Public Prudence and its Support for Counter-Terrorism Initiatives

Transcription

1 Public Prudence and its Support for Counter-Terrorism Initiatives George Shambaugh School of Foreign Service & Department of Government Georgetown University William Josiger Department of Government and Security Studies Program Georgetown University Written for the 2004 Annual Meeting of the American Political Science Association Chicago, IL September 2004 DRAFT: Comments welcomed. Do not cite without permission.

2 Abstract The public is generally assumed to rally around the flag in the event of national emergencies, often granting the Executive expanded support to respond as needed to promote national security. Since September 11, 2001, the Executive Branch has presented terrorism to the public as a matter vital to national security. It has linked much of its policy agenda to the persistent threat of terrorism and, more recently, it has focused much of its reelection campaign strategy on the same. Public support for the President and U.S. policy in the aftermath of terrorist events has been high, but it is not as uniform or long lasting as these reelection strategies might suggest. Our analyses indicate that public support for the President and his policies varies as a function of the national value of the materials or people targeted by the attacks; the public s assessment of the risk of a subsequent attack affecting them; and, to some degree, presidential cues. This variation lends support to the work of Bruce Jentleson and others suggesting that, even in the context of a terrorist attack, the public exercises some prudence when offering its policy support to the President. 1

3 I. The Fear Factor is not Uniform The public is generally assumed to rally around the flag in the event of national emergencies, often granting the Executive expanded support to respond as needed to promote national security. 1 The overwhelming level of public and Congressional approval of the Patriot Act shortly after September 11, 2001 is a dramatic example of this phenomenon. Research on public opinion suggests that the public tends to be receptive to presidents in general and when presidents are popular and the integrity of the executive branch is high presidents can lead public opinion on specific issues. 2 Furthermore, when these factors are low, generating public support can be harder and presidents often engage in strategies -- such as making major policy speeches, engaging in or announcing bold policy initiatives, or otherwise raising public awareness of matters like national security over which the Executive has principal authority -- to increase their popularity, perceived competence, and prestige. These strategies rest on an assumption that support for the President and his policies are likely to increase during crises or more generally when he succeeds in linking other policy initiatives to matters of national security. 3 1 John E. Meuller, War, Presidents and Public Opinion (New York: John Wiley and Sons, 1973). Bruce Russett, Controlling the Sword: The Democratic Governance of National Security (Cambridge, MA: Harvard University Press, 1990). For a review of this literature and a counterpoint, see John Oneal and A. Bryan (1995) The Rally Round the Flag Effect in U.S. Foreign Policy Crises, , Political Behavior 17: George Shambaugh and Paul Weinstein, The Art of Policy Making: Tools, Techniques, and Processes in the Modern Executive Branch (New York: Longman Publishing, 2003): 89, Jeffry Cohen (February 1995) Presidential Rhetoric and the Public Agenda, American Journal of Political Science 39, 1: 87-88, George Edwards, The Public Presidency: The Pursuit of Popular Support (New York: St. Martin s Press, 1983), and Samuel Kernell, Going Public: New Strategies of Presidential Leadership, 2 nd ed. (Washington, DC: CQ Press, 1993). These strategies may, however, have mixed effects, especially if public perceptions of the President are mixed. See Paul Brace and Barbara Hinkley, Follow the Leader (New York: Basic Books, 1992) and R. E, Neustadt, Presidential Power and the Modern Presidents (New York: Free Press, 1990). 3 For example, Denise Bostdorff demonstrates that the framing and rhetoric used by the Reagan Administration played a significant role in promoting and managing the crisis and eventual invasion of 2

4 Since September 11, 2001, the George W. Bush Administration has linked much of its policy agenda to the persistent threat of terrorism. More recently, it has focused much of its reelection campaign strategy on the same. Public support for the President and U.S. policy in the aftermath of terrorist attacks has been high, but it is not as uniform or long lasting as the presumed rally around the flag effect suggests. Indeed, while the public still considers terrorism to be an important issue, the importance it gives to terrorism has declined relative to other issues (like the economy) back to levels similar to those which existed before 9/11. In this study, we find that public support for the President and his policies varies as a function of the characteristics of recent terrorist attacks. Two important factors are the national value of the materials or people targeted by the attacks and the public s assessment of the risk that they will personally be affected by a subsequent attack. Our findings also indicate that presidential cues may matter but their affects are fleeting as the duration of the rally around the flag effect varies depending on the characteristics of past terrorist attacks. Our findings lend support to a growing body of empirical research that the U.S. public discriminates between policies, offering its support to some but not others. Bruce Jentleson and John Oneal, Brad Lian and James Joyner argue, for example, that the public is sufficiently attentive to discriminate between the use of American military force against states engaged in undesirable or threatening activities and the American use of force to promote internal political change. 4 Our study builds on this Grenada. Denise M. Bostdorff (Fall 1994) The Presidency and Promoted Crisis: Reagan, Grenada and Issue Management, Presidential Studies Quarterly 24, 4: J. Hurwitz and M. Peffley (1987) How are Foreign Policy Attitudes Structured? Hierarchical Model, American Political Science Review 81: ; Bruce Jentleson (1992) The Pretty Prudent Public: Post Post-Vietnam American Opinion on the Use of Military Force International Studies Quarterly 36: 49-74; M. Nincic (1992) A Sensible Public: New Perspectives on Popular Opinion and Foreign Policy, Journal of Conflict Resolution 36: ; John Oneal, Brad Lian, and James Joyner (1995) Are the 3

