The Dynamics of Foreign Policy Agenda Setting Author(s): B. Dan Wood and Jeffrey S. Peake Reviewed work(s): Source: The American Political Science Review, Vol. 92, No. 1 (Mar., 1998), pp. 173-184 Published by: American Political Science Association Stable URL: http://www.jstor.org/stable/2585936 . Accessed: 05/02/2013 17:18 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

.

American Political Science Association is collaborating with JSTOR to digitize, preserve and extend access to The American Political Science Review.

http://www.jstor.org

This content downloaded on Tue, 5 Feb 2013 17:18:05 PM All use subject to JSTOR Terms and Conditions

Vol. 92, No. 1 March 1998

American Political Science Review

TheDynamicsof ForeignPolicyAgendaSetting B. DAN WOOD and JEFFREY S. PEAKE T

Texas A&M University

heoreticaland empiricalworkonpublicpolicyagendasettinghas ignoredforeignpolicy. Wedevelop

a theory of foreign policy agenda setting and test the implications using time-series vector autoregression and Box-Tiao (1975) impact assessment methods. We theorize an economy of attention to foreign policy issues driven by issue inertia, events externalto U.S. domestic institutions,as well as systemic attention to particular issues. We also theorizethat the economy of attention is affectedby a law of scarcityand the rise and fall of events in competing issue areas. Using measures ofpresidential and media attention to the Soviet Union, Arab-Israeli conflict, and Bosnian conflict, we show that presidential and media attentions respond to issue inertia and exogenous events in both primaryand competing issue areas. Media attention also affectspresidential attention, but the president does not affect issue attention by the media.

arious theoretical and empirical analyses have focused on understanding how domestic policy issues reach the systemic or institutional agenda in the United States (e.g., Cobb and Elder 1972; Downs 1972; Jones 1977; Walker 1977; Anderson 1978; Kingdon 1984; Peters and Hogwood 1985; Light 1991; Baumgartner and Jones 1993; Jones 1994; Flemming, Wood, and Bohte 1995). There have also been case studies that explore how single issues develop, also all domestic (e.g., Bosso 1987, Carmines and Stimson 1989, Glick 1992). As of yet, however, there has been no systematic analysis exploring the dynamics of U.S. foreign policy agenda setting. This neglect of foreign policy by students of American politics is not a matter of chance but is due to practical reasons. Foreign policy does not readily fit the theoretical mold most scholars associate with domestic issues. Theories and analyses of domestic policy agenda setting usually focus on concepts like problem perception, issue definition, mobilization of interests, subsystem formation, venue shopping, and institutional attention.1 The first two and last of these concepts are certainly relevant to foreign policy agenda setting. It is less certain that interest groups, subsystems, and venue shopping are of much importance. The conventional wisdom is that for many foreign policy issues subsystems do not develop, since these issues do not typically involve the material or solidary benefits required to invoke strong interest group participation

committees are generally considered to have less influence in foreign policy than in domestic policy. Moreover, the president is the primary venue through which new foreign policy issues enter the U.S. system.3 There can be little doubt that the presidency has evolved through time as the primary actor responsible for U.S. foreign policy. The Constitution limited presidential authority in this area with respect to war, making treaties, and appointing official representatives to foreign nations. Short of war, however, the president has traditionally received considerable latitude. The president is the only actor in the U.S. system who can legitimately speak for the nation as a whole or be spoken to by other nations (usually through ambassadors). The increasing involvement of the United States in the international community, the public's desire for strong personal leadership, and the political ramifications of foreign policy mean that the presidency attends more heavily to foreign policy issues than does any other U.S. institution. Theories and analyses of domestic policy agenda setting also tend to describe the process as proceeding in a slow secular fashion, occasionally spurred by focusing events or the policy entrepreneurship of important actors or groups (Anderson 1978, Cobb and Elder 1972, Kingdon 1984). Yet, foreign policy issues do not often arise in such a gradual fashion. Rather, they tend to burst onto the scene as a result of crises or

(e.g., Ripley and Franklin 1991).2 Congress

issues. The degree of involvement obviously depends on the type of issue. Trade policy, for example, would provide the cement for subsystem relations to develop. Congress may also become involved in human rights or strategic arms control. Ripley and Franklin (1991; see also Ripley and Lindsay 1993, chapter 2; Lindsay 1994, chapter 7) provide a framework for considering different types of foreign and defense issues. 3 The view stated here has long been the dominant one among presidential scholars. For example, historian Arthur Schlesinger states: "In foreign policy the inclination is to let the Presidency have the responsibility-and the power" (1989, 420; see also Halperin, quoted in Doherty 1993, 323). Likewise, Barbara Sinclair (1993; see also Hinckley 1994) argues that while congressional leaders have been more active in foreign and defense policy since the Vietnam War, their influence is limited to the extent they are acting as independent policy entrepreneurs, rather than as agents of an active and involved membership. In contrast, see Lindsay (1994) for an extended discussion asserting congressional influence on U.S. foreign policy making.

V

and its

B. Dan Wood is Professor of Political Science, Texas A&M University, College Station, TX 77843-4348, and on the faculty of the George Bush School of Government and Public Service. Jeffrey S. Peake is a Ph.D. candidate, Department of Political Science, Texas A&M University. The authors appreciate comments on earlier versions of this manuscript by George C. Edwards III, Kenneth J. Meier, Steven A. Shull, Nehemia Geva, and four anonymous reviewers. This research was supported financially by the Center for Presidential Studies, Texas A&M University. 1 Venue shopping means that because of the U.S. system of fragmented representation, potential agenda setters have multiple institutions through which to pursue policy goals. 2 We do not mean to imply that interest groups and congressional committees never become involved in foreign policy or have no influence, but their involvement is typically less than for domestic

173

This content downloaded on Tue, 5 Feb 2013 17:18:05 PM All use subject to JSTOR Terms and Conditions

The Dynamics of Foreign Policy Agenda Setting other dramatic occurrences, and once they are on the scene attention is driven by a continuous stream of policy events. While events are occasionally important in driving attention to U.S. domestic issues (e.g., a Martin Luther King speech and civil rights, a thalidomide controversy and drug safety, an attempted presidential assassination and gun control), they are of much greater importance in driving attention to issues in foreign policy.

