Surprising Events and Surprising Opinions

Abstract

The academic debate concerning public opinion about war focuses upon two explanations: cost/benefits and partisan cues. Both sides of this debate use laboratory experiments to estimate the influence of events and cues, but Gelpi is notable for using a well-designed experiment to compare the theories simultaneously. He argues that his results support the cost/benefits explanation as “surprising events” that counter individuals’ prior attitudes have significantly more effect than “surprising opinions” upon people’s attitudes toward the Iraq War. His analysis, however, considers only the direction, but not the strength, of people’s attitudes toward the war. Additionally, the measure of source credibility for determining the influence of cues is not optimal. When the analysis accounts for attitude strength and uses a better measure of source credibility, the results show little support for the effect of surprising events and markedly greater support for the influence of partisan cues.

Keywords

foreign policy war domestic politics belief structure

Public Support for War

Foreign policy is often a secondary focus of public opinion research, but it is important because the public’s attitudes about foreign policy, especially during wartime, can influence leaders’ policy choices, elections, and even political polarization. Many studies indicate that political leaders are attentive to public opinion when formulating foreign policy (Foyle 2004; Holsti 1996 but also see Jacobs and Page 2005). Research on retrospective voting (Fiorina 1981) and issue salience (Aldrich, Sullivan, and Borgida 1989) in elections has shown how foreign affairs can influence voters’ decisions. More recent studies argue that the Iraq War had a major effect upon the outcomes of the 2004 American presidential election (Gelpi, Reifler, and Feaver 2007; Karol and Miguel 2007), the 2006 midterm elections (Grose and Oppenheimer 2007; Jacobson 2007), and 2005 British elections (Clarke et al. 2006). Finally, wars can be divisive events that contribute to political polarization. Strong attitudes concerning the Vietnam War helped reinforce existing divisions within American society (Layman, Carsey, and Horowitz 2006). Recent data reveal significant polarization concerning whether the United States was right to take military action against Iraq, with 61 percent of Republicans thinking it was right compared to only 29 percent of Democrats (CBS News Poll, March 1–3, 2013) and whether the Bush Administration deliberately misled the American public prior to intervention, with 76 percent of Democrats saying the country was misled and 75 percent of Republicans taking the opposite view (CNN/ORC Poll, March 15–17, 2013). In short, attitudes toward wars can have tremendous consequences because they can stir strongly held beliefs.

Given the political importance of the public’s support for wars, it is not surprising that there is an evolving literature on this topic. Research on the public’s attitudes toward war conducted since the early 1970s offers a cost/benefit explanation, arguing that attitudes toward war are shaped primarily by rational consideration of the potential net gains from military involvement. The earliest work examined the Korean and Vietnam Wars, focused upon casualties, and concluded that the American public’s support for war decreased as the number of casualties increased (e.g., Mueller 1971). More recent research (e.g., Althaus, Bramlett, and Gimpel 2012; Gartner 2008) has further examined the subtle nature of these effects. Other research supporting a cost/benefit explanation, however, challenges the idea that casualties influence support for war, contending that the effects attributed to the number of casualties were conflated with the public’s perception of the war’s progress. Rather, people are willing to support wars that are initiated for the appropriate motives and are moving toward victory, even as casualties increase (Gelpi, Feaver, and Reifler 2009). Boettcher and Cobb (2009) claim that the effects of casualties upon support depend upon both how losses are framed and people’s initial attitude toward the war. In all, the cost/benefit approach presumes that the public uses information about the war’s conduct for weighing the cost and benefits of initiating or continuing military action.

A competing explanation emphasizes the influence of partisan cues and arises from Zaller’s (1992) research on the influence of political information and elite cues upon attitudes. This approach (e.g., Berinsky 2007; Larson 1996) argues that most people lack the information necessary to evaluate the costs and benefits of war and instead rely upon their political predispositions and partisan elites’ attitudes. This research starts from Zaller’s (1992) observation that attitudes toward the Vietnam War shifted among the politically informed segment of the public and followed splits in leaders’ views. This explanation argues that the public does not need specific information about the war, only information about partisan leaders’ positions on the war.

Until the recent Afghanistan and Iraq Wars, the absence of good cases for replication made it difficult to study the microfoundations of this question thoroughly.¹ The Afghanistan and Iraq Wars are valuable for examining variables that influence the public’s support for war because while both started off similarly, with the enemy regimes quickly driven from power, the United States and its allies became entangled in a protracted struggle to stabilize each country against hostile insurgent forces. With these conflicts, it became possible to examine the public support for continued engagement by taking into account casualties, the progress of the war, elite cues, and even the rationale for starting the conflict.

Disentangling the many factors that influence the public’s attitudes toward the war, however, is a problem for this research. While the number of casualties and length of each war allow for an examination of the effect of new information from the battlefield and elites upon people’s attitudes toward war, it can still be difficult to sort out the various influences because elites’ attitudes may change in correspondence with battlefield events. As a result, some researchers are using experiments to isolate how casualties, domestic and international opinion, and the rationale for the war influence the public’s attitudes toward war (e.g., Berinsky 2007; Boettcher and Cobb 2009; Gartner 2008; Gelpi, Feaver, and Reifler 2009).

While many of these studies are designed to test aspects of both explanations, Gelpi’s (2010) experimental study pits these two explanations directly against one another. Subjects were randomly assigned to one of the nine conditions and received positive, negative, or no information about the progress of the war, as related by events in Iraq, and positive, negative, or no information about President Bush’s assessment of the war. The study’s aim was to determine whether events, as the cost/benefit perspective predicts, or messages from President Bush, as the partisan cue theory predicts, have more influence upon public opinion about the Iraq War. Gelpi argues that his results show that changes in people’s perceptions of the costs and benefits of war through information about “surprising events” have the most influence upon people’s attitudes about war and that cues have very little effect. While Gelpi recognizes that partisan cues do shape differences between Democrats’ and Republicans’ views on the war, he concludes, “the experiment yielded relatively little direct support for arguments suggesting that elite partisan opinion–whether ‘surprising’ or ‘reinforcing’–had much impact on attitudes toward Iraq” (p. 108).