5 research by examining a parallel, but inverted question. Specifically, we ask how prudent the public is in its support for the President and U.S. policy when the United States is under attack. That is, does prudence persist when the United States is the recipient rather than the initiator of the use force? We analyze this question by examining public support for the President and U.S. policy in the aftermath of terrorist attacks from 1980 through We examine polling data about public support for the President and U.S. foreign policy, what the public considers the most important policy problems facing the country, and its opinion about specific Presidents and their policy responses to certain terrorist events. These polling data are combined with information about terrorist attacks from the ITERATE data set on terrorist activities. 6 Using the combined data set, we test the impact that terrorist incidents with various characteristics have on public opinion. To analyze the duration of these effects and account for a possible lag in public responsiveness to terrorist attacks, we analyze the impact of terrorist incidents which take place in the two months prior to each poll as well as in the month in which each poll was conducted. American People Pretty Prudent? Public Responses to U.S. Uses of Force, , International Studies Quarterly 40: For previous discussions of public opinion polling in the aftermath of terrorist attacks see: Leone Huddy, Nadia Khatib, and Theresa Capelos (Fall 2002) The Polls Trends: Reactions to the Terrorist Attacks of September 11, 2001, Public Opinion Quarterly 66, 3: ; Marc Hetherington and Michael Nelson (January 2003) Anatomy of a Rally Effect: George W. Bush and the War on Terrorism, PS: Political Science and Politics 36, 1: 37-42; Theodore Downes-Le Guin and Bruce Hoffman, The Impact of Terrorism on Public Opinion, 1988 to 1989 (Santa Monica, CA: RAND, 1993); Ronald H. Hinckley, People, Polls, and Policymakers (New York: Lexington Books, 1992). 6 The International Terror: Attributes of Terrorist Events project (ITERATE) provides quantitative data on the characteristics of transnational terrorist groups, their activities which have international impact, and the environment in which they operate. It covers events from 1968 through The ITERATE data set was developed by Edward F. Mickolous, Todd Sandler, Jean M. Murdock and Peter A. Flemming. 4

6 II. Pubic Support and U.S. Policy under the Threat of Terrorism Bruce Jentleson highlights three general factors that are likely to enhance the rally around the flag effect in support of U.S. policy regarding the use of force: presidential cues; the importance of the national interests pursued; and public aversion to risk. 7 The importance of presidential cues is compelling. Particularly in the case of a terrorist attack, the way that the President presents and justifies policy matters because the general public and apparently many in Congress lack the information and expertise necessary to make definitive judgments about terrorist threats and attacks. Much ado has been made recently about intelligence and information available to the Executive, Congress and the American people in advance of the terrorist attacks on 9/11. For the purposes of this project, we make a simplifying assumption that public perceptions of the asymmetry of information between the Executive and the public regarding terrorism remains constant across the cases we analyze. Given this assumption, if the public responds to presidential cues, then the impact of terrorist incidents should be consistent across type. Furthermore, as long as the President continues to link policy initiatives to the ongoing threat of terrorism, the impact of terrorist attacks on public opinion should not decline in the near term. If, on the other hand, the public is prudent albeit not necessarily wise we would expect the impact of terrorist attacks on popular support of the President to decline despite ongoing efforts by the Administration to highlight the persistent threat of terrorism. To test this effect, we include an indicator for terrorist attacks of various types in the two months prior to the public opinion poll. If the non-lagged indicator of terrorist attacks is significant, but the lagged indicator is not, then the impact of terrorism on 7 Bruce Jentleson, (1992) The Pretty Prudent Public: Post Post-Vietnam American Opinion on the Use of Military Force International Studies Quarterly 36:

7 public opinion is considered transient and Presidential efforts to maintain that link ineffective. If both variables are significant, it suggests that rallying effect lasts at least three months. This may be due, in part, to the president s ability to maintain the link in the public mind. Second, Jentleson argues that public support is likely to increase with the importance of the national interests being pursued. In the context of a terrorist attack, we argue in parallel that public support for the President is likely to vary with the national value of the materials or people targeted by the attacks. Thus, public support is more likely to increase in response to a terrorist attack against a U.S. citizen, territory or target than an attack against non-u.s. targets. If the public is not prudent, terrorist attacks anywhere should bolster the President s ability to mobilize public support for his counterterrorism policies. We code variables related to the terrorist attacks as Pretty Prudent (PP) if they involve U.S. targets (either attacks inside the U.S or against U.S. targets abroad). Third, Jentleson posits that political support will also vary as a function of the public s aversion to risk. While Oneal, Lian and Joyner do not find support for Jentleson s thesis that the public s risk aversion varied during the post-vietnam and postpost-vietnam eras, the uncertainty of a future terrorist attack suggests that the public s assessment of the risk of a terrorist attack might have a significant impact on its support for the President and his counter-terrorism policies. If the public s perception of risk matters, the level of support it grants should vary based on the potential costs imposed by the terrorist attack, the likelihood that an individual respondent will be harmed in a future attack, and the likelihood that a future attack will take place. We posit that the cost 6

