens impose upon the president? How do theSe‘ expectations influence the decision to approve or disapprove of presidential performance? (3) Presidential Behavior What are the goals Of the president and his administration? What perogatives or choices are available to the president? How do the goals Of the president influence the exercise Of his perogatives? (4) The Support Management Strategy Given the character of citizen and presi- dential behavior, to what degree can the criteria governing citizen choice be in- fluenced by the actions Of the president? What are the levers by which support can be managed? 15 These sets Of questions constitute the primary concerns of this thesis. Essentially, my Objective is to develop: an integrated model Of citizen-president decision making, to show how this nodel leads to particular hypotheses pertaining to presidential management Of public support, and to test several of these hypotheses. This research strategy is designed to demonstrate that (1) there is an incentive for presidents to manage public support, (2) citizen evaluations Of presidential performance can be influenced by the chief executive, and (3) the behavior Of the president must be incoporated into empirical analyses Of citizen performance evaluations. To elaborate, consider Figure 1.1 which presents a diagrammatic representation of the model of citizen and president interaction which will be developed in the subsequent chapters of this thesis. With respect to the first issue on the research agenda, it has been suggested that public support constitutes a valuable presidential resource and, for this reason, provides an incentive for presidents to attempt to control these assessments. The assertion demands a test. In Chapter Two, empirical evidence is presented to demonstrate that citizen assessments Of presidential performance do influence a variety of political outcomes of import to the president. Particularly important in this regard are the results which show that public support exercises an appreciable impact on the fortunes of the president's legislative program in Congress. Such evidence thus enhances the plausibility Of assuming that presidents have an incentive to control this commodity. This argument is depicted in Figure 1.1 by the lines connecting public support to resources, resources to presidential decision making, and resources to Congress. As such, my emphasis upon the value Of public 16 support in these initial chapters serves as both a prelude and justifica- tion for examining the relationship between presidential behavior and citizen decision making. Chapter Three is designed to establish a foundation for examining this relationship by addressing the second and third sets of questions on the research agenda. The purpose of the chpater is to formulate two models -- the citizen calculus of presidential evaluation and a calculus of presidential action. The model is designed to specify the criteria and decision rules employed by citizens to evaluate the president as well as the manner in which public support impinges upon presidential behavior. With respect to citizen decision making, it will be argued that support decisions are grounded upon a number of specific performance dimensions (for example, handling or managing the economy). This is illustrated in Figure 1.1 by the line connecting handling to public support. It will then be shown how these specific performance evaluations are a function of a set of observable outcomes (e.g. unemployment, inflation, real income) and presidential actions. The line connecting public expectations to handling in Figure 1.1 denotes that both outcomes and presidential actions are evaluated in light of citizen preferences' and expectations about what constitutes acceptable outcomes and appropriate presidential behavior. The second portion of Chapter Three Offers a calculus Of presidential decision making. First Of all, the set of presidential actions is de- fined and shown to consist Of two general types: those whose execution depends on the consent of other government institutions and decision makers and those which represent a unilateral exercise Of presidential authority. It is then argued that the choice of presidential actions 17 has the capacity to influence outcomes (directly in the case of uni- laterial actions and via Congress and the legislative process in the case of joint actions) as well as public assessments on the relevant performance dimensions. It is in this fashion that the choices of presidents are linked to the criteria according to which they are evaluated. Finally, a spatial representation is employed to illustrate how the president's choice of actions is governed by: (l) the policy and ideological goals of the presidents, (2) public expectations and the incentive to manage presidential support, and (3) the necessity of cultivating congressional support. The intent Of Chapter Four is twofold. First of all, it develops a line of argument which culminates in the assertion that public support is, in principle, a manageable commodity and demonstrates that, given the characterization of citizen and presidential decision making form- ulated in Chapter Three, the primary levers Of presidential control are unilateral actions, joint actions, and attempts to alter the salience attached to the specific performance dimensions by the public. In other words, it is asserted that a presidential management strategy is comprised of actions -- their content, timing and attendant publicity -- designed to influence outcomes, specific performance assessments, and public expectations. The second objective of the chapter is to demonstrate how several testable hypotheses emerge as implications of the discussion presented in Chapters Three and the first part_of: Chapter Four. The purpose Of Chapter Five is to evaluate several of the hypotheses pertaining to the impact of presidential behavior on the performance eValuations of the public. To accomplish this objective, an aggregate- level model of public dissatisfaction with the war management of 18 Presidents Johnson and Nixon is estimated. The performance of this model is compared to several alternative specifications based upon the characterization of the president in prior research as a passive bystander. The conclusion of the thesis summarizes the discussion and results presented, compares this work to existing conclusions about public support and its manageability, and argues that the dynamics of public support are governed not by mechanistic influences but by the inter- play between presidential decisions and outcomes of import to the public. 19 Notes tO'Chapter 1 For example, from the May 1962 Gallup Poll: "What is your Opinion about the action which President Kennedy took to get the steel companies to change their plans about raising steel prices?" (Gallup, 1972, p.1768). For example, from the June 1970 Gallup Poll: "DO you approve or disapprove of the way President Nixon is handling the Cambodian situation?" (The Gallup Opinion Index, June 1970, Number 60, p. 4). For example, from the June 1953 Gallup Poll: "In your opinion do you think President Eisenhower is taking too much time off from his job to play golf or not?" (Gallup, 1972, p.1151). Dennis (1975) presents a discussion of these two concepts of support. This distinction and the measurement of diffuse and specific political support is also emphasized in the Miller (1974a,1974b) and Citrin (1974) debate. This Washington establishment is described by Neustadt (1976:126) as follows: These doers comprise what in spirit, not geo- graphy, might well be termed the 'Washington community.’ This community cuts across the President's constituencies. Members of Congress and of his Administration, governors of states, military commanders in the field, leading politicians in both parties, representatives of private organizations, newsman of assorted types and sizes, foreign diplomats (and principles abroad) -- all these are 'Washingtonians' no matter what their physical location. Citizen Evaluations of Presidential Performance 20 Table 1.1 The Frequency of Measurement By the Gallup Opinion Poll President F.D. Roosevelt H.S. Truman D.D. Eisenhower J.F. Kennedy L.B. Johnson R.M. Nixon G.R. Ford J.E. Carter TOTALS Years 1938-1940 1941-1944 1945-1948 1949-1952 1953-1956 1957-1960 1961-1963 1963-1964 1965-1968 1969-1972 1973-1974 1974-1976 1977-1980 1938-1980 1938-1980 Number of Gallup Polls Including the Performance Item 19 24 15 26 51 57 40 16 67 59 34 34 96 538 Yearly Average 12.75 14.25 13.71 14.77 16.75 14.75 20.40 14.07 24.00 12.8 21 mucuomm nzocowoxm moaoouao use>oaox cowuuaoawoa.lllllw. zflacaucooamoum L uuoeesm aa:Oammouwcou m:me«ooo Huaucoofimmum e e mOOHOOmox .t Hmwucovfimoum Ali cofiuo< new accow< i..- .....-iIV Hmmusooamoum mHmoo Homucorfimoum asaxrz cofimwomc ucmvwmmumlconduwu mo amocz < ~.— musxdu 1//////JP er mcowuu< mcoauo< maven Hummumawcs uuoeeam owfinnm mucoeumouu< moonshOuuom owwuoonm :w=«av:w:: - accuumuoonxm uufinsm mocoemo Hmcoauunufim mucosa: amuouooam wormeoa MeccausuaumcH q mconfiuwu CHAPTER TWO PUBLIC SUPPORT AS A PRESIDENTIAL RESOURCE: AN EMPIRICAL ASSESSMENT The emphasis upon the theme Of public support as a presidential resource in Chapter One was based primarily upon the Observations and anecdotes of Washington insiders privy to the day-to-day conduct of politics in the nation's capital. The argument which can be drawn out Of such journalistic and historical accounts is straightforward -- the public standing of the president is an integral component of presidential influence. As such, it can be expected that the actions of those who are the targets Of presidential persuasion will vary according to the level of public support. If such is the case, evidence that public support pays off should not be limited to the insights proferred. by washington insiders. MOreover, standard textbook descriptions of the presidency emphasize that because of the visibility and stature of the office, the president often serves as an important point of reference and stimulus in the formation of citizen attitudes about national politics. Such observations suggest that the president and particularly popular sentiment regarding how well the president has performed in office will operate as both a source of information and a decision cue for citizens. If this expectation is found to hold, then the usefulness of public support as a commodity will not be restricted to the Washington community. In sum, it is my contention that public support has a demonstrable value as an influence on a variety of political outcomes and that such influence can be detected in the choices made by citizens and 23 Washington decision makers. More specifically, it is hypothesized that public evaluations of presidential performance will operate as a determinant of (1) electoral outcomes, (2) congressional roll call outcomes, and (3) citizen evaluations of government institutions. My Objective in the following sections is to summarize the existing research concerning these propositions and to present results which Offer additional evidence about the import of public support as a presidential resource. Presidential Support and Electoral Outcomes Evidence garnered from a number of individual and aggregate level analyses can be marshalled to support the assertion that the assessment of a president's performance is an important consideration in the voting decisions of the citizenry (see, for example, Boyd, 1971; Tufte, 1975; Piereson, 1975; Kernell, 1977; Miller and Miller, 1977; Fiorina, 1979). Let us first consider the character of the relationship between per- formance evaluations and voting in the context Of midtern congressional campaigns. The traditional Surge and Decline explanation (Campbell, 1966) of citizen behavior in such elections holds that because the short term partisan forces generated by presidential contests are largely absent at midterm, not only is the impetus to vote curtailed but, for those who do turnout, the choice of a congressional candidate becomes primarily a function of the individual's partisan affiliation. Kernell's analysis of survey data gathered by the American Institute of Public Opinion amends that standard interpretation Of midterms by uncovering the influence of presidential evaluations on both the 24 decision to vote and candidate choice. Tables 2.1 and 2.2 present a reformulated summary of the results presented by Kernell together with analogous information from the 1974 election survey conducted by the Center for Political Studies.1 The tables depict the relationship between evaluations of the president's performance and midterm voting, both turnout and choice, among members of the president's party, the opposition party and Independents. When this control for partisanship is taken into account, it can be seen that the value of favorable per- formance evaluations comes from two quarters. The first source Of value concerns the partisan division of the vote (Table 2.1). Stated most succinctly, citizen evaluations of the president's performance alter the strength of the relationship between partisan affiliation and the choice of House candidates in midterm elections. For those who identify with the party of the president, approval of the chief executive's conduct translates to a greater proportion of votes for the congressional candidates of his party so that, in this instance, favorable evaluations stimulate party voting. Among those who identify with the opposition, presidential approval depresses party voting. Satisfaction with the current administration's performance reduces the tendency for out-party members to cast ballots in favor of their party's congressional hopefuls. Although midtern elections do evidence a substantial degree of party voting, the results in Table 2.1 intimate that the Surge and Decline explanation must be modified to recognize that evaluations of the chief executive impinge upon or condition the expected association between party identification and voting behavior. Satisfaction or dissatisfaction with the performance 25 of an administration does seem to Operate, in tandem with party identification, as a criterion employed in congressional vote decisions.2 Coupled with this capacity to condition candidate choice is a second source of value for presidential support in midterm elections (Table 2.2). Although there is no consistent pattern to the relation- ship between approval and turnout among members Of the president's party, this is not generally the case for out-party identifiers and independents. In fact, the data of Table 2.2 reveal quite clearly that dissatisfaction with a president's conduct of Office Operates as a stimulus to vote for both of these groups and, as noted above, such dissatisfaction typically results in a larger proportion of the vote for congressional aspirants of: the opposition party. Where the Surge and Decline thesis characterized midtern elections by their absence of short term forces and their electorates as largely a group of habitual or core voters, Tables 2.1 and 2.2 lead to an altered characterization. In addition to its impact on candidate choice, evaluations of a president possess the capacity to shrink or swell the size of the mid- term electorate and must, for these reasons, be regarded as an important short term force during these electoral campaigns. To summarize, favorable evaluation of the president act as a resource in the sense Of reducing (1) turnout among opposition identifiers, (2) defections within the president's party and (3) the vote margins amassed by opposition candidates among out-party identifiers and independents who disapprove of the president's conduct. As such, a higher level of public support permits the party of the president to minimize the loss of House seats typically suffered in off-year elections. 26 Do such results hold for presidential elections as well? The election year surveys conducted by the Center for Political Studies provide us with the opportunity to assess the import of performance evaluations during the presidential election campaigns Of 1968, 1972 and 1976. Tables 2.3 to 2.5 display the relationship between citizen evaluations of presidential performance and voting behavior in these national elections.3 Three conclusions may be drawn from these results. First, turnout is largely unaffected by presidential evaluations. In only one instance, for the Democrats in 1976, do negative assessments of a president generate a significantly higher rate Of participation than positive exaluations; further, the absence of a negative impact is matched by the lack of any positive association between approval and turnout (the exception being the Democrats: in 1968). The decision to cast a ballot in presidential elections, unlike the midterm contests, appears to be largely insensitive to presidential performance considerations. However, the influence of presidential assessments upon candidate perference is unmistakable. Across each of the partisan groups, differences in the evaluation of the incumbent's performance produce sizable differences in vote proportions. This in turn suggests that in addition to the comparative evaluation Of prospective policies Offered by presidential aspirants, citizen vote decisions are dependent upon retrospective criteria such as the performance item analyzed here. Of particular interest in this regard is the 1968 contest where the electoral strength of the Democrate nominee, Hubert Humphrey, was strongly affected by voter sentiments toward the Johnson Presidency. In a fashion similar to congressional candidates, the electoral fortunes 27 of a president's heir apparent revolve about the resevoir of good and ill feeling accumulated during the chief executive's stay in office. My second conclusion then is that, as in the case of midterm competition, positive assessments of presidential performance pay off at the polls. Third, the data also reveal that the evaluation of administration performance has a substantial spillover effect in voting for other offices. Among Democrats in 1968, for example, the belief that the Johsnon Administration had performed poorly produced significantly lower vote proportions in House, Sentorial, Gubernatorial as well as other state and local contests. Moreover, in all three election years, preferences for House candidates among all partisan groups are related to performance assessments. Perhaps most intriguing is the discovery that the proportion of individuals casting straight ticket votes in state arr!" local races is determined in part by assessments Of the president. Based upon this evidence, presidential support appears as a resource whose usefulness operates in a variety of electoral arenas. Presidential Support and Copgressional Outcomes The electoral data examined here emphasize that the value of presidential support flows from the fact that presidential performance is a criterion employed by citizens in formulating their voting decisions. Because the conduct of the president may ultimately influence the partisan balance in Congress, there are three noteworthy implications produced by the relationship between presidential evaluations and Congressional voting. First, the relationship imposes at least a 28 weak form of partisan accountability on the conduct Of national politics. The accountability is indirect since incumbents of the president's party "are not responsible for what the president does or how their party is perceived by the electorate, but ... pay the price for such assessments" nevertheless (Arsenau and Wolfinger, 1973:17). As such, the relationship joins the fortunes Of the presidential and congressional wings of the party. Second, since the partisan composition of the House and Senate is one consideration used to assess the likelihood of success for legislative proposals, the relationship may operate to place an a prior limitation on what the president chooses to submit to the Congress. Of particular import is the relationship between switched seat con- gressional districts and the ability Of the president to forge and maintain the coalitions necessary for bringing about landmark changes in public policy. A number Of studies have demonstrated that instances of significant policy change in Congress, such as the Great Society legislation enacted in 1965 and 1966, was due largely to the influx of members who defeated incumbents of the Opposition party in the pre- ceeding presidential election (Brady and Lynn, 1973; waldman, 1967; Shannon, 1968; and Martin, 1976 all touch upon this theme). It is the additiOn of such members which Often provides the numbers required to convert a minority into a majority policy coalition. To illustrate, consider the cOmparison between the 37 marginal Democrats who won their House seats in 1964 and the Republican incumbents they replaced. In his analysis of congressional elections, Mbrris Fiorina (1977:13) reports that the switched seat Democrats were decidedly more liberal and more supportive Of an activist federal role in combating social 29 problems than their defeated Republican opponents. Undoubtedly, the entry of these liberal Democrates played no small part in supplying Lyndon Johnson with the numbers required to make the Great Society a legislative success. What underscores the import of the relationship between presidential performance and congressional voting is that it is the first-time incumbents and especially the marginal incumbents of the president's party who are most vulnerable to adverse assessemnts of presidential performance. To continue with the example of the 89th Congress, 29 of the Democrats elected in 1964 were defeated in the mid- term elections of 1966 and, as expected, their Republican replacements were appreciably unsympathetic to any further extension of the Great Society (Fiorina, 1977:13). Thus, a low level of popular support may not only doom any bold attempts at policy change from the outset but also may lead to the dismantling of congressional policy coalitions through electoral defeat. The third implication of the citizenry's reliance upon presidential performance as a decision cue concerns the reelection strategies of congressional incumbents. The most significant feature Of midterm campaigns, for example, is that the president himself is typically a political issue (Tufte, 1976:813-814; Kernell, 1977:46-50). Since the policies and actions of the administration are likely to be prominent topics of debate among the participants in the campaign, candidates will ultimately be confronted with the choice Of whether to support or disassociate themselves from the president. For incumbents Of the president's party, association with the initiatives of the chief executive can represent either an impediment or boost to their pro- spects for reelection. Similarly, aspirants from the opposition party 30 must weigh? the potential costs and rewards of attacking presidential policies. It is likely that the association between how the public evaluates the president and how the public votes does not escape the notice of incumbents during the course of such deliberations. Because public assessments of the president's conduct may operate as an asset or liability to congressional incumbents, it can be execpted that the willingness of these Office holders to support legislation sponsored or endorsed by the administration is sensitive to the level of public support. If such is the case, the public standing Of the president will ultimately influence administration success in Congress and thus register an additional impact on the policies produced by the legisla- tive body. The remainder of this section constitutes a preliminary attempt to ascertain whether presidential fortunes in Congress are, at least in part, a function of the president's standing among the public. The proposition under consideration actually involves two ques- tions. DO legislators respond to fluctuations in the level of presi- dential support? If so, is this response strong enough to alter the outcomes of roll call votes in the Congress? Accordingly, the following analysis will focus upon the relationship between public support for the president and (l) the level of roll call support among members of Congress and (2) presidential roll call victories in the chambers of Congress. Let us first consider the extent to which public support for the president influences his support among members of Congress. To examine this question, a yearly data set coverning the period from 1953 through 1978 has been assembled (N-27).4 As a convenient summary Of this information, Table 2.6 presents the yearly level of public 31 support for the president together with several measures of each administration's legislative activity and how administration initiatives fared in the chambers of Congress. Let us consider this information more closely. Presidential support, both for the population as a whole and the Democrat, Republican and Independent subsets of the population,5 is simply the proportion obtained by averaging all Gallup polls administered in a given year. Table 2.6 also displays the average presidential support scores of House members and Senators. These scores were Obtained in the following manner. Each year the Congressional Quarterly organization analyzes all presidential messages and statements relevant to pending legislation in Congress and publishes a list of roll calls on which the chief executive took a clearly identifiable position. Hereafer, these roll calls will be referred to as Presidential Support Votes. According to the scheme employed by Congressional Quarterly, the presidential support score of a particular member of the Senate or House is defined as the proportion of all presidential support votes on which the legislator voted in favor of the position adopted by the president. Similarly, a presidential Opposition score represents the proportion of support votes which finds a legislator voting against the president's accounced position. To correct for the effects of abstention con- tained in the individual and party scores reported by Congressional Quarterly, the following quantity Corrected Support - Z Support 2 Support + 2 Opposition 32 produces a measure of presidential support for a given legislator on those roll calls in which the individual participated. Table 2.6 dis- plays the average revised level of support given the president by Democrat and Republican members of the House and Senate. The previous discussion implies that this measure Of average legislative roll call support should be sensitive to the level of public support enjoyed by the president. Before turning to the analysis of this data, it is necessary to introduce an additional refinement. Essentially the expectation of a relationship between public and legislative support is grounded upon the assumption that members of Congress utilize this element of public opinion as a voting cue. The importance of this cue arises from the connection between citizen evaluations of presidential performance and their congressional voting choices. However, the level of public support for the president among the entire population may provide an inaccurate measure of this cue. It is likely that congressional incumbents are more concerned with opinion among those who comprise what Fenno (1977:844-889) has termed the primary and reelection constituencies than among those within the entire geographic constituency. As Edwards (1976:107) has argued, ... congressmen do not necessarily define con- stituencies as all the voters who live within the legal boundaries of their districts. In effect, congressmen choose which groups within their constituencies they will represent. These groups will generally be part of their success- ful electoral coalitions. ... Thus, it may be that rather than responding to overall presi- dential prestige, congressmen respond to presi- dential prestige among the groups which are most likely to support them. 33 It thus seems reasonable to expect that members of Congress are most attuned to the assessments of their fellow partisans and that the use Of public support among the partisan subsets of the citizenry should provide a more satisfactory measure of this one than would the general measure of public support. The assumption that the partisan breakdown of presidential assessments displayed in Table 2.6 constitutes an acceptable indicator of this Opinion leads to the following specification for Democrats and Republicans in the House and Senate: CSt = Bo+ B St + e 1 t (2.0) where t ‘ 1,2, 00. g 26 CSt a Congressional roll call support; i.e. the average presidential support score among Democrats (Republicans) in the House (Senate) 8 = Public Support; i.e. the average annual level Of presidential support among citizens who identify themselves as Democrats (Republicans) The estimates for Equation 2.0 are presented in Table 2.7. Note, first of all, that the residuals for the Republican equations indicate the presence of serial correlation. As Ostrom (1978: especially 16-17 and 25-31) has shown, the presence of serial correlation introduces a number of inferential problems. Most relevant to this discussion is that while the coefficients for public support remain unbiased, the 2 estimates of the regression variance -c: - and thus the variance of the parameter estimates -- Var (B) - ‘will be biased. If the ordinary 34 least square residuals are positively correlated, for example, the estimated variance of the regression coefficients is likely to be significantly smaller than the actual or true variance. In this circumstance reliance on the conventional t-test may produce the erroneous inference that public support exerts a significant influence on roll call support. To overcome such difficulties, the Republican equations were restimated by employing the Cochrane-Orcutt (pseudo- Generalized Least Squares) procedure.6 These results are displayed in Table 2.8. The estimates reported in Table 2.7 (Democrats) and Table 2.8 (Republicans) indicate that, in the aggregate, the public standing of the president among partisan groups does serve as a relevant cue when members of Congress are faced with roll call decisions involving policies endorsed by the president. In all four instances the estimated coefficients are statistically significant and relatively sizable. According to these results, a four to six point drop in average roll call support can be anticipated for every ten point decline in support for the president among partisan identifiers. Although a more detailed analysis Of this relationship is strongly recommended, these first cut results do Provide reasonable evidence that the level of public support for the president is employed by members of Congress as a decision cue and, for this reason, must be regarded as an important presidential resource. The results presented in Tables 2.7 and 2.8 suggest that legisla- tors do respond to changes in the president's public standing. The question which.must now be addressed is whether public support influences 35 the outcomes of congressional roll call voting. The proposition under consideration is that public support exerts an impact on presidential success in Congress. To define presidential success, let PS - the number of Presidential Support Votes W = Presidential Victories; i.e. the number of support votes in which the position assumed by the president was victorious LS - the presidential legislative success rate so that LS 8 W/PS represents the proporiton of presidential roll call victories. Re- cognizing that the procedure of yearly averaging is likely to Obscure significant fluctuations in both public support and presidential success, a quarterly data set has been constructed (NI-105).7 Public support, again denoted by St’ is defined as the proportion which is Obtained by averaging all presidential support polls administered by the Gallup organization in a given three month period. In working with such quarterly measures of success, we periodically encounter the problem of an abnormally small number of roll call votes in a given quarter. The problem caused by such occurrences if that the resulting time series may contain several unrepresentative peaks and troughs. To compensate for the possible distortions which may occur in employing 36 such a series, the following measure of legislative success is defined for use in this analysis: L8 ..2 vvl'z PS ttttt where t 8 an index representing the quarters of a presidential term; i.e., for a full four year term, t - 1,2, ... , 16 L8 8 a cumulative presidential success rate Thus, LSt represents the proportion of presidential roll call victories through quarter t. The set of presidential support votes was then divided into domestic and foreign subsets and cumulative success rates were calculated for House Domestic policy, House Foreign policy, Senate Domestic policy and Senate Foreign policy.8 It is expected that the success rates will vary over time with the level of public support for the president. To evaluate this expectations, the following specification was estimated for each of the cumulative success measures: LSt a B0 + Blst + at (2.1) As in the case of the roll call support specification, confidence in the results generated by ordinary least square estimation is limited by the presence of serial correlation. For this reason Table 2.9 displays the estimates which were obtained by applying the Cochrane-QrcOtt, pseudo-GL8 procedure to Equation 2.1. Examination of these results 37 reveals the public support does exert an impact upon presidential success but that this impact is conditional upon which chamber of Congress is involved as well as the substantive character of the roll call issues. In the House Of Representatives, public support for the president is significantly related to foreign policy success but exerts little influence on roll calls involving issues of domestic policy. Interestingly enough, presidential success in the Senate is dependent upon public support on domestic rather than foreign policy. Such findings are rather intriguing particuarly when the conventional wisdom regarding the status of relations between president and Congress is considered. In the "Two Presidencies", for example, Wildavsky (1975) asserts that the ability of presidents to influence and control the course of policy making during the post-war period has been much more substantial in the foreign as opposed to the domestic realm. This argument is reinforced by studies of policy making in House committees which emphasis the deference accorded to executive branch foreign policy expertise by such committees and the willingness of committee members to act as advocates of the president's foreign policy program (See, for example, Fenno, 1973:27-30 and 69-73; Dexter, 1971). More generally, post-war relations between president and Congress Unmatters of foreign policy were grounded upon a bipartisan consensus with the rejection of isolationism, the resistance of Soviet expansionism and the development of sophisticated weapons systems serving as the linchpins of the consensus. Because of such underlying agreement, members of Congress were generally willing to follow the lead Of the president and approve programs and policies formulated by executive branch decision makers (this theme is 38 further developed in DeClerico, 1979: 37-44). Under such circumstances one would expect public support to exercise little or no influence upon foreign policy success rates in Congress. These conclusions are generally reflected in the results Obtained for the Senate; the co- efficient for public support in the case of foreign policy is not significant and the intercept term for foreign policy is substantially larger than the domestic policy intercept. The discovery that public support operates as an influence upon foreign policy success in the House of Representatives runs counter to these conclusions however. While highly speculative and certainly dependent upon a more sophisticated consideration of the data, there does exist a plausible line of reasoning which may help to account for this result. The data presented by Wildavsky (1975: 449) shows that congressional action on presidential proposals was decidedly more favorable in matters Of foreign policy during the 1948-1964 period. The data utilized in this chapter corroborates this observation. The information contained in Table 2.10 indicates that, as measured by the average yearly pro- portion of roll call victories between 1953 and 1978, the president was indeed more successful on foreign policy issues. However, the additional information presented in Table Zgulsuggests that the foreign policy advantage enjoyed by the president in the House has diminished over time. Between 1953 and 1966, the foreign policy success rate of the president exceeded the domestic success rate by 15 percentage points; for the 1967-1978 period, this difference fell to approximately 6 points. Although both domestic and foreign policy success rates in the House were lower in the 1967-1978 period, the decline in average foreign policy success (-14.63) far exceeded the drop registered in domestic policy. 