In this article, I argue that a better test of the influence of events and cues upon people’s attitudes about the war considers how new information can influence both the direction and the strength of people’s attitudes. In particular, I reanalyze Gelpi’s data and show that surprising events have little influence over attitudes when we also consider the strength of people’s attitudes. Moreover, cues have a greater effect than in Gelpi’s study once an alternative measure of source credibility is used.

Surprising Events and Attitudes about War

Gelpi’s (2010) treatments present realistic news stories that convey information, in the events treatments, indicating either a positive or a negative direction of the situation in Iraq and, in the cues treatments, a message from President Bush expressing either his confidence with the war’s direction and intention to maintain the US forces in Iraq or his concern with the war and intention to reduce troops. While these treatments could affect all respondents uniformly, Gelpi proposes that the effects of events are conditional upon “individuals’ current attitudes” toward the war and argues that this allows him to compare the influence of events that either reinforce or contradict respondents’ existing beliefs (p. 92). Party identification is a proxy for current attitudes in the analysis, so, under the “reinforcing events” hypothesis, negative event information has more influence upon Democrats’ attitudes about the war, whereas the surprising events hypothesis predicts that positive information has more influence for Democrats, with opposite relationships expected for Republicans. Independents are expected to react new information like Democrats because they are also assumed to have negative attitudes toward the war based upon results from another survey conducted in February 2008 (p. 96).

Gelpi argues that the effect of elite cues depends upon the source’s credibility with the recipient and whether the cue is consistent or inconsistent with the source’s predispositions. In the absence of specific information about respondents’ perceptions of President Bush’s credibility, Gelpi proposes that President Bush’s influence is limited to Republicans because only they are likely to see Bush as a credible source (p. 97). Gelpi acknowledges that a single cue-giver might not have much effect because Zaller’s (1992) theory of elite influence highlights the effects of competing elites providing similar or opposing views, but Berinsky (2007) argues that even a single cue-giver with a known predisposition can provide an informative signal. Beyond the source’s personal credibility with the respondents, Gelpi argues that a message that is inconsistent with the source’s predisposition can have more effect than a consistent message because an unexpected message is thought to provide new and more credible information (see Eagly, Wood, and Chaiken 1978; Petty et al. 2001; Priester and Petty 1995). A cautious assessment (surprising opinion) from President Bush, therefore, is hypothesized to have more influence on Republicans’ attitudes than a confident one (a “reinforcing opinion”). In contrast, the reinforcing opinions hypothesis is that a positive message about the war from President Bush bolsters Republicans’ support through motivated processing (Taber and Lodge 2006).

Gelpi uses binary logit to estimate the relationship between the experimental treatments and the three measures of support for the war: the belief that the “surge” improved the US situation in Iraq, the belief that the US efforts would succeed, and disapproval of a timetable for the withdrawal of troops. He argues that the data provide strong support for the surprising events explanation of attitude change (see his Figure 3), as positive events have a significant effect upon Democrats’ attitudes for two of the three measures of support for the Iraq War and Independents’ attitudes on all three measures. Likewise, negative events have a significant effect upon Republicans’ attitudes for two of the three dependent variables. At the same time, there is very little support for the influence of elite cues. A cautious statement from President Bush influences only Independents’ attitudes about the likelihood of success in Iraq.

Attitudes toward War and Attitude Strength

Gelpi examines the influence of events and cues upon the direction of people’s attitudes toward the Iraq War but ignores the strength of people’s attitudes. The question measuring beliefs about whether the war in Iraq would be successful allows respondents to say that “very likely,” “somewhat likely,” “not very likely,” or “not at all likely” (p. 99).² The question, therefore, measures not only whether respondents think the war in Iraq will succeed but also the certainty with which they hold their opinions. Similarly, the question about a timetable allows respondents to express the strength of their approval or disapproval of a timetable for withdrawing troops, thereby providing information about the extremity of their views. Measuring the likelihood of success in Iraq and approval of a timetable by giving respondents the option to express the strength of their attitudes provides valuable information about participants’ preferences, permits better tests of the hypotheses, and may yield greater insight about the process and consequences of the formation of attitudes toward war.

Attitude strength is a theoretically important concept in both political science (e.g., Miller and Peterson 2004; Peterson 2004) and social psychology (see Visser, Bizer, and Krosnick 2006) and is conceptualized along dimensions such as intensity, extremity, certainty, importance, and accessibility. This research emphasizes that attitude strength is relevant for four characteristics of an attitude: persistence, resistance, impact on information processing and judgments, and guiding behavior (Krosnick and Petty 1995). Stronger attitudes are more resistant to change, more likely to influence how new information is processed, and more likely to influence people’s political behavior than weaker attitudes. The importance of the strength of party identification for attitude formation, information processing, and political behavior is widely recognized (e.g., Campbell et al. 1960). Stronger attitudes on the Iraq War, similarly, could be expected to have a greater impact upon presidential approval, both in general and with respect to the president’s handling of the war. These, in turn, could influence elections held during wartime and the president’s war policy. While the dimensions of attitude strength arise from different factors and have different consequences for the four characteristics of attitude strength, attitudes that are more extreme or held with greater certainty are more likely to resist persuasion and influence behavior (see Krosnick et al. 1993). Examining the strength of attitudes, therefore, can enrich our understanding of both the durability of attitudes in the face of new information and their effects upon political behavior.

While there has been renewed interest in attitude strength, Verba and Brody (1970) showed that stronger attitudes affected the relationship between people’s attitudes toward the Vietnam War and their behavior. Creating a scale from questions about Vietnam, they classified respondents as adopting “dovish” or mildly dovish, “middle,” or “hawkish” or mildly hawkish views on Vietnam and found that hawkish and dovish respondents were more likely to participate politically concerning the war than either mild hawks or mild doves. Their study provides evidence that the extremity of people’s attitudes about war is important for understanding how public opinion toward war can guide the public’s actions directed at influencing political leaders’ decisions about the war.