8 imposed by the terrorist attack increases in conjunction with the number of victims, the level of damage, and the level of violence. The likelihood that an individual respondent might be harmed is estimated in terms of the type of victim and the type of attack. Past attacks against the U.S. military or government personnel do not increase the probability that non-government targets will be selected in the future. On the other hand, attacks involving business people, tourists or other noncombatants increase the likelihood that the next attack may involve civilians as well as government or military personnel. Similarly, we posit that respondents will rally in response to forms of attack that are more likely to target civilians like kidnappings, hostage taking, airplane incidents, and bombings and to be more circumspect of forms of attack that are more likely to be used against military personnel like armed attack or weapons of mass destruction. Finally, the likelihood that a past attack will be repeated is presumed to increase if the past attack was state sponsored or conducted by a known adversary (like Al Qaeda). If, as Kantian theory and Democratic peace scholars would likely argue, the public is risk averse, then we would expect its support for the President and U.S. policy to decrease as the costs, personal threat, and likelihood of a future attack increase. 8 If, however, we adopt prospect theory s assumption that people are willing to take greater risks to avoid loss, then we would expect U.S. support for presidential policy to increase as the costs, likelihood of becoming a victim, and likelihood of an attack increase. 9 8 For a critical overview of the Democratic Peace literature see Sebastian Rosato (November 2003) The Flawed Logic of the Democratic Peace Theory, American Political Science Review 97, 4: Daniel Kahneman and Amos Tversky (March 1979) Prospect Theory: An Analysis of Decisions Under Risk, Econometrica 47, 2:

9 III. Definitions and Data Sources The dependent variables in the study are based on readily available public opinion data from over 40 different polling organizations and poll sponsors (A complete list is provided in Appendix Table 1). Our primary source for these questions was the Roper Center as available online through Lexis-Nexus. These questions include support for the President and the President s handling of U.S. foreign policy, proportion of people selecting a particular issue as the most important policy problem the country is facing, and support for a particular President or a specific policy following a terrorist attack. Where possible we carefully combined similarly worded questions from different polling organizations to increase the date range of these dependent variables. Unless otherwise specified, our dependent variables are net indicators the proportion of respondents who answers were equivalent to somewhat supportive, supportive or very supportive to the questions asked. 8

10 Table 1: Subset of Dependent Variables used in this Study Dependent Variable N Polling Organizations Presidential Foreign Policy Approval 764 ABC, CBS, Fox, Gallup, Harris, LA Times, NYT, PSRA, WP, Roper, Penn & Schoen, Gordon Black, ATS, CPS, USNWR, Quinnipiac, ABC/WP, CBS/NYT, NBC/WSJ, PSRA/Newsweek, PSRA/Pew, Gallup/CNN/USA Approval of George W Bush s handling of the aftermath and military response to the attacks of 9/11/ ABC, CBS, Fox, Harris, LA Times, PSRA, Time, TIPP, Zogby, Quinnipiac, ABC/WP, CBS/NYT, NBC/WSJ, PSRA/Newsweek, PSRA/Pew, Gallup/CNN/USA Approval of Ronald Reagan s handling of the situation in Lebanon 29 ABC, Gallup, Harris, NBC, Roper, ABC/WP, CBS/NYT, NBC/AP Favor US taking direct military action in Afghanistan 29 ABC, Gallup, Harris, NBC, Roper, ABC/WP, CBS/NYT, NBC/AP Worry about someone in your family becoming a 28 ABC/WP, NBC/WSJ, Gallup/CNN/USA victim of terrorism Worry about another terrorist attack occurring soon 100 ABC, CBS, Fox, Gallup, Harris, LA Times, PSRA, Confidence in the US government to prevent further terrorist attacks against US citizens Time, WP, AP, Gordon Black, Yankelovich 31 CBS, ABC/WP, CBS/NYT, Gallup/CNN/USA 9

11 The independent variables include a variety of characteristics of terrorist attacks recorded in the ITERATE data set. The ITERATE data provides descriptive information on over terrorist attacks since 1968 organized by individual attack. 10 We aggregated this data for each month based on different characteristics of the attacks. For example, the following table describes some of the different characteristics of terrorist attacks that took place in January (A complete list of terrorist attack characteristics is available in Appendix Table 2). Table 2: Sample of Characteristics for Terrorist Incidents in January of 1980 Characteristic of Terrorist Attack Number of Attacks Terror Month 52 Total number of terrorist attacks for January 1980 Violent attacks 23 Attacks involving airplanes 6 Kidnappings or hostage takings 6 Bombings 23 Armed attacks 11 Attacks that resulted in US victims 4 (**not the number of US victims**) Attacks that resulted in US government victims 3 (**not the number of US government victims**) Attacks that resulted in US business victims 1 (**not the number of US business victims**) Attacks that resulted in US tourist victims 1 (**not the number of US tourist victims**) Attacks in the US or involving US victims 12 (**we label these attacks as pretty prudent**) Attacks against NATO allies 18 To merge the public opinion and terrorist event variables, we measured the number of terrorist events with certain characteristics that took place in the month that each poll was 10 For a recent example of ITERATE data in use see Walter Enders and Todd Sandler (June 2002) Patterns of Transnational Terrorism, : Alternative Time-Series Estimates, International Studies Quarterly 46, 2:

