3
the surprise of many, because his campaign shifted the nation’s focus from international issues back to New
Deal issues, where he and the Democratic Party had an advantage (Berelson et al. 1954).
Despite this apparent consensus, there are a number of reasons to remain skeptical about the
evidence for priming. Some conclusions from these studies contradict other findings in political science.
Priming findings often imply that the public places a good deal of weight on political issues when making
voting decisions, but research that explicitly attempts to measure issue voting has failed to produce a
consensus about its extent (cf., Brody and Page 1972; Margolis 1977; Markus and Converse 1979). Even if
one believes that the existing evidence supports some issue voting, the high level often implied by priming
studies seems inconsistent with a large body of research suggesting that voters lack the necessary interest in
and knowledge of policy issues (Campbell et al. 1960; Delli Carpini and Keeter 1996). Furthermore, for
individuals to give greater weight to an opinion, they must hold one. Research suggests that much of the
public fails to hold strong opinions about many political issues (Converse 1964; Zaller 1992), a finding that
seems hard to square with the evidence for priming.
There are also methodological reasons for skepticism about the evidence for priming. Two problems
in particular raise suspicions about the conclusions of priming studies. First, the “treatments” in these studies
are very general: whether in lab experiments or survey studies, they consist of viewing television news,
watching campaign advertisements, reading newspaper articles, etc. Consequently, we do not know what
aspect of the treatments causes the observed effects. While scholars have assumed that the effects arise from
salience and thus reflect priming,
4
other aspects could be responsible. Second, priming studies generally
measure issue attitudes after the treatments. Since the treatments may change subjects’ issue attitudes,
conclusions drawn from these studies are vulnerable to alternative explanations.
In light of these problems, I review alternative explanations for priming findings, focusing on the
possibility that the treatments convey information that lead individuals to change their votes or issue opinions
in a manner that creates the appearance of priming. I then propose a simple approach to test priming against
learning-based alternatives. Using this method, I analyze four cases and show that—instead of causing
priming—campaign and media attention to an issue informs some individuals about aspects of that issue,
causing them to adopt attitudes consistent with their support for a party.
Alternative explanations for priming findings
A disparity exists between the priming hypothesis and tests of this hypothesis. Priming describes
what might be called issue-driven vote change: campaign or media attention increases the salience of an
4
I use the term salience to refer to the extent to which an issue is “in the news” or emphasized by
campaigns. Researchers have presented conflicting evidence on whether salience, using my definition, leads
to priming through accessibility or other mechanisms (e.g., Miller and Krosnick 2000; Nelson et al. 1997;
Valentino et al. 2002).
Convention