THE ECONOMY OF FOREIGN POLICY ATTENTION We explore the determinants of attention to major foreign policy issues in the U.S. system. As noted above, past work on domestic policy agenda setting provides a weak theoretical basis for exploring the dynamics of foreign policy attention. Pluralism and interest groups are generally considered of lesser importance, the process is seen as event driven, and foreign policy making is considered the primary domain of the president. A theory of foreign policy agenda setting and attention, therefore, should necessarily focus on the president. Yet, the president's ability to attend to particular foreign policy issues is often constrained by other responsibilities. The president is governed by an economy of attention in which competition exists for scarce resources. Of course, the proliferation of demands on the modern presidency has produced new institutions within the White House that are intended to alleviate limitations due to scarce presidential resources. Yet, modern presidents and their staffs remain only boundedly rational (Simon 1947). Cognitive limitations mean that they neither can accurately forecast the rise of new foreign policy issues nor possibly attend to the plethora of potential issues that ultimately arise. As a result, presidents purposely restrict the set of issues at any one time to a few. They tend to process these few issues in a serial rather than parallel fashion, attending to one before moving on to the next (Jones 1994). In order to prioritize issues, presidents rely on cues from their environment regarding the relative severity of the problem, as well as political benefits and costs. As political creatures, presidents are ever aware of the risks associated with ignoring or attending to new policy problems. To be sure, this economy of attention is altered when presidents make conscious efforts to give higher priority to some issues rather than others. The president brings to office an agenda and, if possible, will attempt to implement it (Light 1991). For example, President Nixon clearly gave a high priority to relations with China, which necessarily restricted the amount of time he could spend on other issues. This personal dimension the president brings to the office is inherent to the motivations and goals of the individual. Yet, even though most presidents have foreign policy goals, they often have agendas thrust on them by the office and external events.

Indeed, multiple forces move attention to foreign policy issues away from presidential discretion. For one

March 1998 thing, presidents often inherit problems that dominate their attention through time. Chief executives from Truman through Bush inherited the foreign policy problems associated with the Cold War and devoted considerable time to these issues. Similarly, President Nixon inherited Vietnam from the Johnson administration and necessarily devoted more attention to this issue than to any other foreign policy domain. On issues of this sort, presidents must fulfill continuing obligations to the office and nation, regardless of personal agendas. More generally, once issues gain prominence in the economy of presidential attention, attention becomes highly inertial. As a technical matter, the most important foreign policy issues are very persistent and require time and effort to resolve. To be successful, presidents must maintain focus on the difficult issues until they wane or at least are displaced in importance by other concerns. The security of the free world, public approval, and the president's historical legacy all depend on dealing effectively with important problems, so there is strong incentive for the president to maintain focus to the potential exclusion of other issues. This highly inertial character of some foreign policy issues suggests a rationale for the economy of attention to exhibit stability. Presidents tend to concentrate on the most difficult foreign policy problems. The proportion of presidential time devoted to these problems remains fairly stable so long as they require strong consideration and no more pressing problem arises. The duration of attention to foreign policy problems is proportional to the tractability of the problem. As long as the problem remains unresolved, presidential attention remains fairly consistent. Nevertheless, short-term disturbances to this stability may occur due to policy events or changes in perception of the relative importance of the issue. Because the president must ration attention according to which issues are most critical, shifts also may occur as new problems arise. These shifts may be either short or long term. In either case, the chief executive often operates in a responsive mode, reacting to a continuous stream of exogenous events. There can be little doubt, for example, that President Truman preferred not to be dealing with the Soviet blockade of West Berlin in 1948, but this event skewed foreign policy attention even more toward U.S.-Soviet relations through early 1949. Likewise, President Carter preferred not to be attending to the Iran hostage crisis during the 1980 election year, but he had little choice, since it also dominated media and public attention that were critical to the election. Not all foreign policy events are this dramatic, but still there is a continuous progression of world events to which presidents must always attend. The decision as to what constitutes an issue deserving of executive attention may initially arise through developing events, but political factors affect the perceived importance and persistence of foreign policy issues. The president cares about public perceptions. The president has always depended on public ap-

174

This content downloaded on Tue, 5 Feb 2013 17:18:05 PM All use subject to JSTOR Terms and Conditions

American Political Science Review

Vol. 92, No. 1

proval as a source of legitimacy and strength in dealing with other actors, both foreign and domestic (Neustadt 1960). While approval of the way a president is doing the foreign policy job does not fully determine approval, it does at times comprise an important component, especially when foreign policy is salient (Edwards, Mitchell, and Welch 1995). To maintain good standing, the president must attend in some measure to what the public perceives as important. Thus, as the public pays greater attention to an issue, the president must in turn attend more to that issue. Of course, public perceptions of the relative importance of an issue are largely determined by the news media. The media interpret issues, giving them more or less significance through the type or amount of coverage provided. The media, particularly television news, also focus attention on events through this same interpretive function. Americans generally do not pay attention to foreign policy issues until events covered by the news media cause them to do so, and then their attention is slight and deteriorates quickly (Wittkopf 1990). Iyengar and Kinder (1987) show that what the television news covers influences what citizens think is important and how they evaluate their leaders. Likewise, Bosso (1989) argues that the media played an important role in mobilizing the citizenry toward famine relief in Ethiopia during the 1980s. Television news chose to focus on the famine, making an endemic problem in Africa a new foreign policy problem in the United States. Thus, media attention to foreign policy issues and events may drive systemic attention, as well as attention by the president. The direction of this relationship may not run purely from the media or system to the president, however, Presidents are not purely responsive actors bending with the breezes of popular attention. Rather, they also seek to lead on matters of policy. The president often wants to achieve a personal agenda, both domestic and foreign, but especially in the area of foreign policy the president is expected to be the locus of policy influence. Manipulation of the press is critical to the White House operation (Cohen 1963), and all presidents since Franklin Roosevelt have maintained a staff dedicated to informing the media on matters of interest to the president. In efforts to lead, the president attempts to mold how the media treats emerging issues and events. For example, Brace and Hinckley (1992) found that presidential trips, summits, speeches, and other initiatives create rallies in public approval while attracting large amounts of media attention. More generally, each year the president is given the stage to set the policy agenda for the American people and Congress through the State of the Union address. Cohen (1995) showed that public opinion follows the president's lead on which problems are most pressing, as outlined in the State of the Union address. Thus, the media are a principal means whereby presidents attempt to shape the foreign policy agenda and lead

systemic attention toward those issues the president deems as important.