Discarding valuable information about the strength of people’s attitudes should only be done with clear justification for several reasons. First, knowing the extremity of people’s attitudes provides information about whether the public is greatly divided over an issue or whether some type of compromise is possible. Given the recent political environment, it would not be surprising to find that attitudes toward the war in Iraq contributed to partisan polarization, but it is difficult to determine this without examining attitude extremity. While ignoring the strength of respondents’ attitudes could be defended if few people expressed strong attitudes, significant numbers of respondents discriminated between stronger and weaker attitudes on both items (Table 1). Moreover, while the distribution of subjects’ perceptions about the likelihood of success is fairly symmetrical, subjects who disapproved of a timetable were more extreme in their views than those who approved. President Bush, therefore, would have had stronger public backing in resisting any timetable than his opponents would have in calling for one. Given this, it is important to know the type of information that would moderate or polarize subjects’ opinions on these issues, not just change the direction of their attitudes.

Table 1.

Strength of Iraq War Attitudes.

Iraq success		Iraq timetable
Not at all likely	16.95 (169)	Strongly approve	20.80 (207)
Not very likely	34.20 (341)	Somewhat approve	27.44 (273)
Somewhat likely	29.89 (298)	Somewhat disapprove	19.20 (191)
Very likely	18.96 (189)	Strongly disapprove	32.56 (324)
N	997		995

Note: Entries are percentages, with number of observations in parentheses.

Second, Gross, Holtz, and Miller (1995, 232) write that “it is generally accepted that subjective uncertainty creates a predisposition to be influenced.” By examining attitude certainty about the war, we could find that supporters of the Iraq War who hear President Bush provide a cautious assessment of the war’s direction still believe that success is more likely than failure, but become less certain that the United States will achieve its goals. Similarly, a cautious statement might be insufficient to change the direction of people’s opinion about the war, but it could reduce the extremity of their attitudes toward a timetable for troop withdrawal and make people more willing to change their position given additional negative information. Such a finding would tell us about the process through which elites can persuade people to change their attitudes about the war as well as leaders’ incentives to carefully craft their messages about a war.

Finally, ignoring attitude strength makes it more difficult to find even apparent support for the reinforcing events and opinions hypotheses. Table 2 shows that 84 percent of the Republicans in Gelpi’s study who received neither positive nor negative events information disapproved of a timetable. Assuming, through randomization, a similar distribution of initial attitudes among Republicans in the positive events treatment means that we could observe change in support from only 16 percent of the relevant respondent pool by considering only attitude direction. Examining attitude strength almost triples that pool, providing 45 percent of respondents who we could observe becoming more opposed to a timetable after receiving positive information. The reinforcing opinions hypothesis faces a similar limitation. If only Republicans are likely to have positive prior attitudes toward the war and see messages from President Bush as credible, then a positive cue from President Bush cannot affect the direction of most Republicans’ views of the war, only the strength of their support. As it turns out, the positive events treatment displays little difference in the direction of Republicans’ attitudes toward a timetable but does cause more extreme disapproval of a timetable. By examining only attitude direction, this evidence supporting the reinforcing events hypothesis would be overlooked.

Table 2.

Support for a Timetable and Treatments–Republicans.

	Experimental treatment
	No event	Positive event	No cue	Positive cue
Strongly approve	5.68 (5)	4.12 (4)	7.29 (7)	5.75 (5)
Somewhat approve	10.23 (9)	10.31 (10)	12.50 (12)	13.79 (12)
Somewhat disapprove	29.55 (26)	16.49 (16)	13.54 (13)	19.54 (17)
Strongly disapprove	54.55 (48)	69.07 (67)	66.67 (64)	60.92 (53)
N	88	97	96	87

Note: Entries are percentages, with number of observations in parentheses.

A more serious problem that Table 2 illustrates is that it is impossible to truly test the reinforcing events hypothesis unless one considers attitude strength. Assume again that 84 percent of the Republicans in the positive events treatment initially disapprove of a timetable. Disapproval of a timetable among respondents in the positive events treatment must be greater than 84 percent in order to show that a positive event caused a more positive view of the war. By considering only the direction of attitudes, the disapproval of a timetable can increase above 84 percent only if, paradoxically, those who initially approved of a timetable—that is, people who receive information that contradicts their initial attitudes—change their attitude. If people do engage in motivated processing, the 16 percent who approve of a timetable should “selectively ignore” positive event information about the war (p. 92). The absence of any difference in the distribution of the direction of attitudes between the two treatments is actually consistent with motivated processing, whereas an increase in opposition to a timetable among Republicans exposed to positive event information would actually support the surprising opinions hypothesis. Table 2 shows, however, that it is possible to show support for the reinforcing events hypothesis arising from motivated processing by allowing people who already disapprove of a timetable to more strongly disapprove of a timetable after receiving positive events information.

The reinforcing opinions hypothesis can be tested even when ignoring attitude strength because whereas the reinforcing events hypothesis “is based upon individuals’ current attitudes on the issue in question,” the reinforcing opinions hypothesis contends that “individuals will attend to cues that reinforce their party identification rather than their prior attitude on the issue per se” (p. 92). This means that a positive message from President Bush could convince Republican opponents of the war to bring their attitudes back into line with the Republican Party’s position, although considering attitude strength still increases the ability to find support for this hypothesis. The important point is that ignoring attitude strength impedes tests of hypotheses derived from motivated processing (Taber and Lodge 2006, 757).

Table 3 displays ordered logit estimates of Gelpi’s model, changing only the likelihood of success and opposition to a timetable variables to account for attitude strength.³ Once attitude strength is considered, positive events do not have an effect that is statistically different from the “no event” treatment for Democrats and negative events do not have a significant effect upon Republicans’ attitudes. Rather, the effect of the positive event treatment upon Republicans’ disapproval of a timetable, consistent with the data in Table 2, is significantly greater than both the no and negative events treatments, suggesting that reinforcing events strengthened Republicans’ opposition to a timetable.

Table 3.

Replicating Gelpi’s Table 1 for Measures Including Attitude Strength.