12 taken, and the number of events with the same characteristics that took place in the two months prior to the month in which the poll was taken. IV. Analysis and Results a. Support for the President and U.S. Policy In this section, we evaluate the impact of terrorist attacks on the level of public support for the President and U.S. foreign policy. The dependent variable is coded as the level of support for the President and U.S. foreign policy. It is a net indicator (including somewhat supportive, supportive or very supportive ) and comes primarily from ABC, CBS, Fox, Gallup, LA Times, New York Times, and Washington Post polling. This coding varies from Oneal, Lian, and Joyner who define the rally effect as the change in the level of support prior to and after a U.S. policy action. Unfortunately, there are very few period between 1980 and 2003 in which there was no terrorist activity within three months of an opinion poll (see Chart 1, September 2001 is month 260 and is represented by the index line). Thus, there is no general baseline against which to evaluate a change in the level of presidential approval in response to terrorist attacks. The polling data are also not available at a standard interval prior to each terrorist event, so it is not feasible to use public opinion levels at a consistent time period in advance of every terrorist event as a case-specific baseline. To compensate, we have included the level of inflation, as measured by a change in the consumer price index from the previous month, to provide a means of controlling for other factors that are known to affect support for presidential policies. 11 In each model, we also include a dummy variable for events that take place after 9/11/2001 to control for the potentially dramatic shifts that the 11 See among others R. A. Brody, Assessing the President (Stanford: Stanford University Press, 1991):

13 attack on that day had on the relationship between terrorism and public support for the President and U.S. foreign policy. In addition, we supplement the regression analyses of terrorist activity over the time period with more case-specific descriptive statistics about the change in public opinion before and after particular terrorist events. 180 Chart 1: Terrorist Incidents 160 Mean Number of Terrorist Incidents Target of Attacks 20 Attacks against US 0 All Everywhere Months since January 1980 We will focus on three sets of propositions: first, if presidential cues matter and the President continues to link the threat of terrorism to his policy agenda, then we would expect that the type of terrorist attacks that effect public opinion the month in which they take place to also have an impact at least two months after they occur. Thus, at a minimum, if the current month indicator of a particular terrorist variable is significant, we would expect the lag of that variable to be significant as well. Second, if public support varies as a function of the national values targeted by terrorist incidents, then we would expect the public support to be higher when terrorist incident involve U.S. targets or victims, but not with others. Thus, we would expect the pretty prudent variables to be 12

14 significant. Third, if public support varies as a function of risk, we would expect it to vary in term of the costs, the likelihood of an individual respondent being targeted, and the likelihood that the attack would happen again. Table 3 presents the estimates of public prudence based on the importance of national interests being attacked. Pretty Prudent indicates that terrorist attacks involving U.S. targets (either attacks inside the U.S or against U.S. targets abroad) took place in the month that the poll was conducted, while Terror Month indicates the total number of terrorist attacks (against the United States and others) in the current month. The lag on both variables indicates that such attacks took place in the two months preceding the poll. Table 3: Us versus Them: Presidential Foreign Policy Approval Pretty Prudent 0.19 (Attack against US in current month) (3.61) ** Pretty Prudent Lag 0.14 (Attack against US in previous two months) (4.76) ** Terror Month (Attack against anybody in current month) (-2.69) ** Terror Month Lag 0.01 (Attack against anybody in previous two months) (0.51) Post (Post 911 Dummy) (6.87) ** (4.52) ** CPI change (Change in the Inflation from the previous month) (-2.83) ** (-1.62) (Constant) (64.34) ** (53.49) ** R Squared N Sig. F (0.00) (0.00) 13

15 In support of the proposition that U.S. citizens differentiate incidents involving targets with high national value from others, terrorist incidents against the U.S. have a statistically significant and positive impact on public support for the President and U.S. foreign policy, while terrorist incidents in general have a significant but negative effect. In other words, the public tends to rally behind the President when the U.S. or its citizens are attacked by terrorists, but it becomes more circumspect or critical of the Administration when terrorist incidents take place in general. Furthermore, the impact of an attack on the United States on public support lasts at least three months, while the impact of a more general incident is more fleeting, lasting no more than one month. This suggests that Presidential cues may matter or, at a minimum, that the President s approval rating will benefit for at least three months. Tables 4, 5, 6 and 7 evaluate the impact that the risk of terrorist attack has on public support for the President and U.S. foreign policy. As noted above, risk is a function of the costs of a given attack, the likelihood that an individual respondent will be the target of an attack, and the likelihood that an attack will happen again. The costs of a given attack are a function of the number of victims, the level of damage, and the level of violence. The results of the Pretty Prudent (PP) variables in Table 4 suggest that the attacks against the United States and U.S. citizens that are violent, involve U.S. victims dying, and have either high or low levels of damage are associated with higher levels of support for the President and U.S. foreign policy. Interestingly, while the public supports the President when U.S. citizens are killed in an attack, they challenge the President when U.S. citizens are wounded. The effect of such attacks in the preceding two months is more mixed with violent and low damage attacks remaining significant, but U.S. killed or 14

