The acceptability of arguments in favour of and against the Iraq War

This paper reports on research based on an experimental questionnaire designed using the logic and the basic structure of the Moral Judgment Test (MJT) by Lind (1982). The study aims at understanding how respondents' acceptance of political arguments may have contributed to the lack of international support for the Iraq war and/or to the failure to prevent a preemptive attack. It examines factors, which respondents may have relied on in their acceptance of the arguments. The instrument consisted of three components: (1) A brief vignette on the declared beginning (March 19, 2003) and ending (May 1, 2003) of major operations in Iraq. (2) Respondents' own agreement with the need for the war. (3) Twelve arguments from political speeches: Six arguments by Bush and Blair "in favor of" (pro,) and six arguments by Chirac and Schroeder "against" (con) the need for the Iraq war. We scored each argument using Kohlberg's (1984) stage theory and Commons' Hierarchical Complexity Scoring System (HCSS) (Commons et al., 2004), which identifies hierarchical features in the structure. Using Kempf's (2003) Cognitive Escalation and De-Escalation Model (CEDM) we scored each argument sentence for details. Our sample consists of (N = 397) respondents of which were 56.4% women, 24.2% men, and 19.4% did not specify. The largest groups consisted of 71% Germans, 37.2% Catholics, 74.3% between the ages of 18-20 with high school education. Another group consisted of 8.6% Americans with a broader spectrum in age and higher educational level. Using 7 point Likert scales, respondents rated whether they would have rejected or accepted the main point of each argument, whether the main point of the argument seemed illusory or realistic, and whether the emotion evoked by the argument was destructive or constructive. The analysis of the data aimed at determining which within subject and between subject design factors were relevant in predicting respondents' acceptance of arguments and the probability that these factors were significant and valid. We performed tests for Covariance Parameter Estimates for Hierarchical Linear Models proposed by Bryk & Raudenbush (1992) with estimation of Maximum Likelihood (ML), by applying the equation (1 - model covariance/intercept covariance). Applying the standard criteria by Cohen & Cohen (1983), we consider a model as relevant, if it accounts for more than 10% of the proportion reduction in error (PRE). Mixed linear model tests identified relevant predictors with chi square significance of p < .001. Design factors, which were relevant included: "Was the argument realistic? Did it evoke constructive feelings? Did argument details offer beneficent considerations?"