	Success in Iraq			Disapprove of timetable
	Democrats	Independents	Republicans	Democrats	Independents	Republicans
Positive event	0.120 (0.262)	0.645** (0.226)	0.188 (0.274)	−0.278 (0.253)	0.667** (0.223)	0.513* (0.295)
Negative event	−0.301 (0.255)	0.120 (0.235)	0.003 (0.273)	−0.282 (0.247)	0.201 (0.233)	−0.017 (0.283)
Confident Bush	0.272 (0.258)	−0.154 (0.235)	−0.007 (0.282)	−0.091 (0.247)	0.049 (0.233)	−0.150 (0.305)
Cautious Bush	0.360 (0.262)	−0.432* (0.224)	−0.038 (0.263)	0.030 (0.253)	−0.217 (0.222)	−0.318 (0.286)
τ₁	−0.981** (0.244)	−1.469** (0.215)	−2.617** (0.322)	−0.959** (0.239)	−1.085** (0.204)	−2.683** (0.333)
τ₂	1.395** (0.252)	0.265 (0.198)	−1.576** (0.269)	0.797** (0.239)	0.147 (0.194)	−1.425** (0.275)
τ₃	3.647** (0.389)	1.761** (0.221)	0.418* (0.250)	2.179** (0.283)	1.008** (0.201)	−0.465* (0.261)
N	332	374	291	332	374	289
$χ_{4}^{2}$	5.26	11.76	0.66	1.81	10.46	5.50

Note: Entries are maximum likelihood estimates with standard errors in parentheses.

*p < .05, one-tailed.

**p < .01, one-tailed.

Positive events do continue to influence Independents’ attitudes concerning the likelihood of success and disapproval of a timetable. The surprising events hypothesis may, therefore, apply at least for Independents; although when considered along with the positive effect upon Republicans’ attitudes toward the war of positive events, it is possible that the effect of positive events is not conditional upon people’s prior positions on the war. There is, however, one ambiguity concerning how to interpret results for Independents. Partisanship in the analysis was measured only with respect to direction, but the survey documentation shows that the 374 respondents classified as Independents include partisan leaners as well as people who answered “other” or “not sure” to the initial party identification question and indicates that only 156 respondents are pure Independents, with another 94 Republican and 100 Democratic leaners. This is relevant because Independents who lean toward one of the parties are often more like weak party identifiers than true Independents (Keith et al. 1992). Consequently, “Independents” include many respondents whose attitudes toward the war are similar to Democrats, but also many who, like Republicans, support the war. It is even possible that the effect of positive events could reflect reinforcing effects among the Republican leaners who are classified as Independents and are incorrectly assumed to have negative existing attitudes toward the war.

While a poll conducted in February 2008 by the Gallup Organization for USA Today that distinguishes leaners from pure Independents confirms that pure Independents had more negative views of the war, with 60 percent believing that Iraq would be viewed as a failure (See online appendix Table A2) and 64 percent either favoring immediate withdrawal or setting a timetable for gradual withdrawal (See online appendix Table A3) that survey also bears out the expectation that leaners held views much more like their respective partisans. Among Republican leaners, only 35 percent believed that Iraq would ultimately be a failure and 38 percent disapproved of any timetable, compared with 26 percent and 33 percent, respectively, for weak and strong Republicans. In other words, Gelpi groups respondents with very different prior views of the war into the Independent category but makes similar predictions for all of them. The significant effect for Independents of a negative cue from President Bush upon the likelihood of success could also arise from the Republican leaners being incorrectly classified as Independents. It makes sense, therefore, to consider whether respondents’ prior attitudes toward the war and their views of Bush as a credible source are better measured by something other than partisanship.

Attitude Change and Approval of President Bush

Gelpi proposes that the effect of cues from President Bush is conditional upon the tone of his message and respondents’ perceptions of his credibility. While the experiment manipulates the message tone, respondents’ perceptions of Bush’s credibility are not manipulated and, therefore, need to be measured. Gelpi does not directly measure respondents’ perceptions of President Bush’s credibility but instead uses respondents’ party identification as a proxy for credibility, citing Lupia and McCubbins’s (1998) argument that speakers can be persuasive when they are perceived as sharing a common interest with the recipient (p. 90).

Partisanship, however, may be limited as a measure of people’s perceptions of President Bush as a credible source of information about Iraq for two reasons. First, respondents’ other attitudes toward the speaker may influence their perception of the speaker’s credibility, as Gelpi notes, “partisan cues in this study is combined to some extent with their opinions of Bush” (p. 96). This is consistent with social psychology, which views credibility as an attribute of a specific source, based upon the source’s expertise and trustworthiness (e.g., Petty and Cacioppo 1981; Pornpitakpan 2004). Expertise concerns whether the source is perceived as speaking with authority, based upon the source’s access to accurate information about and understanding of the topic. Indicators include judgments of the source’s knowledge and intelligence. Trustworthiness concerns whether the source is communicating in good faith and is based upon personal characteristics, such as whether the source is perceived as honest, sincere, or ethical, along with situational considerations, such as the source’s intent to persuade and similarity with the message recipient. Of these traits, Priester and Petty (1995) find that “honest,” “trustworthy,” and “knowledgeable” are the three primary indicators of respondents’ beliefs that a source would provide an accurate statement, while “similar” ranked seventh among seventeen traits.

It is easy to see how a proxy based upon party identification as the sole determinant of credibility ignores how sources from the same party can be perceived differently with respect to their individual expertise and trustworthiness. For example, Republican supporters of the war interviewed in 2008 may have agreed that President Bush and Sen. John McCain shared the respondent’s interest in continuing American involvement in Iraq but had very different judgments of their trustworthiness and expertise on the matter. Data from the 2008 American National Election Study (ANES) illustrate this. While 46 percent of Republican identifiers rated Bush and McCain as equally honest, 43 percent thought that McCain was more honest (See online appendix Table A6). The distribution for knowledgeable is roughly the same. Consequently, if McCain was seen as more expert or honest than Bush, then McCain may reasonably be seen as a more credible advocate than Bush. Party identification influences perceptions of these traits, but differences can emerge among copartisans’ perceptions of a specific source’s credibility. The research concerning credibility indicates that we should consider how message recipients evaluate the specific personal attributes of a known source to arrive at a judgment about the source’s credibility.