16 wounded and high damage attacks not. Thus, the public tends to rally in support of the President and U.S. foreign policy in response to violent terrorist incidents that involve death and damage to its people or territory, but this rally effect is short lived. In contrast to attacks targeted against the United States, terrorist incidents in general with high levels of violence and high or low levels of damage do not have an immediate effect on public support for the President. Furthermore, while general incidents in the past three months that were violent or involved low levels of damage are associated with a boost in popular support, general incidents in the past three months which involved high levels of damage actually reduced public support for the President and U.S. policy. This bolsters the earlier finding that the public is likely to rally in favor of the President even when high costs are involved as long as the attacks involve U.S. people or territory, but it becomes more critical and its support declines when more generic costly terrorist incidents take place. 15

17 Table 4: Costs and Presidential Foreign Policy Approval Pretty Prudent Violent 0.23 (Violent attack against US) (2.97) ** PP Violent Lag 0.16 (3.88) ** Violent Attack (Violent attack against anybody) (-1.25) Violent Attack Lag 0.04 (2.02) * PP High Damage 2.32 (High levels of damage against US) (2.95) ** PP High Damage Lag 0.02 (0.04) PP Low Damage 0.37 (Low levels of damage against US) (2.01) * PP Low Damage Lag 0.52 (5.16) ** Damage High 0.53 (High levels of damage against anybody) (0.98) Damage High Lag (-1.93) * Damage Low 0.19 (Low levels of damage against anybody) (1.76) Damage Low Lag 0.25 (4.45) ** US Wounded (Wounded US citizens) (-2.25) * US Wounded Lag (-4.55) ** US Killed 1.14 (Killed US citizens) (2.82) ** US Killed Lag 0.55 (1.73) POST (Post 911 dummy) (6.45) ** (5.62) ** (7.17) ** (6.89) ** (6.01) ** CPICH (CPI change from previous Month) (-2.39) * (-1.72) (-2.44) * (-2.16) * (-2.00) * (Constant) (70.94) ** (56.73) ** (72.65) ** (63.80) ** (69.30) ** R Squared N Sig. F (0.00) (0.00) (0.00) (0.00) (0.00) 16

18 Table 5: Type of Victim and Support of Presidential Foreign Policy US Victims Govt or Military (US Govt or Military Personnel) (-0.80) (-0.57) US Victim Govt or Military Lag (-1.45) (-0.93) US Victim Business (US Businesspeople) (2.66) ** (2.89) ** US Victim Business Lag (1.52) (1.97) * US Victim Tourist (US non-combatants) (2.97) ** (2.48) ** US Victim Tourist Lag (4.64) ** (4.46) ** US Victims High (High levels of US victims) (0.97) (2.15) * US Victims High Lag (-0.78) (0.08) NATO (Attack against a NATO member) (-2.11) * (-2.11) * NATO Lag (-2.31) * (-0.54) Damage People (Damage primarily to people) (-6.91) ** (-7.18) ** Damage People Lag (-6.54) ** (-6.76) ** Damage Property (Damage primarily to property) (2.00) * (1.39) Damage Property Lag (6.32) ** (6.57) ** Damage High 1.60 (High damage levels) (3.16) ** Damage High Lag (-0.24) POST (Post 911 dummy) (6.93) ** (5.70) ** (5.17) ** (1.18) (1.33) CPICH (Change in CPI) (-2.37) * (-2.30) * (-1.53) (-0.44) (-0.10) (Constant) (62.38) ** (58.98) ** (67.84) ** (55.63) ** (55.43) ** R Squared N Sig. F (0.00) (0.00) (0.00) (0.00) (0.00) 17

19 Table 6: Type of Incident and Approval of Presidential Foreign Policy PP Airplane (Airplane incident against US) (-1.67) PP Airplane Lag (-0.60) PP Kidnap (Kidnapping against US) (-0.82) PP Kidnap Lag (-0.75) PP Bomb 0.14 (Bombing against US) (1.37) PP Bomb Lag 0.22 (3.30) ** PP Armed Attack 0.34 (Armed attack against US) (1.58) PP Armed Attack Lag 0.06 (0.42) PP WMD (WMD against US) (-1.41) PP WMD Lag (-2.91) ** Incident Airplane 0.09 (Airplane incident against anybody) (0.28) Incident Airplane Lag 0.25 (1.10) Incident Kidnap and Hostage (Kidnapping against anybody) (-2.59) ** Incident Kidnap and Hostage Lag (-3.23) ** Incident Bomb 0.03 (Bombing against anybody) (0.71) Incident Bomb Lag 0.10 (4.55) ** Incident Armed Attack (Armed attack against anybody) (-4.65) ** Incident Armed Attack Lag (-2.66) * Incident Exotic WMD (WMD against anybody) (-0.61) Incident Exotic WMD Lag (-2.55) * POST (Post 911 dummy) (6.40) ** (2.05) * CPICH (CPI change from previous month) (-2.04) * (-1.95) * (Constant) (50.82) ** (49.13) ** R Squared N Sig. F (0.00) (0.00) 18