39 Moreover, the correlation coefficients reported in the table provide some indication that success on foreign policy roll call votes in the House has become more sensitive to the public standing of the president. This apparent shift in the behavior of the House has been noted by several recent studies (Ornstein and Rohde, 1977; Franck & Weisband, 1979). Ornstein and Rohde (1977: especially 252-261), for example, emphasize that because of the Vietnam experience, membership turnover and procedural reform, the House and especially its Committee on International Relations has become more assertive in matters of foreign policy and more willing to challenge the proposed policies of the executive branch. It thus seems that foreign policy in the House has become more politicized. It may be conjectured that one manifesta- tion Of this renewed assertiveness is that the willingness of House members to accede to presidential wishes in foreign policy may have grown more dependent upon the public standing of the president. My next concern pertains to the possibility that the relation- ships between presidential support and legislative success exhibited in Table 2.9 may be spurious results arising from the failure to incorporate other relevant influences into Equation 2.1. There are two particularly prominent factors which could Operate to influence presidential success in Congress. First, the level of presidential success may be sensitive to the.number of presidential initiatives which are placed on the congressional agenda. That is to say, it may be the case that success varies in some inverse fashion with the size of the president's legislative program. What would lead us to expect such a relationship? One factor pertains to the time and personnel costs involved in the administration's lobbying effort. As the number of bills introduced or endorsed by the president increases, the ability 40 of administration liaison personnel to devote adequate time to consulta- tion and planning for each bill is likely to decline. If such is the case, many bills could come to the floor of the House or Senate lacking the requisite groundwork for constructing a winning coalition. An associated reason for expecting such a relationship concerns the character of the legislative process. It is only under the most un- usual Of circumstances that the congressional response to a presidential initiative is relatively instantaneous; instead, congressional treatment of presidential initiatives and other legislative proposals follows a particular rhythm -- a leisurely pace which grows more hectic as adjourn- ments and the end of sessions approach. This normal rhythm often results in a tendency for presidential initiatives to pile up at particular points in a congressional session. It is plausible to expect that the accumulation of presidential initiatives and the quickening tempo of the legislative process provide opponents of the president with greater Opportunities to exact concessions, force tradeoffs and block legislation. To the extent that its attention and resources are dispersed over a variety of policy fronts, the administration is likely to grow more vulnerable to congressional factions which concentrate their efforts on defeating specific bills endorsed by the president. Moreover, exhorting the public to exert pressure on Congress or launching an intense personal lobbying effort are limited presidential strategies in the sense that increasing reliance on such tactics is likely to dilute their dramatic impact and thus reduce their effectiveness. These are certainly the lessons which emerge from DeClerico's (1979:103-4) descrip- tion of the Carter Administration in 1977: 41 ... the President's legislative program suffered from an excess of proposals. In a relatively short span of time, Carter sent to the Hill legislation dealing with such matters as energy, election and welform reform, government reorganization, the Panama Canal Treaties, Social Security, and agriculture. In so doing he failed to provide Congress with any indication Of what should take priority. As one congressman noted, 'With.almost no exceptions every issue that has come down from the White House or agency has been viewed as THE big issue.‘ Moreover, by thrusting all of this legislation upon Congress --- most of which was highly controversial -- the President was forced to deal with opposition from many dif- ferent quarters all at once. Both the pace of the legislative process and the demands imposed upon the administration lobbying effort suggest that presidential success rates may be adversely influenced as the size of the administration's legislative agenda increases. What is required is a meausre which not only incorporates the number of administration proposals placed on the congressional agenda but also captures the extent to which bills endorsed by the president accumulate over the course of a given Congress. While not completely satisfactory, reliance upon the cumulative sum of presidential support votes - i.e. PSt -- provides the best available basis for including these aspects of legislative activity in the analysis. The preceding discussion should not be interpreted to imply that success rates will decline monotonically with each additional bill endorsed by the president however. That is to say, there are few grounds for expecting a linear relationship between success and the size of the administra- tion's legislative program. Rather, the problems Of adequate lobbying and the opportunities for opponents of the president are likely to 42 increase as the legislative agenda grows substantially larger (say from 50 to 200 bills and roll call votes). MOreover, the impact of size on success rates may level Off at some point so that further additions to the legislative agenda will not carry the same consequences as earlier increases. Such considerations lead to the hypothesis that the relationship between the size of the administration's legislative agenda and legislative success is curvillinear. To examine this proposition, define At' as a measure of the administration's legislative activity such that At. '3 1n ( E Nt') where t' - 1, ... , 8 and indexes the number of quarters which have elapsed during a particular Congress (two year period) N - the number of presidential support t. votes in quarter t' At' thus represents the natural logarithm of the cumulative number of presidential support votes where the accumulation occurs over the two year period of a specific Congress (83rd, 84th, etc.). A second potential influence on presidential success rates concerns the general ideological character of the policies initiated or endorsed by the president. It is plausible to expect that the degree to which presidential initiatives are perceived as representative; of a liberal or conservative viewpoint provides not only an additional piece of information useful for legislator decision asking but also a focal point 43 around which supporters and Opponents can attempt to forge winning or blocking coalitions. TO illustrate the potential import Of the presi- dent's positions, consider once again the information presented in Table 2.6 and, in particular, the legislative success rates of the Nixon Administration. A comparison of public and legislative support for President Nixon uncovers one strikingly counterintuitive result; the proportion of Nixon roll call victories increased between 1973 and 1974 despite a decline in public support from 42% to 26%. What factors might account for such improved legislative fortunes in the face of an otherwise politically ominous situation? One explanation was offered by Senator Barry Goldwater who suggested that President Nixon shifted his legislative positions in a decidedly liberal direction so as to "appease his would-be impeachers" (Congressional Quarterly Weekly Report, 1974:150). Cited as an example of such a shift was Nixon's 1974 budget message which, according to Goldwater, was ... abandoning such concepts as a reduction in social programs, the liberal use of the presidential veto, the rigorous enforcement of spending ceilings and the impoundment of monies unwisely voted for by the Democratic Congress. The remarks Of James Collins (R, Texas) underscored this description (Congressional Quarterly Weekly Report, 1974:150): "I'd say the president must have gone to the left." (See also, Glass, 1973). Although the circumstances surrounding the relationship between president and Congress in 1973 and 1974 were unique, this episode does point to the general proposition that the character of presidential positions may be related to the level of presidential success in Congress. 44 Although the operationalization and interpretation of such terms as liberal and conservative is often hazardous and must be approached with caution, it is possible to develop a measure which conveys some indication of the ideological character of the president's legislative positions. Recall, first of all, that P8 was defined as that subset of congressional roll calls in which the president took a clearly identifiable position. Also available from Congressional Quarterly Reports is a subset of roll calls labeled Conversative Coalition Votes. These votes represent that set of roll calls in which a majority of Southern Democrats united with a majority of Republicans to Oppose the legislative position favored by a majority of nonsouthern Democrats. The positions endorsed by the Southern Democrat-Republican alliance can be treated as the conservative position on the issues in question. Let this subset of all roll call votes be designated by CC, Next, define CP as the intersection of subsets CC and PS; that is, CP = CC firs represents that subset of presidential support votes in which there exists a reasonably precise description of the liberal and conservative alternatives. If cp is defined as the number of such.votes in which the position advocated by the president matched the Southern Democrat- Republican position, then cp/CP denotes the proportion of votes in which the president adopted a conservative position. As with the legislative success rates and the measures of legislative activity, a cumulative version of this measure is defined for use in this analysis: 45 P = 2 cp / 2 CP t t t t t where t-1,...,16 P - a measure of legislative position taking by the president cp . the number of roll call votes in which the president and the Conservative Coalition were in agreement CP - the number of presidential support roll call votes which were also Conservative Coalition votes The magnitude of Pt thus provides at least a rough approximation of the degree to which a president's legislative positions tend toward a liberal or conservative direction. An increase (decrease) in the value of Pt indicates that the legislative positions advocated by the president are becoming more (less) conservative. The following specification is designed to ascertain whether legislative success rates remain responsive to shifts in public support when controls for legislative activity and the character of presidential positions are introduced: .. + + + . Lst Bo B st + B At, B Pt et (2 2) 1 2 3 Because reports on the Conservative Coalition votes are available only since 1959, the sample size of the data set used to examine Equation 2.2 is reduced from 105 to 81. The results obtained from estimation via OLS are again unreliable because of serial correla- tion; Table 2.11 presents the estimates Obtained by applying the Cochrane-Orcutt procedure to the legislative success equations for domestic and foreign policy in the House and Senate. 46 Examination of these estimates reveals, first of all, that, as measured, legislative activity exercises a negligible impact on the rate of legislative success. Only in the case of House foreign policy does the activity coefficient approach statistical significance. With the exception of Senate foreign policy, the relative conserva- tism of the legislative positions adopted by the president exerts a pronounced negative effect on the cumulative success rates. The estimates indicate that roll call success declines as the president's positions grow more conservative. The result is not unsurprising given that the Democrats held clear majorities in both chambers throughout the 1959-1978 period. Nevertheless, this analysis does constitute one of the first instances in which the relationship be- tween legislative success and the relative liberal or conservative nature of presidential policies has been examined. The finding suggests that presidential fortunes are likely to suffer when electoral or other considerations force presidents to adopt more conservative positions. The results also emphasize that the positions adopted by the president be viewed not as a fixed characteristic in analyses of roll call activity but as a manipulable component of administration strategy. The results in Table 2.11 reinforce the previous conclusion that public support is an important influence on the administration success rates. Despite the introduction Of control variables, the estimates obtained for public support are quite similar to those in Table 2.9. That is, public support exercises a significant influence on foreign policy success rates in the House Of Representatives and -domestic policy in the Senate. Moreover, a comparison Of Tables 2.9 and 2.11 reveals that the introduction of legislative activity 47 and positions does not appreciably alter the magnitude of the public support coefficients; the coefficient for public support in the case of House foreign policy drops from .409 to .351 when activity and positions are incorporated into the specification while the coeffi- cient obtained for Senate domestic policy increases from .384 to .584. While the necessity for further analysis of the data should be underscored,9 the results reported in this section do emphasize that public support is a presidential resource which can be expected to pay off in the legislative as well as in the electoral arena. Presidential Performance and Other Citizen Evaluations A final line of research sounding the theme of public support as a resource revolves about the association between approval of the president and citizen satisfaction with the institutions of the national government. One proposition formulated by such research is that dissatisfaction with the performance of Congress is influenced by citizen assessments of the chief executive's performance (David- son, Rovenock, and 0'Leary,1966; Davidson and Parker,1972; Fenno, 1975; Parker,1977). The basis for this proposition pertains to both the centrality of the president in the political assessments of the citizenry and the general difficulty of evaluating the performance of a collective decision making institution such as the Congress. To examine this contention, I have arrayed in Figure 2.1 available eval- uations of congressional performance (percent disapproving) since 1943 together with the Gallup disapproval scores which correspond to peach congressional sampling point. While the levels of dissatis- faction have fluctuated substantially, the general pattern displayed 48 by the undulations, particularly the upward movement of both measures during the 19603 and early 19705, are quite similar. The product moment correlation between congressional and presidential dissatis- faction for this period is .33. It was such similarity in the pattern of evaluations which prompted Davidson and his associates (1966:62) to conclude that "Congress is not evaluated directly but rather through the prism of the presidency." It has also been suggested that assessments of a specific pres- ident's performance are related to both the level Of public confi- dence in the institution of the presidency (Dennis,1975) and to the level of trust in government (Miller, 1974a,1974b,1974c; Miller, Brudney,JOftis,1975). In a longitudinal context, such findings carry the implication that evaluations of a president's performance may play a substantial role in accounting for changes in the public's more general and long-term predispositions toward the political system. For example, as Operationalized in cross-section, survey analyses, the attitude of political trUSt is treated "as a basic evaluative or affective orientation toward the government" (Miller, 1974a:952). Trust therefore represents the degree to which indivi- duals believe that the national government is operating in accordance with their judgement about how a democratic government ought to function in meeting the fundamental expectations of the citizenry. This usage implies that political trust, a general attitude, is in- variably connected to how citizens evaluate more proximate, short-term aspects of government behavior. Thus Miller (1974a,1974c) has argued that the Observed decline in political trust between 1964 and 1970 was, 'in part, a result Of a persistent sense of citizen dissatisfaction with the performance of Presidents Johnson and Nixon. Because of 49 disagreements concerning the operationalization of political trust and the validity of the trust measure, the interpretation Of such results must be approached with caution (see, for example, Citrin,l974 and Miller's rejoinder in 1974b). Nevertheless, this literature does establish a plausible connection between assessments of presidential performance and the more general sense of satisfaction with the national government. Conclusion This discussion has established a reasonably convincing case for the prOposition that public support is a valuable presidential resource. In Chapter One, the belief that public support shapes the political environment of the Washington community was shown to be part of the conventional wisdom which underlies journalistic and insider descriptions of presidential politics. This conventional wisdom was then subjected to empirical scrutiny in this chapter. The results demonstrate that public assessments of a president's performance exercise a clearly discernible impact on outcomes of import to the president and his party. Perhaps the centrality of presidential support can best be conveyed and summarized by the following diagram: Congressional Election Outcomes (Turnout and Candidate Choice) Presidential Election Outcomes R 11 C 11 S t Am M Public Support --"""" o a “FPO? Ong Cs for the -::::::::::‘Presidential Success in Congress President \\\\\$ Public Evaluations of Congress Confidence in Institutions Trust in Goverment 50 Although no claim is made that assessments of a president's perform- ance act as the sole determinants of such behavior, these results do have two important implications. First, it is clear that public support stands at a crossroads of research in American politics. That is, they demonstrate that the evaluation of a president's performance operates as one conduit through which citizen satisfactions and grievances register an impact on electoral and legislative outcomes. Consequently, contemporary research is pointed toward deveIOping more sophisticated explanations and models which link these decisions and outcomes in a time-dependent system of reciprocal influence. Second, and more important to this analysis, the results esta- blish that public support does operate as a presidential resource. When combined with the findings of past research, the evidence shows that the influence of public support operates in both the electoral and congressional arenas, at both the individual and aggregate levels of analyses, and finally in both the cross-section and time series contexts. Such evidence provides a rather strong rationale for a president to be concerned about his public standing. Because presidents find it useful to have their party control the chambers of Congress and because presidents are dependent upon congressional cooperation, support serves as an instrument in the pursuit of presidential goals. To this point, my discussion has served to identify the reasons why presidents will be concerned with their public standing. It is in this sense that Chapters One and Two serve as a prelude and justification for the discussion which follows. Having identified a rationale for support management, questions arise as to the ability 'of the president to influence public support and how such attempts to 51 to control are pursued. However, the investigation of such issues first requires that we have some understanding Of how citizens eval- uate a president's performance and how public support impinges upon presidential decision making. These are the tOpics which will be addressed in Chapter Three. 52 Notes to Chapter 2 The cross-section data used in this chapter were made available by the Inter-University Consortium for Political and Social Research. The data for the American National Election Studies were originally collected by the Center for Political Studies of the Institute for Social Research, the University of Michi- gan, under a grant from the National Science Foundation. Neither the original collectors of the data nor the Consortium bear any responsibility for the analyses or interpretations presented here. Interestingly, the works of Cover (1977), Ferejohn (1977), and Fiorina (1979) reach a similar conclusion concerning the import of House incumbency in modifying the relation- ship between partisanship and congressional voting. In effect, existing research has identified three decision cues employed by citizens during congressional election campaigns - party identification, incumbency, and presiden- tial performance. The task which now confronts the enter- prise is to assess the relative import of the three criteria. Because the presidential evaluation item was administered to only half of the 1972 survey respondents, the resulting N's for the analysis of senate and gubernatorial vote were quite small. Consequently, these 1972 results were excluded from consideration. It should also be noted that the item probing state and local voting behavior was not included on the 1976 CPS interview schedule. The number of cases for this yearly analysis totals 27 because separate presidential support roll call scores were calculated for Richard Nixon (January - July) and Gerald Ford (August - December) in 1974. The roll call support scores were gathered from the relevant volumes of the Congressional.Quarterly Almanac. The presidential support scores for Democrats, Republicans and Independents during the 1953-1971 period were provided by John Mueller of the University of Rochester. The data for the 1972-1978 period were gathered from the relevant issues of the Gallup Opinion Index. 6. 9. 53 The pseudo-GL8 technique is comprised of the following steps. First, the parameters and residuals Of the model were estimated using OLS. Second, the residuals from this OLS regression were analyzed to ascertain the nature of the serial correlation and a first-order autoregressive scheme was speci- fied and the autoregressive parameter (RHO) was estimated. Fourth, the RED estimates were employed to transform the data such that the disturbance term satisfied the assumptions of the least squares estimation technique. Fifth, the model was then reestimated using this transformed data.' Finally, the resulting parameter estimates were used, in conjunction with the original, untransformed data, to generate the goodness-of- fit statistics. This technique is described more thoroughly and illustrated in Ostrom (1978). All information pertaining to presidential support votes was gathered from the relevant editions of the Congressional Quarterly Almanac. As with the analysis of member support scores, separate success rates were calculated for Nixon and Ford in 1974. This results in two scores for Quarter III of 1974 and a total N of 105. Between 1955 and 1970, the Congressional Quarterly organization reported separate domestic and foreign policy support votes. This CQ classification was employed in constructing the data set for this period. The foreign-domestic breakdown for the 1953-1954 and 1971-1978 periods was performed by myself. In classifying presidential support votes, I tried to follow as closely as possible the criteria evident in the CO break- down of the 1955-1970 votes. In general, I placed all roll call votes pertaining to treaties, other diplomatic issues, and military matters into the foreign policy category. There was one general exception. Any votes pertaining to veterans affairs (e.g. hospital construction, government payments) were placed in the domestic category. This was done because such issues are generally regarded as classic pork barrel concerns (see, for example, Fenno,1973). One recommendation is to improve the measure of presidential roll call success. For example, the consensual and largely pro forma roll call votes can be eliminated from the set of presidential support votes. Because the focus would be narrow- ed tO the set of more controversial issues, a more robust measure of success would result. Indeed, it can be argued that public support is likely to be more important in the case where a presidential position is actively contested. 9. 54 (Continued). Another consideration is to expand the definition of legislative activity to include measures of how much effort the president devotes to the passage Of his legislative program. Finally, the results of this analysis need to be compared to those produced by alternative explanations of presidential success in Congress. 55 .HO>OH mo. one on ucmufiwwswfim ma owumfiumum mumnvmufino onu umnu mouocmn .xuume m.uuoowmoue ecu mo oumofioemo ammo: Ono new oouo> on: musoocoamou mo :Owuuoaoua Osu ucmmouaos moauusm HHOU may .zaeo musoneoamou amoro50mcoc Ou swooped muasmou m.HHocuox ammo peace on nanosw OH .Aoo .av m ouuwfim one awn .av m .Anm .av c .Aom .ev m moaamh ma Asmaav Haoequ an oOfiHQEH no omuuoeou m.z ozu wnqxoaeam he omeaoupo mums coma one coma emozuom meowuooam Susana: me ozu HOW mowumwumum zumeenm new muaowmm one Hmsoauooam assume: memo: ea OOHOLU oumowonmo our mommauowmm Hmfiucooamoum mo meowumzam>m coufiufiu H.N «Home llfi EA is anma coma Noaa wmaa «mad omma coma snow 00. ~.wm H.¢¢ «ma. H.0H w.o~ «ma. H.nm m.nn who so. o.wm n.¢¢ «ea. o.~ n.¢ era. w.Hn w.Hm omH< «RN. o.NH ~.om «ca. H.N N.m «Ha. m.Ho o.om omH< «mm. o.aa o.~m «ea. o.m m.oH «ma. N.mm N.mm omH< «mu. w.m~ m.~m «0H. ~.m m.o~ «0.1 o.ooH o.c¢ omH< «Hm. w.w~ m.~c HA. H.N m.m «NN. m.mn h.mo omH< «ma. o.m~ H.ma «ma. o.N m.oH «ma. N.oq Nw.mm omH< magnum 335d 33%3' mhzmnzmmmnzH >8mOH no. sea on unmofimaswfim ma Ogumfiuwum osmoomlfimo Ono umnu mouocmo Ila .ooOO> one musoosoemou mo sOfiuuoaOue onu ueomouaou mowuuso HHOO OLE IIN .maco musoocoamou suozusomso: Ou nfimuuoe muaamou m.aaosuox more pogo: on oasonm OH .Aoo .aV n museum was awn .aV s .Aem .aV m mosses as AssaHv immense an saunas“ to emotoamo m.z wen wcssosaam an oomwmuno mums coma new oeoH consume moOfiuomHO ammuofie xwm ecu pom mowumfiumum appease new muasmmu any Ila no. m.mm m.H¢ «mH.I w.oo o.~m No.l w.mo $.50 Ho.l m.om o.mm «q~.| o.wm H.m© koH.I a.om H.m sonuuuu N.N wanna who «mag omH< coma omH< Nome omH< wmoa omH< enma omH< onma omH< coma muma\umo> 57 muse axon co JOLACL m.N amuse Ache moons Axes x~xv ionic some Ammo issue AsoNV .su. h.m o.ml ¢.- «mu. ~.~ m.~ m.m was. n.m~ o.os o.mc An~v Anne AGNO asNO AsNV Axle AONO “may floods c~. s.on n.- m.sm so. e.c~ ~.cN ~.- .eN. o.ms u.ua m.om Acme x~sv iamv Aces Axes ance ions Amos Asses so. o.os ~.ns s.ss no. m.su m.m~ c.s~ .mu. c.mm a.im s.wm some some moss lose a-lv ARGO lose maize x~m~v .ou. a.w~ m.zs s.~m .mz. m.o~ m.su m.o~ .no. G.Re m.mm m.um “new house Acme some sense Aces Acme locus AolmO .an. n.u c.s~ s.os «an. s.« a.o n.m_ .mm. a.m~ o.mo o.mw Asouv Asses menus muons Amaze Amos Anus issue Asonv so.- u.o~ o.ns s.~o cc. s.¢n s.~m ~.mm .co. ~.oo s.oh a.m~ 0 sub noon sash nose 0 ask uOOm Cash vooo No one soon came oooo eucoocomovcu memouunsmom compasses Ecoauoeqquu< cementum an .cosuoodm saccuoaz moa~ «so as nod>szon acuuo> was sceczon .m some»: mo accuuu3~m>m coauuuo .m.~ oases Ou mmuoz lie muo> HmOOA\Oucum auo> and Incuseuoezo Ouo> museum muo> mouse: actress: ecu Ouo> uzoeush 58 Notes to Table 2.3 1-- Citizen assessments of President Johnson's performance were obtained from the following CPS survey item: "In general, how do you feel about how President Johnson has done his job. Would you rate his handling of America's problems over the past four years as very good, good, fair, poor, or very poor?” In this analysis, the very good/good responses as well as the poor/very poor categories were collapsed to form the ”Good- Fair-Poor" scheme. 2- The measure of association employed here is Kendall's Tau B (for asymmetric tables). The parenthesized figures are the relevant column N's. 3- The cell entries for the House, Senate and Gubernatorial votes are the proportion of respondents who voted for the Democratic candidate. 4-- The entries for state and local voting are the proportion of respondents who reported voting a straight Democratic ballot in these contests. *-- denotes that the chi-square statistic is significant at the .05 level. 59 .HO>O~ no. org us ucuuuuwcwuo nu caucuumuu Oucsvunfiso any use» ocuocon use .OOOOueOO Omega cu nodasa csoqaaamuz usuuuuua o meauouu seasons» as: mucovcoamou a show no :Oauuoeoue as» one usuuo> dosed use muons sou uuuuuuo 0:9 lac .Ouuvuvcuu mono: coauansaom one new vOuO> or: oucovsoeoou H loom mo ecuuuoeoua usu aeoeounou ouo> mono: use new nouuueo undo oak Inn .eoHnuu ecuuuasaeuooouu usu>o~ou ozu aouu n.z sundae say one nouswuu vonuuosusouse 0:9 Ila .vovaduxo coop O>az nuances unau0uscuonao vac museum or» .u.z games we monsoon .Aoaeaom -a~ ouuuco ozu no «n.0m "Nemalzv sane aueoveoamou H anon cu tuneueueuevu on: .uo>ero: .auou mask :«ueooquoua an no« out mcu~oeus mu coxuz an: emu mo o>ouenuuqo no o>ounno so» on: “luau xo>u=a mmo «caroAHOu use ecuu condense one: OOcQILOHuon u.=ox«z acupunoum no euuoaoooou< eonauuo Ilg “Name Aisle x~muv Aimee A-O Aom~v com. ~.- <.—~ «AN. n.m c.w cc~. o.od ~.m< cauo> mono; new venom Amie Acme Annie xzo_v AQNO x~e~O «an. n.m~ q.ec «mL. m.m m.o~ «efi. o.mm 0.05 nauo> ooze: Andy Ancv Acefiv AooNv ALNV Adomv soxqz sew. ~.L~ n.mo «mm. w.m o.cc «mm. «.mm ~.¢o uOu weuuo> aeoouom Anne mass scams cause Acne Algae $0.: c.oc m.om no. m.nn o.n~ cc. ~.e~ m.~a Nuaocuss mlflmm a mam a mug mufl m 22. amass duo. mszuazeamozu weuzon mcmuo> new sex“: .2 cumsoqx LO msouuo=~m>m couwuuu e.N munch 6C) .Ho>o~ no. use as accuse—swan mu Oauouuuuo ouuzvmluzu us» away oeuocon Inc .Ousouvcuo cuouansaux ago new vOuO> 0;: aucoocoauou we ecuuuoaoua use one eu~zoou ~s«u0uscuonsu use museum .oosoz ecu new measure one tin .e.z sundae one nouswdu vonuaozucouun 0:9 IIN :sueooqsoue no can our wcuavcs: as once .ut as: ecu we o>ouaemeqv no O>Ouaeu sou on: "emu“ xo>usu mac mauso-0u one scum nocuuuno one: oucmBuOuuon o.vuom acouumoum uo oucoeaooaou nonauao Ind Aide mm. ~.o Aawv -. c.a~ Aonv can. ~.- some «on. o.o~ Anne o~.l m.oo m ash museum mucovcomooc— Amie o.~n Adsv n.~e Acne ~.om Aoo~v o.w~ Aoodv n.cm mufl a~ov ac. N.- AmANO is. s.~i momma awN. c.s xousv ass. m.m loamy so.u o.cs fl 3GB. mmmmdo mumuuosoo a-v u.w~ Ammav ~.m~ Ammmv ~.~m scene o.os RNOGV u.o~ mm« No. -. sew. «mm. co. m ssh ~=ovum__~we< cmn_auma a: .:o~uoo_m .r:oammz o~o~ are :— uc_>m;om x:.uo> use once .x o~muoa L: m.~ O~an zcouumsdm>m Ame aside m.as ~.uw Anne saunv u.mm n.ss ions amass z.ms s.ma Aime Asmmv c.1n o.~o Away assoc o.ms s.ow unease age mcmun~nzmmx conduuu Ouo> AuduOuusuonso Ouo> museum neuo> more: when .¢.o new muo> mcuuo> Nucouuom 61 Table 2.6 Public and Congressional Support for the President,1953-1978 Public Support Congressional Support House House Senate Senate Year Total Dems Reps Inds Domestic Foreign Domestic Foreign 1953 68.7 55.4 86.7 67.1 88.9 93.7 82.6 92.3 1954 65.3 49.9 87.2 68.2 78.1 83.3 75.0 86.4 1955 71.7 55.9 90.6 73.0 41.7 94.1 71.4 96.8 1956 72.4 56.2 93.1 76.2 71.0 99.9 54.0 85.7 1957 63.5 46.7 85.4 65.4 53.2 84.6 57.7 96.8 1958 54.2 36.0 81.4 55.1 74.3 71.4 73.4 89.5 1959 62.8 45.4 86.8 63.4 44.2 99.9 42.4 79.3 1960 61.7 43.5 87.4 64.4 56.2 90.9 56.7 84.6 1961 75.2 87.0 58.3 71.1 75.6 99.9 78.4 86.1 1962 71.9 85.8 48.8 68.0 80.0 95.0 82.9 90.7 1963 63.4 79.2 40.1 58.8 89.8 68.2 88.6 91.1 1964 73.7 84.0 61.2 68.0 91.4 81.2 88.6 80.8 1965 66.3 79.2 49.3 60.0 91.1 99.9 93.1 89.1 1966 50.7 66.9 32.9 44.3 91.4 78.3 68.8 65.6 1967 44.0 59.4 26.0 38.2 75.5 76.2 83.6 80.7 1968 42.1 57.6 27.4 33.9 82.1 88.9 62.1 83.7 1969 61.5 49.3 82.0 61.1 70.0 85.7 58.5 96.7 1970 56.7 41.5 82.3 58.5 84.2 87.5 67.1 93.3 1971 50.0 35.7 79.0 48.8 75.0 95.0 55.0 83.3 1972 57.4 40.0 85.1 58.1 76.0 91.7 50.0 59.1 1973 43.0 27.9 69.7 44.1 46.7 51.5 38.7 80.3 1974N 25.9 13.3 52.1 25.4 70.4 55.6 46.0 80.0 1974F 54.4 44.0 69.0 52.8 60.5 56.2 51.6 56.8 1975 43.8 32.7 63.7 43.9 42.4 73.9 60.4 85.0 1976 48.3 36.2 70.0 50.0 42.9 50.0 62.9 66.7 1977 62.2 72.4 45.9 60.9 73.2 78.3 72.9 89.5 1978 46.5 58.0 30.0 47.0 77.0 71.4 75.0 94.7 62A .cc—um—stucc ~s~toe we socoeoua osu maumoaveu In. .~o>m~ me. can u: acme—uusxem av tugs—umumnu ozu umxu emuocmo Ila KN.N Oc.~ u-.— §0N.~ O :Omuc3lc~nhsn no.0 c~.m cN.N mm.o coduwu>wa vhuvcmum woe. ace. woo. ocm. kuuwhuou .N1 the.“ ¢~N.m nc.m mw.c UdumfiuauWIPII oo.€ N~.m Nm.v m0.m heuhm .umll en.0n m0.¢~ n¢.mm no.o~ ucuumcco e~m.h t~m.n On.h -.m Uuumuuuumlhll N50. ~06. cos. cwo. houhfl .umll ououuuucoou 02. So. as. 23. 5:8 98.2 28 8°33 32.... museum Were: Mum:om memo: mP<¢UOZmD m2OH no. use um unmowwwnwfim ma m¢.~ Hm.5 men. «qw5. «me.w mH.5 mq.oe «O5.n OHH. mos. smfiosom A>Hv w5m~ I AHV mmma .oumcom one memos on» ad oeumoaoo mmaom. sm.~ m~.oH mmo. Hmm5. «mc.o No.o o~.~o 5mm. NmH. 0mg. Ofiumwumumlu osu umsu mouoaon Ila somumzlsanuan scaumfi>on uncommon oOuoouuoo .Nm omm owumfiumumue u: uouum .um II usmumsoo unassumbmus nosum .um II some uuoaasm coupes meauOuoa> Hamu HHOm Hmwuuoofimoum one uuonemm ouanwm nooauom ease:OHumHom one o.~ canny Average Success Average Success Average Success 65 Table 2.10 A Comparison of Presidential Success on Domestic and Foreign Roll Call Votes in the House of Representatives Yearly Presidential Rate, 1953-1978 Yearly Presidential Rate, 1953-1966 Yearly Presidential Rate,1967-1978 Difference in Success Rates (1953-66) vs. (1967-78) Pearson Correlation Between Public Support and Legislative Success, 1953-66 Pearson Correlation Between Public Support and Legislative Success, 1967-78 House Domestic Policy Issues 70.47% 73.35% 67.37% -5098 O 188 .185 House Foreign Policy Issues 81.58% 88.62% 73.99% .364 .609 66 5c~. —m5<. «5o.o 5N.» co.o5 m~.~: oqc. omo.l qnm.l ofio. «05.: c~.~ «Na. emd. cwuouom ._O>O# mo. mru um c~.~ #~.~ c¢.5 om.5 «mm. mam. ~n_5. ce~5. «xc.c so~.a em.w co.m ~o.