Another reason why partisanship may not be the best measure of credibility is that shared interests can also be based upon attitudes toward specific policies about which the source is speaking. When President Bush’s message makes the war in Iraq salient, Republicans who oppose the war may see Bush primarily not as a fellow partisan, but as someone with an opposing interest in the war whose credibility on Iraq is questionable. Data from the 2008 ANES show that Republicans who had a negative view of the war were less likely to perceive President Bush as honest (See online appendix Table A7). The 2008 Gallup poll described above goes further and asks respondents whether they thought the Bush Administration had deliberately misled the public about Iraq. Respondents’ perceptions of Bush’s trustworthiness on Iraq were strongly related to their attitudes toward the war, even controlling for party identification (See online appendix Table A8–A10).

Gelpi does not ask about credibility, but approval of President Bush’s performance was asked before the experimental treatments and so provides an alternative measure of source credibility (p. 113). Unlike party identification, approval represents a direct assessment of respondents’ attitudes toward President Bush, which, as with perceptions of honesty and knowledge, can vary among Republicans. Republicans who express disapproval of President Bush’s job performance may not consider him to be a persuasive source regardless of any interest or similarity arising from their common partisanship. In fact, Republicans who depart from their partisanship to disapprove of President Bush may be especially unreceptive to his arguments. At a minimum, cross pressures between approval and partisanship make contestable any strong assumptions about these respondents’ willingness to accept a message from President Bush. Rather, cross-pressured Republicans’ probability of seeing President Bush as a credible source should be lower than for Republicans who approve of Bush. Consequently, the probability of accepting a message from President Bush and the subsequent probability of attitude change should be smaller, consistent with Zaller’s (1992, 132) model.

The 2008 Gallup survey provides support for making a distinction between partisans whose approval of President Bush is consistent with their partisanship and those who are cross-pressured. Republican identifiers who disapproved of President Bush’s job performance were more likely (44 percent) to believe that the Administration had misled the public on Iraq than both Republicans who approved of the president’s performance (8 percent) and even Democrats who approved of President Bush (37 percent; Table 4). In addition, incorporating information about approval may help distinguish Independents who lean toward one of the parties, with 77 percent of Independents (including leaners to mirror the three-point scale in Gelpi’s study) who disapprove of Bush’s performance thought that the Administration had misled the public on Iraq, compared with only 22 percent of Independents who approved of the president’s performance. Overall, analysis of the Gallup data shows that using only approval would allow a correct prediction of Bush’s credibility for 81.8 percent of the respondents (See online appendix Table A11), compared to 76.3 percent of the public the three-point measure of party identification (See online appendix Table A4).⁴ Approval, therefore, provides at least as accurate a way of determining respondents’ perception of President Bush as a credible source on Iraq.

Table 4.

Party Identification, Approval, and Iraq Attitudes.

	Disapprove of Bush			Approve of Bush
	Democrats	Independents	Republicans	Democrats	Independents	Republicans	Total
President Bush’s credibility on Iraq
Administration misled on Iraq	86.98 (570)	76.93 (323)	44.20 (52)	37.37 (14)	22.48 (37)	7.66 (25)	56.88 (1027)
Administration did not mislead	13.02 (85)	23.07 (97)	55.80 (65)	62.63 (23)	77.52 (127)	92.34 (418)	43.12 (778)
How will Iraq be viewed?
A total failure	29.90 (197)	27.98 (116)	13.59 (17)	7.10 (3)	3.39 (6)	1.40 (6)	19.03 (345)
Mostly a failure	49.11 (324)	50.61 (210)	45.76 (56)	27.51 (11)	18.50 (32)	13.39 (53)	37.96 (688)
Mostly successful	20.59 (136)	18.92 (79)	38.54 (48)	64.08 (25)	70.12 (122)	75.42 (300)	39.20 (710)
A total success	0.41 (3)	2.48 (10)	2.11 (3)	1.31 (1)	7.99 (14)	9.79 (39)	3.81 (69)
Withdrawal of troops from Iraq
Withdraw as soon as possible	28.81 (185)	28.20 (117)	17.07 (21)	8.19 (3)	5.32 (9)	2.35 (10)	19.15 (344)
Gradual withdrawal	57.82 (371)	49.95 (206)	46.00 (57)	38.86 (16)	26.26 (45)	20.32 (83)	43.25 (777)
Maintain troops	13.37 (86)	21.84 (90)	36.94 (46)	52.96 (22)	68.42 (117)	77.33 (316)	37.59 (676)

Note: Entries are column percentages. Number of respondents in parentheses. Data from the February 2008 Gallup survey for USA Today. Data weights are used.

The idea that approval provides an acceptable proxy for credibility receives empirical support both elsewhere and within the literature concerning attitudes toward war. Mondak (1993) uses experiments to determine whether supplying a cue by naming the governor who appointed members to the California Supreme Court influenced respondents’ attitudes toward whether the judges should be retained and finds that the interaction of the cueing condition and respondents’ approval of the appointing governor (either Jerry Brown or Ronald Reagan) had a greater effect upon respondents’ attitudes than an interaction between party identification and receiving the cue. Gelpi also ran his analysis using approval instead of party identification (p. 106). Similarly, Berinsky (2007, 989) uses approval of Franklin Delano Roosevelt in examining the influence of cues on attitudes toward World War II. And while he uses the complete seven-point measure of party identification for cue-taking in the experimental part of the study, he notes, “In retrospect, I might have measured the respondent’s approval of George Bush to more directly account for the process of cue-taking” (Berinsky 2007, 980).

Approval may also be a good proxy for existing attitudes toward the war because it combines partisan attitudes with current information about both domestic and foreign political conditions (Krosnick and Kinder 1990). First, 82 percent of respondents in the Gallup survey who approved of President Bush thought that the United States would succeed in Iraq, compared to only 23 percent of those who disapproved (See online appendix Table A14). There are similar results for attitudes concerning a timetable, with 73 percent of respondents who approved of President Bush opposing a timetable, compared to 18 percent among disapprovers. Among Republicans (excluding leaners), belief in success in Iraq was 74 percent (See online appendix Table A2) and disapproval of a timetable was 67 percent (See online appendix Table A3). For Democrats and Independents (including partisan leaners), 31 percent believed that the United States would be successful and 25 percent disapproved of a timetable. In all, the three-point party identification scale allows for correct predictions of respondent’s attitudes toward the chances of success in Iraq and disapproval of a timetable in 70 percent (See online appendix Table A2) and 73 percent (See online appendix Table A3), respectively, of the observations, whereas approval correctly categorizes respondents in 79 percent and 79 percent (See online appendix Table A14) of the cases. If we want to maximize the accuracy of respondents’ initial beliefs about Iraq and their views of President Bush’s credibility, approval is a better measure than the three-point party identification scale.