20 Table 7: State Sponsorhip, Al Qaeda and Support for Presidential Foreign Policy State Sponsorship All (Any type of State Spons (-0.04) (-1.15) (0.21) Sate Sponsorship All Lag (-3.59) ** (-5.45) ** (-3.53) ** Pretty Prudent 0.23 (Attacks against US) (4.44) ** Pretty Prudent Lag 0.17 (5.75) ** TerrorMonth (Attacks against anybody) (-2.49) ** Terror Month Lag 0.01 (1.12) AQ Attacks (Attacks by Al Qaeda) (3.19) ** (3.21) ** AQ Attacks Lag (2.16) * (2.24) * US Victims High 0.79 (High levels of US Victims (2.18) * US Victims High Lag (-0.08) POST (Post 911 dummy) (5.06) ** (5.96) ** (4.30) ** (1.90) (1.99) * CPICH (Change in the CPI) (-1.48) (-2.75) ** (-1.50) (-0.91) (-0.73) (Constant) (80.40) ** (65.83) ** (53.89) ** (80.46) ** (77.28) ** R Squared N Sig. F (0.00) (0.00) (0.00) (0.00) (0.00) 19

21 The respondent s perception that he or she or his or her family will be targeted by the next terrorist event is presented in Chart 2. The findings indicated that about 25 percent of people expressed at least some worry before 9/11 and that 47 percent expressed at least some worry in September of 2001 (September 2001 is month 260 and is marked by the index line). 50 Chart 2: How worried that you or your family will become a victim? 40 Percentage of People Worried Month Since January 1980 Unfortunately, data for this question are scarce. To compensate, we estimate the likelihood that an individual respondent will be targeted in a future attack in terms of the type of victim targeted and type of weapon used in a current terrorist incident. Table 5 presents the impact that type of victim in an attack has on public opinion. We predict that respondents will assign a lower probability of themselves being attacked if a previous attack involved U.S. military or government victims as opposed to business people, tourists, or other noncombatants. They are also expected to assign a lower probability of 20

22 themselves being attacked if the past attack involves property rather than people. In support of these propositions, the public supports the President when attacks involved business people or tourists and other non-combatants. These effects last beyond two months and remain significant when controlling for attacks that involve high levels of U.S. victims in general. These results suggest that the public may be risk acceptant and will rally behind the President when it perceives the threat against itself to be high. On the other hand, counter to the general assumption that the U.S. public will rally to support its troops, incidents involving U.S. military and government officials had no effect on public support. An observation we discuss at further length below. Furthermore, while attacks involving property tend to increase the level of support, attacks involving people make the public more critical of the president. These effects hold when controlling for events that involve high levels of damage and they also persist for at least two months. Chart 3 provides a summary of the types of terrorist incidents which took place during the time period under review. Table 6 expands this analysis by examining the impact of different types of weapons on the level of public support. Based on the proposition that the general public will rally when it perceives itself to be a target, one would expect it to rally in response to forms of attack that are more likely to target civilians like kidnappings, hostage taking, airplane incidents, and bombings and to be more circumspect of forms of attack that are more likely to be used against military personnel like armed attack or weapons of mass destruction. Interestingly, with the exception of attacks in the last three months involving bombings and weapons of mass destruction, attacks of different kinds against U.S. targets (Pretty Prudent) had no significant effect on public opinion. In contrast, as expected by our argument armed 21

23 attacks and incidents involving weapons of mass destruction against general (US and non-us) targets made the public more critical, while bombings against general targets made it more supportive of U.S. foreign policy. In contrast to expectations, kidnappings in general made the public more critical rather than supportive. This may be due, in part, to the relative rarity of kidnappings in comparison to other forms of attack. The mixed results lend further support to the proposition that the public differentiates between attacks involving U.S. targets and those that do not, but the varying effect of different types of weapons is unclear and worthy of future research. Chart 3: Types of Terrorist Incidents Mean Number of Incidents Exotic, WMD Armed Attack Bombing Kidnapping, Hostage Airplane Year The final component of risk is the probability that a past attack will be repeated. We posit that this probability increases with knowledge of state sponsorship and the involvement of a known adversary, like Al Qaeda. The results in Table 7 are mixed. On one hand, state sponsorship has no immediate effect on popular support, but state sponsorship of an attack in the past three months makes the public more critical of the 22

24 President and U.S. foreign policy. This effect holds when controlling for attacks against the United States and attacks in general. On the other hand, the involvement of Al Qaeda increases the level of public support and this holds true when controlling for high numbers (10 or more) of U.S. victims in the attack. Chart 4 provides some additional insight into the concept of risk by presenting the public perceptions of the likelihood of another attack. Ironically, both the proportion of people worried that there will soon be another attack and those confident that the United State can defend against another attack rises sharply in September of 2001 (month 260) and decline thereafter. After peaking in September, they tend to move in opposite directions, each varying between approximately 15 and 25 percent. Regression analyses using net positive indicators of How worried are you there will soon be another terrorist attack? (n = 100) and Confidence in the Ability of the U.S. government to prevent further terrorist attacks against its citizens (n = 31) indicate that similar factors are associated with both variables. 12 In particular, overall the responses to both questions are not affected by whether the incidents involved the United States or U.S. citizens. Furthermore, incidents with high levels of violence tend to decrease perceptions of worry as well as confidence, while those involving Al Qaeda tend to increase both. Also, neither perception was affected by the type of person attacked or weapons used. 12 As in the other analyses, all estimates include a post-911 dummy and control for the change in CPI from the previous month. 23

