March 14, 2006 (4:30pm) / 19
These results speak to the separate effects of the treatment on Democrats and Republicans.
But what of the hypothesis that the treatment exerted a different effect on the two groups—that
partisanship moderated belief perseverance? Some evidence in favor of the hypothesis is
provided by the third column of Table 1, which shows that the difference between the treatment
effects on the logit scale is nearly four times its standard error (p < .001). The difference is large,
too: evidence of its size can be seen in the bottom panels of Figure 1, which show that the gap in
disapproval rates between Democrats and Republicans grew from 20% in the control condition to
60% in the treatment condition.
Additional perseverance can be found in subjects’ attitudes toward Congressional
hearings about the abuse of detainees. In the control group, 69% of subjects called for hearings.
In the treatment group, 80% did. The finding was significant overall (p = .02) and for Democrats
(control 78%, treatment 86%, p = .09) but not for Republicans (control 65%, treatment 69%,
p
= .35).
In summary, both hypotheses were borne out. Even after learning that Newsweek
disavowed its own claim, professing to understand that disavowal and seeming to accept it,
subjects who were exposed to the claim continued to be affected by it. And the extent and
direction in which they were affected depended on their party identification.
But consider two important details. The information provided to all subjects at the
beginning of the experiment, although quite minimal, may have caused treated subjects to attend
more to the claim about the Qur’an than they otherwise would have. (Of course, most articles
about the controversy contained at least as much background information about the Qur’an’s
place in Islam. See, e.g., Kurtz 2005, Seelye and Lewis 2005.) Moreover, Newsweek’s retraction
is just that: it retracts the claim that a military investigation had uncovered Qur’an abuse. It
certainly does not say that the allegation was untrue. And although it is a clear and total retraction
of the information originally provided, it is not inconceivable that subjects might have viewed it
as ambivalent. This is especially true in light of widely publicized criticisms of Newsweek from
the Pentagon and the White House. If subjects were aware of those criticisms, they might have
Convention