Analysis Using Party Identification and Presidential Approval

Analysis indicates that both party identification and presidential approval have independent effects upon both attitudes toward Iraq and perceptions of the president’s credibility on Iraq (See online appendix Table A12–A19). If the goal is to best evaluate both respondents’ prior attitudes and perceptions of credibility, then combining their party identification and approval of the president provides a superior proxy for these unmeasured attitudes than either separately. The data in Table 4 suggest a natural ordering that runs, in increasing levels of support for the war and perceptions of President Bush’s credibility, from Democrats, then Independents, and then Republicans who disapprove of President Bush to Democrats, Independents, and then Republicans who approve of President Bush. A simple index, coded from 0 to 5, can be used for examining differences in the effects of the various treatments upon attitudes toward the war that can more accurately capture the degree to which respondents have positive or negative prior attitudes toward the war and President Bush.

When the data are reanalyzed using both the direction and the strength of the respondents’ attitudes toward the war and using the index combining partisanship and approval of President Bush as proxies for respondents’ prior attitudes toward Iraq and beliefs about Bush’s credibility, there is little evidence for the effects of surprising events (Table 5). While the effects of positive events remain statistically significant for the surge among people who strongly disapprove of President Bush (coded 0), events are no longer significantly related to these people’s attitudes about success in Iraq and their approval of a timetable.⁵ All of this indicates that surprising events have only limited influence upon attitudes toward the war. And while negative cues from President Bush do not significantly influence the attitudes concerning the surge, negative cues are associated with significantly less certainty about success and less extreme disapproval of a timetable among respondents most likely to have positive views of the war and see President Bush as a credible source.

Table 5.

Iraq War Attitudes, by Partisan-approval Index.

	Surge	Success	Timetable
Positive events	0.845** (0.302)	0.235 (0.217)	0.008 (0.207)
Negative events	0.345 (0.314)	−0.138 (0.211)	−0.142 (0.203)
Confident Bush	−0.134 (0.296)	0.265 (0.215)	0.040 (0.205)
Cautious Bush	−0.080 (0.296)	0.173 (0.215)	0.099 (0.207)
Positive events × index	−0.079 (0.111)	0.011 (0.074)	0.118 (0.075)
Negative events × index	−0.170 (0.106)	0.032 (0.073)	0.059 (0.074)
Confident Bush × index	0.055 (0.109)	−0.084 (0.075)	−0.018 (0.078)
Cautious Bush × index	−0.045 (0.102)	−0.126* (0.072)	−0.144* (0.074)
Partisan-approval index	0.812** (0.096)	0.871** (0.070)	0.580** (0.068)
τ₁	−2.106** (0.279)	−0.423* (0.192)	−0.484** (0.187)
τ₂		1.879** (0.203)	1.074** (0.189)
τ₃		4.231** (0.245)	2.185** (0.201)
N	997	996	994
χ²	407.7	571.3	355.9

Note: Standard errors in parentheses.

*p < .05, one-tailed test.

**p < .01, one-tailed test.

Comparing the predicted probabilities of subjects’ responses based upon each treatment shows that negative cues from a credible source also have a substantial effect upon people’s responses. Using the estimates from Table 5 to calculate predicted probabilities shows that Republicans who approved of President Bush, but received a cautious statement, had a predicted probability of responding that success in Iraq was very likely that was .11 lower than those not receiving any treatment. The corresponding reduction in the probability of “strongly” disapproving of a timetable for Iraq was even larger, .15. While the predicted probability of saying that success in Iraq was not very likely for approving Republicans was only slightly greater for those in the negative cue treatment than those who did not receive any treatment, there was an increase in the predicted probability that an approving Republican would also support a timetable, .23, compared to a similar Republican who did not receive a treatment, .14. The effect of a negative cue upon approving Republicans’ beliefs about ultimate success in Iraq was only a weakening of their certainty that the mission would succeed, whereas a negative cue both increased approval of a timetable and reduced the extremity of opponents’ attitudes toward a timetable.

We can also see that the predicted probabilities for respondents of any party who disapprove of President Bush are not affected by any of the treatments. This supports the hypothesis that respondents who do not see President Bush as a credible source of information on Iraq would not be influenced by any statement he would make, even one that countered expectations. Similarly, given a general negative perception of the war, more negative event information does not provide any reinforcement of those attitudes. Finally, contrary to the surprising events, positive events did not have any effect upon the predicted probability of increased support for the Iraq War.

Some readers may believe that party identification is the only appropriate proxy for determining respondents’ prior attitudes and their receptivity to a message from President Bush. Limiting analysis only to Republicans who approve of the president—those most likely to see him as a credible source—and comparing the independent effects of the four treatments shows that a negative statement from President Bush still reduced the opposition to a timetable from Republicans who approved of President Bush’s performance. There is also evidence for the surprising opinions hypothesis among Independents who approved of President Bush concerning their attitudes about the likelihood of success in Iraq, although it should be noted that the effect of a negative message from President Bush was not significantly different than the corresponding effect for Independents who disapproved of President Bush’s performance. We also see that positive event information increased their opposition to a timetable, supporting the reinforcing events hypothesis, although it should be noted that positive event information had similar effects upon all Independents’ and Republican disapprovers’ opposition to a timetable, so it is quite possible that the effect of positive information merely suggested to these respondents that positive developments in the war should provide troops with more time to complete their mission.

Discussion

This article shows that attitude strength is an important concept for understanding people’s attitudes toward the Iraq War. When attitude strength is considered, support for the surprising events hypothesis is limited. The analysis highlights the importance of understanding that people’s attitudes can consist of more than just the direction of the attitude when doing public opinion research. The implications of this have been shown previously in social psychology and even political science research on attitudes toward war. This article provides another example that attitude strength should generally be considered and may even be necessary for testing some hypotheses.