~m 5<.e5 cn<.N| a~m.¢| 5mo. one. on—.: ne~.I Emm.u «mo.~t moo. m—m. n55.| mo.~t «n~.¢ «no.5 o~—. -~. sew. mmm. Ouuwoeoa cwfiouom acacueucu_m a. Oaueuucucuu or» uczu emuocoa Inc o_.~ 5~.o— one. moan. Om.c Nc.- ~m.m5 aoo.oo eunoeuum vauoouuoo .«z on: mayoraunm ouueuumumlh II touum .um I: acmuocoo o—uauuuumnh II nouum .um I: Aumv ucOwuunom o>uuu~oumoa Ouuowuuumlh II uouuu .um II A.osuu< -.N OHAOP O>uuc~e¢woa .uuoeasw Ou~a=a mo scuuocsm a an noduOuou> ausu -O¢ duaucooqooum 67 AP r a 9.155 5.5” u p . L o w Hm>oueesmua amuunovumosm _ p m m — N H C Hm>cuaemmuc ~m:o«mmmuwcou NI-sxm F" Nun-(Np: H50 ~_VO°‘ u—oI-AOQ N —I"‘Df\ 5 N 00 dhom «‘ch “MOM q—e o‘om cur-moo «June _ cor-«5‘1 57!- m5¢~ I meo— :O«uom_m«ummm«: ~n:o«mmOuwcou rum Hnfiucmuimmua cmo3umm :«LmCOwum~mz one _.~ museum I'NI-lnm N-Q‘D 0557“ new» Gh¢¢ MTG!” ‘3' race: oe TI 05 .I CC a co I DC— is wcw>ounasmun ucoouoe CHAPTER THREE: THE CALCULUS OF PUBLIC SUPPORT CITIZEN AND PRESIDENTIAL DECISION MAKING The discussion Offered in Chapter One and the empirical results presented in Chapter Two inform us that as a shaper of perceptions, asean influence upon public attitudes toward the political system, and as a cue for citizen and legislator decisions, public support is a valuable presidential resource. In effect, these portions Of the analysis provide a strong justification for presuming that presidents will employ the perogatives at their disposal in an attempt to control the level and pattern of public support. As noted in Chapter One, however, the investigation of support manage- ment must rest upon a specification of the components which under- lie citizen and presidential choice -- goals, valuations, and decision rules. This is the objective of this chapter. More specifically, this chapter formulates a model of citizen-president interaction which is based upon the premise that presidents are evaluated according to two general sets Of criteria -- the actions taken by the president over the course of the term and a set of outcomes for which the president, by virtue of his position, is held accountable. It is assumed, moreover, that citizen decisions are essentially two-tiered. First, citizens make specific management evaluations (denoted by the term Handling in Figure 1.1). Such decisions are assessments of whether a citizen is satisfied or dissatisfied with how the president has managed such concerns as 68 69 the economy, foreign relations, or particular domestic problems. It is then argued that such specific evaluations rest upon a comparison between expected and actual presidential performance where such comparisons are based upon the enduring concerns of citizens, the promises of action made by the president during the election campaign. and problems generated by domestic and inter- national crises. It is then asserted that the second component of the support decision -- the choice to approve or disapprove of the president's general or overall performance -- is a weighted function of these specific evaluations. Thus, upon entering office, presidents are confronted by a variety of outcome-based and action-based demands. Yet the incen- tive to satisfy such demands and thereby influence support is but one consideration which impinges upon presidential decision making. Presidents will also be motivated to pursue their own preferences regarding what outcomes are desirable and what actions and programs are necessary to realize these outcomes. Furthermore, -the ability of the president to direct the machinery of government toward the states desired by either himself or the public is constrained. Not only may certain combinations of outcomes be unrealizable but, more important to this analysis, particular actions which the president considers necessary for the control of outcomes will depend upon the consent and cooperation Of competing institutions and decision makers. To capture this constraint, the role of Congress will be explicitly incorporated into the model by drawing a distinction between presidential actions that are unilateral and those whose execution depends upon Congress. It will also be assumed that the capacity of the president to influence 70 outcomes is, in part, a function Of congressional support. Here one dynamic aspect of the model becomes apparent (see Figure 1.1) for, as the results of Chapter Two demonstrate, congressional support is itself dependent upon the level of public support and the ideological character of the actions and programs submitted to the Congress by the president. Clearly, the basic rudiments of this model - specific performance evaluations of the president, public support, presi- dential actions, and congressional support - are interdependent. The subsequent sections of this chapter are designed therefore to state this dependence formally and to develop a model of citizen- president interaction which conforms to the pictoral representation presented in Figure 1.1. The Criteria of Citizen Choice It is assumed that the citizen support decision is funda- mentally a function of three influences: a set of presidentially- relevant outcomes, a set of presidentially-relevant actions, and public expectations of presidential performance. With respect to outcomes, in any given period Of time there is a sizable amount of information emanating from the national government which includes economic reports, administrative decisions, proposed legislation, and countless statements by executive branch officials. This output provides the public with numerous indicators of government performance and, for this reason, is labeled the government outcome set. The elements of this set are not of uniform import to either the public or the president however. 71 There are variations, first of all, in salience or the degree to which particular outcomes are perceived to impinge upon citizen satisfaction. A second source of variability within this outcome set pertains to what office holders and/or institutions respon- sibility for achieving or failing to achieve a particular outcome is assigned. Now define the column vector (For convenience, brackets ... will be employed to represent column vectors): z , ... , z } (3.1), 2 = {21, 22, ... , n N as the presidentially-relevent outcome (PRO) set. This set, whose typical element is zn, represents those elements of the government outcome set which satisfy the following criteria. First, the outcome is perceived as salient by the public; that is to say, the outcome is perceived by citizens as important to their sense of well-being and security. Second, the president is viewed as being primarily responsible for the character of the outcome. The presidentially-relevant action set can be defined in an analogous manner. During any given time period, the president can be expected to engage in a variety of activities which include the introduction of legislative proposals in Congress, the appointment of individual office holders, the conduct of news conferences, and travels both within and beyond the United States. Each of these activities will not, however, attract an equal amount of attention from the print and broadcast media; neither will they be perceived as equally important by the public. Accordingly, let the vector P denote the set of presidentially- 72 relevant actions (PRA); i.e. Palpls P29 00- a pm: °°° 9 FM} (3'2) This set is defined to include those presidential actions which are visible, salient, and associated by the public with some element of the presidentially-relevant outcome set. The salience criterion employed in the definition of both the PRO and PRA sets is critical for it implies that not only will particular outcomes be monitored by the public but that individuals have preferences about what outcomes ought to occur and about how the president should behave in order to ensure that the preferred outcomes are realized. With respect to the criterion of presi- dential responsibility, Ostrom and I (1978b, 1980) have argued that there exists a set of identifiable and measureable outcomes for which the president is held responsible. More specifically, it will be assumed in this thesis that there are essentially three types of outcomes and associated actions which define the set of public expectations depicted in Figure 1.1 and which ultimately influence the level Of public support for the president: (1) institutional expectations or demands that are placed upon the Office of the presidency regardless of the identity of the incum- bent; (2) electoral expectations or demands that are molded during a specific incumbent's effort to capture the presidency; and (3) situational expectations or demands which arise from the context of crises and other unanticipated events. The outcomes associated with these expectations constitute the elements of the presidentially-relevant outcome set while the presidentially- 73 relevant action set consists Of those actions perceived to be connected to the values assumed by these outcomes. Institutional Demands Throughout their tenures of office, presidents have traditionally been confronted with three public expectations or demands. First, it is expected that the administration will act so as to maintain a reasonably PIGSperOus economy. A second demand is that the president will utilize the military and diplomatic expertise of the government to avoid war. Finally, it is presumed that the behavior of the president will not undermine the integrity and credibility of either the institution of the presidency or the national government in general. These demands are termed institutional because they are directed not so much toward a specific occupant of the White House but rather toward the institution of the presidency itself. In other words, the influence of criteria such as peace, prosperity, and government integrity on assessments of presidential performance follows from the citizenry's belief that the power and capabilities which reside in the presidency provide any incumbent with the necessary means to ensure the preservation of these desired states. Several strands of evidence can be marshalled to support the assertion that the demands of peace, prosperity, and integrity form a distinct set of evaluative criteria. First of all, the results Obtained from investigations of public hopes and fears emphasize the ubiquity of these concerns. Elsewhere (Ostrom and Simon, 1980: Table 1), it has been shown that with only one exception 74' during the period from 1953 to 1976, problems directly linked to prosperity (unemployment, inflation), war (Korea, Vietnam) and government integrity (watergate) have dominated the political consciousness of the public. These concerns thus share a distinc- tion as issues most critical to the body politic. A second commonality is that peace, prosperity, and integrity Operate as prominent valence issues during electoral campaigns. Whereas most campaign issues are bounded or defined by liberal and conservative alternatives, the endpoints of the dimensions which underlie valence issues do not necessarily carry such ideological connotations. Instead, the underlying dimensions of evaluation are defined by pairs of conditions that are positively and negatively valued by the electorate. Among the most electorally decisive of such ordered pairs are war and peace, integrity and scandal, as well as economic good and bad times (Stokes, 1966: 170-171). Moreover, the distribution of Opinion on such issues is likely to be degenerate or collapsed about the positively-valued alternatives of peace, prosperity, and integrity. Because of the nature of the preference distribution on these valence issues, candidate strategy is likely to focus upon tactics designed to convince the electorate that the candidate should be associated with those conditions most valued by the citizenry. As labels such as the Party of Depression and Party of war imply, the likelihood that such persuasive efforts will succeed and the electorate will associate a candidate with the positions of peace, prosperity, and integrity is limited by the prior performance of a candidate (in the case of presidential incumbents) or a candi- date's political party. Not only do these demands share the 75 properties of valence issues but as a result, compel prospective occupants of the White House to address these concerns and thus operate as criteria according to which candidates for the office of the presidency are assessed. The idea that peace, prosperity, and integrity constitute basic demands which the public requires of any presidential admin- istration gains further credence when we consider what occurs when one or more of these demands are not satisfied. Regardless of the incumbent president's party affiliation, his specific campaign pledges, or the division of partisan control between executive and legislative branches, presidents and their parties have been held electorally accountable for economic downturns, wars, and scandals. A variety of empirical research and historical accounts of the post-war presidency has shown that failure to satisfy such fundamental citizen expectations meets with retribution (for example, Campbell, et. a1., 1960, especially pp. 555-556; Kramer, 1971; Bloom and Price, 1975; Tufte, 1975, 1978; Kernell, 1977; Miller and Miller, 1977; Sundquist, 1968). The citizenry, in effect, assumes that peace, prosperity, and integrity are normal states of affairs, demands that this normality be maintained, and punishes its disruption. It is in this sense that these expecta- tions are imposed by the public as the givens of presidential politics. A final argument which suggests that public expectations of peace, prosperity, and integrity form a distinct set of evaluative criteria revolves about the implications which arise from the prolonged absence of one or more of these conditions. While the citizenry's short-term, electoral reaction to economic distress, 76 wars, and scandals has been emphasized, other research reveals that the frustrations produced by these ills have long run con- sequences and may ultimately lead to the deterioration of such public attitudes as trust in government, confidence in the insti- tution of the presidency, political efficacy, and confidence in the electoral process (on this theme, see Miller, 1974a, 1974b, 1974c; Miller, Brudney, and Joftis, 1975; Converse, 1972; and Abramson, 1977). To summarize, the assertion that peace, prosperity and integrity form a distinct set of evaluative criteria according to which the performance of all presidents will be assessed is grounded upon evidence which emphasizes that these concerns (1) are persist- ently salient, (2) operate as compelling valence issues in electoral campaigns, (3) constitute standards which, if unmet, result in electoral misfortune, and (4) are related to the public's long term political predispositions. Electoral Demands The second set of criteria employed to evaluate presidential performance consist of electoral expectations or citizen demands which are formed during a specific incumbent's effort to capture the presidency. It is my contention that along with the noise of a campaign, the rudiments of an agenda for presidential action are put forth. That is to say, attention and debate are directed toward particular problems and alternative plans of action for attacking these problems are presented. 77 During campaigns, candidates behave as entrepreneurs in the sense that their strategies are designed to convince the public that particular outcomes ought to be considered presidentially- relevant and that, if elected, a particular set of actions will be pursued to guarantee that more desirable outcomes will prevail. For example, during the election of 1968, much import was placed by Richard Nixon on the issue of law and order. Nixon's campaign themes, designed to attract the support of disgruntled Johnson supporters and potential Wallace voters, emphasized rising crime rates as an outcome of import to the public's sense of security and proposed a plan of attack which included the appointment of judges who would treat criminals more harshly and legislation designed to weaken then—recent Supreme Court decisions pertaining the the rights of accused criminals (Page, 1978: 82-83; see also White, 1969; Chester, Hodgson, and Page, 1969). More generally, a candidate's selection of issues and stated plans for dealing with those issues are treated in both substantive and theoretical accounts of campaigns as forms of political currency employed to attract electoral support. To elaborate, consider the interpretation of electoral competition as an n-person coalition game (see, for example, Riker and Ordeshook, 1973). In this scheme citizens or distinct groups of citizens are assumed to be utility maximizing players and candidates are seen as inter- mediaries who, by offering alternative agendas or platforms, seek to form a majority coalition of citizen-players. This perspective reveals that the electoral strategies of presidential candidates are not grounded exclusively upon universal appeals to all citizens 78 but include focused efforts to solicit the votes of particular subsets of the electorate in the hopes of piecing together a winning coalition. Victory in both the primary and general election phases is achieved by fashioning a set of proposed actions and programs which appeal to specific issue publics as well as specific ethnic and demographic subsets of the citizenry. As such, a president's electoral majority consists of a number of minorities or factions whose loyalty to the president is based upon specific elements of the campaign agenda. It is in such fashion that the act of competing for office compels presidential aspirants to address and propose remedies for problems of import to the public; this, in turn, leads the public to expect certain behaviors from the president-elect. The winning candidate assumes office with a number of specific commitments which are, fundamentally, promises to perform. It is in this sense that a proposed agenda for action is transformed, after inauguration, into a standard against which the success or effectiveness of an administration can be judged. Unlike the institutionally based demands of peace, prosperity, and integrity, such electoral expectations are specific to a given administration. This consideration of electoral competition thus emphasizes that presidents will be evaluated according to criteria over which they themselves exercise control. The strategies formulated by candidates to capture the presidency - choices regarding issue emphasis and proposed actions -- will thus determine, in part, how the subsequent performance of the successful candidate is evaluated. 79 Situational Demands Thus far two sets of demands which operate as criteria employed by citizens to evaluate the president have been identified - a set of concerns which, because of their special import, are imposed upon the office of the president and a variable set of expectations produced by electoral competition for this office. What remains to be considered are the set of outcomes and actions which are generated by events, crises, and their specific domestic and international contexts. The third set of expectations which confront a president can be labeled situational or episodic. Over the course of a term, there are specific domestic and international events which cannot be anticipated a priori. What can be anticipated, however, is that, such events will occur and that the public will expect a presiden- tial response. Indeed, as Neustadt (1976:5) has observed, the modern president has been increasingly confronted by emergencies in decision making despite the backdrop of politics as usual. The notion of a situational set of criteria is designed to capture the impact of the emergencies and unanticipated events which arise during the tenure of every president. The impact of events and crises on the level of public support will depend upon individual assessments of the outcome as well as the extent to which presidential actions are perceived by the public to be consistent with expectations of how a national leader ought to perform in such circumstances. That is to say, it is assumed that there exist certain standards of behavior which citizens presume will be met as the president reacts to events and crises. 80 What considerations might form the substance of these under- lying standards? During periods of crisis, it seems reasonable that citizens will expect the response formulated by the president to be cautious but firm. With respect to events which generate changes in the nation's foreign or domestic policy, it is antici— pated that the actions initiated by the president will serve the best interests of the nation. Note, however, that the fundamental character of such expectations is that seldom will a public consensus exist over what constitutes a "firm response" or the "best interests of the nation." For example, although pleasing to the liberal wing of the Democratic Party, Richard Nixon's overtures to both the Chinese and Soviets also aroused the ire of staunch anti-communists within the president's own Republican Party. Further, there were many who considered the actions of Lyndon Johnson during the Dominican Republic crisis to be neither cautious nor well-reasoned but an exaggerated response designed to reap political benefits at home. The fact that some individuals do disagree with the responses selected by presidents as well as the difference in the proportion of such disapprovers as we move from episode to episode appears then to imply that events (i.e. the components of the situational set of outcomes and actions) will not exert a uniform and positive impact upon assessments of presidential performance. This is not to deny that the balance of public opinion typically favors the president's actions, but to emphasize that the influence of the situational criteria ultimately rests upon each citizen's assess— ment of the outcome and the president's behavior. 81 To summarize, I have argued that presidents will be evaluated according to three sets of outcomes and actions which occur over the course of an administration. Because of the import of these criteria, citizens can be expected to have well—defined preferences over all values which a given zn or pm may assume. Accordingly, let i i 1._ 1 1 1 1 X = {x1, ... , xm, ... , xM, xM+1, ... , xM+n’ ... , XM+N} (3.3) summarize citizen i's most preferred state on each offithe M action and N outcome criteria. Further, define f(Xi) as the distribution of preferred states across the citizenry. For the purposes of argument, it is assumed that such preferences remain constant over the period of an administration; that is, 1 1 1 1 xo=x1=...=xt=...=x48 (3.4) f(X:) -—- 101.1) = = fcxi) = = 150128) (3.5) The Criteria of Citizen Choice: A Spatial Representation The model of citizen decision making developed in this chapter rests upon three additional premises. The first is that citizens have or will be induced to form expectations about presidential performance. Next, it is assumed that assessments of how a president has handled or managed problems associated with a specific performance dimension evolve from a comparison of these expectations with the actual outcomes and actions relevant to that management dimension. The final premise asserts that the choice to approve or 82 disapprove of the president's overall performance - the second stage of the decision process - is a weighted function of the judgements made on the specific performance dimensions. To represent such expectations and comparisons, several concepts typically associated with the spatial model of electoral competition - a multidimensional space, locations in that space, and the distance between spatial locations - will be utilized (see, for example, Downs, 1957; Davis, Hinich, and Ordeshook, 1970; Riker and Ordeshook, 1973; Ordeshook, 1977). The use of these concepts implies that the range of alternative values on each of the M action and N outcome criteria can be arrayed on an underlying continuum and that a specific action or outcome can be represented as a point on that continuum. For example, the prevailing rate of unemployment can be located on a continuum which ranges from zero to 100%. With respect to presidential actions, the continuum may be defined with respect to ideological or programmatic considerations. To illustrate, consider the criterion of action to combat unem- ployment. One standard against which potential actions can be compared pertains to the degree of economic intervention that is associated with the presidential action. In this instance, the underlying scale would range from a hands-off, laissez faire point to such activist options 3 a government-sponsored jobs program. Another example is supplied by the pursuit of the SALT II agree— ment during the Carter Administration. It is reasonable to presume that the assessment of the presidential actions associated with this treaty (e.g. the meeting with Brezhnev in Vienna, the national 83 publicity campaign sponsored by the administration) were based upon beliefs regarding how the United States ought to conduct its relations with the Soviets. Such beliefs were likely to vary from a preference for detente and cordial relations to a hardline position which disdains any diplomatic approach as misguided. More generally, it is assumed that there exists some underlying standard or metric by which a particular presidential action can be measured. Together, the (M+N) continua define a multidimensional space (hereafter referred to as the action-outcome space) and specific values of these criteria are said to specify a location or position in that space. For example, the definition of X1 in Equation 3.3 denotes the location or ideal point of citizen 1 in this action- outcome space. It must be emphasized that the distinction between actions and outcomes is employed to investigate the manner by which presidents may attempt to manage or regulate public support. The distinction is not meant to imply that there are two distinct spaces. However, the treatment of the PRO and PRA sets as components of a single space requires that the definition of these vectors be modified for the purpose of formal precision. It is assumed, therefore, that the relevant action-outcome space is of dimension M+N and can be represented by a single vector, 0, in the following manner 3 O n+1’ooo, e e } 9: 9 T ‘ {1w 9 9 my“ M+N (3.6) m M’ ={Pla-o': pmavoog PM, 21,...,Zn,...,ZN} 84 Thus, the first M entries of 9 denote the presidentially-relevant actions and the next N elements represent presidentially—relevant outcomes. Given this representation, the formal definitions of the P and Z vectors are given by: P = {p1, ... , pm, ... , pM, o, ... , 0} Zn, ... , ZN} such that e = P + 2. Depending upon the degree of specificity required, both the general and the (P,Z) representation of the evaluative criteria will be employed in the following discussion. Throughout, it is understood that the last N elements of P and the first M elements of Z are zero entries. Having characterized the criteria which underlie the support decision and citizen preferences over these criteria, what remains to be defined is the third component of citizen decision making —- expectations of presidential performance. Let us first consider the role of election campaigns as a series of events which provides the public with an opportunity to form an assessment of how candi- dates can be expected to perform on the institutional, electoral, and situational dimensions. First and as noted in the discussion of the institutional criteria, candidates are generally compelled to address the pervasive concerns of peace, prosperity, and inte- grity. Certainly Richard Nixon's 1968 allusion to a plan for peace in Vietnam, Jimmy Carter's emphasis on the themes of honesty and integrity in 1976, and Ronald Reagan's exploitation of economic discontent in 1980 provide ample illustrations of this point. 85 Second, the discussion of electorally-generated criteria implies that performance expectations are formed as a result of the attention devoted by candidates to other problems and proposed remedies in the hopes of constructing an electoral majority. Finally, election campaigns provide voters with an opportunity to reach a judgement regarding how the candidates are likely to respond to crises and rapidly changing domestic and international situations. Indeed, negative public reaction toward the candi- dacies of Barry Goldwater in 1964 and George McGovern in 1972 were based, in part, upon the perception that neither of these individuals would respond to such concerns in an appropriate manner. For these reasons, it is assumed that citizens form estimates of the expected criteria. Given the definitions of the PRO and PRA sets, let * * * 9: * * * e ={91, ooo, 9m, ooo, GM, GM+1’ ooo, GM, ooo, GM+N} * * * * P ={P1, ooo, pm, ooo, pM, O, ooo’ 0} (3'7) * * * * Z ={0, ..., 0, 21, ..., zn, ..., zN } designate the expected location of the president in the action— outcome space.2 Together, P* and 2* constitute the public's perception of presidential promise - the expectation that certain outcomes will be achieved (or maintained) through the president's initiation and execution of particular actions. As such, 9* summarizes how the president will be expected to respond to demands directed toward the presidency itself, to problems discussed by the president during the campaign, and to the crises which arise 86 * during the term. It is in this sense that O operates as a standard of evaluation when the president assumes office. This is depicted in Figure 1.1 by the lines connecting the institutional, electoral, and situational criteria to public expectations. It must also be recognized that campaigns for the presidency are conducted against the backdrop of a status quo defined by prevailing outcomes and those actions previously executed or currently pursued by the incumbent administration. That is to say, campaigns for the presidency can be seen as a process wherein presidential aspirants promise, in effect, to move the status quo to a different and presumably more acceptable location in the outcome-action space. Now define this status quo point as: 0° = {e?, ..., 9° 0 O m, ..., 6M. eM+1, ... , e§+n, ... , e§+N } p° . {p§, ..., ;, ..., pg, 0, ... , o} (3.8) O z 2 {0, ooo, O, 21, ooo, Zn, ooo, ZN} where Z0 describes those outcomes which prevail at the time of the presidential campaign and Po represents the set of actions which characterize the behavior of the incumbent administration. In a similar fashion, the current performance of the admin- istration can be depicted as: G - A t {0... ---.em., ---.%%I,..