25 Chart 4: Future Terrorist Attacks Percentage of Respondents Worried will soon be another attack Confident US can prevent attack Month Since January 1980 In sum, we find that the U.S. public exercises some prudence when assessing the President and his foreign policy. In particular, in support of the proposition that the national value of the targets matter, the public is supportive when attacks involve U.S. citizens or U.S. territory, but it is critical when attacks in general take place. Furthermore, the public is not skittish in contrast to expectations based on Kantian theory and the Democratic Peace, the public rallies rather than runs from incidents involving high levels of violence or death. Indeed, in further support of the proposition that the public is risk acceptant, our finding indicate that it rallies when terrorist incidents target private citizens as opposed to government or military personnel. Public support also increases when Al Qaeda is involved. This lends support to prospect theory, since the involvement of a known adversary increases the probability that an attack will take place in the future. Finally, our data provide more mixed support for the proposition that 24

26 Presidential cues are important. Although our results do not invalidate this proposition, many of the effects of terrorist incidents on public opinion do not last more than a month. b. The Two Most Important Policy Problems Facing the United States Public identification of the two most important policy problems facing the United States provides useful assessment of the impact that terrorist events have on public perceptions. Charts 5 and 6 below provide a summary of the proportion of people who chose particular issues to be the most important then facing the United States. 13 The responses are provided on a monthly basis. The reference line in each chart marks October of Chart 5: Most Important Problem 40 Mean Proportion of Respondants Terrorism Fear of War Foreign Affairs By Month Since January Although many different polling agencies ask similar questions for consistency we limited ourselves to data from Gallup and LA Times. 25

27 Chart 6: Most Important Problem Mean Proportion of Respondants Terrorism Fear of War Foreign Affairs Economy Ethics and Morality Healthcare Months Since January 1980 As expected, the importance of Terrorism peaks after the attacks on September 11, When compared with other factors, however, the importance of Terrorism subsides very quickly and returns to a level of roughly 10 percent which is comparable to other issues including ethics and morality, health care, the budget, and other matters in general (not all of them are shown on the table). Despite the decline of Terrorism, two other issues retain public attention Fear of War and the Economy. As expected, the Fear of War peaks in the run-up to the war in Iraq during January and February of 2003 and increases again beginning in December of 2003 when the news of the prison scandal and continuing insurgent attacks bring the matter back to public attention. The economy, in contrast, generally remains more important than other factors after September 11 and only begins to decline in December of Ironically, public concern about Terrorism continues to decline even as its Fear of War increases. This suggests that the public differentiates the war on terror from the 26

28 war in Iraq. In a separate poll before the war in Iraq began, 50 percent of respondents agreed that a war in Iraq would be part of the Global War on Terror. 14 This percentage jumped as high as 77 percent towards the end of the war, but declined thereafter, returning to approximately 50 percent by March of 2004 (See Appendix Table 3). 15 Furthermore, popular support for the Bush Administration s actions in Afghanistan decreased in response to general terrorist attacks, attacks against NATO and attacks by Al Qaeda (the regression analyses are included in Appendix Table 4). At the same time, although attacks in which Americans were killed or involved large numbers of victims increases support, terrorist attacks against U.S. targets (pretty prudent) had no significant effect. Finally, terrorist attacks with no evidence of state sponsorship had a negative impact on support for US military action in Afghanistan. To examine the impact of terrorist attacks on the public s perception of the most important problems facing the United States, we repeated the same regression analyses conducted above on the importance of Terrorism, Foreign Affairs, the Fear of War and the Economy. Table 8 summarizes the analyses of public prudence. Neither attacks on the United States (Pretty Prudent) nor attacks in general (Terror Month) had significant effects on the number of people who designated any of these issues to be among the two most important facing the United States. In addition, while a terror attack against the United States in the past three months decreases the importance of Terrorism, a general terrorist attack in the previous three months increases its importance. This result is likely the function of the general decline in the public s perceived importance of terrorism since 14 Poll sponsored by Gallup/CNN/USA, fielded September 20-22, The high point was in a poll sponsored by ABC/WP, fielded April 16, The March 2004 poll was also sponsored by Gallup/CNN/USA. This poll was fielded March 26-28, For a complete list of responses to this question see the table in the Appendix. 27