In addition, this article shows that determining how people view a source’s credibility is important for evaluating a source’s persuasiveness. There is consistent support for the influence of negative messages from President Bush among respondents most likely to see him as credible once a more varied measure of credibility is employed, regardless of how the models are specified. Partisanship undoubtedly influences people’s willingness to accept a message from that speaker and evaluation of traits associated with credibility, but when people are evaluating a single, well-known speaker, other perceptions of the speaker’s personal attributes and message sharpen our ability to estimate the effect of credibility upon attitudes.

Regarding the specific question of the cost/benefit versus cues explanations of public opinion on war, this analysis also shows that, placed head-to-head in an experiment, the effects of cues perform at least as well the influence of events. None of this should be interpreted to mean that cues always matter and events never do. It is worth considering that these experiments were conducted at the fifth year of the war. It would be remarkable if an article with an account of changing events would greatly influence people’s attitudes. The differences between partisans in their attitudes toward Iraq reflects the accumulated influence of partisan cues from the beginning of the war. Moreover, it is also not surprising that effects of events would be limited mostly to a short-term assessment of how well a particular military strategy, the surge, was performing, without having much effect upon attitudes about either the likelihood of success or policies concerning troop withdrawal, where the longer-term implications may be less clear.

In this vein, these results provide a valuable addition to our understanding of the factors that influence the public’s attitudes toward war. One clear implication of this is that when a leader who supports a war provides a surprising opinion, the main effect is to reduce the certainty and extremity of supporters’ positive attitudes toward the war. This may help us understand why leaders have incentives to remain relentlessly positive, even in the face of discouraging news from the battlefield, because negative information may weaken attitudes that make constituents more likely to move their attitudes from support to opposition or, at the very least, make them less willing to actively show support on the leader’s behalf.

Finally, is worthwhile considering how different factors could influence the public’s attitudes about the war differently over the course of a long war. One finding from the research on attitude strength is that stronger attitudes influence how new information is processed. Baum and Groeling (2010) suggest that patterns emerge where elite influence is felt first, followed by events, and then a stable period where attitudes change little until there is a sustained change in the events on the ground, at which time elite influence reemerges as the dominant influence over attitudes. An implication of the findings about attitude strength is that when elite attitudes polarize opinions on the war, modest changes the direction of the war that are not reflected in updated elite opinions are unlikely to influence public opinion because new developments that support strong attitudes will merely bolster existing opinion, whereas those that run contrary to strong attitudes will be discounted. Nyhan and Reifler’s (2010) research using information about the absence of weapons of mass destruction in Iraq illustrates how hard it can be for new information to correct misperceptions, at least among individuals who attach strong importance to an issue. In the midst of a war when information about the war’s progress may be quite difficult to determine, it should not be surprising that new information about events would have little influence, especially among people with strong attitudes.

Footnotes

Author’s Note

All information necessary for replicating this study can be found at the Journal of Conflict Resolution website as well as the author’s dataverse () page.

Acknowledgments

The author would like to thank Adam Berinsky, Jon Krosnick, David A.M. Peterson, two anonymous reviewers and the editor for their comments on earlier versions of the article. The author would also like to thank Christopher Gelpi for the information he provided about his study.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Supplementary Material

Supplementary material for this article is available online.

Notes

References

Aldrich

John H.

Sullivan

John L.

Borgida

Eugene

. 1989. “Foreign Affairs and Issue Voting: Do Presidential Candidates ‘Waltz Before a Blind Audience’?” American Political Science Review 83 (1): 123–42.

Althaus

Scott L.

Bramlett

Brittany H.

Gimpel

James G.

. 2012. “When War Hits Home: The Geography of Military Losses and Support for War in Time and Space.” Journal of Conflict Resolution 56 (3): 382–412.

Baum

Matthew A.

Groeling

Tim

. 2010. “Reality Asserts Itself: Public Opinion on Iraq and the Elasticity of Reality.” International Organization 64 (3): 443–79.

Berinsky

Adam J.

2007. “Assuming the Costs of War: Events, Elites, and American Public Support for Military Conflict.” The Journal of Politics 69 (4): 975–97.

Berinsky

Adam J.

2009. In Time of War: Understanding American Public Opinion from World War II to Iraq. Chicago: University of Chicago Press.

Boettcher

William A.

Cobb

Michael D.

. 2009. “‘Don’t Let Them Die in Vain’: Casualty Frames and Public Tolerance for Escalating Commitment in Iraq.” Journal of Conflict Resolution 53 (5): 677–97.

Campbell

Angus

Converse

Philip E.

Miller

Warren E.

Stokes

Donald E.

. 1960. The American Voter. Chicago: University of Chicago Press.

CBS News, 60 Minutes, Vanity Fair. 2013. CBS News/60 Minutes/Vanity Fair National Poll, March #1, 2013 ICPSR34995-v1 [computer file]. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor].

Clarke

Harold

Sanders

David

Stewart

Marianne

Whiteley

Paul

. 2006. “Taking the Bloom Off New Labours Rose: Party Choice and Voter Turnout in Britain, 2005.” Journal of Elections, Public Opinion and Parties 16 (1): 3–36.

10.

Eagly

Alice H.

Wood

Wendy

Chaiken

Shelly

. 1978. “Causal Inferences about Communicators and Their Effect on Opinion Change.” Journal of Personality and Social Psychology 36 (4): 424.

11.

Fiorina

Morris P.

1981. Retrospective Voting in American National Elections. New Haven, CT: Yale University Press.

12.

Foyle

Douglas C.

2004. “Leading the Public to War? The Influence of American Public Opinion on the Bush Administration’s Decision to go to War in Iraq.” International Journal of Public Opinion Research 16 (3): 269–94.

13.

Gallup Organization. 2008. USA Today/Gallup Poll # 2008-11: Presidential Election 2008/War in Iraq/Steroid Use in Baseball/Pet Food Recall/Relations with Cuba [computer file]. Storrs, CT: The Roper Center, University of Connecticut [distributor].

14.

Gartner

Scott Sigmund

. 2008. “The Multiple Effects of Casualties on Public Support for War: An Experimental Approach.” American Political Science Review 102 (1): 95–106.

15.

Gelpi

Christopher

. 2010. “Performing on Cue? The Formation of Public Opinion toward War.” Journal of Conflict Resolution 54 (1): 88–116.