-wem+n,t » ---.GM+N,. } Pt “{plt, ooo’ pmt, ooo, th, O, ooo’ 0} 2 é{0, ..., n, z (3.9) t It, ..., znt’ ooo, th} 87 where Pt and Zt respectively denote the presidentially—relevant actions and outcomes which are observed at time t. This implies that the current performance of the president is associated with the adoption of a location in the multidimensional action-outcome space. With respect to presidential actions, it is assumed that prominent presidential decisions are associated with the adoption of a position on one or more of the action dimensions P. The vector Zt simply records the values of the PROS which are observed at time t. With these definitions, the nature of the citizen comparison between promise and performance can be stated as follows. During the campaign for office, the president will be associated with a position in the action-outcome space, 9*. At the time of his inauguration, however, the government rests at the status quo point defined by 9° and, in general, it is likely that 9* f 9°. Therefore, from the perspective of managing public support, the problem confronting the president is to move the status quo point toward the location defined by the set of public expectations. At each time point then, citizens are assumed to evaluate the new presidential location (St) in comparison to these expectations. The question of concern at this juncture is the calculus by which the effort to reach.6* is evaluated. To investigate this question, the definitions of the status Iquo, presidential promise, and presidential performance will be employed to construct a model of citizen support decisions and to formulate an aggregate level representation of the presidential Support function. The logic of the decision model will be illustrated 88 by reference to the spatial analogy. That is, the support decision will be modelled as a function of the distance whidh separates the citizen from the expected and actual locations of the president in the multidimensional action-outcome space. The Formation of the Initial Support Coalition Let us first consider citizen decisions to join the president's support coalition at the outset of the term and the resulting initial distribution of public support (labeled So). This decision is deserving of attention for, as illustrated in Figure 1.1, it establishes the initial resource base for the president and thus represents the starting point of a dynamic, interactive process of decision making. Table 3.1 presents this initial distribution for both the public as a whole and along several political and socio- economic dimensions for the administrations from Eisenhower to Carter. There are three noteworthy features in this table. First, the proportion of the public who support the president at the outset of the term has been relatively constant across adminis- trations. With the exception of the Nixon terms, the initial level of support falls between 68% and 722. (The initial level of support for Franklin Roosevelt in January 1941 was 712 and 60% for Harry Truman in 1949.) Second, it can be seen that an essential characteristic of the initial support coalition is its heterogeneous character. Presidents begin their terms with sizable reservoirs of support among each of the subsets of the citizenry. Indeed, it is a rare 89 instance when the chief executive does not receive the support of a majority of citizens who fall into each demographic and political classification. Third, the univariate breakdowns presented in Table 3.1 also reveal that the proportion of each group who initially approve of the president's performance varies both within and across presi- dential administrations. According to the discussion of the elector- ally-generated expectations, the fact that presidents attract larger proportions of support from some groups than from others is unlikely to be accidental. That is to say, the presence of group differences in the level of initial support may be interpreted as a reflection of the particular electoral strategy followed by the president wherein the loyalty of political, economic, and geographic groups was actively cultivated through proposals and pledges designed to acknowledge the problems and meet the specific demands of these groups. Additional information about the character of So can be obtained from Table 3.2 which offers a comparison between the pro- portion of the Gallup sample who reported voting for the president and the proportion who initially supported the president. The gain/ loss quantity presented in the table is calculated as the difference between initial and voting support for each of the relevant subsets. Although the electoral results are not adjusted for turnout (as in the analysis of Axelrod, 1972), this comparison provides one opportunity to examine how the president's electoral coalition is augmented to produce the initial support coalition. Inspection of these data reveal that the largest increments in job support 90 occur among those subsets who were the least supportive in the electoral sense. This is particularly clear on the partisan dimension.where the largest gains for the president are among the out-party identifiers. This differential gain is also evidenced, albeit in less consistent fashion, on the education, occupation, and regional variables. The characterization of citizen decision making incorporated in the Spatial model can be employed to help account for these features of the initial support coalition. According to this model, voting behavior is assumed to depend upon the relative proximity between the citizen and the candidates. In other words, the partisan division of the vote as well as turnout are cast as functions of the distance which separates citizen ideal points from the perceived spatial positions of the candidates. As such, the differential assessment of candidate locations induces some citizens to vote for the Democratic candidate, others to vote for the Republican competitor, and still others to abstain due to indifference or alienation (see, Riker and Ordeshook, 1973:323-24). NOw’what occurs when this distance or proximity argument is applied to the initial support decision? It may be the case that the choice to join So is based simply upon the spatial position of the citizen via-advis the president-elect. To illustrate, let 3: denote the initial support decision facing citizen 1 such that: 1 if citizen i approves of presidential performance 31 - , (3.10) o (‘0 if citizen 1 disapproves of presidential performance 91 According to the proximity or distance thesis, this decision can be depicted as: 8: ' fIL1(9*)J = f[(x1-e*)'(xi-e*)]15 (3.11) where L1 (9*) denotes the loss of citizen 1 and represents the value or benefit associated with the president's expected performance on the institutional, electoral, and situational criteria (for an elaboration of the loss concept, see the discussion associated with expressions 3.13-3.16 below). The initial level of public support for the president can therefore be written as: N - ( z s: )/N (3.12) 1-1 So where N is the number of citizens. Because 6* is, in part, generated by the electoral strategies of the candidates, the proximity argument would also lead us to expect that the assessment of 9* and thus the initial level of support will vary among the demographic and political subsets of the public. To account for the vote-support differential, it should be noted that two sources of abstention -- indifference (Riker and Ordeshook, 1973:323-24) and satisfaction (Fiorina and Weisberg, 1977) - suggest that So should include some proportion of abstainers. Thus, the use of the proximity argument leads to the interesting conjecture that the president's initial support coalition may be comprised of four distinct groups of individuals: (1) those citizens 92 who voted for the president, (2) those citizens who abstained because of indifference, (3) those individuals who abstained because of satisfaction with the likely performance of both candidates, and (4) those citizens who voted for the opposition candidate on the basis of a comparative evaluation that only marginally favored that candidate over the president. This last group may help to account for the gain in support among such sets as out-party identifiers. This manner of coalescence appears to be quite general. That is to say, the four groups, while varying in size from election to election, may sum to a relatively constant proportion of the population. As such, the assumption that the initial decision to support the president is a function of a citizen's assessment of the president's expected performance in tandem with a consideration of electoral abstention not only provides a plausible account of the formation of So but also suggest that, ceteris paribus, the size of 80 may be relatively constant across presidential admin- istrations. A Model of the Citizen Support Decision Having assumed that the initial distribution of support is based upon expectations of presidential performance, our attention turns to how this initial distribution is altered in subsequent time periods. It is assumed that the individual support decision is, first of all, a function of the discrepancy between presidential promise and performance. That is to say, it is plausible to 93 presume that the evaluation of presidential performance depends upon two factors: (1) the assessment of current actions and outcomes and (2) how this assessment compares to the president's expected performance. I With respect to the assessment of current performance, it is assumed that citizens evaluate the elements of the presidential performance vector in light of their own preferences. This implies a comparison between X1 and at. In other words, it is assumed that citizens compare current outcomes such as unemployment and inflation to the values which they believe ought to prevail. The assessment of presidential actions will depend upon the specific standards or continua which are activated by the presi- dent's behavior. This, however, does not undermine the assertion that citizens assess presidential behavior in light of their beliefs of what actions should be initiated by the chief executive. Such comparisons can be represented by the distance which separates the citizen from the president in the multidimensional action-outcome space. Here, a Euclidean metric will be employed suCh that the distance between any two points in a k-dimensional coordinate system is given by the formula: d (A,B) = [ : (ak—bk)21If (3.13) or, in matrix notation d (A,B) = [(A-B)'(A-B)];§ 94 where d(A,B) = the distance between points A and B A,B = k-element column vectors with entries ak and bk In tandem with this distance measure, the concept of loss is employed as the standard underlying citizen assessments. Therefore, it is assumed that the loss or valuation which citizen 1 associates with current actions and outcomes is a function of the distance between X1 and 9t. This loss function can be written as: __ 1 , 1 11 Li(®t) - [(X -6t) (x -®t)] (3.15) where Li(et) denotes the loss of citizen 1 based upon the presi- dentially-relevant actions and outcomes observed at time t. Next, it is necessary to formulate an expression for the expected location of the president in the action-outcome space. To accomplish this task, define the loss citizen 1 associates with the expected performace of the president as: * * * Li(0 ) = [(Xi-O )‘(Xi-O )]L2 (3.16) Both L1(e*) and Li(9t) can be interpreted as measures of the satis— faction or benefit derived from presidential locations. For instance, the loss function Li(9*) represents a statement of the satisfaction which individual i expects to obtain as a consequence of the president's electoral triumph. Observe also that loss is minimized when the citizen's ideal point and the president's 95 location coincide. Hence, these formulations yield a satisfactory representation for citizen valuation since they capture the intuitive notion that expected or actual presidential performance should become less favorable as outcomes and actions diverge from those most preferred by the citizen. While Li(e*) represents the valuation of the president's campaign platform, it offers no information about performance expectations or expected loss as the term progresses; as a result, there is no indication about how presidential attempts to alter the status quo will be assessed. However, consider the situation wherein the president successfully moves 0° to 6*. In this circum- stance, the loss of citizen 1 would be equal to Li(0*) and the change in loss experienced by the citizen would be given by the quantity, Li(9°) - L1(9*) . At issue then is what type of progress in the movement of 9° toward 9* is expected or, equiva- lently, the manner in which expected loss, denoted by EL1(Ot), varies over the course of a 48 month term. One formulation is given by *+ 48 ° * EL1(9t) 11(9 ) 11-(t/ )1 [L1(e ) - L,(e )1 (3.17) or, upon algebraic manipulation, o o * EL1(9t) = Li(9 ) - t/481LiCO ) - L1G) )1 c=1,2,... .48 96 The expression implies that the difference in loss associated with the status quo and presidential promise will be reduced in uniform fashion - (1/48)th of the distance between eoand 9* per month - over the course of the term. In other words, it is assumed that the public expects a pattern of steady and regular movement in at from 9° toward 6* (i.e. the change in expected loss is equal for all t). Given this characterization of public expectations, the comparison of expected and actual performance in month t can be written as: 2 2 2 _ i_ _ i- o - _ i o _ Li(Qt)—EL1(Ot)- [IX atll (llX 9 II t/48IIIX -@ ll 2 llx'-®*1l 1) (3.18) where IIX-Gllzirepresents the shortened expression for the Euclidean distance measure. The logic underlying the development of Expression 3.18 also suggests that the aggregate level of public support will be determined, in part, by the distance which separates the president's current location (9t) from the expected location at time t. Applying the same argument used to define the expected loss of individual i, this expected location can be written as: 8 met) = Il-(t/48)][(e°- $)'(e°- (3‘)) (3.19) so that the aggregate level of presidential support can ultimately be represented as a function of the distance between at and E';)1 - 2 xj (Li(et_j)-ELi(et_j)1)z o In effect, the cumulative term operates as a reservoir in the sense that it summarizes the accumulated grievances of good will created by past outcomes and administration actions. For example, if prior administration performance has exceeded expectations, a departure from EL1(9t) will be offset on the basis of the commendable cumu- lative performance. Similarly, favorable outcomes and actions at time t may be insufficient to overcome a poor performance record prior to time t. Because it incorporates the notion of an inter- action or tradeoff between short and long run performance assess- ments, this characterization of individual behavior is more general than many of the descriptions of citizen decision making found in prior work on public support. The decision rule specified in Equation 3.25 implies that the aggregate level evaluation of presidential performance on dimension k can be written as: 105 N . Hkt = ( '2 h1 )/N (3.26) where Hkt is the proportion of the public who approve of the manner in which the president has handled problems associated with dimension k. In aggregate terms, the proportion can be written as a function of expected and actual performance in the following manner: k t )1' kk,k 1/2)-jk_k f (([Gt‘E(et>] [at-E(e )1 z 1 ([et_j E(et_ j k k 1/2 t_j-E(ot_j)] )) (3.27) Hkt= [O Recalling that 3: denotes the overall performance evaluation or support decision of citizen i at time t and that s: is equal to one (zero) if the individual approves (disapproves) of the president's performance, it is assumed that the choice to approve or disapprove of the president's performance is a weighted function of the specific performance assessments; that is to say, . K . 1 1 s = f ( X h ) (3.28) t k=1 1kt kt where w denotes the weight or import which citizen i associates ikt with performance dimension k at time t. The aggregate level of public support can then be written as: N N K 1 (3.29) 5"“25:)=f[i:1(k:1“1kthkt) m 106 or, by letting Wit denote the approximate salience of performance dimension k at the aggregate level, as K = 2 St k=1 Wkt H'kt (3° 30) To summarize, the discussion of choice criteria, loss functions, and decision rules in this section provides a complete and plausible model of citizen decisions to support or oppose the president over the course of a four year term. This decision is modelled as a function of the distance which separates the citizen from the expected and actual location of the president in a multidimensional action- outcome space and thus implies that presidential choices of actions will determine, in part, changes in the distribution of public support. That is to say, the choice of actions maintains the capacity to alienate current supporters as well as attract individuals who presently disapprove of the president's conduct of office. Presidential Decision Making According to Figure 1.1 and the decision model formulated in the previous section, the primary levers of control for the management of public support reside in the president's choice of actions. Now from the president's perspective, the set of possible actions may be divided into two subsets. The first consists of those actions which involve unconstrained exercises of presidential authority. Such actions are characterized by the fact that execution 'of the president's decision will not depend upon the consent of 107 other decision makers within the national government. As Figure 1.1 illustrates, such choices will be labeled unilateral presidential actions. Prominent examples of such unconstrained exercises of presidential authority include Lyndon Johnson's 1965 decision to bomb the North Vietnamese oil dumps in Hanoi and Haiphong, Richard Nixon's removal of two top Justice Department officials during the "Saturday Night Massacre" in October 1973, and the ill-fated rescue mission attempted by the Carter Administration in April, 1980. The second subset includes those presidential actions whose execution is contingent upon the assent of other institutions and decision makers, particularly the United States Congress. This subset, denoted in Figure 1.1 by joint actions, will include such activities as the introduction of legislative proposals by the president, the submission of treaties and cabinet-level nominees to the Senate, and various presidential decisions (e.g. reorganization plans, troop commitments, executive agreements) over which the Congress may exercise a legislative veto (on the legislative veto, see Dodd and Schott, 1980; Fisher, 1978). With respect to presidential decision making, the set of possible behaviors on each of the M action criteria constitutes the choice set. That is to say, in each time period presidents are faced with the decision of what actions to initiate or execute on each of the M action criteria. Phrased in terms of the spatial representation developed in the previous section, presidents must select or adopt a location in the action portion of the multi- dimensional action-outcome space. 108 Examination of Figure 1.1 also illustrates that such decisions exert a dual impact on the criteria employed by citizens to evaluate presidential performance. First of all, such actions will, in principle, influence outcomes. In the case of unilateral or unconstrained exercises of presidential authority, such influence will be direct; for actions whose execution requires congressional assent, the impact upon outcomes is of a second-order variety and depends upon the relationship between presidentially-relevant out— comes and the policy instruments whose values are set by congressional legislation. Second, the choice of actions, regardless of their ultimate impact on outcomes, will influence citizen assessments of how well the president is handling the set of problems in question (i.e. the hit)' Given the hypothesized impact of such actions, the objective in this portion of the chapter is to formulate a calculus posited to underlie the choice of these actions. To summarize the context of such choice, the president enters office with, from the public's perspective, the expectation that the prevailing set of outcomes and actions (O°=P°+Z°) will be moved toward an alternative set (0* = P*+Z*). It has also been argued that public preferences toward 0* underlie the initial distribution of public support so that, hypothetically, if public expectations were perfectly satisfied in the sense implied by Equations 3.25 or 3.27, this initial level of support will be preserved. It is not reasonable to assume that the president's choice of actions are governed solely by a concern with regulating the level of public support however. Public support must be regarded not ‘ as an end in itself but rather as an instrument or resource to be 109 utilized in the pursuit of other administration objectives. In other words, the choice of actions by the president will also be influenced by his own preferences and ideology. Moreover, presiden— tial attempts to influence outcomes will, in the case of jointly dependent actions, depend upon the satisfaction of congressional preferences and by the complex set of relationships between the PRO set and congressional legislation. Thus, as denoted in Figure 1.1, there are several influences which impinge upon the choice of presidential actions - the desire to manage support via the satis— faction of public expectations, the president's own goals and objectives, and the necessity of cultivating congressional support. As such, the task in this section is to represent how these consider- ations impinge upon presidential decision making, the value accorded to each, and the rules according to which actions are chosen. Presidential Goals The motivations which influence the decision of individuals to compete for the presidency and the choices that are made once the office is secured are obviously numerous and complex. Undoubtedly, the assumptions made about office seekers in general —- power, limelight, ego—satisfaction -- are applicable. In other words, it can be expected that presidents are motivated by a desire to be recognized as a strong leader and competent chief executive, a desire to play a prominent role in the affairs of the nation, and a desire to be accorded a respected place in the history of the ‘country. 110 It is also plausible to assume that presidents have strong beliefs regarding how such desires are to be achieved. Such beliefs will include preferences regarding what goals are most beneficial to society and therefore ought to be pursued, what specific outcomes are important to the realization of such goals, and what means are necessary for the regulation of these outcomes. Indeed, as intimated by Barber (1978) in his discussion of a president's . world view, individuals enter the presidency with a philosophy that specifies what ought to be done and a theory of action explaining how to implement this philosophy. For these reasons, it is assumed that presidents are motivated by a desire to achieve particular outcomes through the pursuit and execution of particular actions. Accordingly, it is also assumed that presidents have a complete preference order over all possible values of the PROs and PRAs that is based upon their philosophy and theory of action. In a fashion similar to the definition of the promise and performance concepts, let the following vectors define the set of presidential goals: ** { ** 9** O** 0** ** } e 91 ’ ... , m , ... , m1, ... , 1" , ... ’ OWN ** as ** ** P 8 {pl ’ ooo ’ Pm , ooo , pM ’ 0, ooo , O} (3.31) ** ** ** ** Z = {0, ... , 0, 21 , ... , zn , ... , ZN } This definition implies that a set of prevailing actions and goutcomes, at, will be preferred by the president to some alternative 111 1 set, Ot’ if and only if St is more proximate to the presidential ** ideal point, 0 . A Calculus of Presidential Action Although the set of presidential goals includes both actions and outcomes, it must be emphasized that only the choice of actions is under the immediate control of the president. This does not imply that presidents lack influence over outcomes; rather, the argument developed in this chapter asserts that presidents cannot straight- forwardly manipulate the values of the PROs. Neither can they unilaterally determine their overall location in the action-outcome space. Rather, it is argued that presidents attempt, via their choices of joint and unilateral actions, to influence the path of the PROs in a desired direction. My objective in this section then is to illustrate how the personal preferences of the president together with the desire to cultivate congressional and public support determine the president's location in the action portion of the multidimensional space. Now to represent the valuation of alternative actions, define L + b22(fiz‘ 92) + b12(fi1’ C2) 3 A * 2 . 2 % Iw-rlu Hr-og (3.468) where c1 and c2 denote the prevailing sentiments of Congress on action dimensions 1 and 2 and the remainder of the terms conform to definitions - offered previously. 125 An attempt to solve for the optimal values of pland p2 by setting the partial derivatives in Equations 3.46A and 3.46B equal to zero produces expressions which are fourth—degree polynomials in p1 and p2. Rather than attempting to manipulate such complex expressions, the presidential loss function specified in Equation 3.45 will be simplified by eliminating the square root function. That is to say, it is assumed that L(Pt) = ( [Pt—E(Pt)]'A[Pt-E(Pt)] + [Pt—CJ'B[Pt-C] ) (3.47) Thus, the valuation which the president associates with any location in the action space is a function of the weighted and squared discrepancies between his current location in the action space and (a) the location expected by the public as well as (b) the location of Congress. Although this simplification is introduced solely for the purposes of mathematical tractability and ease of exposition, the reformulation of the objective function does not alter the substantive character of the decision problem confronting the president -- the necessity of satisfying contradictory public and congressional expectations.7 * A A A Recalling that E(P = P and letting P = (p1, p2) denote the 48) optimal presidential locations on action dimensions 1 and 2 at t—48, the partial derivatives of L(P) are: aL(§)/ah a 2 (A *) + A * A A p1 811 p1 ' p1 2a12(1’2 ' 92) + 2b11(P1' c1) + 2b12(P2‘ C2) ’ ‘ "(3.48A) 8L(P)/8A = 2 A — * A _ * A A p2 a22(p2 p2) + 2a12(P1 P1) + 2b22(P2‘ c2) + 2b12(91‘ C1) (3.483) 126 A Setting each of these partials equal to zero and solving for p1 in 3.48A and for p2 in 3.48B produces the following two simultaneous equations in p1 and p2: x * * A A p1 = a11P1 + b11c1 + 812(p2' P2) + b12(°2‘ P2) (3.49A) 811 + b11 *+b * " A * a22P2 22°2 + 312(p1' P1) + b12(°1‘ P1) (3 49B) p2 = a + b 22 22 A solution for p1 can be obtained by substituting Equation 3.49B for p2 in 3.49A; similarly, p2 is calculated by replacing p1 in 3.49B with 3.49A. Such algebraic manipulations produce the following A solutions: P1 3 (822+ b22"811p1+b11c 1) (312+b12)(a12P1+b 12 c1) + (a12b 22 b12 a22)(P2 c2> (311+b11)(822+b22) ' (""12"'°12)2 (3.50A) A P2 = * * (811+b11)(322P2+b22 2) (812+b12)(a1292+b12° 2) + (a12b11a11b12)(91 C1) (311+b11)(322+b22) ’ (a12+"’12)2 (3.50B) Note that if the president reaches the judgement that Congress and the public regard the issues as mutually independent (i.e. 312=b12'0)’ the expressions for the solutions reduce to: * * _ a11p1 + b11°1 p1 - a + b (3.51A) 11 11 . * + b p2 = 82292 22°2 (3.513) + b 127 Moreover, if it is believed that Congress and the public weigh the issues identically (i.e. AFB), then the optimal positions can be written as: p1 = P1 + C1 (3.52A) 2 p2 = p2 c2 (3.523) 2 To examine the character of this solution, consider the two— dimensional example where P*=(5,5), Po-(20,20), C=(30,15), and AFB such that the diagonal entries of the matrices are unity and the off-diagonal cells are equal to zero. According to Equations 3.52A and 3.52B, the optimal vlaues at t=48 are: p1 . 17.5 (Example 1) A p2 ' 10.0 This implies that the optimal trajectory, as illustrated in Figure 3.2, lies along the line segment connecting P°=(20,20) and P=(17.5,10). It should be noted that the best location available to the president at any given game point t can be obtained by calculating Eo 138 sa RN ol No o mm ol so on On ol mo 0 oo HI mm Nu on N oo o1 no HI on ol oo H so an co m us on mm m1 on Aooo>uomo om mnlumaa soxaz mucooumoum smusaosdom MA mo um Om on as oo mm no so mo Hm oo no mo mm no No ouo> so so 3N am so me ma No «N om as me w an mm mm me No NA we as mm on an as we ma an mm as s. cm ma an Aooosuomo om ao1wean aoxwz NH ms mm as as on us ms mm us on mm ms sm ss NH ow ss ouo> XX NXXX XX Aooo>uomo mmlomma uosozswmsm xx NXXK XX ao No no no om on on as no um am on oo on no mo on on oo On ma om mm ouo> am oH MA Amuo>uomv mmINmaH uoaossomfim no On mo am so mo sm oo on so mo om am so ss mo on as mm mm ms oo so as no oo no mm mm mm ouo> "msoaufiamoo uuoaasm HmHuHcH ocm Homewooam mo consumqaoo < N.m pansy ouaoslcoz «use: ussoeomo acmumououm has: nusom “moses: ummm umHHoo moan umHHoo moans Hmcosmoomoum Hoocom compo Hooeum swam owoaaou nucoocoaoosH muouooson osmosaoneom Annoy 138A NI MN NH mm on oH oN HN nH oN sn HH NH mm NM #1 ms NN x4u3>10mc on HR s5 H5 H5 oh sh Nu mm as on mo Hm on oh Ho on Nu om sulohaH mousse Oufiflflfimflhm ufiHUOEQG no os so os os so as Hm on on Ns on so «s on No m on ouo> K K n so Ho oo T1V1 no No an H as Hason NI oo s so mH so NI so 5 mo NH so o no oHI ms Hm Hm o no Amuosuomv om molsoaH concoon "ocOHuHHooo uuoedsm HoHuH=H was Honouoon mo somHuoesoo s Aeosasuaouv ~.n «Home so an on on Co an Ho mo H5 no so oo no No on no ON Ho ouo> sN sH on HN K N «N c on a.HN ms no sh sn )4 N no so on NB Huuo>aomv om HotoomH homecox mo as on on as Hm as no co os Ns on no on ms so H.on ouo> oanslsoz ouH:3 usaoeuao acouoououm sum: eusom nausea: swam uaHHoo «use Hudson muse: HgOfiQDOHOHN Hoocom oomuo Hooeum ems: oonHoo cocoonoeoocH muouooaon mcooHHoonom House 139 Table 3.3 The Impact of Differential Public and Congressional Salience Weights on the Optimal Presidential Location: Example 1 811 = 822 Example 4 b11 = b22= = 2; a 1 22 Example 5 = b22= = 1; a 22 = b = b 11 = 1 = 2 22 An Example p1 p2 17.5 10.0 21.7 10.0 17.5 11.7 13.3 10.0 17.5 8.3 d(P,P*) 13.5 17.4 14.2 12.8 13.3 25 20 15 10 140 Figure 3.1 The Representation of Expected and Actual Presidential Positions in the Action-Outcome Space: An Example 0 1 . (10,20) 0 (20,20) )11 (1,8) 0* (5.5) 1) CD H 80 75 70.. 65 1 75- 70.. 65.. 141 Figure 3.2 A Comparison of Public Support During the Initial Six Months of the Presidential Term Kennedy,1961 \ ’ \ \ ' '\ I \ \ I \ \ / \\/ Johnson,1965 'r' Eisenhower,l953 \. Eisenhower,l957 \ \ '0 0" \ \ \ \ \ 'o \ ‘,..—"' O I I l fl I T I I TF—F I l r U r I 7‘ January February March April May June 75 7O 65 65 60 55 50 45 142 Figure 3.2 (cont'd.) Carter,1977 Nixon,1969 \ I \ Nixon,1973 \fl / \ \ \ l \. l \ 1 I \ ‘ \ \ \ ‘ \ \ l i l 1 l r t 1 I I I :--:--:--;—l January February March April May June 25 20 15 10 143 Figure 3.3 The Optimal Trajectory of Presidential Actions: An Example P° (20,20) P (17.5.10) * P (5,5) C (30,15) I Pl 20 15 10 144 Figure 3.4 The Impact of Varying Salience Assumptions on the Optimal Trajectory of Presidential Actions An Example P°=(20,2o) E4'(17-"»”-6) . o 523(17.1,13A3) C=(30,15) El-(17.5,10) . E3=(17c9,6.7) ' E5=(17.4,5.4) * P '(5,5) 25 15 10 145 Figure 3.5 The Impact of Differential Public and Congressional Salience Weights on the Optimal Presidential Trajectory An Example p°=(20,20) C=(30,15) E3=(17.5,11.7) E(P) . t o 0 0E =(2lc7,10) F 2 E4=(13.3,10) '1 O E5t(17.5,8.3) * P =(5,5) 11; p1 CHAPTER 4 THE MANAGEMENT OF PUBLIC SUPPORT Toward an Alternative Explanation of Public Support The previous chapter was designed to identify the criteria and characterize the valuations and decision rules which underlie public assessments of presidential performance and the president's choice of actions. The resulting models of citizen and presidential decision making are novel in several respects. First of all, the evaluation of presidential performance is characterized as a two- stage process. It was asserted that public assessments of presi— dentially-relevant outcomes and actions - the basic criteria of choice -- lead to judgements about the president's conduct on specific performance dimensions (the hk). Such judgements are then weighed by their perceived import to produce an overall or general performance evaluation. Second, the calculus of presidential decision making formulated in Chapter 3 departs from typical characterizations of government as a unitary actor pursuing the goal of reelection (see, for example, Nordhaus, 1975; MacRae, 1979). Instead, the president is cast as a decision maker motivated by a desire to achieve particular goals but constrained in this pursuit by the necessity of cultivating public and congressional support. The latter part of Chapter 3 illustrates how these potentially contradictory considerations impinge upon the substantive or ideological character of the president's actions. 146 147 The implications of this discussion are most easily discerned if we first consider the conventional wisdom arising from prior research on public support for American' presidents. In effect, the objective of this research is to identify determinants of public support and to estimate an aggregate-level presidential support function. Because each purports to offer a comprehensive explanation of how citizens evaluate presidential performance, the work of Mueller (1970; 1973), Stimson (1976) and Kernell (1978) constitute the core of the support literature. These explanations have two commonalities. First, the econo- metric models formulated by Mueller, Stimson, and Kernell share the same underlying structure. That is, each model partitions the hypothesized influences on public support in the following manner: Public 8 Constant . Dynamic Within Environmental Random Support Level Term Component Factors Error The actual determinants which appear in each model are summarized in Table 4.1. Although the authors differ with respect to the specific factors included in the models and how these factors are operationalized, the environmental term is designed to assess the influence of prominent outcomes generally believed to effect citizen evaluations of public officials - unemployment, inflation, wars, and scandals. The inclusion of the within-term dynamic component is based upon the premise that' there exists a driving force to presidential support which is akin to a cycle and responsible for the time path over the course of a four year term (for a more extensive review and critique of these specifications, see Ostrom and Simon, 1980). 148 A second commonality is that the empirical analyses presented in these works lead to the conclusion that public support is largely beyond the control of the president. Thus Mueller reasons that (1970:22; emphasis added): ...the concepts underlying the coalition of minorities and rally round the flag variables predict that the president's popularity will continually decline over time and that inter- national crises and similar events will explain short term bumps and wiggles in this otherwise inexorable descent. A similar conclusion is offered by Stimson (1976:10; emphasis added): The president, in this theory, is largely a passive observer of his downsliding popularity. Its causes lie more in the misperception of the electorate than in the president's actions. .25 would therefore occur no matter what he did. New presidents pick up the support of many who did not prefer them - much less vote for them - at election time. This easily acquired approval, based as it is on unrealistic expectations of the man or his policies, is then regularly lost over the early years of the term. In such a manner, these works result in a rather mechanistic explanation grounded upon the conclusion that public support is an inherently unmanageable commodity.1 The characterization of citizen and presidential decision making offered in Chapter 3 challenges not only the conclusion of presidential helplessness but also the research strategies and econometric models which produced this conclusion. The model of interaction between citizen and president rests upon the proposition that public support 149 is responsive to presidential behavior. It is argued in this thesis that salient presidential decisions - be they exercises of joint or unilateral authority —- will influence the specific performance evaluations of the public and, in turn, the overall level of public support. In tandem with the evidence regarding the value of public support, the discussion of Chapter 3 suggests an alternative theory of public support based upon the following premises: Proposition 1: Public support is amenable to control or management by the president. Proposition 2: The primary levers for managing public support consist of the actions, both joint and unilateral, initiated by the president. Proposition 3: Because of the value of public support, presidents will attempt to exercise control over public support. Proposition 4: The impact of such attempts will be most detectible in the specific performance assessments of the public. It must be emphasized that this explanation does not imply that presidents can regulate the level of public support with perfect precision; nor does it assert that erosion of support will not occur over the course of the term. This is because the connection between presidential or governmental action and the outcomes for which presidents are held responsible is uncertain. That is, public expectations toward presidentially-relevant outcomes may be mutually unrealizable; outcomes themselves may be jointly dependent and subject to influences beyond presidential control. Moreover and as illustrated in Chapter 3, presidential decisions are formulated in the face of tensions among presidential goals, public expectations, and congressional 150 preferences. It is quite conceivable then that actions likely to generate increases in public support may be rejected because they are inimical to the president's goals or judged to be infeasible in light of prevailing congressional sentiments. Nonetheless, this explanation does assert that the study of public support is simultaneously a study of presidential decision making. It demands, therefore, that the prevailing research agenda on public support be expanded to include a more explicit consider- ation of presidential behavior. More specifically, the hypothesized link between presidential and citizen decision making requires that attention be directed toward (1) how presidential actions operate as the levers of control, (2) how the impact of such actions can be coordinated in a support management strategy, and (3) how such strategies impinge upon the time path of public evaluations. These concerns serve as the focus of this chapter. In contrast to prior research which, in the operational sense, has ignored the role of the president, it is my contention that presidential behavior can and ought to be incorporated into the empirical analysis of public support. Consequently, this chapter is designed to show how attention to these issues leads to a number of testable propositions which, if supported empirically, would dissolve the image of presi- dential helplessness created by the conclusions of prior research. Presidential Actions as Levers of Control The discussion presented in Chapter 3 argues that assessments of presidential performance can be influenced by the president's 151 .choice of joint and unilateral actions and that the impact of these actions is two-fold. First, actions may effect those outcomes deemed important to the public. In other words, the execution of an action may move one or more of the presidentially-relevant outcomes toward or away from the states (2*) most valued by the public. Second, presidentially-relevant actions will exercise a more direct impact upon the specific performance assessments of the public (the Hk's). According to the model of the support decision developed in Chapter 3, citizens will evaluate a given presidential action in light of their expectations of what actions ought to occur. Along with assessments of presidentially-relevant outcomes, individual judgements regarding how well the president has performed on a particular dimension will depend upon the perceived discrepancy between expected and actual presidential behavior. Such comparisons will induce some citizens to approve of the action, others to disapprove, and some proportion to be indifferent Cor express no opinion). Stated in aggregate-level terms, the model of citizen decision making leads to the expectation that: Proposition 5: Assessments of the president's conduct on a given performance dimension (hkt) will vary with the proportion of the public who approve of the action(s) initiated by the president at time t. This proposition asserts that presidents are evaluated according to the substantive or ideological content of their actions. It is in this sense that the choice of actions constitutes one lever by which support can be managed. However, it must be recognized 152 that the choice of actions generates additional consequences which, given the model of citizen and presidential interaction developed in the previous chapter, are important to the study of support management. First of all, presidential actions can activate a particular criterion of evaluation which in previous time periods played little or no role in the assessment of presidential performance. A prominent example of issue activation which appears to have generated an increase in public support was Richard Nixon's choice to establish diplomatic relations with the People's Republic of China and his visit to mainland China in February 1972. Some circumstantial evidence of the beneficial impact of this initiative is provided by data from the Harris Poll (Louis Harris and Associates, Inc., 1973:5,8,9). During the period from 1969 to 1973, the Harris organization measured public assessments of the president's overall performance and his conduct on a number of specific performance dimensions.2 Now consider the data displayed in Table 4.2. These poll results suggest that the activation of the China issue was largely responsible for the increase in overall support observed between February and March 1972. Note that prior to the actual visit, a sizable proportion (19%) of the public expressed no opinion about Nixon's decision. The occurrence of the visit reduced this figure to 9% and also generated a 14 point increase in positive assessments of the China visit. The conclusion regarding the impact of the visit seems all the more reasonable given that assessments of Nixon's handling of such concerns as the economy, civil rights, and crime declined during this period. 