EVALUATINGTHE ECONOMY OF ATTENTION TO FOREIGN POLICY To what extent is presidential attention to foreign policy issues a result of issue inertia or a continual progression of stochastic events? To what extent is systemic attention to foreign policy issues a result of issue inertia or stochastic events? Does the president respond to changing systemic attention to foreign policy issues? Does the system respond to changing presidential attention to foreign policy issues? These are important questions for understanding the president's role as foreign policy leader, as well as for extending political science theories of agenda setting to the foreign policy domain. We address these questions by developing measures of systemic and presidential attention for three foreign policy issues and related events. In particular, we look at issues involving the Soviet Union, the Arab-Israeli conflict, and the more recent Bosnian conflict. We chose these three because they have been the most highly visible and enduring foreign policy issues for the president over the last 15 years. Foreign policy issues of lower visibility and endurance (such as the South China sea, chemical and biological weapons) are also part of the president's agenda and may be of considerable importance. As a result, we suggest that future research examine a diverse range of issues with differing visibility and interest to the president. We used weekly time-series data to capture the fine time dynamic associated with responsiveness by both the media and president to world events and also to enable greater confidence in the findings due to the larger sample size. We measured systemic attention for issues involving the Soviet Union and Arab-Israeli conflict as the weekly broadcast time devoted by television news from the 27th week of 1984 through the 23d week of 1994. The Bosnian conflict, which has been less protracted and is confined to the 1990s, is measured as the weekly broadcast time from the 25th week of 1991 through the 11th week of 1995.4 Rather than counting the number of stories, as has been done in some past research (e.g., Baumgartner and Jones 1993; Flemming, Wood, and Bohte 1995), we counted the number of minutes devoted to each issue on the three nightly television news programs. Specifically, we searched the VanderbiltTelevisionNews Abstracts using keywords (listed in the Appendix) to capture stories concerning the Soviet Union and Arab-Israeli and Bosnian conflicts.5 We examined each hit of the keyFor post-Soviet Union observations after late 1991, we examined relations between the United States and the states of the former Soviet Union. The time frame studied was driven entirely by data availability. The beginning and ending dates were constrained by the PANDA events data (to be discussed below), which are continuously being updated and extended toward the present by the project directors. At the time we constructed the Soviet and Arab-Israeli measures, the PANDA data were available only through mid-1994; at the time we constructed the Bosnia measure, the PANDA data were available only up to early 1995. Abstractsare available on the web at the following 5The Vanderbilt address: http://www.TVnews.Vanderbilt.edu. The information logged here comes in the same format as the hard copy versions

4

175

This content downloaded on Tue, 5 Feb 2013 17:18:05 PM All use subject to JSTOR Terms and Conditions

The Dynamics of Foreign Policy Agenda Setting words for validity concerning whether it dealt with some facet of the foreign policy issue or whether some isolated event affected the measure. Although the nightly news programs are not the only television news sources that may influence or be influenced by the president, they do provide a consistent sample of coverage and are also continuous for the period under study. Each network spends about 22 minutes each day delivering the news in half-hour programs (Tyndall 1995). This leads to a typical weekly news coverage of roughly 150 minutes, or 450 minutes total for the three networks. Measuring presidential attention to these three issues is more difficult. The president's public face may not be a true reflection of what is actually being attended to behind the scenes. Unable to know beyond question what the president is doing, we must assume that what he does and says publicly from week to week reflects what is on his agenda.6 Cohen (1995) made this assumption and measured presidential attention to domestic policy issues by looking at State of the Union addresses. These speeches are only made annually, however, and may give an incomplete image of attention as it changes due to events and shifting circumstances during the year. It is also unclear that the relative issue emphases in the State of the Union address provide an accurate picture of presidential priorities on foreign policy. Presidents make their priorities known through a number of outlets, including the State of the Union address, legislative proposals, news releases, position taking, briefings, speeches, press conferences, and letters. To capture presidential attention registered through all these diverse outlets, we used Public Papers of the President, an annual compilation of activities during each year. As with the TV news search, we used keywords (listed in the Appendix) in the subject index in each volume and searched the text for relevant activities. Specifically, we counted the number of paragraphs during each week of each year of the Public Papers devoted to some facet of Soviet relations, the Arab-Israeli conflict, and the Bosnian conflict. We also read each entry to assure that it was pertinent to the keyword entry and the concept under investigation. We measured international events using the PANDA events data set provided by the Program on Nonviolent Sanctions and Cultural Survival from Haravailable in most university libraries and has the advantage of allowing electronic searches. We used NETSCAPE as a search engine, which virtually eliminated the problem of double counts that may plague the use of electronic searching in other contexts. The cache file in NETSCAPE changes the colors (by default from blue to pink) of all links previously accessed, until it is reset. Each story was rounded to the nearest half minute. 6 We can argue the relative merits and demerits of this assumption from either perspective. On the one hand, the president wants to appear "on top of things" in order to garner public approval. As a result, his public activities and statements should actually reflect what is being attended to behind the scenes. On the other hand, there may be occasions when secrecy is important, in which case public activities and statements may be a facade that masks what the administration is actually doing.

March 1998 vard University (Bond and Bond 1995). PANDA uses an artificial intelligence program (KEDS) to code discrete events from Reuters news leads.7 "An event is an interaction, associated with a specific point in time, that can be described in a natural language sentence that has as its subject and object an element of a set of actors and as its verb an element of a set of actions, the contents of which are transitive verbs" (Bond and Bond 1995, 5). In other words, PANDA events are reports of discrete actions by a source country toward a target country or discrete actions completely within a single country, such as a civil war or rights violation. Using keywords (listed in the Appendix) relating to the three issues, relevant events were separated from the entire data set. Of course, we could have coded a dummy variable for the presence or absence of international events based on the PANDA data set. Using a weekly count of events, however, enabled the measure to reflect the relative seriousness of developing situations, as well as their presence. The events were then counted by week to create a consistent measure of exogenous events to match the other two series. Both the TV news networks and the president use Reuters to monitor world events, so it is almost prima facie evident that there should be some effect from this measure on media and presidential attention to these issues. The question is whether, independent of world events, effects exist from the national media to the president or from the president to the media.8 With measures in hand for systemic and presidential attention to foreign policy issues, as well as exogenous events, we used vector autoregression (VAR) methods (Freeman, Williams, and Lin 1989; Simms 1980) to evaluate the causal direction of attention and also to provide evidence of the temporal dynamics associated with the policymaking system. VAR is the most appropriate method for circumstances in which theory provides a weak rationale for imposing restrictions on the parameters of a structural equation system. As our theoretical discussion above indicates, we have no a priori reason for imposing parameter restrictions in either direction; yet, without them a structural equation system is not identified. Thus, structural equations are inappropriate. This does not mean, however, that the approach we use is devoid of theory. Indeed, all aspects of the reported VAR model have some theoretical rationale. We merely ask the data to tell us 7KEDS stands for the Kansas Events Data System. This program was developed by Philip Schrodt of the University of Kansas at Lawrence. Schrodt and Gerner (1994) describe the program in detail and also report validity tests of the machine-coded events. 8 In a strict sense, not all PANDA defined events are strictly exogenous to the presidency. The president may take initiatives that form "historical threads" which become part of event progressions. In addition, some PANDA events may not be causally prior to presidential or media attention. The president and media may anticipate such events as elections, scheduled meetings, and the like. In theory it would be possible to separate out such potentially confounding influences if one had access to the text used to create the PANDA data. Given that we do not have such access, we simply acknowledge these potential confounding effects.

