Friday, 25 July 2003

Who thinks Saddam was involved in 9/11?

One of the more bizarre questions revolving around the Iraq war is that there is a large proportion of the American public who believe that Saddam Hussein was involved personally in the 9/11 attacks on the World Trade Center and Pentagon and the foiled attack on the White House.* This belief persists despite there being no evidence of a direct link, no statements by any credible source that there is a direct link, and repeated refutations of a direct link. (Many leftists want to pin this belief on the Bush administration, but I don’t think the charge sticks without showing that Bush et al. deliberately fostered this belief; there’s simply no evidence of that.)

Being a good empirical social scientist, I was curious about who would believe this assertion. Again, I used the second March 2003 CBS/New York Times poll (conducted March 5-7, 2003). The poll doesn’t have much useful data for testing any psychological theories, but a sociological model seemed to work fairly well. I produced both maximum-likelihood (ML) and MCMC estimates; since the ML estimates were basically identical to the MCMC estimates (the missing data problem was less acute in this model), for ease of interpretation I stuck with them. Here are the probit results (I’m too lazy to build a proper table from the R output; sue me):

            Estimate Std. Error z value Pr(>|z|)
(Intercept) -0.01797    0.26186  -0.069 0.945277
pid         -0.09596    0.04594  -2.089 0.036699 *
college     -0.27348    0.14702  -1.860 0.062858 .
male        -0.62078    0.14587  -4.256 2.08e-05 ***
black        0.17079    0.17603   0.970 0.331934
catholic     0.05115    0.11446   0.447 0.654921
jewish      -1.18340    0.49424  -2.394 0.016649 *
atheist     -0.41736    0.14938  -2.794 0.005208 **
haskids     -0.04056    0.10507  -0.386 0.699462
agecat       0.01696    0.05475   0.310 0.756710
libcon       0.25993    0.07314   3.554 0.000379 ***
pid:college -0.10990    0.05790  -1.898 0.057699 .
pid:male     0.12066    0.05629   2.144 0.032059 *
---
Signif. codes:  0 `***' 0.001 `**' 0.01 `*' 0.05 `.' 0.1 ` ' 1
N: 782
Percent correctly classified: 64.83%
Proportional reduction in Error: 27.25%
McKelvey/Zavonia Pseudo-R^2: 0.214

Jewish and atheist voters are significantly less likely than Protestant voters (the omitted reference category) to believe Saddam was personally involved in 9/11, while there is no difference between Catholics and Protestants. Conservatives are significantly more likely to believe in the Saddam-9/11 link than liberals.

The other significant effects are expressed in interactions between multiple variables. I estimated interactions between gender (male) and party identification (pid) and between level of education (college) and party identification. These effects are shown in this graph. The horizontal axis is the respondent’s party identification, where 0 is “strong Republican”, 1 is “independent leaning Republican”, 2 is “true independent”, 3 is “independent leaning Democratic”, and 4 is “strong Democrat.” (Age is set to the mean value; other variables are set to the modal category.)

Males in general, particularly male Republicans, are much less likely than females to believe the Saddam-9/11 link, regardless of education level. However, among Democrats, the primary difference is between the college educated and the less-well educated, with the gender difference being relatively small.

What does this mean? There are a few possibilities. The most compelling one is that people who don’t know are guessing, drawing on some vague association between Saddam Hussein and radical Islam. The demographic variables may be indicators of attentiveness to the media; those who pay more attention to the media may have a more nuanced understanding of Middle Eastern politics. The partisan effects suggest that some voters may be projecting their own belief systems onto the question; strong Republicans may be projecting hawkish attitudes onto questions about Saddam, while strong Democrats may be projecting a belief that Saddam isn’t a threat onto him, at least among the better-educated.

More generally, the results suggest that trying to argue Saddam wasn’t linked to terrorism may be a losing strategy among their own base for Democratic presidential candidates that opposed the war or are having second thoughts now. Strongly Democratic voters without a college education are more likely than not to believe that Saddam was involved in 9/11, and it will be difficult to reeducate them on this point. These findings suggest that however candidates like Bob Graham and Howard Dean try to spin things, many Democratic voters think Saddam Hussein was a legitimate target in the war on terror, and they will cross these voters at their own peril.

Then again, maybe all these people think Saddam was involved in 9/11 because his regime actually was, at least to some degree.

* One could quibble over the definition of “personally involved.” I personally believe it is possible—perhaps even probable—that Saddam Hussein’s government, or elements of it, provided limited logistical and materiel support to Al Qaeda. There is evidence that the Ba‘athists had aided other terrorist groups with professed Islamist sentiments (e.g. the Abu Nidal organization and a number of Palestinian-based terrorist groups), so it is hardly beyond the realm of possibility that Saddam would also provide aid to Al Qaeda. However, I don’t believe Hussein’s regime or any elements of it were directly involved in the 9/11 operation.