153 This example illustrates how presidential actions maintain the capacity to activate a standard of evaluation in a manner which favorably influences public support. A second consequence of presidential actions pertains to their impact on the salience which the public associates with particular criteria of evaluation (the elements of the A matrices) and the specific performance dimensions (the weights, Wit)“ The potential for influencing salience arises from the fact that presidential actions serve to focus or direct public attention. Indeed, by virtue of their symbolic position as national leader and the continuous media coverage they receive, presidents have considerable latitude in determining the amount of publicity or exposure which a given action will receive. For example, the arrival and departure of Richard Nixon during the China trip in 1972 were arranged to occur at 6:30 p.m., Eastern time, so as to insure that these events would receive feature coverage on network news broadcasts. More generally, presidents can request nationwide television and radio time to announce actions, schedule press conferences and background briefings to underscore the import of the action, and dispatch administration spokespersons to publicize the action and rally public support. Rather than dismissing such activities as public relations gimmicks, it is argued that they operate as mechanisms which can be employed to increase awareness of a presidential action and to enhance the import of the performance dimension to which the action pertains. The ability of the president to direct public attention con— stitutes an additional lever of control. Recall from Chapter 3 that 154 the specific performance assessments of the public are determined in part by the import associated with presidentially-relevant outcomes and actions; further, the overall level of public support is assumed to be a function of the weight associated with each specific performance dimension. Thus it is hypothesized that: Proposition 6: The impact of a presidentially- relevant action upon the specific performance assessments of the president will vary with the salience of the action. Proposition 7: The impact of a specific performance evaluation ( t) upon the overall level of publ c support (S ) will vary with the import attached to that dimension. Support for these propositions would imply that presidents can influence their public standing by attempting to direct public attention toward those outcomes, actions, and performance dimensions on which the president is, in the spatial sense, more proximate to public expectations. It is in this sense that presidential actions and the manner in which they are orchestrated and publicized can be employed to heighten the import of particular concerns and to displace public attention from problems on which performance ratings are unfavorable. This is not to assert that salience is wholly determined by presidential behavior or that the import associated with performance dimensions or criteria of evaluation can be manipulated with ease. For one thing, it is likely that salience varies as a function of outcomes. To investigate this possibility, I have gathered available 155 sample responses to the Gallup organization's Most Important Problem question for the 1953-1980 period.3 A casual inspection of these data strongly suggests that public concerns do fluctuate in response to outcomes and environmental conditions. This is particularly clear when the institutional expectations of peace and prosperity are considered. For example, listed below are the average yearly proportions of Gallup respondents who cited Vietnam as the most important problem facing the country. These proportions were obtained by averaging the relevant polls adminis- tered in a given year. The data indicate that public concern over Vietnam grew and declined in rough proportion to the level of U.S. involvement in the war. Average Proportion of Number of Respondents Citing Vietnam 3222. Polls as Most Important 1965 5 28.0% 1966 50.5 1967 50.0 1968 52.0 1969 40.0 1970 27.0 1971 25.2 1972 28.2 Next, consider the relationship between public concern and economic outcomes depicted in Figure 4.1. The aggregate measure of public concern presented in the Figure is the average yearly 156 proportion of Gallup respondents who identified the economy (high cost of living, jobs, threat of recession for example) as the most important problem facing the country. The measure of economic outcomes displayed in the figure is the so—called Misery Index obtained by adding the annual inflation and unemployment rates.4 The graph illustrates quite clearly that public concern with the economy varies with economic outcomes; the simple product-moment correlation between the two measures is .87. Both of these examples suggest that the import attached to a dimension of presidential evaluation will fluctuate with the degree to which presidentially-relevant outcomes diverge from public expectations. I Nevertheless, the responses to the Gallup item also provide indications that public concern is sensitive to the agenda and actions of the president. Although the sporadic appearance of the Gallup question makes any systematic analysis difficult, circumstantial evidence regarding the impact of presidential actions on public concerns can be extracted from this data. For example, the following Gallup results illustrate the degree to which public attention in 1964 and 1965 focused upon problems pertaining to civil rights and integration: Percent of Gallup Respondents Citing Civil Rights 8 Integra- tion as the Most Important Date Problem April 1964 412 June 1964 47 August 1964 40 September 1964 35 March 1965 52 May 1965 23 September 1965 27 November 1965 17 157 Observe, first of all, that concern over civil rights peaked at 52% in March 1965 following Lyndon Johnson's introduction of the historic Voting Rights Act. This proposal was unveiled by Johnson in his dramatic March 15 speech before a joint session of Congress and a nationwide television audience. The Gallup survey conducted between March 18 and March 23 revealed not only a high level of concern about civil rights but that an overwhelming proportion of the public approved of the president's initative.5 It should also be noted that the increases in concern registered in June 1964 and September 1965 coincided with the passage and signing of the Civil Rights Act of 1964 and the Voting Rights Act of 1965 respectively. These data thus suggest that public concern with civil rights was in part a response to Lyndon Johnson's agenda of action on the civil rights front. Another example which indicates that presidents have the capacity to influence the perceived import attached to performance dimensions pertains to Richard Nixon's imposition of wage and price controls in August 1971. This action was announced before a national television audience at 9:00 p.mi, Eastern time, Sunday, August 15. According to William.Safire's account of administration deliber- ations, this day and time were chosen in a deliberate attempt to create a sense of urgency among the public (Safire, 1977:668—671). In addition to the 90-dayfreeze on wages, prices, and rents, President Nixon's actions also included detaching the dollar from the gold standard, a 10% surcharge on imported goods, and such legis- lative initiatives as the repeal of the 7% excise tax on automobiles, an enlarged investment tax credit, and acceleration of scheduled 158 hikes in income tax exemptions and deductions (Congressional Quarterly Weekly_Report, 21 August 1971, 1759-1764). Public reaction to what Nixon labeled as his New Economic Policy was clearly favorable. Both the Gallup (Gallup Opinion Index, Number 75, 1-3) and Harris organizations (Louis Harris and Associates, 1972, 175) reported that 732 of the public approved of the use of wage and price controls by Nixon. To illustrate the impact such actions exerted upon the import of economic concerns, consider the following information: Percent of Gallup Percent of Gallup Respondents Citing Respondents Citing Misery Date Vietnam Economy Index February 1971 28% 242 9.7 June 1971 33 22 9.6 September 1971 25 45 9.1 November 1971 15 41 9.1 The data reveals a twofold increase in public concern about the economy in the month following the announcement of Nixon's New Economic Policy. The Misery Index readings are included here to show that the growth in concern cannot be attributed to an abrupt shift in economic outcomes. Rather, it appears that the shift was largely induced by the actions of the Nixon Administration. Note as well that Nixon's actions appear to have influenced the level of public concern about the Vietnam war. This illustrates that presidential actions may be employed not only to focus but to divert public attention. 159 To summarize, I have presented several bits of circumstantial evidence to suggest that the import associated with particular criteria and dimensions of presidential evaluation may shift in response to presidential actions. This is not to deny the general expectation that import or public concern will vary with the discrepancy between presidentially-relevant outcomes and public expectations. Instead, my objective is to suggest that there‘ remains room for presidential maneuver. Instances such as the civil rights agenda pursued by Lyndon Johnson and Richard Nixon's rejec- tion of Republican orthodoxy in managing the economy illustrate that presidents can alter, if only momentarily, the relationship between perceived salience and outcomes. Because of this capacity and because salience is hypothesized to influence public support, attempts to capture and direct public attention constitute potentially important components of the president's support management strategy. Thus far I have argued that the president can exercise control over the substantive content of an action as well as the manner in which such actions are publicized. With respect to the consequences of such choices, it has been hypothesized that the aggregate-level impact of a presidential action upon performance assessments will vary with the prominence of the action (for example, the proportion of the public who are aware of the action) and judgements as to whether the action is consistent with expected presidential behavior (for example, the proportion of the public who approve of the action). It must also be recognized that presidents possess substantial discretion in determining when an action will be initiated 160 or executed. That is to say, presidents can exercise control over the timing of their actions. While subject to the rhythms of the legislative calendar and budget cycle, presidents can nevertheless determine when prominent legislative initiatives will be formally introduced into the Congress. The president's discretion seems even greater as far as unilateral actions are concerned. Hence the decisions as to ghep_wage and price controls would be frozen or when the operation to rescue hostages in Iran would be launched were essentially within the hands of executive branch officials. The import of this control over timing has not gone unnoticed. Studies of recent presidencies reveal that the timing of actions is often a subject of strategic concern and planning within adminis- tration councils. For example, Lyndon Johnson's view of relations between president and Congress clearly underscores the import of timing in the legislative process GMcPherson, 1972:268): You've got to give it all you can, that first year ... Doesn't matter what kind of majority you come in with. You've got just one year when they treat you right, and before they start worrying about themselves. The third year, you lose votes ... The fourth year's all politics. You can't put anything through when half the Congress is thinking how to beat you. To cite another instance, analyses of presidential decision making during the Vietnam war suggest that orders to halt and resume bombing raids on North Vietnam were timed so as to influence the conduct of the Hanoi regime on the battlefield and at the bargaining table 161 (see, for example, Gelb and Betts, 1979:144-180; Lewy, 1978: 374-417; Milstein, 1974:39-56). It is also plausible to expect that the timing of presidential actions can be utilized for more political purposes. For example, episodes such as the Tonkin Gulf Crisis in August 1964, the Nixon visits to China and the Soviet Union in 1972, and the attempted rescue of U.S. diplomatic personnel held in Iran in April 1980 certainly provide prima facie evidence that control over the timing of actions can be exercised to benefit a president’s standing and reelection prospects. Indeed, the incentive to orchestrate presidential actions for political benefit is commonly acknowledged by washington journalists and, at times, by administration officials. Consider, for instance, the opening passages of Jack Anderson's report (Lansing State Journal, September, 1980) on a probable attempt by the Carter Administration to free the American hostages: Jimmy Carter studies the polls the way gypsies read tea leaves. The secret soundings of his favorite pollster, Pat Caddell, go directly to the president for his personal scrutiny. These polls show a national crisis would produce a rally—round-the—flag reaction. This is the real reason, in the opinion of insiders, that Carter has ordered preparations for a limited invasion of Iran in October. Hugh Sidey's account of foreign policy decision making in the Carter Administration sounds a similar theme (Time, 12 June 1978, p. 20): 162 Talking tough was a way to rally American voters and foreign leaders, a bit of sabre rattling that almost seemed to fulfill a script lightly pondered last fall by National Security Adviser Zbigniew Brzezinski. Talking to some congressional aides, Brzezinski said it might be good for Carter if he were to have a 'Mayaguez,' recalling the ship seizure by Cambodians in which Gerald Ford counterattacked with marines and raised his prestige. It might, suggested Brzezinski only partially seriously, show Carter's resolve. In sum, it is argued that presidents can control the timing of their actions, that the decision of when to take action is often a matter of concern to administration decision makers, and that such decisions, whether intended.or not, will exert an impact on public support. In fact, the president's ability to control the timing of actions has been acknowledged, albeit in backhanded fashion, by prior research on public support. Thus, in explaining the upturn in support associated with the forgiveness phase of his cyclical model, Stimson (1976:10) observes that: The President probably plays a more active role in rebuilding his support at the end of the term. ... (T)wo related sorts of influences seem to be operative. For one, incumbent Presidents have shown some ingenuity in using their office to control events, particularly symbolic events, to give a good impression of their administrations during the crucial third and fourth years. Richard Nixon's travels to China and Russia may be the best example of the sort of 'events' Presidents may manipulate to their electoral benefit. With somewhat greater controversy,‘ and probably less success, Presidents are alleged to time the ending of wars and the warming of the economy to the election schedule. 163 Despite this recongition, questions regarding the timing of presi— dential actions have been, at best, a tangential concern of the support literature. In contrast to such treatment, it is my contention that presidential control over the timing of actions constitutes an additional mechanism which can be employed to influence public support. The model of presidential decision making developed in Chapter 3 suggests that the timing of presidential actions will be influenced by three considerations. First of all, the choice to initiate an action may be a response to changing outcomes. That is, it can be expected that the president will act on those problems which are judged to be most urgent. Now because it was argued that the perceived urgency of problems, as reflected in the president's agenda, will be revised in light of current outcomes, the timing of actions is likely to be a function of the discrepancy between presidentially-relevant outcomes and those most preferred by the president. In other words, it is hypothesized that: Proposition 8: The likelihood of presidential action on a given performance dimension will increase as the outcomes relevant to that dimension diverge from the states most valued by the president. Second, the value of public support as a resource implies that the timing of presidential actions will also be determined by changes in public support. Because declines in support are interpreted by members of the Washington community as a dimunition in presidential influence, presidents may time their actions so as to replenish this resource base. This line of reasoning leads to: 164 Proposition 9: The likelihood of presidential action will increase as a function of declines in public support. Third, it can also be argued that the timing of presidential actions will be influenced by the electoral cycle. As noted in the discussion of Chapter 3, the import or weight which the president places upon public support and the satisfaction of public expecta— tions (i.e. the coefficient r t in Expression 3.42) may increase 2 with the approach of a national election. If such is the case, it can be expected that presidents will be more inclined to initiate actions designed to benefit their public standing as well as the electoral fortunes of themselves and their parties. Thus, it is hypothesized that: Proposition 10: The likelihood of presidential action will increase with the proximity of a national election. The primary research question which underlies these rather loosely formulated propositions concerns the extent to which the timing of presidential actions is subject to systematic and measure- able influences. Any desire to formulate these hypotheses in a more precise manner and to describe the evidence necessary for falsification requires that attention be directed toward defining the relevant set of presidential actions. My objective in the dis— cussion which follows is not to undertake such an exercise; rather, it is to suggest how a test of these hypotheses can be conducted 165 and to illustrate how a concern with timing as a lever of presidential control demands that we reconsider the manner in which authors such as Mueller, Stimson, and Kernell have incorporated events and crises into their explanations of public support. The influence of events and crises on public support was first examined by John Mueller (1970; 1973). His use of "rally points" was grounded upon the argument that "certain intense international events generate a rally-round-the-flag effect which tends to give a boost to the president's popularity rating" (Mueller, 1973:209). Rally points have subsequently appeared in most alternative models of public support and their import has become an accepted bit of ‘wisdom in the public opinion literature. To identify such events, most analyses have generally adopted the definition offered by Mueller (1973:209) which requires the following conditions to hold in order for an event to be considered a rally point: In general, a rally point must be associated with an event which (1) is international and (2) involves the United States and particularly the president directly; and it must be (3) specific, dramatic and sharply focused. Given my concern with the timing of actions as a lever of control, there are three criticisms which can be lodged against the manner in which such rally points have been treated. First of all, the question of whether a presidential action is itself responsible for the appearance of a rally point has been ignored. That is, no attention has been devoted to the manner in which asuch events are triggered. It is possible, however, to draw a 166 distinction between two types of events commonly included in any list of rally points. The first are those crises and episodes where the behavior of other actors and prevailing circumstances force the president to take action. The distinguishing features of these exogenously-induced rally points are that (l) the event provides the stimulus for presidential action. Rally points such as the Berlin Wall crisis in September 1961 and the Cuban missile crisis in October 1962 fall into this exogenously-induced category. In both instances the event in question preceeded presidential action and decision time was limited. Now compare such episodes to the Bay of Figs Invasion of 1961, the Nixon visits to China and the Soviet Union in 1972, and the Camp David Summit of 1978. In these cases, presi- dential action was instrumental in producing the event. That is to say, it is reasonable to believe that in the absence of administration action, the event wouldinot have occurred. Further, in each of these cases the president played a prominent role in determining when the event would take place. Let us label the events and crises which share these characteristics as presidentially—induced rally points. Table 4.3 presents my preliminary effort to separate those events and crises commonly regarded as rally points into the exogenously- induced and presidentially-induced categories.6 Although any scheme for the selection and classification of events is liable to charges of arbitariness, the exercise is nevertheless instructive. A plausible argument can be made that of the 37 episodes listed in the table, 21 (57%) were presidentially—induced.7 These include unilateral decisions regarding U.S. military actions during the "Vietnam war, visits to the United States by Soviet leaders, and presidential trips to summit meetings held abroad. 167 The purpose of this exercise is not to produce an exhaustive list of relevant events or an authoratative classification; rather, it is intended to illustrate that events and crises which have been treated in past research as exogenous occurrences are germane to the question of support management. Indeed, the exercise was designed to suggest that many of these rally points seem to have been triggered by the actions of the president. In this sense, the classification has identified a number of actions and associated events which can be employed as evidence to evaluate Propositions 8-10. By themselves, however, this set of events is not sufficient for evaluating the hypotheses. Essentially, the list of presiden- tially-induced rally points displayed in Table 4.3 excludes those actions which pertain to domestic concerns and those which produced no observable increases in public support. These omissions illustrate the second and third criticisms which can be made against the way events have been analyzed in prior research -— a marked inattention to domestic actions and events and the assumption that events exert a positive and uniform impact on public support. Both criticisms are relevant to testing the hypotheses offered in this section. With respect to the former criticism, there is no a priori reason to expect that prominent domestic actions will not influence support in a manner analogous to those which pertain to the international arena. Indeed, according to Propositions 5 and 6, public reaction to both domestic and international 'actions should be similar in the sense that their impact is 168 hypothesized to depend upon salience and judgements about the appropriateness of the action. Note that such propositions also challenge the assumption of uniform impact employed by Mueller, Stimson, and Kernell. 'Moreover, the distinction between presi- dentially-induced and exogenously-induced actions seems no less relevant in the domestic context. Thus, it is argued that a more complete and general consideration of events demands that the international criterion imposed in prior research be expanded to include domestic events. This discussion also implies that an appropriate test of Propositions 8-10 requires examination of presidentially-induced international ggd_domestic events. Finally, let us consider the inattention of prior research to presidential actions and events which seemingly produced no observable increases in public support. While such an omission is of minor import to the results produced by Mueller, Stimson, and Kernell, a satisfactory test of the timing propositions requires that such episodes be included in the set of actions to be examined. This is because questions about the timing of actions are separate and distinct from those pertaining to the impact of actions. Although presidents may attempt to time actions so as to trigger bursts in public support at important points in the term, such attempts may fail; support may remain unchanged or the action may exert a negative impact. For example, let us examine the case of the Salt II Treaty signed by President Carter and Premier Brezhnev on June 18, 1979. The signing of the treaty would seem to be a likely candidate for a rally point; Jimmy Carter's travels to Vienna, his first meeting 169 with Brezhnev, and his declarations that the treaty served as the linchpin of the administration's foreign policy all received thorough coverage in the media. However, consider the following information compiled from the Gallup polls:8 Percent Percent of Those "Aware" Who: Thware Date of Treaty Approve Disapprove No Opinion March 16-19 '79 58% 52% 17% 31% June 22-25 '79 82 48 27 25 Awareness of the Salt Treaty increased from 58% to 82% after the event but this increase actually results in a less favorable distribution of sentiment toward the treaty. As the following data indicate, this decline occurred even among those individuals knowledgeable enough to cite a specific advantage or disadvantage of the treaty. Percent of those Citing an Advantage or Disadvantage Percent Citing A who: Specific Advantage Date or Disadvantage Approve Disapprove No Opinion March 16-19 '79 45% 60% 20% 20% June 22—25 '79 64 53 30 17 The most pertinent point registered by this example is that events such as the Salt II Summit must be included in the set of episodes used to assess the hypotheses about the timing of 170 presidential actions. In other words, the data required to evaluate Propositions 8-10 will consist of all salient foreign and domestic presidential actions regardless of their ultimate impact upon public support or specific performance assessments of the president. Although such a requirement calls for a more careful scrutiny of administration histories than has heretofore been the case, I have tried to suggest that such an undertaking promises to result in a more refined treatment of presidential decision making. To summarize, the discussion in this section was designed to explore two issues. The first concerns the consequences of presidential actions. Relying upon the model of citizen decision making developed in Chapter 3, I have argued that presidential actions (1) will be evaluated in light of public expectations and will influence the path of performance assessments, (2) can activate performance criteria and dimensions which previously played little or no role in the evaluation of the president, and (3) can alter the salience which citizens associate with particular criteria and dimensions of evaluation. Second, the discussion has attempted to identify the levers or mechanisms which can be employed by the president to trigger such consequences and thus influence public support. It is argued that presidents can exercise control over three aspects of a decision —- the substantive content of an action, the effort devoted to publicizing the action, and the timing of the action. Control over content, publicity, and timing are thus the levers which can be utilized by presidents to manage public support. 171 Although motivated by their own theory of action and subject to the constraints imposed by Congress, presidents can be expected - because of the value of public support —- to exercise control over these levers so as to influence performance assessments in a beneficial manner. The next section of this discussion is designed to consider an example which illustrates how the use of such levers can be coordinated to produce a relatively successful strategy of support management. The Pursuit of a Management Strategy: Richard Nixon and Vietnam Because it represents a violation of a fundamental expectation directed at the institution of the presidency, wars invariably pose a severe threat to the public standing of presidents. As conflicts drag on and human costs grow, a war weariness develops among the public which begins to erode both the influence of and confidence in the president. Such was the effect that the Vietnam war exercised upon the presidency of Lyndon Johnson. As Figure 4.2 illustrates, dissatisfaction with the manner in which Johnson handled the Vietnam war increased throughout the period from 1966 to early 1968.9 The political damage of this war and the public dissatisfaction it stimulated were substantial. It weakened the president to the point where he was forced to initiate actions contrary to his stated philosophy —- the proposal of a tax increase in 1967 and the abandonment of further extensions of his Great Society programs, 172 the cornerstone of his domestic agenda. As Kearns (1976:300) reports: After hesitating for nearly two years to recommend a tax increase, Johnson finally asked the Congress for a 6 percent surcharge in September, 1967, but the price of the legislation -— as Johnson had feared all along - was a crippling reduction in domestic programs. 'You have to go on TV,‘ Wilbur Mills notified the President in 1967, and explain that 'because of Vietnam we must cut domestic spending and pass a tax increase, and if you take this position you can count on me to go with you all the way ... I also want some committments made in executive session of the ways and Means Committee on major slashes in domestic spending'. Ultimately, the war led to Johnson's abdication and left the Democratic Party bitterly divided over how the stalemate in Southeast Asia was to be resolved. Lyndon Johnson's legacy to the new Republican administration was no less threatening. In January 1969, the number of U.S. troops in Vietnam stood at 549,500; total casualties from 1965 to 1968 were 29,807 with an average rate of 1200 per month during 1968. Further, public opinion regarding the war in Vietnam was sharply divided. An October 1968 Gallup poll revealed that 44% of the public considered themselves "hawks" on the war with 42% opting for the "dove" category;10 in the same poll, 52% believed that the Unitd States had made a.mistake in becoming involved in Vietnam and 37% thought the Vietnamese policy was correct. The problem confronting Richard Nixon was essentially how to preside over a nation heavily involved in an Asian land war and polarized at home. 173 A consideration of histories and memoirs from this period suggests that the strategy of action followed by Nixon consisted of two components. (See, for example, Nixon, 1978; Kissinger, 1979; Safire, 1977; Evans and Novak, 1971; Schell, 1975.) The first might be labelled the strategy of Nixon the hawk. This element of the administration's strategy was characterized by an adamant refusal to abandon the primary American objective in Vietnam.