176

This content downloaded on Tue, 5 Feb 2013 17:18:05 PM All use subject to JSTOR Terms and Conditions

American Political Science Review

Vol. 92, No. 1

(given that theory does not) which, if any, parameter restrictions are appropriate. Another important advantage of VAR modeling over the structural equation approach is that it provides a strong control for history by including multiple lags of each variable in all equations. This means that problems of specification error are of less concern, since the VAR disturbances are random with respect to time.9 One way of viewing VAR modeling is as a multivariate extension of the Granger (1969) approach to causal inference. Each dependent variable is regressed on lagged values of itself, as well as lagged values of the other dependent variables in the system.10 Causal relations are evaluated by conducting joint hypothesis tests for blocks of lags associated with each variable. What is interesting for us theoretically is that the control for history provides a representation of the degree to which issue inertia affects attention by both the media and president. A significant block of coefficients on the dependent variable in each equation implies that issue inertia is strong. VAR methods typically exhibit high colinearity due to the multiple lags included for each variable in the system. For this reason analysts do not typically try to interpret coefficient estimates but instead do simulations to track out the system dynamics. We report Granger tests and simulations using the moving average response approach. That involves introducing a shock to a variable in the system and tracking out movements in the other variables using the VAR estimates for computing a forecast. Shocking a variable means successively multiplying the matrix of vector autoregressive estimates by a simulation vector containing changes in the variable of interest.1" Based on theory, we also entered international events into the VAR system as a single exogenous variable with no lags. Such an approach is sometimes 9 We do not mean to imply here that the omission of relevant independent variables is not a problem in VAR. It can be. Disturbances can be random with respect to time but not random with respect to the dependent variables. Relevant to this point is whether the VAR should contain a congressional variable. Of course, theory suggests that continuous congressional influence on foreign policy attention should be less than that for the president. Edwards and Wood (1997) tested for the presence of congressional influence for two of the same issues and found none. See also Bartels (1996) for similar null findings. 10Based on Simms' (1980) methods for determining the appropriate lag length, we included in the Soviet and Arab-Israeli VARs four weekly lags for the media and presidential variables. We included five lags of these variables for the Bosnian conflict VAR. The Simms procedure entails sequentially adding lags to the VAR system and testing the statistical significance of each additional lag using a modified F test. We tested lags from 1 through 8 for the VAR system and arrived at the lag lengths reported below. We also did sensitivity testing to determine the effects of including additional lags up to eight weeks. Longer lag lengths diminished marginally the probabilities associated with the reported findings, but in all cases relations were stable and consistent. Residuals from final analyses were nonautocorrelated and normally distributed. 11 To facilitate interpretation of the moving average responses, and because the variables had no natural metric, we standardized all variables prior to the analysis. All initial shocks are one standard deviation in magnitude. Because the innovations are correlated between variables, we plotted Choleski orthogonalized responses to simulated shocks of one standard deviation.

termed ARX modeling (Judge et al. 1988, 776). We used no additional lags on the events variable, because events are discrete occurrences that should affect the media and president immediately, but once the events have occurred, the stimulus is no longer present. The system should be free to work itself out as the president and media interact to interpret the relative significance of the events. Thus, the events variable provides a theoretical test for the manner and extent to which events instantaneously shock attention by the media and president to foreign policy issues. The events variable is also a statistical control for the hypothesis of spuriousness. That is, we can exclude the possibility that events definitively shape systemic or institutional attention, rather than the processes discussed above. The question we address after controlling for events is whether persistent attention by either the media or president has long-term effects on attention by the other actor.

THE DETERMINANTS OF U.S. FOREIGN POLICY ATTENTION The determinants of attention to Soviet, Arab-Israeli, and Bosnian issues are so similar that a single discussion can pertain to all three analyses.12This similarity in findings suggests a certain theoretical consistency for the three issue areas. In particular, the analysis shows that events were extremely important in driving contemporaneous attention movements by both the president and media. Attention inertia was strong for both actors, and a definite pattern of influence emerged from media attention to presidential attention, but not in the opposite direction. In all three vector autoregressions reported in Table 1, contemporaneous events are strongly significant at less than the 0.01 level. Of course, this consistently strong relationship shows that the economy of attention is often determined by occurrences external to both the media and presidency. Controlling for exogenous events, how did issue inertia, media attention, and presidential attention play out over time? Table 1 also reports the Granger tests for attention to the Soviet Union, Arab-Israeli conflict, and Bosnian conflict, respectively. In each case, past media attention to an issue Granger causes current media attention to the same issue, and past presidential attention also Granger causes current presidential attention.13 Thus, as our 12 The results were also reasonably consistent through time. In testing for coefficient stability, we did Chow tests as well as separate VARs for different presidencies. There were no substantively important differences across presidencies, but there were some differences in the magnitude of responses. Generally, President Reagan was somewhat less responsive on all three issues to the media than were presidents Bush and Clinton. All responses were in the same direction, however, and fully consistent with results reported below. 13 A variable X Granger causes a variable Ywhen, controlling for the history of Y, past lags of X covary with current values of Y. The standard approach to Granger testing involves selecting an appro-

priatecommonlag lengthto simulatean infiniteorderautoregressive disturbanceprocessandthen doingsignificancetestingfor the effects of laggedX on Y.

177

This content downloaded on Tue, 5 Feb 2013 17:18:05 PM All use subject to JSTOR Terms and Conditions

The Dynamics of Foreign Policy Agenda Setting

TABLE 1. Events

March 1998

Granger Tests for Attention to Three Foreign Policy Issues, Controlling for Related

Dependent Variable MediaAttention

Coefficient Block Media President Events

PresidentialAttention

Media President Events

Soviet Union 6.64 (0.00) 1.49 (0.20) 371.71 (0.00) 2.63 (0.03) 5.21 (0.00) 134.56 (0.00)

Arab-Israeli Conflict 14.92 (0.00) 0.89 (0.47) 129.96 (0.00) 2.81 (0.02) 2.40 (0.05) 36.24 (0.00)

Bosnian Conflict 13.51 (0.00) 1.61 (0.16) 70.73 (0.00) 1.95 (0.09) 5.31 (0.00) 26.63 (0.00)

Note: The numbers in the tables are F statistics; p values are in parentheses. The Soviet and Arab-Israeli VARs contain four lags. The Bosnian conflict VAR contains five lags. Events at lag 0 are included as an exogenous variable to control for the contemporaneous effect of events on both the media and president. There were 514 weekly observations in the Soviet and Arab-Israeli series, running from the 27th week of 1984 to the 23d week of 1994.

