Mass Deception: Moral Panic and the U.S. War on Iraq

4. How the Bush Administration Sold the Iraq War to the U.S. Public

quotes by the president in the month prior to the polls increases the log odds of supporting the invasion by .16. The word “evil” is not significant in model 2. In model 3, quotes from president Bush in three-month lags is not significant. However, the key word “evil” is positive and significant in model 3, controlling for quotes from President Bush. A one-unit increase in articles containing M a s s D e c e p t i o n 74 Table 10 Multivariate Logistic Regression Results for “Favor Invasion of Iraq”a : Number of Articles with Quotes from Bush and “Evil” Variable Model 1 Model 2 Model 3 Model 4 Bush 2-wk. Lag .310* – – – Odds Ratio 1.363 (.163) 1-mo. Lag – .156* – – Odds Ratio 1.169 (.090) 3-mo. Lag – – .000 – Odds Ratio 1.000 (.021) 5 mo. Lag – – – .016 Odds Ratio 1.016 (.011) “Evil” 2-wk. Lag .424 – – – Odds Ratio 1.528 (.297) 1-mo. Lag – .121 – – Odds Ratio 1.128 (.167) 3-mo. Lag – – .317* – Odds Ratio 1.374 (.191) 5-mo. Lag – – – ⫺.014 Odds Ratio .986 (.041) Constant ⫺4.006* ⫺3.902 ⫺6.634* ⫺.005 (2.197) (2.570) (4.007) (1.497) Wald χ2 13.152** 10.034** 10.366** 2.430 (df) (3) (3) (3) (3) log-likelihood 11.412 14.530 14.198 22.134 Note: N⫽24. Standard errors in parentheses. a Same or higher than previous poll⫽1. *p ⬍ .10 **p ⬍ .05 the word “evil” in the three months preceding the polls increases the log odds of support for the invasion by .32. Neither of the coefficient estimates in model 4 (five-month lag) is statistically significant. In summary, a series of logistic regression analyses were conducted to examine the relationships between public support for invading Iraq and time, volume, and tone. The analyses for whether “support of invasion” of Iraq changed on volume and tone of rhetoric took into account the amount of rhetoric in specified time intervals prior to when the Gallup polls were conducted . The results revealed that punitive rhetoric increased the likelihood that public support for the invasion remained stable or increased (in three out of four time lags), as expected. Communitarian rhetoric increased the probability of supporting the invasion, as expected. These findings are highly consistent with moral panic processes which state that moral panics can begin when punitive and communitarian rhetoric are used together by the president (Hawdon, 2001). Also, greater numbers of articles with the key word “evil” increased the odds of support for the invasion in three out of four time lags, while the word “threat” was positive and significant in one time lag. Interestingly, the word “terrorism” or “terror” had no influence on support for invading Iraq. As expected, source of rhetoric was very important for public support of invading Iraq. Aggregated administration sources (not shown here) and President Bush each independently increased the odds of support for invasion in three out of four time lags preceding Gallup polls, as anticipated. Quotes from Vice President Dick Cheney were positive and significant in one time lag, while articles with quotes from Defense Secretary Donald Rumsfeld and National Security Advisor Condoleezza Rice (not shown here) were positive and significant in two time lags each. As expected, nonadministration sources had negligible impacts on public opinion. For example, the volume of articles with quotes from Democratic policy makers (not shown here) was unrelated to support for invasion, despite the fact that the volume of rhetoric from this group significantly increased post- versus pre-9/11. Similarly, although frequently quoted in the news media, articles with statements from U.N. Chief Weapons Inspector Hans Blix (not shown here) were not related to support for the invasion of Iraq. When other sources or tones of rhetoric were added as independent variables in models along with President G. W. Bush, the number of articles with quotes from Bush continued to have important impact on public support for invasion. That is, a greater number of articles with Bush quotes increased the probability of support in two out of four time-lagged models, controlling for each type (tone) of rhetoric in the study, including counterclaims, while the tones of rhetoric were not related to support in any of the models analyzed Empirical Evidence of an Elite-Engineered Moral Panic 75 (results not shown...