- to prevent a North Vietnamese takeover of the South (on the nature of American objectives, see Gelb and Betts, 1979, especially pp. 181-200). Rhetorically, the objective was most evident in Nixon's reliance on such themes as Peace with Honor and "preserving American credibility as an ally." Indeed, these themes appear throughout the eleven speeches which Nixon delivered before nationwide television audiences (see Table 4.4) during the 1969—1972 period. Operationally, adherence to this objective is illustrated by such administration actions as the Cambodian incursion of April 1970, the Laotian operation of 1971, the Christmas bombing of 1972, and the mining of North Vietnamese harbors in May 1972. Regardless of their military or diplomatic value, such actions certainly protected Nixon from charges of softness by the more militant segments of American public opinion. The second component can be characterized as the strategy of Nixon the dove. It consisted primarily of an effort to curtail direct U.S. involvement in the land war and to reduce the most immediate human costs of the war - the dead and the injured. The strategy was primarily pursued through Nixon's Vietnamization plan. As announced to the American public on November 3, 1969, the 174 plan was designed to let South Vietnamese ground forces assume an increasingly larger share of the responsibility for prosecuting the war while American forces were gradually withdrawn. The manner in which this program was implemented is summarized in Table 4.5. This information shows that by December 1972, 95% of the troops stationed in Vietnam when Nixon assumed office had been withdrawn. With respect to the publicity and timing of this policy, it is instructive to note that four of the withdrawal announcements were made before nationwide television audiences, that the rate of troop withdrawal peaked during the early months of the 1972 election campaign, and that the June 1972 withdrawal announcement also included a directive that henceforth no American would be sent to Vietnam involuntarily. Such actions and policies enabled to Nixon to claim that, in comparison to his predecessor, the pain imposed on the American public by the war had been substantially reduced. The point is illustrated in Figure 4.3 which displays the casualties incurred by U.S. forces during the months from June 1965 to December 1972. It can be seen that casualties generally declined throughout the Nixon Administration and that at no time between 1969 and 1972 did the levels approach those recorded during 1968. Both the withdrawal of troops and the reduction of casualties were utilized during the campaign of 1972 as evidence of Nixon‘s committment to peace in Southeast Asia. Examination of Figure 4.2 suggests that these policies and actions also enabled Richard Nixon to avoid to the sizable growth Pin public dissatisfaction toward the conduct of the war which 175 characterized the tenure of Lyndon Johnson. Although the pertinent Gallup and Harris results differ with respect to question wording and the manner in which the results are tabulated, we can compare the shape or pattern of these specific performance assessments. For Johnson, the proportion of the public expressing disapproval with the way he conducted the war steadily increased from 22% in late 1965 to 60% in the month following the TET offensive of January 1968. Such a pattern is not evident in the evaluations of Nixon. Rather, there is a pattern of undulations about the 50% level of dissatisfaction. The available Gallup results concerning Nixon's war management are quite similar. Although the increase in dissatisfaction recorded in early 1970 is more pronounced than in the Harris results, the shape of the Gallup pattern for Nixon is quite similar to the Harris readings and both are clearly distinct from the Johnson pattern. A Strategy for Testipg the Alternative Explanation Although it was argued in Chapter 3 that war represents a violation of public expectations and thus will lower the president's public standing, the model of citizen-president interaction and the discussion in this chapter implies that presidents can limit the damage. Presidents, via their actions, have the capacity to influence outcomes pertaining to the war; moreover, choices regarding the content, publicity, and timing of actions can be employed to influence performance assessments directly. The manner in which Lyndon Johnson and Richard Nixon managed the war in Vietnam appears 176 to be a case in point. In contrast to the gradual buildup of U.S. forces and the misleading manner in which the magnitude of this committment was revealed to the public (Gelb and Betts, 1979; Halberstam, 1972), the policy of disengagement and military actions seems to have enabled Nixon to avoid the erosion of public support experienced by Lyndon Johnson. Such an interpretation is specu- lative however. This is particularly true in light of the conclu- sions produced by prior analyses of public support. Consequently, this section is designed to review such conclusions, to compare these conclusions to the hypotheses advanced in this chapter, and to outline the manner in which these hypotheses and the alter- native explanation of public support can he conducted. According to the results presented by Mueller and Stimson, the Vietnam War exerted no appreciable impact on public support for Lyndon Johnson. Instead, the rather precipitous erosion of general support observed between January 1965 (71%) and August 1968 (35%) is attributed to the operation of the within-term.dynamic components in their economietric specifications. Kernell, on the other hand, does find that U.S. casualities in Vietnam exercised a significant influence on Johnson's support; curiously, however, Kernell makes no attempt to assess the impact of the war on Nixon's public standing. Thus, the empirical results produced by prior research create a rather ambiguous picture. Our understanding of the impact of Vietnam is further limited because to date no attempt has been made to analyze how~the public evaluated the war management of Johnson and Nixon. 177 Now consider the arguments offered in this discussion. In contrast to the conclusions of prior research, the model of citizen— president interaction developed in Chapter 3 and the propositions formulated in this chapter lead to the following arguments. First, the Vietnam war - because of its relevance to the institutional expectations of the public -— did exert a significant impact on the overall level of public support for both Lyndon Johnson §£d_ Richard Nixon. Empirically, it is expected that, given proper controls for other specific performance assessments of these presidents, the mangitude of this impact will vary with (a) the proportion of the public who disapproved of the way the war was handled and (b) the proportion of the public who regarded Vietnam as the most important problem facing the country. Second, public disapproval of how Johnson and Nixon handled the war will be a function of (1) war-related outcomes, (2) salient presidential actions initiated during this period, and (3) exogenously-induced, war-related events. Third, as stated more generally in Propositions 5 and 6, the impact of presidential actions on war disapproval will be a function of the salience of the action as well as the proportion of the public who approve of the action. In effect, the second and the third arguments imply that the actions taken by Johnson and Nixon underlie the distinct patterns exhibited by their war disapproval ratings. Given adequate data and a proper measurement strategy for operationalizing relevant outcomes, actions, and events, an econometric analysis of the war disapproval series will provide one method by which the propositions regarding I presidential control can be evaluated. This will be the task undertaken in Chapter 5. 1. 178 Notes to Chapter 4 Technically, the question of presidential control is not expli- citly addressed by Kernell. Rather, the conclusion emerges upon careful inspection of the statistical results presented by Kernell. Ironically, Kernell's discussion is intended to refute the Mueller and Stimson arguments which contend that commonly cited factors such as wars and recessions do not exercise an impact on presidential performance evaluations. However, consideration of the most reliable statistical results presented by Kernell (1978:522, Table A.2) reveals little support for this attempt to resurrect conventional political wisdom. Excluding the lag support term and the regression intercepts, Kernell estimated twenty-one coefficients. Of those, only eight proved to be statistically significant. The following illustrates the results reported by Kernell for each of the presidents included in the analysis: Pres ident Number of Coef f ic ients Ins ignif icant Estimated Coefficients Truman 4 3 Eisenhower 4 4 Kennedy 3 2 Johnson 6 3 Nixon 4 1 Thus, the reader is led to conclude either that support is governed by a poorly explicated autoregressive process beyond the scope of presidential control or that because of methodological difficulties, Kernell's analysis provides no useful insights into the problem of support management. The general performance question administered by the Harris sur- vey is: "How would you rate the job President Nixon is doing as president - excellent, pretty good, only fair, or poor?" The "excellent" and "pretty good" responses are then summed and reported under the "Favorable" category; similarly, the "only fair" and "poor" responses are added to produce the "Unfavorable" score. Specific performance assessments are measured by asking respondents to apply the same rating scheme to a variety of presidential actions and problem areas. Both the general and specific evaluations are reported in the annual Harris survey volumes (Louis Harris and Associates, Inc., 1971- 1974). 179 The responses to this iten were obtained from the results reported in the three volume anthology edited by Gallup (1972) and relevant issues of the Gallup Opinion IndeX. The survey question - "What do you think is the most important problem facing this country today?" - has been administered an average of three to four times annually since 1953. The unemployment and inflation rates were obtained from the U.S. Department of Commerce publication entitled Business Conditions Digest (Series #43 and Series #320C). See Gallup (1972:1933). The item administered was: A law has been proposed that would allow the Federal Government to send officials into areas where the turnout of eligible voters in the last presidential election was so low that it suggested that some persons were denied the right to vote. These officials would make sure Negroes and whites are given an equal opportunity to register and to vote. WOuld you favor or oppose such a law? The responses were: 76% Favor, 16% Oppose, 82 No Opinion. The list of events displayed in Table 4.3 is an updated version of the set employed by Mueller (1973:211). Two sets of events proved to be particularly difficult calls -- the events which marked the close of hostilities (the Korean truce, the Vietnam Peace Accord) and prominent American shows of force (Lebanon, Gulf of Tonkin, Dominican Republic). Both were classified as exogenously induced although it can be argued that the latter set presents an example of presidents using events as a pretext for action. Such were the circum- stances surrounding the U.S. bombing of North Vietnam in 1965. As Gelb and Betts have reported (1979:105): Months later, when the United States began bombing North Vietnam in retaliation for a Vietcong attack on American installations at Pleiku, McGeorge Bundy would acknowledge the extent to which the incident served as a . pretext for escalation by saying, ‘Pleikus are streetcars'; such incidents could be relied on to happen regularly, offering opportunities to implement decisions already made. 180 8. This information was obtained from the following issues of the Gallup Opinion Index: May 1979, Number-166, p.13; June 1979, Number 167, pp.10-11; and October 1979, Number 171, pp. 19-20. 9. The data on dissatisfaction with the conduct of the war by John- son and Nixon were gathered from the relevant issues of the Gallup_9pinion Index and the Harris Survey Yearbook of Public Opinion. 10. See Gallup (1972:2222). The question used to generate these responses was: People are called hawks if they want to step up our military effort in Vietnam. They are called doves if they want to reduce our military effort in Vietnam, How would you describe yourself - as a hawk or a dove? Dynamic, Within-Term Component Environmental Factors 1-- This variable is operationalized as follows. 181 Table 4.1 Econometric Models of Public Support A Summary Mueller Stimson Time into Term Unemployment Change1 (ECOSLUMP) Korean War Dummy Vietnam war Dummy International Crisis3 (RALLY) Time into Term Time Squared Unemployment Change (ECOSLUMP) Korean Var Dummy Vietnam War Dummy International Crisis (RALLY) Kernell Lag Value of Support Unemployment Change Inflation Change U.S. Casualties in Korea U.S. Casualties in Vietnam U.S. Bombing Missions International Crisis“ Early Term SurgeS Watergate Dummy Define Uo as the rate of unemployment which prevails in the month of inauguration and U: as the rate during month t. ”t is greater than U0 and is set to zero otherwise. ECOSLUMP is defined by the quality (0: - U0) if 2-- Unemployment and Inflation Change are based upon a six month moving average comparison. 3- The variable is defined as the length of time, in fractions of years, since the last rally point. designated a rally point). (The initial month of the term is 4- It is assumed that rally points have a duration of five months and a (S,4,3,2,1,0) scheme is employed to measure their impact. S- A variable which takes on the value of 5 in the first month of the term and declines by one unit per month until it reaches a value of zero. Unlike Mueller and Stimson, Kernell does not include the first month of the term in his set of rally points. 182 meow uoz mm 50 no on co «a m>fiumwmz Name somaunma mm mm an as am Nmm "ooqum>auo< mammH mom wewxmz scamfiuoo Hmaueoofimmum .o.w.m .oe .mm¢~ ..oeH .moumfioomm< use mayhem mass; mama “cowefieo seapom «0 xoonumow >o>usm means: one "ransom ea o>aufimom meow uoz m>Humwoz um mm um um om ma meme xumoemh «a mm mm om mm Nae o>HuHmom uuoaesm season so uomeeH was no oaoamxm e< ~.¢ oHan mango sameness on easy mouuo mom and a waste measeem: magmas Ha>ao message: was ameuma> was massage: mauamom aaoeoum osu wewoomx oversuomuwm Hamum>o 183 Table 4.3 Presidentially and Exogenously Induced Rally Points: Date August, 1953 *November, 1953 July, 1954 *July, 1955 October, 1957 August, 1958 *September, 1959 *December, 1959 May, 1960 *April, 1961 *June, 1961 September, 1961 November, 1961 October, 1962 August, 1964 *February, 1965 May, 1965 *June, 1966 *June, 1967 January, 1968 *March, 1968 *October, 1968 *November, 1969 *May, 1970 *October, 1970 *February, 1972 *May, 1972 January, 1973 *June, 1973 May, 1974 *June, 1974 June, 1975 *November, 1975 *November, 1978 November, 1979 December, 1979 *April, 1980 A Preliminary Classification Event Korean Truce Signed Eisenhower's "Atoms for Peace" Speech at United Nations Indochina Truce Signed Big Four Summit Conference in Geneva Sputnik I Launched U.S. Troops Dispatched to Lebanon Kruschev Visits the United States Eisenhower ”Goodwill Tour" of Three Continents U-2 Incident/Paris Summit Cancelled Bay of Pigs Invasion Kennedy-Kruschev Summit in Vienna Berlin Wall Constructed U.S.-Soviet Confrontation in Berlin Cuban Missile Crisis Gulf of Tonkin Crisis Retaliatory Bombing of North Vietnam Begins U.S. Tr00ps Dispatched to Dominican Republic Bombing Extended to Hanoi and Haiphong Glassboro Summit Pueblo Seized/ TET Offensive Partial Bombing Halt/ LBJ Withdraws from Campaign Full Bombing Halt Announced Nixon Vietnamization Speech Cambodian Incursion EurOpean Tour Nixon Visits Mainland China Nixon Visits U.S.S.R./U.S. Mining of North Vietnamese Harbors Vietnam Settlement Announced Brezhnev Visits U.S. War in the Middle East/U.S. Enters Negotiations Nixon Visits U.S.S.R. and Middle East The Mayaguez Seizure Vladivostok Accord Signed/Ford Visits China Camp David Summit Iran Militants Seize U.S. Embassy Soviet Union Invades Afghanistan U.S. Attempts Rescue of Hostages Held in Iran *-- Denotes that the event had been initially classified as presidentially-induced. 184 Table 4.4 Televised Reports to the Nation on Vietnam by Richard Nixon, 1969-1972 Date Subject 14 May 1969 Nixon presents an eight-point plan for ending the war in Vietnam. 3 November 1969 Nixon announces a Vietnamization Program - a secret but ”orderly scheduled timetable" for withdrawing U.S. troops from Vietnam; he appeals to Silent Majority for support of his program. 15 December 1969 Nixon announces that 50,000 U.S tr00ps will be withdrawn from Vietnam by 15 April 1970. 20 April 1970 In a progress report on the war, Nixon announces that 150,000 U.S. troops will be withdrawn from Vietnam by 1 May 1971. 30 April 1970 Nixon announces that American and South Vietnamese troops have invaded Cambodia 3 June 1970 Nixon reports on the completion of the Cambodian operation and labels it the most successful of the war. 7 October 1970 Nixon calls for expanded peace talks on Vietnam and asks Hanoi and the Viet Cong to join in a cease-fire. 25 January 1972 Nixon makes public a plan for ending the war which includes a general cease-fire, U.S. withdrawal within 6 months, and supervised elections in South Vietnam. 26 April 1972 Nixon announces that 20,000 U.S. troops will be withdrawn from Vietnam by 1 July 1972. 8 May 1972 Nixon announces that North Vietnamese ports of Hanoi and Haiphong have been mined. 185 ooo.- ooo.mm ooo.as ooo.ao coo.oma ooo.swa ooo.sm~ ooo.sms ooo.sms cem.s~m mam>oq oo\H\H mo osmouom m< moooue wowefimEom mmoouh weacfimaom "Naoauaooa .smcuoas as amouoe .m.= mo coeuosewe may ooo.s coo.m ooo.o~ oom.m~ cem.- oom.qa cow.~a oom.~a cos.sa ooo.o~ Hmsmuveuaz mo some seeucoz humaaom < m.¢ magma ~a\H\NH ~m\a\m ~e\a\e ~e\m\m Nm\a\~ aa\a\ms He\a\m ce\ma\s ao\ma\aa ¢o\HM\w mama oowuoaaaou ooo.~a coo.oi ooo.o~ ooo.o~ coo.ms ooo.ooa ooo.cms coo.0m com.os ooo.n~ osmuueuaz meooue mo amnesz Nm\m~\w NN\wN\o N5\@N\¢ NN\mH\H HN\~H\HH H~\e\¢ ON\ON\¢ mo\mH\NH ao\o~\¢ me\w\c mama 186 1A awesome uemuuoaau umoz mo zsoccom wnu wewuuo oHoEmm mouamo mo couuuonoum xoucH Assam: one ——pp—pp-—bppbppplphb——-hLl~ pp- aco~ mead ~co— nmm~ mmo~ m.~ m.n o.o~ m.- c.m_ m.nd c.o~ xomc~ scammeum=-~ :< (a zummvz Homecomoc o_Ec:ocm 1cm xEoeoom on“ uses< ensues: c_~a=m csosumm magnesium—mm was l.s .t=a_t c~ ON on oc om co oh cw ucmouma esfifimo 187 Nno~ .mm. c~o~ mood mca~ “cog coo“ moo~ K p — — — — — b NON «on. ./ \, Zoe // . x 3:8 .. aeoz / x z . \ / \1/ . N3 \ T e I / \ Ron Haom HHom m«uum: I eoxuz madame I couscow Moo 1:2 who“ . moss czasomaa 6cm .coxmz .:Cmc:cw Hzmummm "ucoaoxmcoz um: #mmuemcamoum mo mes—.o:~m>m owflasm N.¢ ataxia rsofi ~nm~ cmw— moo~ womd 5cm— 188 mes—unco~ .Emeume> smocm c. nounscch mmfiuumsmmo oduumm :mofiumE< zflzucoz n.e wtswaa mood voo— f oon ooc~ OOm~ oooN CHAPTER FIVE MANAGING PUBLIC SUPPORT: AN EMPIRICAL ASSESSMENT At several points in the last chapter it was asserted that the model of citizen and president behavior developed in Chapter 3 offers a rather sharp alternative to the prevailing mechanistic explana- tions of how citizens evaluate the performance of presidents. Rather than characterizing the behavior of presidents and citizens as de- tached and unrelated, the model developed in Chapter 3 asserts that the judgements made by citizens and the decisions emanating from the Oval Office are, in effect, linked by'a mutual concern with outcomes and actions. For citizens, presidentially-relevant outcomes and actions touch upon matters of personal satisfaction or security and, for this reason, serve as the criteria of choice in evaluating a president's conduct of office. For presidents, actions and the consequences which they exert on outcomes Operate as the means employ- ed to pursue the substantive goals of the administration and to maintain public and congressional support, the resources necessary for the achievement of presidential goals. Accordingly, the formal argument developed in Chapter 3 was designed to illustrate how the value of public support impinges upon the pursuit of presidential goals and the content of the president's actions. Chapter 4 sought to explore how the choice of actions -- their content, timing, and publicity -- serve as levers which can be utilized to influence public evaluations of presidential performance. This discussion is summarized by the alternative explanation of 'public support contained in Propositions 1-4 and the hypotheses 189 190 pertaining to the choice and impact of presidential actions on the specific and general performance assessments of the public (Prop— ositions 5-10). In the closing portions of Chapter 4 it was suggest- ed that the Vietnam War and the manner in which Lyndon Johnson and Richard Nixon attempted to manage the conflict provides one oppor- tunity for assessing several of these propositions. The purpose of this chapter is to conduct such an assessment. My intent is to illustrate how the discussion presented in Chapters 3 and 4 leads to a general representation of an aggregate-level ‘model of specific performance assessments and to apply this model to public evaluations of presidential war management during both the Johnson and Nixon Administrations. The chapter consists of five sections. The first summarizes both the manner in which prior studies have investigated the impact of war on public assessments of the president and the conclusions generated by this research. In the second section, the arguments advanced in Chapters 3 and 4 are utilized to formulate a model of public dissatisfaction with President Johnson's and President Nixon's conduct of the Vietnam war. In contrast to the general treatment of war and performance assessments in previous research, this model rests upon the prOp- osition that the impact of war on public evaluations is influenced not only by outcomes but also by the salient presidential actions initiated to influence the course of the war and by exogenously- induced, war-related events. The third section of this chapter develops a measurement strategy to operationalize the influences identified in the prior section and ' describes the data which will be used to estimate the model. This 191 discussion is also designed to rectify several measurement problems found in prior research and to illustrate the connection between the measurement strategy pursued and the arguments advanced in previous chapters. In the fourth section, the model of war dis- approval is estimated and evaluated. To assess the model, its performance will be compared to the results produced by three alternative models formulated on the basis of prominent arguments in the literature on presidential evaluations. The final section of the chapter is intended to summarize the results of this empir- ical analysis, to show how these results reflect upon the question of support management, and to compare the resulting conclusions to the current wisdom produced by prior research. war and Public Support: The Conclusions of Prior Research If the manner in which prior research has studied the relation- ship between war and public support is examined, a rather confusing picture emerges. Most authors have acknowledged the potential im- port of wars as a criterion of presidential evaluation. Thus Mueller sought to assess the impact of Korea and Vietnam which, he argued, were "unpopular, puzzling, and indecisive wars" (1970:23). Similarly, Kernell includes these conflicts under the general rubric of the chronic problems which challenge administrations and threaten the public standing of presidents. Beyond the recognition that war should operate as an important influence on presidential support however, the literature is characterized by disagreement over the measurement of war and confusion about the estimated impact of war ' on public support. 192 The situation is summarized in Table 5.1. The table depicts the manner in which the Korean and Vietnamese Wars were incorporated into the aggregate-level models of public support and the resulting coefficient estimates for the war variables. It is clear that no consistent conclusions are produced by this research. For example, Mueller reports that while the Korean conflict exercised an appre- ciable impact on the public standing of President Truman, Vietnam proved to be inconsequential in accounting for the erosion of support experienced by Lyndon Johnson.1 The conclusions are highly question- able, however, because of the presence of serial correlation in Mueller's results; indeed, Hibbs' reanalysis of the Mueller data (1974) produces the revised inference that neither Korea nor Vietnam exerted a significant impact on public support. A similar conclusion is offered by Stimson (1976)2. The remaining studies reach somewhat different conclusions. Substituting measures of American casualties and bombing missions for the binary war variable employed by Mueller and Stimson, Kernell finds that the casualty rate exerts a significant impact during the Vietnam period only; the reported estimates thus reaffirm the conclusion of Hibbs and Stimson regarding the Korean war. In a rather peculiar analysis,3 Stimson and LeGette (1975) conclude that the duration of both the Korean and Vietnam wars exercised a neg- ative influence on the performance evaluations of Presidents Truman and Johnson. Given such results, it can be concluded that this literature offers very little insight regarding the relationship between the occurrence or conduct of wars and public assessments of the president. 193 Absent is any agreed upon method for measuring the impact of war and, when viewed as a collective body of empirical results, there is no unambiguous conclusion regarding the direction and magnitude of a war's influence on public sentiment. Even more puzzling is the inconsistency between the explanations which are used to justify the inclusion of war variables in econo- metric models of public support and the manner in which the models are applied. To elaborate, the measures employed to capture the impact of war are typically grounded upon one of two justifications. The first, a variant of the war weariness argument articulated by Richardson (1960) and Boulding (1962), asserts that negative opinion toward a war and presumably toward those viewed as responsible for the conduct of the war will increase with the duration of the con- flict. Thus Stimson and LeGette reason that without ruling out such factors as the ugliness of war on television or the horror of body counts, we postulate a public reaction keyed to the length of time presidents pursue wars without Whining 0 Here, time is employed to approximate the manner in which public opinion is transformed from enthusiasm to weariness as the human and material costs associated with a war accumulate.‘ While not directly concerned with the question of a president's war management, Mueller's analysis of support for the wars in Korea and Vietnam is based on a dynamic similar to the war weariness proposition (Mueller,1973:42-6S). To predict the proportion of the public who regarded these wars as a mistake,4 Mueller relies upon a more concrete indicator of the severity and costs of the war -- 194 a cumulative war casualty index. Mueller justifies the use of this indicator on the grounds that such a measure captures the amount of pain caused by the war. As Mueller notes (1973:59), "the assumption is that peeple react more to the cumulative human costs of the war than to its duration." The second rationale employed to justify the inclusion of war- related factors in models of public support is articulated by Kernell and rests upon what can be termed an intensity argument. Implicit in Kernell's classification of Vietnam and Korea as chronic prob- lems is the supposition that any war, because of the pain imposed on the public, will erode the support enjoyed by the president. It is further argued that the rate or magnitude of such decay will vary with the intensity of the conflict. Therefore, Kernell's selection of casualties and bombing missions rests upon the argument that these indicators measure the "overall level of military conflict" and the "war's intensity" (1978:510). The obvious difference between these justifications is that the former envisions the impact of war as cumulative in character while the latter envisions a more immediate impact and emphasizes contem- poraneous measures of a war's intensity. More important is the commonality shared by these rationales. Both lines of reasoning assert that, regardless of its measurement, war imposes pain upon the public and therefore threatens to undermine support for the president. In other words, the rationales suggest that war represents a rather universal criterion of presidential evaluation. However, this premise is contradicted by the manner in which the models of public support ‘ are specified and estimated by these authors. Although war is char- acterized.as a general criterion of assessment, the war variables 195 are defined and their impact assessed in a selective manner. Thus no attention is directed to the import of the Korean War and its resolution on the public standing of President Eisenhower. A.more critical omission in the works of Stimson and Kernell is.the failure to examine the relationship between the Vietnam war and public support for Richard Nixon. While Stimson and LeGette (1975:10—11) acknow- ledge this problem, Kernell is mysteriously silent about why a war which continued throughout Nixon's first term and produced over 15,000 additional casualties is excluded from consideration. Simply put, such oversights clearly reveal that the empirical analyses upon which current understanding of the relationship between war and public support rests were conducted in a fashion that is inconsistent with the substantive explanations advanced by these authors. On the basis of this review and the discussion presented at the outset of Chapter 4, the current wisdom regarding the overall level of public support and the impact of war on this level can be summar- ized as follows. First, there is the conclusion that the general level of public support is beyond the control of the president. The fundamental expectation is that support for the president will erode over the course of a term. Second, it is agreed, in principle, that war operates as a criterion of presidential evaluation. As such, the occurrence of war is expected to exacerbate the expected decline in public support. Third, the impact of war on the overall level of public support will be a function of either its duration or its intensity. Note that in both cases, however, the denial of meaning- ] ful presidential control is affirmed. According to the research of 196 Stimson and LeGette, for example, the expectation-disillusionment- forgiveness cycle governs public opinion toward the war and presiden- tial behavior cannot be expected to alter this dynamic. The implication arising from Kernell's analysis is that the president may influence public support only to the degree that he can limit a war's intensity. Fourth, because of empirical results which are at best ambiguous (see Table 5.1) and because of a variety of specification and measure- ment problems (Ostrom and Stimon,1980), the empirical impact of war on public support cannot be described with any degree of precision or confidence. Having characterized the results of prior studies, let us now consider how the discussion presented in Chapters 3 and 4 challenges both the substantive explanations and the research strategy contained in this literature. A Mbdel of Specific Performance Assessments: The Case of Vietnam The manner in which prior studies have attempted to ascertain the impact of war on public support can be depicted in diagram form as follows: Duration and/or Intensity of a waE“-.._‘_~“““ ' Overall Presidential Performance Evaluations Within-Term Dynamic/ T Component Other Environmental Influences This scheme differs from the explanation of citizen evaluations ' offered in Chapters 3 and 4 in two fundamental senses. The first 197 pertains to the decision model posited to underlie citizen choice. According to prior research, overall public support for the president is a result of a vaguely specified decision process in which the impact of the relevant war measures is assumed to be direct. The model of citizen-president interaction formulated in Chapter 3 asserts that such overall performance assessments are the result of a two-stage decision making process. In the first stage, the relevant citizen decisions are whether to approve or disapprove of how the president handled or managed the problems associated with a particular dimension of perform- ance (e.g. war, the economy, civil rights). The discussion presented in Chapters 3 and 4 further asserts that such specific performance assessments are a function of a set of presidentially-relevant out- comes, a set of presidentially-relevant actions, and exogenously- induced events perceived as relevant to a particular performance dimension. In the second stage, the specific performance evaluations are weighted by their perceived import to arrive at an overall perform- ance assessment. This model is depicted for the case of war on the following page. The second difference follows from the first and pertains to both the determinants of presidential evaluations and the research strategy employed to assess the impact of war. As noted previously, the arguments advanced by prior research to justify the inclusion of war measures in their support specifications imply that, aside from preventing its outbreak, the impact of war is largely beyond the control of the president. As a result, presidential behavior is summarily dismissed from consideration. Rather than excluding the (president, the explanation advanced in Chapters 3 and 4 asserts that 198 presidents have the capacity to limit the adverse impact which war is hypothesized to exert on their performance evaluations. A Two Stage Model of Citizen Evaluations Presidential ‘K Salience Behavior ’ of war war-Related Presidential \ Actions *x_ \\\\3 Evaluation of ' Overall War-Related President's \_ Performance Outcomes 7 Evaluation // Hang: ment T Exogenous “ -Other Specific war-Related/ Performance Events Evaluations Furthermore, it is argued that such attempts by the president will be most detectible in the specific performance evaluations of the public. Thus, in contrast to to the analyses which search for a direct or immediate relationship between war variables and overall support, the arguments of Chapters 3 and 4 imply that attention be directed to public assessments of the president's conduct of war, the intermediate step between war-related variables and overall public support. Attention to such specific performance assessments promises to yield . both a sharper test of the president's ability to influence public opinion and a richer understanding of the impact of war on public support. 