theory predicts, much of what the media and presidents attend to through time is determined by issue inertia and what they have attended to in the past. As new issues emerge, political actors are not free to choose attention levels. Rather, they must attend to issues of continuing importance until they are resolved or displaced in the economy of attention. The Granger tests are also consistent in showing that media attention to the three issues is exogenous to presidential attention. The president does not Granger cause media attention to the Soviet Union, ArabIsraeli conflict, or Bosnian conflict. Theoretically, this implies that the president is not very successful in directing media attention to foreign policy issues; rather, issue inertia and the continual progression of world events seem to drive media attention. Yet, the media do Granger cause presidential attention for all three issues. This is, of course, consistent with a model of presidential attention that views the president as responsive to those issues on the systemic agenda. It also suggests a view of the president as responsive to cues from the policy environment in deciding which issues should receive priority in the economy of presidential attention. As the media and mass public become more concerned with an issue, the president also becomes more concerned. The Granger tests provide evidence concerning the joint effects of lagged media attention while controlling for attention inertia and exogenous events. They provide no indication of the polarity of relations, however, or the dynamics associated with evolving attention processes. We can explore the dynamics of presidential responsiveness to the media by tracking out simulations from the VAR through time. Figure 1 plots the moving average responses of the media and president, respectively, when each variable is subjected to a simulated shock for all three issues. The first row of Figure 1 shows that a simulated shock to media attention produces an increase in presidential attention. In initial response to this shock, presidential attention to the issues shifts upward by about 0.20 (for the Bosnian

conflict) to around 0.40 (for the Soviet Union and Arab-Israeli conflict) standard deviations in the week after. The increase in presidential attention due to the media lasts for about two (for the Soviet Union) or three (for the Arab-Israeli and Bosnian conflicts) weeks, eventually returning to the preintervention equilibrium. In the other direction, the second row of Figure 1 shows that a shock of one standard deviation in presidential attention produces no systematic response by the media in either direction for any of the three issues. Thus, the president responds to the media, but the opposite relation does not occur.

COMPETITION IN THE ECONOMY OF ATTENTION If presidential attention is a scarce resource, and if there is competition for presidential attention between multiple issues, then attention to an issue should change not only as a function of events, inertia, and media coverage of that same issue but also as a function of events associated with other issues. More specifically, as competing issues become more prominent in the economy of attention, attention to the primary issue should decline. A similar logic should also pertain to television news coverage and the media. Given the limited amount of time the TV evening news can devote to any one issue, it would not be surprising that as one issue becomes more visible another is squeezed out. For the president it is less clear that the squeeze should be as serious as for the TV evening news. The chief executive is under less pressure to balance coverage between domestic policy issues and foreign policy, so there may be attention to more than one foreign policy at a time by simply diminishing attention to domestic policy issues. The president also has much more than the standard half-hour of coverage for attending to foreign policy issues and may speak to many topics in one briefing, news conference, speech, or statement. We can test the notion that competition exists

178

This content downloaded on Tue, 5 Feb 2013 17:18:05 PM All use subject to JSTOR Terms and Conditions

American Political Science Review

Vol. 92, No. 1

FIGURE 1. Simulated Moving Average Responses Autoregressions

1.2

1.2

1.2

1

1

1

Media

0.6

to

0.6

0.6

Ien

-0.4-

0.4 -0.4

0.2 -0.2-

0.2

-0.2

Shockto

-

-0.8

0.8 -0.8

Media

Bosnian Conflict

Arab-IsraeliConflict

Soviet Union

Shock

for Foreign Policy Attention Vector

0

2

4

6

8

10

-0.2

0

2

4

6

8

10

-0.2

0

2

4

6

8

10

-0.2L 0

2

4

6

8

10

1.2

1.2

1.2

0.8 -

0.8 -

0.8

06Media0.06

PresidentPren -0.4-

0.4 -0.4

0.2

0.2

0.2

0

1.

-0.21 0

2

4

6

8

10

-0.2

0

among issues in the economy of attention using two different approaches. First, we can enter the events variable for competing issues exogenously into the VAR system of interest. If competition affects attention, then events from competing foreign policy issues should Granger cause and reduce attention to the primary issues. Second, we can identify some critical event associated with a competing foreign policy issue and do intervention analysis to determine whether the event reduced systemic and presidential attention to the primary issue. Using either approach, care is required in designing the test to assure that issues are orthogonal and actually competing for attention. For example, the stream of events associated with the Bosnian conflict may actually increase, rather than decrease, attention to the Soviet Union because of concerns that the former Soviet Union, due to long cultural and political ties to the Serbs, would react to U.S. involvement. Similarly, a critical event like the Persian Gulf War could increase attention to the Arab-Israeli conflict because of inseparability of the war from long-term conflict between the Arabs and Israelis. With these design constraints in mind, we employed both approaches to test for competition in the economy of attention. We entered events for the Arab-Israeli conflict exogenously into the VAR for the Soviet Union and events for the Soviet Union and the Bosnian conflict exogenously into separate VARs for the Arab-Israeli conflict. Table 2 reports Granger tests from these

(

2

4

6

8

10

analyses. In the analysis for the Soviet Union, reported in column 1, there is clear evidence that competing events from the Arab-Israeli conflict affected attention by the media and president to the Soviet Union. The evidence is less clear from the Granger tests for the Arab-Israeli conflict. Competing events associated with the Soviet Union did Granger cause presidential attention to the Arab-Israeli conflict, but there is little evidence that competing events from the Bosnian conflict caused either media or presidential attention to the Arab-Israeli conflict. The Granger tests are suggestive but do not reveal the direction of relations, which should be negative if the theory is correct. The simulation analyses reported in Figure 2 show the dynamics associated with shocks of one standard deviation to events from competing systems. Consistent with the Granger tests, column 1 of Figure 2 shows that both media and presidential attention to the Soviet Union diminished for some time after a positive shift in Arab-Israeli events. Column 2 demonstrates that media attention to the Arab-Israeli conflict diminished sharply when there was a positive shift in Soviet events.14 It also shows an initial decline in presidential attention to the Arab-Israeli conflict in response to Soviet events but suggests that three or four weeks later there was a resurgence of presidential attention. Of course, this 14 Note that this large negative response was invisible to the associated Granger test. Such a disparity may occur due to contemporaneous feedback through the covariance matrix of errors.

179

This content downloaded on Tue, 5 Feb 2013 17:18:05 PM All use subject to JSTOR Terms and Conditions

March 1998

The Dynamics of Foreign Policy Agenda Setting

TABLE 2.