29 9/11/2001 despite an increase in the number of attacks, the persistence of the economic recession that followed, and continued efforts by the Bush Administration to emphasize the continued threat of terrorism and link it to a variety of policy initiatives. (See Charts 5 and 6). Table 8: Most Important Problem Terrorism Fear of War International Affairs Economy Pretty Prudent (Attacks against US) (-1.14) (-0.37) (-1.37) (-0.21) Pretty Prudent Lag (-2.33) * (-0.46) (-0.10) (2.17) * TerrorMonth (Attacks against anybody) (0.26) (0.44) (-0.96) (1.51) Terror Month Lag (-1.95) (0.96) (-0.24) (2.26) * POST (Post 911 dummy) (4.26) ** (2.63) ** (1.94) (2.46) (4.83) ** (3.98) ** (6.36) ** (6.81) ** CPICH ** (Change in CPI) (-1.55) (-0.89) (0.52) (0.29) (0.92) (0.74) (1.16) (0.89) (Constant) (2.56) * (2.26) * (3.46) ** (1.60) (8.24) ** (6.78) ** (5.43) ** (3.32) ** R Squared N Sig. F (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) We argued that risk is a function of the costs, the likelihood that the respondent will be targeted in a future attack, and the likelihood that a future attack will take place. Neither the likelihood of a future attack as measured in terms of state sponsorship nor the involvement of Al Qaeda had a significant impact on the importance of any of these issues. Therefore, we will focus on the respondent s assessment of the costs associated with a terrorist attack and the probability that he or she will personally be targeted by the next attack Given the large number of insignificant results the results are not printed in this manuscript. They are available on request. 28

30 The importance of Terrorism: Violent attacks against U.S. citizens did not affect the importance of terrorism. On the other hand, the public s perception of the importance of terrorism increased in response to violent attacks in general, as well as those involving high or low levels of damage. Thus, in contrast to expectations based on the earlier findings about support for the President and U.S. foreign policy, public perceptions of terrorism were not affected by the costs of attacks against U.S. targets, but they increased in response to violent attacks in general. In support of our findings regarding risk acceptance, the public perception of the importance of terrorism increased in response to attacks involving weapons that are likely to be used against civilians like hijackings and decreased when the weapons chosen are generally used against military personnel as with armed attacks or WMD. At the same time, while the public did not discriminate between the types of victim in the terrorist incident, it responded to terrorist incidents which targeted people rather than property. Yet, curiously, it downgraded the importance of terrorism in response to incidents in which people were wounded rather than killed. In sum, the public is less prudent in its assessment of terrorism as an important policy issue it does not discriminate in terms of the importance of U.S. values, and is less discriminating in terms of costs and the likelihood that they would be affected by the next terrorist attack directly. The importance of the Fear of War: In terms of costs, neither the level of violence or amount of damage of terrorist attacks against U.S. targets had a significant impact on the fear of war. At the same time, general incidents involving high levels of violence or damage in the past three months increased the fear of war. In terms of the likelihood that the respondent will be the target in the future, large number of U.S. 29

31 victims, and attacks involving U.S. military or government employees increased the importance of the fear of war. At the same time, attacks involving tourists and other noncombatants decreased the importance of the fear of war. In sum, the public appears more prudent in its assessment of the importance of the fear of war than in its assessment of the importance of terrorism. Although the level of violence did not affect the public s assessment, its responsiveness to attacks involving large number of victims and the perceived importance of attacks against military and government personnel suggest that the public perceives war in traditional state-to-state and perhaps military-to-military terms that are of a different nature than concerns about transnational threats reflected in the importance of terrorism. The importance of International Affairs: Strikingly, terrorist attacks had virtually no impact on the proportion of people who considered international affairs to be one of the two most import policy issues. Of all the variables analyzed above, only a high level of damage in the previous three months had an effect. It alone is associated with an increase in the importance of international affairs. This suggests that the public does not consider terrorism to be international or at least not part of what it conceives of as international affairs. The importance of the Economy: We posit that the impact of terrorist incidents on the economy is likely to be highest when the costs of a terrorist attack are high, and when targets involve U.S. territory or people because of the negative impact such attacks are likely to have on the economy. The public is somewhat less discriminating than it is when evaluating the President in terms of whether the attack involves the U.S., or others. For example, armed attacks and those which were violent or affected NATO allies 30

32 increased the perceived importance of the economy regardless of whether they were exclusively focused on the United States. At the same time, while a high number of U.S. victims in the current month is associated with greater importance of the economy, higher levels of US victims in general and higher levels of U.S. military or government victims in the preceding two months decrease the importance of the economy. This suggests that the public views violent attacks and those against others as costly, but it focuses on other matters like security or terrorism when attacks take place against the U.S. government or military. c. Effectiveness of President s Handling of Terrorist Attacks and Aftermaths Although there were few periods free from terrorist incidents during the time frame of this study, there were several attacks and responses that clearly dominated the news coverage at the time they took place. Analyzing public perceptions around these specific events further illuminates their impact on the level of public support for the president s handling of foreign policy. In this section we examine several different notable terrorist attacks that took place on US soil or involved US citizens during the period Events during the Reagan Administration include the crises in Lebanon between 1983 and 1984, the hijacking of the Achille Lauro in 1985, the bombing of La Belle Disco in 1986 and the bombing of Pan Am 103 in Events during the Clinton Administration include the bombing of the World Trade Center in 1993, the Murray Federal Building in 1995, the Khobar towers in 1996, the US embassies in Kenya and Tanzania in 1998 and the USS Cole in Finally, in addition to the attacks on September 11, the Bush Administration faced anthrax attacks in Washington, 31