199 Finally, although my discussion in Chapters 3 and 4 argues that war is a salient criterion of evaluation and can be expected to gener- ate declines in public support, the selective analysis of this proposition that appears in prior research.must be rejected. That is, because war constitutes a violation of the fundamental public expecta- tion of peace, it is my contention that war will operate as a criterion of assessment during any administration presiding over such conflicts. In the case of Vietnam, this implies that any worthwhile empirical analysis must include both the Johnson and Nixon Administrations. Having articulated the basic differences between prior research and the explanation of citizen and presidential behavior advanced in this thesis, let us next consider the determinants of specific per- formance evaluations in the general case and the instance of Vietnam. According to the discussion of Chapter 3, public approval of how the president has managed the problems associated with a particular performance dimension are a function of three types of influences: (1) presidentially-relevant outcomes, (2) presidentially-relevant actions, and (3) exogenously-induced events which activate the sit- uational set of performance criteria. Recall that from the discussion of Chapter 3, the PRO set is defined to include those outcomes perceived as salient to the public's sense of well-being or security and for which the president is held responsible. The PRA set consists of those presidential actions which are visible, salient, and associated with one or more of the outcomes in the PRO set. Scuh actions were further defined to include either uni- lateral or joint exercises of presidential authority. Because the intent of Chapter 3 was to examine the consequences of presidential 200 decision making in the abstract case, no further distinctions in the types of presidential actions were drawn. However, such distinctions will prove useful as the context of the discussion shifts from the theoretical to the applied case. First, a distinction is drawn be- tween presidential actions whose intent is clearly programmatic or policy oriented and those designed to exhibit a president's concern with a problem and are thus more symbolic in content. In the former, presidential action consists of a series of imposed or recommended changes in the status quo while in the latter, the objective may be to focus public attention on a problem or reassure the public of the president's commitment to devising future actions. The distinction is apparent if we compare, for example, Nixon's use of the airwaves to announce the imposition of economic controls and other policy changes to the "Malaise Speech" delivered by Jimmy Carter in 1979. Events of the first type will be labeled policy-based actions and those of the second as personal-based actions. It is my contention that both types of actions are relevant to the assessment of presidential performance and that both types will be evaluated by the public in light of their own expectations regarding appropriate presidential behavior. Given the aggregate character of past research and the analysis which follows, it is also necessary to distinguish between actions whose impact is hypothesized to raise the level of support for the president and those expected to exert a negative influence on perform- ance assessments. According to the discussion presented.in Chapter Four, whether a particular action generates a positive or negative impact will depend upon the underlying standards of evaluation which 201 are activated by the decision and upon the resulting aggregate-level distribution of public opinion regarding the action. Consequently, it is argued that the specific performance assessments reached by the public will be determined in part by four types of presidentially- relevant unilateral and joint actions: policy-based actions (negative and positive) and personal-based actions (negative and positive). Finally, it has been argued that specific performance assessments of the president will be influenced by a situational set of criteria. That is to say, presidents will be held responsible for the outcomes which result from.events or crises induced by actors and circumstances beyond the direct control of the president. As noted in Chapter 3, the occurrence of these unanticipated episodes will not only lead the public to expect a presidential response but also will activate a set of performance criteria according to which the event will be. evaluated. Ultimately then, the impact of exogenously-induced events will depend upon whether (1) the event is perceived to be detrimental or beneficial to the citizenry's sense of well-being and (2) the pres- ident's response to the episode is judged to be consistent with public expectations of appropriate presidential behavior. With respect to the types of exogenously-induced events, it will be useful to separate the relevant set of crises and events into those which are domestic in origin (e.g. national strikes, riots and other civil disturbances, civil rights demonstrations) and those which are triggered by the actions of foreign governments (e.g. hostile behavior toward the United States, the outbreak of conflicts). .As in the case of presidentially-relevant actions, it will be necessary to distinguish between events whose expected impact on performance evaluations is, 202 because of the outcomes produced and standards which are activated, positive or negative. As such, it is hypothesized that in addition to presidentially-relevant outcomes and actions, aggregate-level evaluations on a particular dimension of performance will be influenced by exogen- ously-induced domestic (positive and negative) and foreign (positive and negative) events and crises. To summarize, the discussion presented in Chapters 3 and 4 has been employed to argue that specific performance evaluations of the president will be determined by three sets of influences: presidentially- relevant outcomes, presidentially-relevant actions, and exogenously- induced events. Given the abstract character of the discussion presented in Chapter 3 and the aggregate data to be analyzed in this chapter, several distinctions pertaining to the types of relevant actions and events were introduced. The discussion thus implies that aggregate- level evaluations of the president's performance on a given dimension can be depicted as: Specific ' _ ’ I - Performance = PRO's + :giicyzggs;d + :Ri'cyzgasfd Assessments S s 8 e3 Personal-Based Personal-Based Domestic + PRA's (P05) + PRA's (Neg) Events (Pos) (5.0) Domestic + Foreign + Foreign Events (Neg) Events (Pos) Events (Neg) This specification of influences is general in the sense that it can be applied to any specific performance dimension such as the economy, energy, civil rights, or foreign policy. -As such, the next task 203 is to narrow the focus and apply this general representation to the case of war management by Presidents Johnson and Nixon. My objective is to identify those outcomes, actions, and events which are hypothe- sized to influence the pattern of evaluations on this war management dimension. With respect to the selection of relevant war-related outcomes, the impact of war on public opinion is quite complex and consequently judgements about which outcomes are most germane to the citizenry's sense of security are problemmatic. This is particularly true with respect to the Vietnam conflict. For example, research has shown that opposition to this war was based on a variety of underlying prefer- ences and rationales. There were those who opposed any American involvement whatsoever; the basis of such opposition ranged from revulsion at the potential costs of the war to the argument that American involvement was clearly imperialistic. Others expressed dis- agreement not with the military incursion per se but with the policies pursued by Presidents Johnson and Nixon in prosecuting the war. For those individuals, the war was seen as necessary for preventing Communist expansion and preserving American credibility as an ally. Moreover, the most immediate consequences of the war -- the dead and injured -- served to strengthen the commitment and resolve of some and to crystallize or harden the opposition of others. Thus, the selection of relevant outcomes is made in full recognition that they represent only one aspect of a war whose impact was multidimensional. Because of their immediacy however, I have chosen U.S. battle casualties in Vietnam as the presidentially-relevant outcome to be used in this analysis. This decision has two justifications. First, 204 it is argued that casualties measure the degree to which a war effects individual citizens and households. The discussion in Chapter 3 empha— sized that war is important as a criterion of presidential evaluation because it directly impinges upon the citizenry's sense of personal security and well-being. The death of or injury to an immediate family member, a more distant relative, or a friend brings pain and disruption to a household. Indeed, it is for this reason that the expectation of no war is fundamental. Now the likelihood that a citizen or family will experience such distress varies with the casualty rates of a war. That is to say, as casualties mount, so too does the prob- ability that a given citizen will be directly effected by the distress which war creates. It is in this sense that casualties capture the degree to which the pain of war has filtered through society and been felt by individual citizens and households. In addition to its direct impact on citizens, war also imposes costs and pain on society as a whole. The use and loss of the nation's resources -- both human and material -- in war effectively denies society both the current and future benefits which would have been produced if these resources were devoted to other purposes. Herein lies the second justification for focusing upon casualties as a presidentially- relevant outcome. It can be argued that the casualties suffered during the course of a war constitutes an adequate measure of the creative and productive opportunities foregone when society engages in war. During the course of the Vietnam.War, Presidents Johnson and Nixon shared a commonality in the sense that both were confronted with two kinds of policy decisions. .The first concerned.the'gggpguggughg ~war and involved choices about the mission of U.S. troops in Vietnam 205 and the geographic boundaries of the war. For example, U.S. policy under Lyndon Johnson shifted from U.S. troops playing an advisory role to an enclave strategy in which American forces were to remain concen- trated so as to protect U.S. bases and areas within the control of the South Vietnamese government to a strategy of attrition characterized by free fire zones and reliance upon search and destroy missions (see Gelb and Betts,l979; Lewy,l978). The tenure of Richard Nixon was marked by the Vietnamization plan in which South Vietnamese forces gradually replaced American troops in combat zones and operations which acknowledged the extension of the war to neighboring countries. A second set of decisions pertained to the use of American airpower. In this instance, both presidents were faced with almost daily decisions about whether to start, stop, reduce, or increase bombing raids over North Vietnamw It is from this series of decisions that a set of policy-based presidential war actions will be selected. The specific actions chosen for scrutiny are displayed in Table 5.2. Each of these actions received heavy coverage in the media and, with one exception (the full bombing halt of October 1968), was the subject of Gallup or Harris survey questions. The list includes changes in the scope of the conflict (Nixon's announcement of the Vietnamization plan, the Cambodian invasion, the Laotian incursion, the mining of Haiphong harbor) and five publically announced changes in U.S. bombing policy.5 It is hypothesized that these policy-based actions are presi— dentially-relevant and therefore will exert an influence on public assessments of the conduct of the war by Presidents Johnson and Nixon. (Recall that according to PrOposition 5 (page 151) the impact of such 206 decisions will vary with the proportion of the public who approve of the action. To specify whether the actions are expected to exercise a positive or negative influence on performance evaluations, I will rely upon the aggregate-level survey results produced by the Gallup and Harris organizations.6 Thus, because a majority of those express- ing an opinion disapproved of the Laotian incursion, it is hypothe- sized that this action will exert a negative influence on the propor- tion of the public who approved of Nixon's conduct of the war. In the remaining cases, the distribution of opinion leads to the proposition that the impact will be positive.7 In addition to these common policy decisions, there is one set of actions which distinguishes the Nixon period from the Johnson tenure. It was Richard Nixon who promised and subsequently implemented a policy of disengagement and withdrawal of forces from Vietnam. As noted in Chapter 4 (see Tables 4.4 and 4.5), the troop withdrawal announcements were highly publicized by the administration and by September 1972, 952 of U.S. forces in Vietnam when Nixon took office had been removed. The discussion presented in Chapters 3 and 4 suggests that this set of actions moved the status quo point in a direction favorable to public expectations and thus will exercise a positive impact on public approval of Richard Nixon's conduct of the Vietnam war. There were three personal-based presidential war actions that are hypothesized to exert a favorable influence on public assessments: Lyndon Johnson's 17-day tour of Vietnam and other nations in Southeast Asia in November 1966, Johnson's December 1967 tour of Vietnam high- lighted by his visit to the marine base at Cam.Rahn Bay, and the June .1969 wake Island meeting between Richard Nixon and President Thieu of 207 South Vietnam. In addition to their substantive objectives, these actions demonstrated the degree of presidential concern about the war and underscored the commitment of the administrations to resolving the conflict. Examination of several histories of the Vietnam conflict and chronological accounts of the war uncovered no personalvbased actions which could be expected to influence presidential performance assessments in an adverse fashion. Therefore, the set of negative personal-based actions is assumed to be empty in this particular application. Two sets of exogenously-induced events have been selected for use in this analysis. The first is domestic in character and consists of the major anti-war demonstrations which occurred in the United States between 1965 and 1972. These demonstrations were selected after consulting histories of the period and are displayed in Table 5.4. Each of these disturbances received headline coverage in major American newspapers and magazines and thus can be considered the most prominent actions of the anti-war movement. Because such events represent disruptions of domestic tranquility and, as a by-product of war, threaten to exacerbate divisions over war policy, it is argued that the occurrence of a major demonstration will exert a negative impact on public evaluations of the president's conduct of the war. The prominent Vietcong and North Vietnamese combat actions launched during the 1965-1972 period constitute the second set of exogenously- induced, war-related events. Such offensives clearly represent adverse developments in the war. This is particularly true in light of the American firepower which was applied in Vietnam.and the continuous I prognostications of success and victory by administration spokespersons. 208 Thus, the analysis to follow will examine the proposition that the set of major enemy offensives illustrated in Table 5.5 adversely influenced evaluations of Presidents Johnson and Nixon. Note that as in the case of negative personal-based actions, my reliance upon histories and chronologies of the war produced no events, domestic or foreign, which can be argued to have exerted a positive impact on specific performance assessments. To summarize, this portion of the chapter was designed to develop a general representation of specific performance evaluations and to identify the particular outcomes, actions, and events hypothesized to exert an influence on public assessments of how Presidents Johnson and Nixon managed the war in Vietnam. In light of this discussion, Equation 5.0 can be rewritten as: Public Evaluations American Changes in Scape Troo of war = Battle + of war and Bomb- + Withdragals Management Casualties ing Policy (5.1) + Personal AntiAWar Enemy Actions Demonstrations Offensives The objective of the next section is to present a measurement strategy which will be employed to operationalize this model. Qperationalization The data which.will be employed to measure public evaluations of the war performance of Presidents Johnson and Nixon are based upon Ia combination of two sources. The first is the following item 209 administered by the Gallup organization during the periods from June 1965 to April 1968 and from March 1969 to May 1970: Do you approve or disapprove of the way President Johnson (Nixon) is handling the situation in Vietnam? The variable HANDLE is defined for this period as the proportion of the public who disapproved of the manner in which the war was managed by Presidents Johnson and Nixon.8 For the period from June 1970 to October 1972, responses to the following item from the Harris survey will be utilized: How would you rate President Nixon on handling the war in Vietnam - excellent, good, only fair, or poor? Following the convention used by the Harris organization in reporting these results, HANDLE is defined as the proportion of public selecting the "only fair" and "poor" options.9 Combining the two data sources' yields a set of 79 observations over time. Essentially this disapproval time series is constructed by patching together two sets of survey results. Although both the Gallup and Harris items measure the extent to which the public was dissatisfied with the war performance of Presidents Johnson and Nixon, the results are not strictly comparable because of the differences in the format of the survey questions. As Figure 4.2 illustrates, the Harris disapproval scores are generally about nine to twelve points higher than the Gallup results. To account for these differences, the 210 binary variable, HARRIS, is defined such that: 0 if HANDLE is based upon the Gallup Poll (6/65-4/68; 3/69-5/70) 1 if HANDLE is based upon the Harris Poll (6/70-10/72) The HARRIS variable will be included in the econometric model of war disapproval to control for the variation which results from differences in the survey methodology of the Gallup and Harris organizations. In the previous section, it was argued that U.S. battle casualties represent the most immediate and salient outcome produced by the war. The measure of casualties to be used in this analysis is based upon two arguments developed in Chapter 3. First, it was asserted that public evaluations of presidential performance will vary with the degree to which outcomes diverge from the point of public expectations in the action-outcome space. Because the fundamental public expectation is no war and because it captures the extent to which outcomes have deviated from this expectation, a cumulative measure of American casualties will be employed. That is to say, the cumulative number of casualties essentially tells us how far outcomes have diverged from the ideal point of no war and no deaths. The second argument used to define the casualty measure is grounded upon the logic of promise and performance discussed in Chapter 3 and primarily concerns the relevant time period over which casualties will be summed. The promise and performance logic asserts that, upon inauguration, presidents will be expected to move prevailing outcomes toward the states most valued by the public. In other words, pres- 211 idents will be held accountable not for the status quo at the time they assume office but for their success or failure in changing the status quo over the course of the term. In the case at hand, this implies that Richard Nixon was not penalized for the casualties incurred during the tenure of Lyndon Johnson. For this reason, the U.S. casualties incurred in Vietnam are summed for the Johnson and Nixon Administrations separately. Therefore, the variable CASUALTY is defined as the sum of the U.S. casaulties incurred through month t in Vietnam where it is understood that the summation begins anew at the start of the Nixon Administration.10 In Chapter 4 it was observed that prior research has acknowledged the potential impact of presidential actions, albeit in an implicit and incomplete manner, by including rally point variables in the econometric models of public support. Two criticisms can be lodged against the manner in which presidential actions and what I have termed exogenously-induced events have been treated. The first is the exclusion of events which are domestic in character. Second, and most important to this discussion,is the manner in which the rally point notion has been operationalized. For example, the use of "time since the last rally point" by Mueller and Stimson as well as Kernell's modification of this measure presumes, in effect, that.gll events exert a uniform impact on the aggregate level of support. While it is recog- nized that measurement must invoke some assumptions, this premise of uniform impact is particularly problemmatic and will be replaced by more suitable and realisitic assumptions in this analysis. More specifically, the measurement strategy employed to opera- tionalize the policy-based presidential war actions (Table 5.2), the 212 personal-based actions, and exogenously-induced events (Tables 5.4 and 5.5) assumes that the theoretical impact of actions and events is a function of four factors: (1) an a priori likelihood of opinion change (2) the proportion of the public who are aware of the action (Proposition 6, page 154)‘ (3) the proportion of the public who approve of the action (Proposition 5, page 151) (4) the duration or length of time that the action remains salient 0 Let us first consider the likelihood of opinion change. At any given time there is a proportion of the public that disapproved of the president's war conduct during the previous month (i.e. HANDLE ). t-l In effect, these disapprovers constitute a pool of potential supporters who can be "won over" by a favorable event or presidential action. Similarly, there exists a pool of approvers who might be alienated be- cause of an adverse event or because they disagree with an announced presidential decision (i.e. loo-HANDLEt_1).. Dividing each of these proportions by 100 provides an estimate of the a prior probability that any one citizen will be induced to approve or disapprove of the pres- ident's war performance. Letting PATTRACT and PREPEL denote these probabilities, we have: PATTRACT - (100 - HANDLEt_l)/loo PREPEL - (HANDLEt_1)/100 The second and third factors are a rather straightforward applica- tions of the discussion presented in Chapter 4. It can be expected that 213 the influence of actions and events will depend upon the proportion of the public who are cognizant of the action or episode. To approximate this prOportion, I have defined the following quantity: 2 Who Regard Action 2 Who Regard Action ) AWARE 3 ( or Event as Favorable or Event as Unfavorable The impact of an event or action will also depend upon the distribution of public opinion about the event or action. To measure such opinion, let /'2 Who Regard Action If the action or event is or Event as Favorable hypothesized to reduce the level of HANDLE OPINION = Z Who Regard Action If the action or event is or Event as hypothesized to increase the Unfavorable level of HANDLE The fourth factor in specifying the impact of an event pertains to the issues of duration and salience. That is, the measure of public reaction must be adjusted to reflect the amount of time which has trans- pired since the occurrence of the event or action. Here it is assumed that the salienceof an event or action declines exponentially over a period of four months. To capture this notion, the variable SHOCK is defined and takes on a value of one during the month in which an event or action occurs and increases by one until a value of four is registered. Otherwise, SHOCK is set to zero. 214 These four components will be employed to operationalize the sets of actions and events hypothesized to influence the level of war dis- approval. First, consider the policy-based presidential actions which are expected to reduce the level of disapproval (i.e. all actions listed in Table 5.2 except the Laotian incursion and the bombing halt of October 1968). To assess this proposition, the variable POLICY is defined as follows: SHOCK)*100 POLICY = ( PATTRACT * (AWARE*OPINION) Similarly, the variable LAOS is constructed to evaluate the hypothesis that this action increased the level of war disapproval: saocx)*loo LAOS = ( PREPEL * (AWARE*OPINION) Because the personal set of actions and the exogenously-induced events listed in Tables 5.4 and 5.5 were not the subject of opinion poll inquiries, it is necessary to formulate a proxy measure for the (AWARE*OPINION) component of the action and events variables. Conse- quently, as a surrogate for these proportions, I will employ the probabilities associated with the president's war performance in the previous month -- HANDLEt_1/IOO and (lOO-HANDLEt_1)/100. This decision is justified on the grounds that in the absence of other information, the best estimate of opinion regarding an event or action at time t is the level of approval or disapproval at time tel. To assess the impact of personally—based presidential war actions, the 215 following variable is defined: SHOCK PERSONAL - ( PATTRACT * ((IOO-HANDLEt_1)/IOO) )*100 In analogous fashion, two variables, VCOFF and DEMO, were constructed as measures of enemy offensives and anti—war demonstrations respectively. The formulae used to calculate the relevant values are: SHOCK VCOFF = ( PREPEL * (HANDLEt_1/100) )*100 SHCOK DEMO = ( PREPEL * (HANDLEt_1/100) )*100 The final measure to be considered concerns the troop withdrawal policy pursued during the Nixon Administration. As noted in Chapter 4, this policy was comprised of a set of announcements in which President Nixon informed the American public of the number of American troops to be withdrawn from Southeast Asia and the timetable for their removal. My objective then is to develop a measure of public satisfaction with this set of actions. At ten points during the Nixon Administration the Harris organization administered the following item to its nation- wide sample: In general, do you feel the pace at which the President is withdrawing troops from Vietnam is too fast, too slow, or about right? The survey dates and responses to this question are presented in Table_5.3. Assuming that selection of the "too slow" or "too fast" Options represents disagreement with the Nixon policy, the variable IWITHDRAW is based upon the proportion of the sample selecting the 216 "about right" option. As constructed, WITHDRAW assumes its initial nonzero value in June 1969, remains constant at that value until the next poll date, and changes in similar fashion throughout the remainder of the 1969-1972 period. Rather than relying upon a mechanistic explanation-concerning the impact of war on public opinion, this discussion has attempted to develop an alternative model which is firmly linked to theories of citizen and presidential decision making and to operationalize the identified deter- minants in a meaningful and plausible manner. The operational version of the model presented in Equations 5.0 and 5.1 can be expressed as follows: HANDLE = bo + bIHARRIS + bZCASUALTY + bBWITHDRAW + baPOLICY + bSLAOS + b6PERSONAL + b7DEMO + bBVCOFF + et (5.2) where, on the basis of Chapters 3 and 4 as well as the arguments in this chapter, it is anticipated that b b b b b8 >0 and 133,134,136 < o 1’ 2’ 5’ 7' That is, it is hypothesized that CASUALTY, LAOS, DEMO, and VCOFF will increase the level of public dissatisfaction with a president's war performance while WITHDRAW, POLICY, and PERSONAL should produce declines in the level of dissatisfaction. Model Estimation_and Evaluatipn It is my contention that estimation of the model specified in Equation 5.2 provides a plausible test for evaluating the explanation 217 . of public support developed in Chapter 4. Moreover, such an exercise will also supply evidence regarding the capacity of presidents to in- fluence and, in effect, regulate such performance assessments. For example, Equation 5.0 represents the general case of specific performance assessments and clearly the choice of policy-based and personal-based actions included in the expression are subject to presidential dis- cretion. They are, simply put, the levers of control. .In the specific case of Vietnam, presidential capacity to control can be assessed by examining the estimated coefficients for POLICY, LAOS, WITHDRAW, and PERSONAL. The discovery that the estimates are implausible or that the variables do not operate as significant determinants of HANDLE would imply that the model as well as Propositions 5 and 6 are to be rejected. Model evaluation must also take place in a comparative context however. Although public evaluations of war performance have not been scrutinized by prior research, the rationales used to justify the inclusion of war variables in econometric models of overall presidential support do lead to three alternative or baseline models against which the performance of the model in Equation 5.2 can be compared. The first model is essentially an application of the war weari- ness prOposition discussed in the second section of this chapter. Recall that this proposition.implies that disapproval of the president's conduct of the war should increase as a function of the war's duration. This, in turn, leads to the following econometric specification: DURATION + c HARRIS +'e (Model 1: war Weariness) HANDLE = co + c1 2 t 218 Here, DURATION is measured as in the Mueller and Stimson analyses - the amount of time, in fractions of years,elapsed in.the war. Because no argument is offered to suggest that the impact of a war's duration is term-based, this time counter is constructed to increase continually from June 1965 to October 1972. As in the case of Equation 5.2, the HARRIS variable is included to control for the fact that the data for the dependent variable come from two polling sources. The second model to be examined is based upon the Mueller argument (1973) that the public reacts not to the duration of war but to the cumulative costs and pain which war produces. The equation employed to assess this proposition is simply an extension of the model employed by Mueller in his analysis of responses to the "war as mistake question". That is, HANDLE - (10 + d LGSUMCAS + d HARRIS + et (Model 2: Cumulative Pain) 1 2 where LGSUMCAS - the base 10 logarithm of the sum of U.S. battle casualties in Vietnam As in the case of the war weariness model, no distinction is drawn between the terms of Johnson and Nixon. Thus, casualties are summed throughout the 1965-1972 period. The intensity of war proposition advanced by Samuel Kernell pro- vides the foundation for the third baseline model. Recall that according to Kernell, public opinion toward a war will vary with the intensity of the conflict. Applied to the case at hand, this produces the following specification: 219 HANDLE 3 go + ngEATH + ngARRIS + et (Model 3: Intensity Argument) where DEATH - the number of U.S. battle casualties in Vietnam during month t The differences between these three models and what might be termed the Presidential Action Model specified in Equation 5.2 are quite clear. Each of the baseline models excludes exogenously-induced events from consideration; the models and their underlying rationales also imply that the behavior of the president is irrelevant to public Opinion. Thus, if presidents are in fact passive bystanders and can do nothing to influence how they are evaluated by the public, then it should be the case that (1) the coefficients for POLICY, WITHDRAW, LAOS, and PERSONAL in the Presidential Action Mbdel will be insigni- ficant and (2) one or more of the baseline models will outperform the Presidential Action Model in accounting for the time path of HANDLE. Let us first consider the performance of the Presidential Action Model. Table 5.6 displays the OLS results, the results obtained from application of the Cochrane-Orcutt, pseudo—GL8 technique, and the impact ranges for the variables in the equation. The impact range is the minimum and maximum.value obtained from performing the calculation -- b * variable -- for all time points in the sample. As such, the impact range conveys the minimum and maximum impact on war disapproval attri- buted to each variable. The estimates reveal that presidential actions, both personal and policy-based, did indeed exercise a significant Iimpact on war disapproval. For example, the coefficient for POLICY~ indicates that publicized changes in the scape of the war and U.S. 220 bombing policy generally lowered disapproval by approximatelwaive points. That Nixon was penalized for launching the Laotian incursion is another plausible inference drawn from the results. This action increased dissatisfaction with Nixon's conduct of the war by nearly seven percentage points in February 1971. While somewhat smaller in magnitude, the Johnson trips to Vietnam and the Nixon-Thieu meeting were also effective in controlling the level of HANDLE. Note as well that, according to the estimates, Nixon's policy of disengagement appears to account for his ability to avoid the widespread disapproval experienced by Lyndon Johnson. This can be seen more readily if we compare the last time points -- April 1968 and October 1972 - in the two administrations. The cumulative casualties incurred were 21,663 under Lyndon Johnson and 15,153 during the Nixon Administration. According to the coefficient for CASUALTY, this implies that Nixon's estimated disapproval score was six points lower than Johnson's on the basis of casualties alone. However, because satisfaction with Nixon's rate of troop withdrawal stood at 38% in October 1972, disapproval was further reduced by 5.5 points. Despite the fact that the war continued throughout the 1969-1972 period, disapproval of Nixon's war conduct is predicted to be 11.5 points lower than Johnson's because of public sentiment toward the policy of trOOp withdrawals and the effect Of that policy in preventing further growth in the casualty index. Finally, the estimates reveal that, aside from their intended military impact, offensives launched by the Vietcong and North Viet- ~ namese also exerted an impact on Opinion within the United States. This was not the case with anti-war disturbances, clearly the least 221 robust of the coefficients estimated. The GLS results indicate that, contrary to expectations, the occurrence Of an anti-war demonstration tended to lower disapproval of the president's war conduct. Although such a result suggests that the rally effect noted by Mueller (1970, 1973) may extend to domestic episodes, the insignificant coefficient underscores the speculative nature of such a conclusion. With only DEMO as an exception, all coefficient estimates are statistically significant and conform to their hypothesized directions. The results also indicate that the overall fit of the model is satisfactory. The equation accounts for 80% of the variance in the HANDLE series, the 5.59 value for the standard deviation of the regression is well below the standard deviation of 11.96 recorded for HANDLE, and the Durbin-Watson statistic exceeds the Thiel-Nagar Q value of 1.71 for the .05 level. The comparison between predicted and actual war disapproval scores is presented in Figure 5.1 for the John- son sample points and in Figure 5.2 for the Nixon series. The model misses or fails to capture the decline in Johnson's disapproval in mid-1965 and the rise in disapproval in mid-1967. The other major source of error occurs early in the Nixon term where the estimated points are somewhat inflated. Nevertheless, the pictoral representations indicate that the predicted scores track actual war disapproval reasonably well. This assessment seems all the more reasonable when we turn to the comparative context. Both the OLS and GLS estimates for the three baseline models are presented in Tables 5.7, 5.8, and 5.9. It is evident that the war weariness proposition (Table 5.7) fares rather poorly when applied to this data. The coefficient is not only in the 222 in the wrong direction but statistically insignificant as well. The same is true for MOdel 2 (Table 5.8) which operationalizes the cumulative pain rationale. Finally, the results of the war intensity argument do conform to expectations; the coefficient for DEATH (Table 5.9) is positive and statistically significant. While clearly the best of the baseline models,the explanatory power of the war intensity model is, as the summary statistics suggest, limited and not on a par with the results Obtained for the Presidential Action Model. Further illustration Of this point is straightforward. The number of monthly battle casualties in Vietnam reached its highest point Of 2,197 during the TET Offensive Of 1968. According to the estimates reported in Table 5.9, these casualties are predicted to increase dissatis- faction with Johnson's war management by only 2.6 points, clearly a questionable, if not implausible, result. Conclusion On the basis of this comparison, it is concluded that the Presidential Action Model outperforms the three baseline models and thus promises tO provide a richer understanding of how war influences public evaluations than that which is provided by current wisdom. The results also provide support for the alternative explanation of presidential evaluations advanced in Chapter 4 and, in particular, for Propositions 5 and 6 which assert that presidential actions will exert an influence on public assessments and that this influence will vary as a function of the salience and public approval of the action. 