Granger Tests for Attention to Two Foreign Policy Issues, Competing Events Included Arab-Israeli Arab-Israeli Conflict Conflict Soviet Union Competing Events, Competing Events, Competing Events, Soviet Union Bosnian Conflict Dependent Variable CoefficientBlock Arab-IsraeliConflict 14.59 3.14 MediaAttention Media 6.91 (0.00) (0.02) (0.00) 0.96 1.62 President 1.67 (0.43) (0.17) (0.16) 2.14 0.20 1.51 Competing Issue (0.94) (0.20) (0.07) 2.61 1.38 PresidentialAttention Media 2.47 (0.03) (0.24) (0.04) 2.18 0.72 President 5.17 (0.07) (0.58) (0.00) 1.99 0.37 2.95 Competing Issue (0.02) (0.83) (0.09)

Note: The numbers in the table are F statistics; p values are in parentheses. All VARs contain four lags. As in Table 1, events at lag 0 are included as an exogenous variable to control for the contemporaneous effect of events on both the media and president. There were 514 weekly observations in the Soviet and first Arab-Israeli series, running from the 27th week of 1984 to the 23d week of 1994. There were 155 weekly observations in the second Arab-Israeli series, running from the 25th week of 1991 to the 23d week of 1994.

resurgence is consistent with the long-termimportance of Arab-Israeli issues over the period of interest. Column 3 shows a consistently small but negative response by both the media and president to events associated with the Bosnian conflict. Thus, in all three issue areas the simulations demonstrate negative movements, with responses that remain negative with few exceptions.

FIGURE 2.

We also did intervention analyses to test whether single events associated with competing issues affected media and presidential attention to primary foreign policy issues. In order for such critical events to have an effect they must be large in magnitude; they must also be relatively orthogonal to the foreign policy issue of interest. For the Soviet Union issue we

Simulated Moving Average Responses

to Shocks in Exogenous

Arab-IsraeliConflict

Soviet Union 1.2

Arab-IsraeliConflict

1.2 abIsr IConflicMe12a

1 2I 1

V

1.2 1

Sovie UnionMeia 1

H

0.8

0.8 08

Shock to Soviet Union

0.2

-Conflict

0.4

0.2-

0~~~~~~~~~~~~~~~-.