16.

Gelpi

Christopher

Feaver

Peter D.

Reifler

Jason

. 2009. Paying the Human Costs of War: American Public Opinion & Casualties in Military Conflicts. Princeton, NJ: Princeton University Press.

17.

Gelpi

Christopher

Reifler

Jason

Feaver

Peter

. 2007. “Iraq the Vote: Retrospective and Prospective Foreign Policy Judgments on Candidate Choice and Casualty Tolerance.” Political Behavior 29 (2): 151–74.

18.

Grose

Christian R.

Oppenheimer

Bruce I.

. 2007. “The Iraq War, Partisanship, and Candidate Attributes: Variation in Partisan Swing in the 2006 US House Elections.” Legislative Studies Quarterly 32 (4): 531–57.

19.

Gross

Sharon Ruth

Holtz

Rolf

Miller

Norman

. 1995. “Attitude Certainty.” In Attitude Strength: Antecedents and Consequences, edited by Krosnick

Jon A.

Petty

Richard E.

, 215–45. Mahwah, NJ: Lawrence Erlbaum Associates.

20.

Holsti

Ole R.

1996. Public Opinion and American Foreign Policy. Ann Arbor: The University of Michigan Press.

21.

Jacobs

Lawrence R.

Page

Benjamin I.

. 2005. “Who Influences U. S. Foreign Policy?” American Political Science Review 99 (1): 107–23.

22.

Jacobson

Gary C.

2007. “Referendum: The 2006 Midterm Congressional Elections.” Political Science Quarterly 122 (1): 1–24.

23.

Karol

David

Miguel

Edward

. 2007. “The Electoral Cost of War: Iraq Casualties and the 2004 U.S. Presidential Election.” The Journal of Politics 69 (3): 633–48.

24.

Keith

Bruce E.

Magelby

David B.

Nelson

Candice J.

Orr

Elizabeth

Westlye

Mark C.

Wolfinger

Raymond E.

. 1992. The Myth of the Independent Voter. Berkeley: University of California Press.

25.

Krosnick

Jon A.

Boninger

David S.

Chuang

Yao C.

Berent

Matthew K.

Carnot

Catherine G.

. 1993. “Attitude Strength: One Construct or Many Related Constructs?” Journal of Personality and Social Psychology 65 (6): 1132–51.

26.

Krosnick

Jon A.

Kinder

Donald R

. 1990. “Altering the Foundations of Support for the President through Priming.” The American Political Science Review 84 (2): 497–512.

27.

Krosnick

Jon A.

Petty

Richard E.

. 1995. “Attitude Strength: An Overview.” In Attitude Strength: Antecedents and Consequences, edited by Krosnick

Jon A.

Petty

Richard E.

, 1–24. Mahwah, NJ: Lawrence Erlbaum Associates.

28.

Larson

Eric V.

1996. Casualties and Consensus: The Historical Role of Casualties in Domestic Support for U.S. Military Operations. Santa Monica, CA: Rand.

29.

Layman

Geoffrey C.

Carsey

Thomas M.

Horowitz

Juliana Menasce

. 2006. “Party Polarization in American Politics: Characteristics Causes, and Consequences.” Annual Review of Political Science 9 (1): 83–110.

30.

Lupia

Arthur

McCubbins

Mathew D.

. 1998. The Democratic Dilemma: Can Citizens Learn What They Need to Know? Cambridge, UK: Cambridge University Press.

31.

Miller

Joanne M.

Peterson

David A. M.

. 2004. “Theoretical and Empirical Implications of Attitude Strength.” Journal of Politics 66 (3): 847–67.

32.

Mondak

Jeffery J.

1993. “Public Opinion and Heuristic Processing of Source Cues.” Political Behavior 15 (2): 167–92.

33.

Mueller

John E.

1971. “Trends in Popular Support for the Wars in Korea and Vietnam.” American Political Science Review 65 (2): 358–75.

34.

Nyhan

Brendan

Reifler

Jason

. 2010. “When Corrections Fail: The Persistence of Political Beliefs.” Political Behavior 32 (2): 303–30.

35.

Opinion Research Corporation. 2013. Cable News Network. CNN/ORC Poll: Politics/Federal Spending Cuts/International Issues, Mar. 2013 [dataset]. Storrs, CT: Roper Center for Public Opinion Research [distributor].

36.

Peterson

David A. M.

2004. “Certainty or Accessibility: Attitude Strength in Candidate Evaluations.” American Journal of Political Science 48 (3): 513–20.

37.

Petty

Richard E.

Cacioppo

John T.

. 1981. Attitudes and Persuasion: Classic and Contemporary Approaches. Dubuque, IA: Wm. C. Brown Company Publishers.

38.

Petty

Richard E.

Fleming

Monique A.

Priester

Joseph R.

Feinstein

Amy Harasty

. 2001. “Individual versus Group Interest Violation: Surprise as a Determinant of Argument Scrutiny and Persuasion.” Social Cognition 19 (4): 418–42.

39.

Pornpitakpan

Chanthika

. 2004. “The Persuasiveness of Source Credibility: A Critical Review of Five Decades’ Evidence.” Journal of Applied Social Psychology 34 (2): 243–81.

40.

Priester

Joseph R.

Petty

Richard E.

. 1995. “Source Attributions and Persuasion: Perceived Honesty as a Determinant of Message Scrutiny.” Personality and Social Psychology Bulletin 21 (6): 637–54.

41.

Taber

Charles S.

Lodge

Milton

. 2006. “Motivated Skepticism in the Evaluation of Political Beliefs.” American Journal of Political Science 50 (3): 755–69.

42.

Verba

Sidney

Brody

Richard A.

. 1970. “Participation, Policy Preferences, and the War in Vietnam.” The Public Opinion Quarterly 34 (3): 325–32.

43.

Visser

Penny S.

Bizer

George Y.

Krosnick

Jon A.

. 2006. “Exploring the Latent Structure of Strength-related Attitude Attributes.” Advances in Experimental Social Psychology 38:1–67.

44.

Zaller

John R.

1992. The Nature and Origins of Mass Opinion. New York: Cambridge University Press.