'Although.more general conclusions must await the application of 223 Equation 5.0 to other dimensions of presidential performance, I believe that this exercise has succeeded in introducing an empirically-supported and theoretically-based explanation of specific performance evaluations. Nonetheless, there remains a question regarding the degree of presidential control. While rather convincing evidence about the comparative performance of the Presidential Action Model has been offered, the results do not directly convey the extent to which war disapproval can be regulated via presidential actions. To investi- gate this issue, I have constructed three instrumental variables which summarize the impact of outcomes, exogenously-induced events, and those actions under the direct and immediate control of the president. These three variables were formulated as follows: OUTCOME - (.001*CASUALTY) EXOG = (.198*VCOFF) - (.056*DEMO) ACTION = (.425*LAOS) - (.144*WITHDRAW) - (.181*POLICY) - (.140*PERSONAL) Each of the instrumental variables represents the portion of the pre- dicted war disapproval score that can be attributed to events beyond the president's control, to outcomes partially manipulable by the presidents, and to actions whose choice lies completely within the authority Of the president. The values of these instruments are plotted in Figure 5.3. It is quite obvious that the measure of battle cas- ualties was most responsible for the level of war disapproval. Nevertheless, presidential actions are clearly important. This is 224 particularly true during the Nixon term where the ACTION variable partially cancelled out the upward pressure which war casualties exerted on assessments of Nixon's performance. The graph reveals that Nixon's behavior presents a rather vivid example of how actions can be employed to influence and thereby manage public support. The impressions drawn from Figure 5.3 are reinforced by the 11 results Obtained from regressing the instruments on HANDLE. The coefficients, betas, and elasticities Obtained were: b BETA Elasticity OUTCOME 1.61 .70 .35 EXOG .153 .02 .001 ACTION 1.29 .38 .125 Again, it is clear that casualties were the most important influence on war disapproval. Yet this is not surprising given that war constitutes such an abrubt disruption of public expectations. The ACTION coefficient and beta suggest that, despite the impact of casualties, there is much room for presidential maneuver and control. This information together with the prior analysis reveals that presidents are not to be considered passive bystanders. Rather, presdients, via their actions and the manner in which actions influence outcomes, are capable of influencing public evaluations and therefore.should be considered masters of their own fate. 225 NOtes to Chapter 5 It should be noted that the inclusion of the time variable in Mueller's specification precluded, because of the likely multicollinearity problems, the use of a more sophisticated measure of war. For this reason, Mueller is careful to qualify these conclusions. He suggests that the decline in support generated by the coalition of minorities effect (see Table 4.1) may have obscured the impact of Vietnam on Lyndon Johnson's public support. Essentially, Stimson modifies the Mueller specification by adding the square of time (see Table 4.1) to account for the upturn in support Observed near the end of some presidential terms. Because of the manner in which the data was organized, Stimson's estimates represent the impact of both Korea and Vietnam on public support. The methodology employed by Stimson and LeGette consists of two steps. First, a quadratic response function is estimated for each presidential term (Truman through Nixon); second and with- out correcting for the serial correlation, the residuals from this step are employed as dependent variables and regressed on duration measures for wars, recessions, and rally points. The results of this second step serve as the basis for the conclusions regarding the import of Korea and Vietnam, For the Vietnam period, Mueller analyzes the aggregate-level responses to the following Gallup item: "In view of the develop- ments since we entered the fighting in Vietnam, do you think the U.S. made a mistake sending troops to fight in Vietnam?" This item was administered twenty-two times between August 1965 and January 1973. A similarly worded item was analyzed in the case of the Korean War. Actually, there were several salient decisions made by the Johnson Administration during June and July 1965. According to Gelb and Betts (1979:116-130), it was during this period that Johnson and his advisors reached a decision to increase the number of U.S. troops in Vietnam and to expand the mission of those trOOps. Thus, in a televised press conference on July 13, Johnson suggested that American forces might undertake a more active combat role in South Vietnam" On July 28, Johnson announced at a televised press confer- ence that troops levels in Vietnam would be increased from 75,000 to 125,000 and implied that this level would escalate in the future. The transcripts of these press-conferences were obtained from the relevant issues of CongreSSiOnal'QuarterlyWeekly Reports. 6. 10. 11. 226 While admittedly post hoc, this is, given available data, one of the few plausible methods for arriving at such a determination. Because the Gallup organization did not administer the handling question between May 1968 and February 1969, the full bombing halt announced by Johnson in October 1968 will not be included in the analysis. It should also be noted that one other prom- inent action taken by Johnson - the imposition of a 6! income tax surcharge - could be expected to exercise a negative impact on approval of Johnson's war management. According to the Gallup poll of January 1968, 792 of the public disapproved of the proposed surtax. However, the tax was not signed into law until June 1968 and thus its most noticeable impact was likely to occur during the period in which evaluations of Johnson's performance were not measured by Gallup. There were five months during this period in which this item was not administered. Consequently, a disapproval score was interpolated by using the polls on either side of the gap. The "only fair" and "poor" options are generally combined and reported by Harris under the label of "Unfavorable." There are five interpolated disapproval scores in this Harris series. Information pertaining to U.S. casualties in Vietnam was obtained from the relevant appendices in Milstein (1974) and Littauer and Uphoff (1972). Also included in the equation was the instrument for the Harris variable. Because it is not directly germane to the discussion, this portion of the results is not reported. The Impact of War on Public Support for American Presidents: Author Mueller Hibbs Stimsonl Stimson2 6 LeGette Kernell 227 Table 5.1 A Summary Korea Measure Dummy Variable Dummy Variable Dummy Variable Time Elasped During War Time Squared Log of Monthly U.S. Casualties m -18.43* (3.43) 2.49 (4.37) -l.26 (1.06) -20.58 7.68 -1.62 (1.13) Vietnam Measure Dummy Variable Dummy Variable Dummy Variable Time Elasped During War Time Squared Monthly U.S. Casualties. U.S. Sorties Results -.28 (2.79) 6.84 (4.15) -1.26 (1.06) '5.93 1.44 -.0059* (.0016) (.00024) *-- Denotes that the estimated coefficient is statistically significant. The parenthesized figures are the standard errors of the estimates. l- Stimson estimates only a single coefficient for the wars in Korea and Vietnam. 2- The Stimson and LeGette article does not report the standard errors of the coefficients. Sources: Mueller, 1970, p. 28, Table 2, Equation 6. Hibbs, 1974, p. 288, Table 2. Stimson, 1976, p. 17, Table 5, Equation 3. Stimson and LeGette, 1975, Table 2. Kernell, 1978, p. 522, Table A.2. The Set of Policy-Based Presidential War Actions Date June 1965 July 1966 April 1968 October 1968 November 1969 May 1970 February 1971 May 1972 September 1972 228 Table 5.2 Action Retaliatory Air Raids on North Vietnam Resumed U.S. Bombing Extended to Oil Dumps in Hanoi and Haiphong Partial Bombing Halt Announced (Negotiations to Begin; LBJ Withdraws from Campaign) LBJ Declares Full Bombing Halt Nixon Delivers Silent Majority and Vietnamization Speech Nixon Announces Cambodian Invasion Laotian Incursion Nixon Announces U.S. Mining of Haiphong Harbor U.S. Resumes Heavy Bombing of North Vietnam 5 70 64 56 39 * 59 Dis N.O. 22 12 11 19 26 10 6 17 32 12 42 19 24 17 32 13 * Denotes that Opinion Distributions were obtained from the Harris Poll. All other results were gathered from the Gallup Poll. 229 Table 5.3 * Public Opinion Toward Nixon's Troop Withdrawal Policy Date June 1969 October 1969 December 1969 April 1970 May 1970 July 1970 January 1971 November 1971 February 1972 May 1972 Too Slow 42% 29 26 34 27 32 33 53 45 42 Too Fast 16 13 12 About Right 29 49 56 47 46 48 55 38 44 38 Not Sure 13 16 13 11 14 12 *These responses were generated in response to the following Harris Poll question: "In general, do you feel the pace at which the President is withdrawing trOOps from Vietnam is too fast, too slow, or about right?" Date November 1965 April 1967 October 1967 August 1968 October 1969 November 1969 May 1970 April 1971 May 1971 230 Table 5.4 Major Domestic Anti-war Demonstrations 1965 - 1972 Description 25,000 stage "March on washington for Peace in Vietnam" 100,000 demonstrators, including Martin Luther King, stage protest against the war 50,000 march against the war in Washington,D.C. Pentagon stormed by protestors Demonstrations and violence occur in Chicago during Democratic National Convention First Vietnam war Moratorium Second Vietnam War Moratorium 250,000 march in Washington,D.C. Anti-war demonstrations erupt on U.S. campuses; Kent State shootings 200,000 rally against war in Washington,D.C. Washington,D.C. police arrest 12,000 during three days of protests Date March 1966 December 1966 February 1968 March 1968 February 1969 July 1969 March 1970 May 1972 231 Table 5.5 Major North Vietnamese-Viet Cong Offensives 1965 - 1972 Description U.S. Special Forces outpost in Ashau abandoned after 36-hour seige by Viet Cong Major U.S. Air Base at Tonsonhut attacked by Viet Cong TET Offensive North Vietnamese and Viet Cong launch offensives against Saigon and 100 military targets in Saigon region Offensives launched by North Vietnamese and Viet Cong on Hue, Ben Het, and Saigon Spring Offensive launched by North Vietnamese and Viet Cong Major North Vietnamese offensive launched Variable CASUALTY WITHDRAW POLICY LAOS PERSONAL DEMO VCOFF HARRIS CONSTANT RHO R2 St. Dev. Durbin-Watson 232 Table 5.6 Evaluations of Presidential War Performance The Presidential Action Model OLS .0016* (.0001) (.035) -0230* (.089) (.294) -.134 (.111) .048 (.068) .061 (.083) 12.40* (1.97) 27.45* (1.21) .86 4.71 .53 Impact Range GLS Minimum Maximum .001* .145 21.66 (.0004) -.144* 0 '8.06 (.061) -.181* O -5.23 (.056) .425* 0 6.8 (.16) -.140* O -3.5 (.070) “.056 0 -2.2 (.045) .198* 0 7.0 (.048) 12.45* 0 12.45 (2.79) 32.92* (4.84) .835 .80 5.59 1.85 *-- Denotes the estimated coefficient is statistically significant. The parenthesized figures are the standard errors of the estimates. 233 Table 5.7 Evaluations of Presidential War Performance Model 1 Variable OLS GLS DURATION .311 -.560 (.764) (1.91) HARRIS 15.89* 14.38* (3.59) (4.11) CONSTANT 36.17 43.63 RHO -- .884 R2 .48 .35 St. Dev. 8.72 9.66 Durbin-Watson .22 2.02 *-- Denotes that the estimated coefficient is statistically significant. The parenthesized figures are the standard errors of the estimates. Variable LGSUMCAS HARRIS CONSTANT RHO R2 St. Dev. Durbin-Watson 234 Table 5.8 Evaluations of Presidential War Performance Model 2 OLS 4.788* (1.68) 13.48* (2.32) 18.291* (6.65) .53 8.298 .24 GLS -9.375 (11.295) 14.34* (3.87) 83.816 (73.01) .904 ## 13.623 2.08 *-- Denotes that the estimated coefficient is statistically significant. The parenthesized figures are the standard errors of the estimates. ##--The formula for calculating this goodness-of-fit statistic produced a negative value. 235 Table 5.9 Evaluations of Presidential War Performance Model 3 OLS GLS * DEATH .013* .0012 (.003) (.0022) HARRIS 23.61* 14.34* (2.2) (3.73) CONSTANT 28.9* 40.3* (1.95) (3.73) RHO -- .874 R2 .61 .43 St. Dev. 7060 8.95 Durbin‘Watson .22 2.02 *-- Denotes that the estimated coefficient is statistically significant. The parenthesized figures are the standard errors of the estimates. 236 we mm mo hm no um mo Nu No mumcnm m.~ > ooavmawmo: om mamaaonma mom >nncmH Ema cwmmvunonncmH \\\\\llf/ Hoao Homu Hoom 237 mm mo mw mo bm no wm mo mm No wnmawonma ~ \ \ > ooSvmnHmo: 0m wnmmwnnmm man >nncmH 2mm memuvnooncmp Home Howe kuw you» 238 +~m +~o +u memenm m.w arm Harmon 0m cannoamm. >onwosm. man mOHHOZ 4 ><1 >OHHOZ «111 _ <\ CONCLUSION A Summary of Results This thesis has attempted to accomplishthree objectives. The first was to develop the argument that public support Operates as a presidential resource and that because of its value, presidents will be motivated to control or manage this commodity. Accordingly, the discussion in Chapters 1 and 2 presented both empirical and journal- istic evidence to substantiate the assertion that how the public evaluates the president exercises an influence on outcomes deemed important by the president. In particular, it was shown that perform- ance assessments exert an impact in both the voting booth and the halls of Congress. It is my contention that the incentive for pres- idents to manage public support arises from such results. Simply put, public support constitutes a means or instrument which can be employed by presidents to pursue and achieve their goals. The second objective of this thesis was to demonstrate that presidential management of public support is a topic amenable to systematic investigation. To accomplish this task, a model of citizen- president interaction was constructed in Chapter 3. The exercise was designed to illustrate how both the general level of public support and specific performance assessments are a joint function of citizen and president decisions. The mathematical model developed in this chapter reveals the manner in which outcomes and actions link the choices made by citizens and presidents. The discussion also empha- sized that although constrained by potential contradictions among public 239 240 expectations, congressional preferences, and their own goals, presi- dents maintain the ability to choose how these conflicts will be resolved. The third objective of this discussion was to establish that the model of citizen-president interaction provides a useful mechanism for investigating the question of support management.: Thus, Chapter 4 offered an alternative explanation of public support and developed specific propositions regarding how the levers available to the president -- the content, timing, and publicity associated with actions -- can be employed to influence public assessments. Several of these prOpositions were assessed in Chapter 5. The purpose of this discussion was to establish the empirical relevance of the alternative explanation and to demonstrate that public evaluations of how Presidents Johnson and Nixon managed the Vietnam war were influenced by the decisions and behaviors of these presi- dents. In contrast to the denial of presidential control and mechan- istic explanations of citizen evaluations so prevalent in the litera- ture on public support, the results of the econometric analysis strongly support the conclusion that performance assessments are sub- Ject to presidential control. Declines in Support: Inevitable Decay or Presidentially_lnduced? Let us next consider how these results compare to conventional wisdom regarding presidential behavior and the management of public support. Over the past decade the claim that the presidency is an inherently unmanageable office has gained acceptance among those who chronicle and study American national politics. Perhaps the most 241 compelling evidence used to support this assertion is the fate of Ronald Reagan's immediate predecessors. In effect, the previous four presidents were driven from office -- two via the ballot box and two by "voluntary" means. At the time of their demise, each of these presidents had practically exhausted the goodwill and.support of the American public. As Chapter 2 has asserted, the depletion of this resource is synonomous with waning presidentialinfluence and in- effectiveness among government decision makers. Yet recent scholar- ship on the presidency expresses little surprise at the demise of presidents and typically observes that the inability to manage nation- al affairs and to maintain public support is to be expected. Such observations arise from two lines of inquiry. First are the analyses of the "post-imperial" presidency. These studies begin with a general description of recent changes in the political system - the growth of the federal bureaucracy, the erosion of strong party organizations, the dispersion of.decision making power in Congress, and the reassertion of congressional independence vis-advis the executive branch. It is then argued that such changes have fractionalized the political process to the point where it is exceedingly difficult, if not impossible, for any president to exercise control over the affairs of state. The inability of presidents to manage public support is cited as but one manifestation of a more general paralysis confronting the institution of the presidency. The conclusions of such institutionally-based studies are reinforced by public opinion research and particularly the analyses concerned with explaining changes in public support over time. .The general con- clusion which emerges from this literature essentially argues that 242 public demands have exceeded the ability of the national government to provide services and rectify social or economic ills. Simply put, expec- tations about government performance have grown increasingly exaggerated or unrealistic. At the same time the capacity to control outcomes has weakened, demands for such control have increased. One implication of this irony concerns the management of public support. Whether characterized as a coalition of minorities effect or an expectation-disillusionment-forgiveness cycle, the president will inevitably be faced with public frustration and the loss of support during the course of his term. Advocates of this perspective point to the support enjoyed by recent presidents and conclude that, as expected, erosion did occur. In such a fashion the studies of political structure and public opinion converge and lead to the conclusion that the situation which confronted Presidents Johnson, Nixon, Ford, and Carter were to a great extent inescapable. The argument seems to be plausible and appealing. The fundamental problem lies not in the weakness of particular leaders but in the constraints which have been imposed upon leadership in general. While presidents are not completely absolved, the inability to manage outcomes and public support is shown to result from government complexity, poli- tical disorganization, and exaggerated public expectations. The explané ation is reassuring since solutions to the problem are immediately apparent. Structural reform is required so that complexity can be re- duced and manageability reestablished. The demand side of the problem can be rectified by a more realistic rhetoric on the.party.of candidates and officeholders which will convince the public that expectations must be scaled down. 243 This thesis challenges this prevailing view in three respects. First, I have argued that the conclusions about the inevitable erosion of support are grounded upon explanations whose premises have not been systematically examined and empirical analyses which are plagued by conceptual and methodological problems. Second, the thesis has formu- lated a complete and plausible calculus of citizen and presidential decision making which does not depend upon mechanistic or ad hoc characterizations. Third, it has been shown that this underlying theory of decision making leads to a model of performance assessments which includes the behavior of the president, is empirically-supported, and outperforms several alternative models. In contrast to the conclusions that our four previous presidents were simply the victims of inescapable problems, this thesis leads to a different position. Over the past fifteen years, the American public has experienced a seemingly endless war, economic malaise, and a scandal whose scope extended throughout the executive branch. The analysis presented in this discussion implies that presidential actions were, in part, responsible for the eroption of these problems and the erosion of support experienced by these presidents was a consequence of their failure to contain and solve these difficulties. In sum, declines in support are not inevitable but presidentially induced. BIBLIOGRAPHY BIBLIOGRAPHY Abramson, Paul R. 1977. The Political Socializaton of Black Americans: A Critical Evaluation of Research On Efficacy and Trust. New York: Free Press. American Institute of Public Opinion. The Gallup Opinion Index. Numbers 1 (June 1965) - 161 (December 1978). Princeton, N.J. Anderson, Jack. 1980. "Iran Attack Would Aid Carter," Lansing State Journal, September 1980. Arsenau, Robert B. and Raymond E. WOlfinger. 1973. "Voting Behavior in Congressional Elections," A Paper Prepared for Delivery at the An- nual Meeting of the American Political Science Association. Axelrod, Robert. 1972. "Where the Votes Come From: An Analysis of Electoral Coalitions," American Political Science Review, March 1972, 11-20 0 Barber, James David. 1977. The Presidential Character: Predicting Per- formance in the White House. 2nd Edition. Englewood Cliffs, N.J.: Prentice-Hall. Bloom, Howard S. and Douglas Price. 1975. "Comment: Voter Response to Short-Run Economic Conditions: The Asymmetric Effect of Prosperity and Recession," American Political Science Review, December 1975, 1239-1254. Boulding, Kenneth E. 1976. Conflict and Defense. N.Y.:Harper and Row. Boyd, Richard. 1972. "Papular Control of Public Policy: A Normal Vote Analysis of the 1968 Electidn," American Political Science Review, June 1972, 429-449. Brady, David W. and Naomi B. Lynn. 1973. "Switched-Seat Congressional Districts: Their Effect on Party Voting and Public Policy," Mid- west Journal of Political Science, August 1973, 528-543. 244 245 Breton, Albert. 1974. The Economic Theory of Representative Government. Chicago: Ald ine . Brody, Richard A. and Benjamin I. Page. 1975. "The Impact of Events on Presidential Popularity: The Johnson and Nixon Administration," in Aaron Wildavsky, ed., Perspectives on the Presidency. Boston: Little, Brown. Burns, James MacGregor. 1973. Presidential Government. Bostonzfioughton Mifflin. Campbell, Angus. 1966. "Surge and Decline: A.Study of Electoral Change," in Campbell, et. al., Elections and the Political Order. New York: John Wiley. Campbell, Angus., Philip E. Converse, Warren E. Miller and Donald E. Stokes. 1960. The American Voter. New York:John Wiley. Chester, Lewis, Godfrey Hodgson, and Bruce Page. 1969. An American Melodrama: The Presidential Campaign of 1968. New York:Viking Press. Citrin, Jack. 1974. "Comment:The Political Relevance of Trust in Govern- ment," American Political Science Review, September 1974, 973-988. Congressional Quarterly Almanac. Volumes 9(1953) - 34(1978). washington, D.C.: Congressional Quarterly, Inc. Congressional Quarterly Weekly_Report. washington,D.C.: Congressional Quarterly, Inc. Converse, Philip E. 1972. "Change in the American Electorate," in Angus Campbell and Philip E. Converse, eds.,-The Human Meaning_g£ Social Changg. New York: Russell Sage Foundation. Cornwell, Elmer. 1965. Presidential Leadership of Public Opinion. Bloomington: Indiana University Press. 246 Cover, Albert. 1977. "One Good Term Deserves Another: The Advantage of Incumbency in Congressional Elections," AmeriCan Journal of Politi- cal Science, August 1977, 519-531. Davidson, Roger H., David M. Kovenock, and Michael K. O'Leary. 1966. Congress in Crisis: Politics and Congressional Reform. Belmont, Cal.: wadsworth Publishing. ., and Glenn Parker. 1972. "Positive Support for Political Insti- tutions: The Case of Congress," Western Political Quarterly, Decem- ber 1972, 600-612. Davis, Otto A., Melvin J. Hinich, and Peter C. Ordeshook. 1970. "An Expository Development of a Mathematical Model of the Electoral Process," American Political Science Review, June 1970, 426-448. Dennis, Jack. 1975. "Dimensions of Public Support for the Presidency," A Paper Prepared for Delivery at the Annual Meeting of the Midwest Political Science Association. Chicago, Illinois. Dexter, Lewis A. 1977. "The Representative and His District," in Robert Peabody and Nelson Polsby, eds., New Perspectives on the House of QRgpresentatives. 3rd Edition. Chicago: RmmiMcNally. DiClerico, Robert E. 1979. The American President. Englewood Cliffs, N.J.: Prentice-Hall. Dodd, Lawrence C. and Richard L. Schott. 1979. Congress and the Ad- ministrative State. New York: John Wiley. Downs, Anthony. 1957. An Economic Theogy of Democracy. New York: Harper and Row. Edwards, George C. III. 1976. "Presidential Influence in the House: Presidential Prestige as a Source of Presidential Power," American Political Science Review, March 1976, 101-113. . 1980. Presidential Influence in Congress. San Francisco: W.H. Freeman. 247 Evans, Rowland, Jr. and Robert D. Novak. 1966. Lyndon B. Johnson: The Exercise of Power. Cleveland: World Publishing Company. . 1971. Nixon in the White House. New York: Random House. Fenno, Richard F. 1973. Congressmen in Committees. Boston: Little, Brown. . 1975. "If, as Ralph Nader Says, Congress is ‘the Broken Branch,‘ How Come We Love our Congressmen So Much?" in Norman Ornstein, ed., Congress in Change. New York: Praeger. . 1977. "U.S. House Members in Their Constituencies: An Explora- tion," American Political Science Review, September 1977, 883-917. Ferejohn, John A. 1977. "On the Decline of Competition in Congressional Elections," American Political Science Review, March 1977, 166-176. Fiorina, Morris P. 1977. Conggess: Keystone of the Washington Establish- ment. New Haven: Yale University Press. . 1979. "Short and Long-term Effects of Economic Conditions on Individual Voting Decisions," California Institute of Technology, Social Science WOrking Paper, Number 244, June 1979. , and Herbert Weisberg. 1977. "Candidate Preference Under Uncertainty: An Expanded View of Rational Voting," California Institute of Technology, Social Science Working Paper, Number 138, June 1977. Fisher, Louis. 1978. The Constitution Between Friends. New York: St. Martin's. Franck, Thomas M., and Edward Weisband. 1979. Foreign Policy by Conggess. New York: Oxford University Press. ‘ Frey, Bruno and Frederich Schneider. 1978. "Am Empirical Study of Politico-Economic Interaction," Review of EconomicS'and'Statistics, Volume 60, 174-183. 248 Gallup, George H., ed. 1972. The Gallup Poll: Public Opinion 1935- 1971. Three Volumes. New York: Random House. . 1978. The Gallup Poll: Public Opinion 1972-1977. Two Volumes. Wilmington, Delaware: Scholarly Resources, Inc. Gelb, Leslie H. and Richard K. Betts. 1979. The Irony of Vietnam: The System WOrked. Washington,D.C.: The Brookings Institution. Glass, Andrew J. 1973. "Congress Report - Watergate Diminishes Nixon's Leverage, Force Series of Legislative Compromises," National Journal, 21 July 1972, 1049-1056. Golden, David G. and James M. Poterba. 1980. "The Price of Popularity: The Political Business Cycle Reexamined," American Journal of Political Science, November 1980, 698-714. Greenstein, Fred I. 1978. "Change and Continuity in the Mbdern Presi- dency," in Anthony King, ed., The New American Political System. Washington,D.C.: The American Enterprice Institute. Halberstam, Drvid. 1972. The Best and the Brightest. New York: Random House. Hibbs, Douglas A., Jr. 1974. "Problems of Statistical Estimation and Causal Inferences in Time-Series Regression Models," in H. L. Costner, ed., Sociological Methodology: 1973-1974. San Francisco: . 1980. "On the Demand for Economic Outcomes: Macroeconomic Performance and Mass Political Support in the United States, Great Britain, and Germany," A Paper Prepared for Delivery at the Annual Meeting of the American Political Science Association, Washington, D.C. Kearns, Doris. 1976. Lyndon Johnson and the American Dream. New York: Harper and Row. Kenski, Henry C. 1977a. "Inflation and Presidential Popularity," ‘Public OpiniongQuarterly, Spring 1977, 86-90. 249 . 1977b. "The impact of Economic Conditions on Presidential Popularity," Journal of Politics, August 1977, 764-773. Kernell, Samuel. 1977. "Presidential P0pularity and Negative Voting: An Alternative Explanation of the Midterm Congressional Decline of the President's Party, American Political Science Review, March 1977, 44-66. . 1978. "Explaining Presidential Popularity," American Political Science Review, June 1978, 506-522. Kramer, Gerald H. 1971. "Short-Term Fluctuations in U.S. Voting Behavior," American Political Science Review, March 1971, 131-143. Lee, Jong R. 1977. "Rallying Around the Flag: Foreign Policy Events and Presidential Popularity," Presidential Studies Quarterly, Volume 7, Fall 1977, 252-256. Lewy, Guenter. 1978. America in Vietnam. New York: Oxford University Press. Littauer, Raphael and Norman Uphoff, eds., 1972. The Air war in Indochina. Boston: Beacon Press. Louis Harris and Associates, Inc.“ 1971-1974. The Harris Survey Yearbook of Public Opinion'1970-l973. N.Y.: Louis Harris and Associates, Inc. - Machiavelli, Niccolo. The Prince and the DiscOurses. New York: Modern Library, 1950. MacRae, C. Duncan. 1977. "A Political Model of the Business Cycle," Journal of Political Economy, Volume 85, 239-263. Martin, Jeanne. 1976. "Presidential Elections and Administration Support Among Congressmen," American Journal of Political Science, August 1976, 483-489. ' 250 McPherson, Harry G. 1972. A Political Education. Boston: Atlantic- Little, Brown. Miller, Arthur H. 1974a. "Political Issues and Trust in Government, 1964-1970," American Political Science Review, September 1974, 951-972. . 1974b. "Rejoinder to 'Comment'by Jack Citrin: Political Dis- content or Ritualism?" American Political Science Review, Sep- tember 1972, 989-1001. . 1974C. "Change in Political Trust: Discontent with Authorities and Economic Policies, 1972-1973," A Paper Prepared for Delivery at the Annual Meetings of the American Political Science Associa- tion. Chicago, Illinois. ., Jeffrey Brudney, and Peter Joftis. 1975. "Presidential Crises and Political Support: The Impact of watergate on Attitudes Toward Institutions," A Paper Prepared for Delivery at the Annual Meeting of the Midwest Political Science Association. Chicago, Illinois. ., warren E. Miller. 1977. "Partisanship and Performance: 'Rational Choice' in the 1976 Presidential Election," A Paper Prepared for Delivery at the Annual Meeting of the American Political Science Association. Washington,D.C. Monroe, Kristen R. 1977. "'God of Vengeance and of Reward?‘ An Empirical Examination of the Economic Influences on Presidential Popularity," A Paper Prepared for Delivery at the Annual Meeting of the Public Choice Society. New Orleans, La. Milstein, Jeffrey. 1974. The Dynamics of the Vietnam war. Columbus: Ohio State University Press. Mueller, John E. 1970. "Presidential Popularity from Truman to Johnson," ‘American Political Seience Review, March 1970, 18-34. . 1973. “War, Presidents and'Public Opinion. New York: Wiley. 251 Neustadt, Richard E. 1976. Presidential Power: The Politics of Leadership. 2nd Edition. New York: Wiley. Newsweek. washington,D.C.: The washington Post Company. Nordhaus, William D. 1975. "The Political Business Cycle," Review of Economic Studies, April 1975, 169-190. Ordeshook, Peter C. 1976. "The Spatial Theory of Elections: A Review and Critique," in Ian Budge, Ivor Crewe, and Dennis Farlie, eds., Party Identification and Beyond. London: John Wiley. Ornstein, Norman J. and David W. Rohde. "Shifting Forces, Changing Rules and Political Outcomes: The Impact of Congressional Change on Four House Committees," in Robert Peabody and Nelson Polsby, eds., New Perspectives on the House of Representatives. 3rd Edition. Chicago: Rand McNally. Ostrom, Charles W., Jr. 1978a. "A Reactive Linkage Model of the U.S. Defense Expenditure Policy-Making Process," American Political Science Review, September 1978, 941-957. . 1978b. Time Series Analysis. Beverly Hills, Cal.: Sage Publica- tions. Ostrom, Charles W., Jr., and Dennis M..Simon. 1978s.. "A Dynamic Model of Presidential Popularity," A Paper Prepared for Delivery at the Annual Meeting of the Public Choice Society, New Orleans, La. , 1978b. "The Making and Unmaking of the President: A Theoretical and Empirical Study of the Determinants of the Rise and Fall of Presidential Popularity," A Paper Prepared for Delivery at the Annual Meeting of the Midwest Association for Public Opinion Research. Chicago, Illinois. . 1979. "Entry, Exit and Alienation; The Constituency of the Presidential Support Coalition," A Paper Prepared for Delivery at the Annual Meeting of the Midwest Political Science Association. Chicago, Illinois. 252 . 1981. "A Choice-Based Model of Citizen Evaluations of the President," A Paper Prepared for Delivery at the Annual Meeting of the Public Choice Society. San Francisco, California. Page, Benjamin I. 1978. Choices and Echoes in Presidential Elections. Chicago: University of Chicago Press. Parker, Glenn R. 1977. "Some Themes in Congressional Unpopularity," American Journal of Political Science, February 1977, 93-110. Piereson, James E. 1975. "Presidential Popularity and Midterm Voting at Different Electoral Levels," American Journal of Political Science, November 1975, 683-694. Polsby, Nelson W. 1976. Congress and the Presidency. 3rd Edition. Englewood Cliffs, N.J.: Prentice-Hall. Richardson, Lewis. 1960. Arms and Insecurity. Pittsburgh: Boxwood Press. Riker, William H. and Peter C. Ordeshook. 1973. An Introduction to Positive Political Theory. Englewood Cliffs, N.J.: Prentice- Hall. Roll, Charles W., Jr. and Albert H. Cantril. 1972. Polls: Their Use and Misuse in Politics. New York: Basic Books. Safire, William. 1977. Before the Fall. New York: Ballantine Books. Schell, Jonathan. 1975. The Time of Illusion. New York: Random House. Schneider, Friedrich. 1977. "Presidential Popularity Functions of Different Classes: A Theoretical and Empirical Approach," A Paper Prepared for Delivery at the Annual Meeting of the Public Choice Society, New Orleans, La. Shannon, Wayne. 1968. "Electoral Margins and Voting Behavior in the House of Representatives: The Case of the 86th and 87th Congresses,' Journal of Politics, November 1968, 1028-1045. 253 Sidey, Hugh. 1978. "It's A Time of Testing," Time, 12 June 1978, 20. Sorenson, Theodore. 1965. Kennedy. New York: Harper and Row. Stimson, James A. 1976. "Public Support for-American Presidents: A Cyclical Model," Public Opinion anrterly, Spring 1976, 1-21. , and Caroline LeGette. 1975. "Public Support for American Presidents: Does Anything But Time Matter?" A Paper Prepared for Delivery at the Annual Meeting of the American Political Science Association. San Francisco, Cal. Stokes, Donald E. 1966. "Spatial Models of Party Competition," in Angus Campbell, et. al., Elections and the Political Order. New York: John Wiley. Sundquist, James L. 1968. Politics and Policy: The Eisenhower, Kennedy, and Johnson Years. washington,D.C.: The Brookings Institution. Tufte, Edward. 1975. "Determinants of the Outcomes of Midterm Congressional Elections." American Political Science Review, September 1975, 812-826. . 1978. Political Control of the Economy. Princeton, N.J.: Princeton University Press. waldman, Loren K. 1967. "Liberalism of Congressmen and Presidential Vote in Their Districts," Midwest Journal of Political Science, February 1967, 73-85. White, Theodore. 1961 (1965,.1969. 1973). The Making of the President 1960 (l964,l968,l972). New York: Atheneum. . 1975. Breach of Faith: The Fall of Richard Nixon. New York: Atheneum. Wildavsky, Aaron. 1975. "The Two Presidencies," in Wildavsky, ed., Perspectives on the Presidency. Boston: Little, Brown.