0

2

0

1.2

4

Arab-isrr

6

8

10

-0.4

0

-0.6

-0.2

0

2

1.212 1

ConflictPreSen

0.8

4

6

8

10

2

|Bosnia

4

6

8

10

8

10

ConflictPre'ent

0.8

Shock to Bosnian Conflict

06 -

0.4

0.6 -

0.4

~~~~~~~~~0.2 -0.2-

0.2 -

4 ~~~~~~~ ~0 2

0

ovnionPredent

0.8

Shock to Soviet Union

Shock to 0.6 Arab-Israeli Conflict 0.4

1802

l

Shock to 06 Bosnian

0.2

I

BosniaConflictMeia

01\ 0.8

H

0.6

Shock to 06 Arab-Israeli Conflict 0.4

-0.2

Issues

6

8

10

-0.

0

2

4

6

8

10

180

This content downloaded on Tue, 5 Feb 2013 17:18:05 PM All use subject to JSTOR Terms and Conditions

-02

0

2

4

6

American Political Science Review

Vol. 92, No. 1

Effect of Critical Events in Competing Issue Areas on Foreign Policy Attention Soviet Union Arab-IsraeliConflict Presidential Media Presidential Media Attention Attention Attention Variable/Parameter Attention 29.76*** 26.78*** 9.02*** Preinterventionmean , 27.78*** (5.34) (2.61) (7.61) (6.03) -8.26* -28.44 -15.75* Criticalevent coo -28.45*** (-1.96) (-1.33) (-1.68) (-2.58) 0.96*** 1.00*** 1.00** 0.97*** 51 (39.50) (15.05) (38.37) (40.79) ("Death"of the Soviet Union) (GulfWar) 0.26*** 0.58*** 0.21* Autoregressive 0.27* 01 (4.26) (2.58) (1.84) (2.09) 20.20 3.56 26.47 Box-LjungQ with 23 d.f. 4.88 TABLE 3.

Note: There are 100 weekly observations in each series. The Soviet time series runs from the 1st week of 1990 through the 26th week of 1991. The Arab-Israeli time series runs from the 38th week of 1990 through the 33d week of 1992. The interventions were pulses. The values of the dynamic parameters (8) for the Arab-Israeli conflict were rounded upward to 1.00 and did not exceed the bounds of stability. The "death" of the Soviet Union variable was lagged two weeks for the presidential attention equation. The numbers in parentheses are t-statistics. Up < .05, ~*p< .01, i**p < .001 (one tailed).

selected the beginning of the Persian Gulf War as having such characteristics.15 The Iraqi invasion of Kuwait occurred on August 2, 1990, and we hypothesized this date as the time attention should have shifted away from the Soviet Union. For the Arab-Israeli conflict we selected the announcement of the "death" of the Soviet Union and formation of a commonwealth of former Soviet republics. This announcement occurred on December 21, 1991, and dominated the news for the next several months. Accordingly, this competing event should have produced a subsequent decline in attention to the Arab-Israeli conflict. We used Box-Tiao (1975) methods to evaluate how these critical events affected attention to the primary issues. The results for the Persian Gulf War, reported in the first two columns of Table 3, show an immediate and long-term decline in attention to the Soviet Union by both the media and president. Media and presidential attention to the Soviet Union (w0), dropped off to near zero in the week of the invasion and recovered very slowly through time. The results for the "death" of the Soviet Union, reported in the third and fourth columns of Table 3 (w0), show a similar negative decline in attention to the Arab-Israeli conflict by both the media and president following the announcement. In all four analyses the near unity rate of recovery parameters (81) show that the effects would have been near permanent if other events had not occurred later to restore the primary issue to prominence in the economy of attention.'6 Thus, the results uniformly 15 One could argue that this event was not orthogonal and that an increase in attention to the Soviet Union should have occurred at this time. Although the Soviet Union had been an ally of the Iraqi regime prior to the war, the changes occurring in the Soviet Union at about this same time diminished the likelihood of a Soviet reaction. Moreover, the Soviets actually cooperated with the United Nations in removing Sadaam Hussein's forces from Kuwait. Thus, we are comfortable in asserting orthogonality. 16 The level at any particular time after an intervention may be computed recursively usingy, = wo + 8 y,-1 Thus, prior to the Iraqi invasion of Kuwait, the amount of presidential attention to the Soviet Union resulted in about 29.76 paragraphs per week. The week of the invasion the number of paragraphs dropped to yt = -28.44 +

show that critical events associated with competing issues diminish foreign policy attention to primary issues by both the media and president.

CONCLUSION The political science literature on agenda setting has been directed at explaining how U.S. domestic policy issues reach and remain on the systemic and institutional agendas. Yet, foreign policy is an important component of U.S. policymaking and has been largely ignored. This is undoubtedly because foreign policy is fundamentally different from domestic policy and requires a different rationale for explaining the rise and fall of issue attention. In this study we have advanced a theory of attention to foreign policy issues and then tested expectations that arise from that theory using empirical data. In contrast to domestic policy, foreign policy is primarily centered on the presidency due to constitutional obligations, the need for strong centralized leadership, and expectations by the public and other institutions. Yet, the president is not in full control of what receives attention in the foreign policy domain. There is an economy of presidential attention that is governed by the realities of scarce resources and rational efforts by the president to garner favorable public approval and historical treatment. The chief executive can occasionally establish issue priorities in the economy of foreign policy attention (as Nixon did with China) but may also have an agenda thrust on him by problem persistence, a continual progression of events, and a perception that he must attend more heavily to those issues the public deems most important. The economy of attention is stable so long as 29.76 = 1.32 paragraphs per week. Five weeks after the onset of the Persian Gulf War the number of paragraphs had risen to 6.57 per week; ten weeks after it was up to 10.85 per week; 26 weeks after it was 19.92 per week. Of course, other events and interventions can affect this mathematical sequence, moving attention either up or down through time.

181

This content downloaded on Tue, 5 Feb 2013 17:18:05 PM All use subject to JSTOR Terms and Conditions

March 1998

The Dynamics of Foreign Policy Agenda Setting issues persist and problems continue to be defined as important. Disturbances to this stability may occur, however, due to exogenous events or changing public perceptions of the relative importance of foreign policy problems. Our empirical analysis is based on the three most visible and enduring U.S. foreign policy issues over the past 15 years. The limited number of issues that we examine suggests a need for future research which differentiates on the basis of issue salience, importance, or issue type. The empirical findings may also be limited because the measure for the president depends on the president's public rather than private or hidden agenda. Tentatively, however, our study confirms expectations that flow from the preceding theory. Both presidential and systemic attention are highly inertial. The president attends heavily to those issues that have also been heavily attended to in the past, as does the media, which is an indicator of systemic attention. We presume that this observed inertia exists because foreign policy problems are not easy to resolve and because problems that are defined as important at one point are likely to remain so defined. Such problems also demand continuing presidential attention because the system defines certain problems to be important. Thus, presidents who want to be viewed in a favorable light attend to those problems toward which the system directs them. Foreign policy attention is inertial, but this inertia is often disturbed by a continual progression of events associated with the issue itself. As foreign policy crises emerge, attention becomes more intense to the issue in proportion to the severity of foreign policy events. Attention also shifts due to changing media interpretations of events and the perceived relative importance of an issue. Intense media coverage of an issue can extend presidential attention to that issue for a longer period than is normal. Because the economy of attention involves scarcity, issue inertia is also often disturbed by competition from other foreign policy issues. Systemic and institutional attention shifts away from one issue toward another due to world events that alter perceptions of what issues are most important at the time. As issues in one area are perceived as more important, constraints imposed by limited time and resources necessarily mean that issues in other areas receive less attention. In these regards, foreign policy differs sharply from domestic policy, since the progression of events is far more important. There is no analogous stream of domestic policy events similar to those that determine the context for attention to the Soviet Union, ArabIsraeli conflict, and Bosnian conflict. Whereas critical domestic policy events can occur and focus presidential attention for a time on particular issues, the foreign policy agenda operates in the context of a continually unfolding international drama. The drama depends on the media for production and interpretation. The president is drawn into the drama as an integral part,

but he has no script and often must respond in impromptufashion to media interpretationsand the

continuous progression of events affecting U.S. interests.

APPENDIX The following keywords were used to search the Vanderbilt Television News Archive. The list includes the words and the corresponding years in which they were used. Not all the stories that came up with these keywords were counted. We read the abstracts and coded the variables so that only those stories related to the Soviet Union, ArabIsraeli conflict, or the Bosnian conflict were part of the measure. Year

Vanderbilt Key Words

Soviet Union 1984 USSR, Soviet, Russia, Moscow, US-USSR rels., arms control, summit, nuclear weapons 1985 USSR, Soviet, Russia, Moscow, US-USSR rels., arms control, summit, nuclear weapons 1986 USSR, Soviet, Russia, Moscow, US-USSR rels., arms control, summit, nuclear weapons, Chernobyl, Gorbachev 1987 USSR, Soviet, Russia, Moscow, US-USSR rels., arms control, summit, nuclear weapons, Gorbachev 1988 USSR, Soviet, Russia, Moscow, US-USSR rels., arms control, summit, nuclear weapons, Gorbachev 1989 USSR, Soviet, Russia, Moscow, US-USSR rels., arms control, summit, nuclear weapons, Gorbachev 1990 USSR, Soviet, Russia, Moscow, US-USSR rels., arms control, summit, nuclear weapons, Republic, Gorbachev 1991 USSR, Soviet, Russia, Moscow, US-USSR rels., arms control, summit, nuclear weapons, Gorbachev, Yeltsin, Republic, Commonwealth 1992-94 USSR, Soviet, Russia, Moscow, US-USSR rels., arms control, summit, nuclear weapons, Yeltsin, Gorbachev, plus all 16 former Soviet states, besides Russia: Latvia, Lithuania, Estonia, Azerbaijan, Mongolia, Kazakhstan, Tajikistan, Moldova, Belarus, Armenia, Ukraine, Uzbekistan, Turkmenistan, Kyrgzstan, Georgia, Kurdistan Arab-Israeli 1984-94 Israel, Jerusalem, Tel Aviv, Syria, Lebanon, Jordan, Egypt, PLO, Arafat, Palestine, Palestinian, Arab-Israeli, terrorism, intifada, W. Bank, Gaza, Golan, occupied Bosnia 1991-95 Yugoslavia, Bosnia, Bosnia-Herzegovina, Serbia, Croatia, Slovenia, Sarajevo, civil war, embargo, ethnic cleansing, Moslem, Croat, Serb, Balkans The keywords for the index of the Public Papers of the President are provided in the following list. All the entries mentioned in the index were read to ensure validity for inclusion as attention to the Soviet Union, Arab-Israeli conflict, and Bosnian conflict.

182

This content downloaded on Tue, 5 Feb 2013 17:18:05 PM All use subject to JSTOR Terms and Conditions

American Political Science Review

Year

Vol. 92, No. 1

Public Papers Key Words

Soviet Union 1984-90

1991-94

United Soviet Socialist Republics (all concurrent listings), nuclear weapons (all concurrent listings), arms control, Afghanistan, any mention of Soviets in any other listing in the Index (i.e., Middle EastSoviet Role) Same as above, except no Afghanistan, and included: Commonwealth of Independent States (all concurrent listings), Estonia, Latvia, Lithuania

Arab-Israeli 1984-94

Israel, Jordan, Syria, Egypt, Lebanon, Middle East (having to do with conflict or terrorism), terrorism (dealing with the Middle East), and all "see also"

Bosnia 1991-95

Bosnia (all subheadings), Macedonia, Serbia, Yugoslavia, Montenegro

Croatia,

The following keywords were used for searching the PANDA events data set within a database environment (Microsoft Access). Abbreviations were used per the codes. These correspond to targets, sources, and places. Year

PANDA

Key Words

Soviet Union 1984-94

Armenia, Azerbaijan, Belarus, Estonia, Georgia, Kazakhstan, Kurdistan, Kyrgystan, Latvia, Lithuania, Moldova, Mongolia, Russia, Tajikistan, Turkmenistan, Ukraine, USSR, Uzbekistan

Arab-Israeli 1984-94 Bosnia 1991-95

Israel, Jordan, Palestine, Syria, Egypt, Lebanon

Bosnia, Yugoslavia, Croatia, Serbia, Montenegro, Macedonia

REFERENCES Anderson, James E. 1978. Public Policymaking: An Introduction. Boston: Houghton Mifflin. Bartels, Larry M. 1996. "Politicians and the Press: Who Leads, Who Follows?" Paper delivered at the annual meeting of the American Political Science Association, San Francisco, CA. Baumgartner, Frank R., and Bryan D. Jones. 1993. Agendas and Instability in American Politics. Chicago: University of Chicago Press. Bond, Doug, and Joe Bond. 1995. "Protocol for the Assessment of Nonviolent Direct Action (PANDA) Codebook for the P24 data set." Harvard University working paper. Bosso, Christopher J. 1987. Pesticides and Politics: The Life Cycle of a Public Issue. Pittsburgh: University of Pittsburgh Press. Bosso, Christopher J. 1989. "Setting the Agenda: Mass Media and the Discovery of the Famine in Ethiopia." In ManipulatingPublic Opinion, ed. M. Margoles and G. Mauser. Pacific Grove, CA: Brooks-Cole. Box, George E. P., and George C. Tiao. 1975. "Intervention Analysis with Applications to Economic and Environmental Problems." Journal of the American StatisticalAssociation 70(March):70-9.

Brace, Paul, and Barbara Hinckley. 1992. Follow the Leader. New York: Basic Books. Carmines, Edward G., and James A. Stimson. 1989. Issue Evolution: Race and the Transformationof American Politics. Princeton, NJ: Princeton University Press. Cobb, Roger W., and Charles D. Elder. 1972. Participation in American Politics: The Dynamics of Agenda-Building. Baltimore, MD: Johns Hopkins Press. Cohen, Bernard. 1963. The Press and Foreign Policy. Princeton, NJ: Princeton University Press. Cohen, Jeffrey E. 1995. "Presidential Rhetoric and the Public Agenda." American Journal of Political Science 39(February):87107. Doherty, Caroll J. 1993. The Reluctant Warriors. Congressional QuarterlyWeeklyReport 13(February):323. Downs, Anthony. 1972. "Up and Down with Ecology-The IssueAttention Cycle." Public Interest 28(Winter):38-50. Edwards, George C., III, William Mitchell, and Reed Welch. 1995. "Explaining Presidential Approval: The Significance of Issue Salience." American Journal of Political Science 39(February):10834. Edwards, George C., III, and B. Dan Wood. 1997. "Who Influences Whom? The President, the Media, and the Public Agenda." Presented at the 1997 meeting of the Midwest Political Science Association. Chicago. Flemming, Roy B., B. Dan Wood, and John Bohte. 1995. "Policy Attention in a System of Separated Powers: The Dynamics of American Agenda Setting." Presented at the annual meeting of the Midwest Political Science Association. Chicago. Freeman, John R., John T. Williams, and Tse-min Lin. 1989. "Vector Autoregression and the Study of Politics." American Journal of Political Science 33(November):842-77. Glick, Henry. 1992. The Right to Die. New York: Columbia University Press. Granger, Clive W. J. 1969. "Investigating Causal Relations by Econometric Models and Cross-Spectral Models." Econometrica 37(July):424-38. Hinckley, Barbara. 1994. Less than Meets the Eye: Foreign Policy Making and the Myth of the Assertive Congress.Chicago: University of Chicago Press. Iyengar, Shanto, and Donald R. Kinder. 1987. News That Matters: Television and American Opinion. Chicago: University of Chicago Press. Jones, Charles 0. 1977. An Introduction to the Study of Public Policy. Monterrey, CA: Brooks/Cole. Jones, Bryan D. 1994. Reconceiving Decision-Making in Democratic Politics:Attention, Choice, and Public Policy. Chicago: University of Chicago Press. Judge, George G., W. E. Griffiths, R. Carter Hill, Helmut Lutkepohl, and Tsoung-Chao Lee. 1988. Introduction to the Theory and Practice of Econometrics. New York: Wiley. Kingdon, John W. 1984. Agendas, Alternatives, and Public Policies. Boston: Little, Brown. Lindsay, James M. 1994. Congress and the Politics of U.S. Foreign Policy. Baltimore, MD: Johns Hopkins Press. Light, Paul. 1991. The President's Agenda. Baltimore, MD: Johns Hopkins Press. Neustadt, Richard. 1960. PresidentialPower. New York: Wiley. Peters, B. Guy, and Brian W. Hogwood. 1985. "In Search of the Issue-Attention Cycle." Journal of Politics 47(February):239-53. Ripley, Randall B., and Grace A. Franklin. 1991. Congress, the Bureaucracy,and Public Policy. 5th ed. Pacific Grove, CA: BrooksCole. Ripley, Randall B., and James M. Lindsay. 1993. "How Congress Influences Foreign and Defense Policy." In Congress Resurgent: Foreign and Defense Policy on Capitol Hill, ed. Randall B. Ripley and James M. Lindsay. Ann Arbor: University of Michigan Press. Schlesinger, Arthur M., Jr. 1989. The Imperial Presidency. Boston: Houghton-Mifflin. Schrodt, Philip A., and Deborah J. Gerner. 1994. "Validity Assessment of a Machine Coded Data Set for the Middle East, 19821992." American Journal of Political Science 38(August):825-54.

Simms, Christopher A. 1980. "Macroeconomics and Reality." Econometrica 48(January):1-48.

183

This content downloaded on Tue, 5 Feb 2013 17:18:05 PM All use subject to JSTOR Terms and Conditions

March 1998

The Dynamics of Foreign Policy Agenda Setting Simon, Herbert. 1947. AdministrativeBehavior. New York: Macmillan. Sinclair, Barbara. 1993. "Congressional Party Leaders in the Foreign and Defense Policy Arena." In Congress Resurgent:Foreign and Defense Policy on Capitol Hill, ed. Randall B. Ripley and James M. Lindsay. Ann Arbor: University of Michigan Press. Tyndall, Andrew. 1995. "The Tyndall Report." Transmission on

September 25, 1995, dealing with coverage of the 0. J. Simpson trial. Walker, Jack L. 1977. "Setting the Agenda in the U.S. Senate." British Journal of Political Science 7(October):423-45. Wittkopf, Eugene R. 1990. Faces of Internationalism, Public Opinion and American Foreign Policy. Durham, NC: Duke University Press.

184

This content downloaded on Tue, 5 Feb 2013 17:18:05 PM All use subject to JSTOR Terms and Conditions