Page 117 - Media Effects Advances in Theory and Research

106 D. ROSKOS-EWOLDSEN, B. ROSKOS-EWOLDSEN, F. DILLMAN CARPENTIER

War), which were of the longest duration (highest intensity), were signifi-
cantly smaller than the priming effects from shorter-duration (less-
intense) media primes. However, these apparently contradictory results
are confounded by the lag between the priming event and the measure-
ment of the prime’s effect. The lag between the priming event and the
measure of the prime’s effect was substantially longer in the campaign
studies than in the other media priming studies.
Despite the lack of direct evidence for the two characteristics of prim-
ing, the meta-analysis has provided evidence that media priming has
these two characteristics. Therefore, for models of media priming to be
adequate, they must incorporate the two characteristics of priming. Of
course, they must also be able to explain existing media priming results.
For example, political priming effects last considerably longer than the
typical priming effects found in psychological experiments (Iyengar &
Simon, 1993; Krosnick & Brannon, 1993; Pan & Kosicki, 1997; Roskos-
Ewoldsen et al., in press). In the next section, we discuss the current mod-
els in each domain, with a focus on their ability to incorporate the two
characteristics of media priming and to explain the media priming results.

Models of Media Violence Priming

One of the most prominent explanations of the consequences of media
violence is Berkowitz’s (1984, 1990, 1994, 1997) neo-associationistic model.
Berkowitz’s model draws heavily from network models of priming. The
model hypothesizes that depictions of violence in the media activate
hostility- and aggression-related concepts in memory. The activation of
these concepts in memory increases the likelihood that a person will
engage in aggressive behaviors and that others’ behavior will be inter-
preted as aggressive or hostile. Without further activation, however, the
activation levels of these hostile and aggressive concepts, and their associ-
ated likelihood of influencing aggressive behavior, fades with time.
Anderson, Deuser, and DeNeve (1995) proposed as an extension of
Berkowitz’s (1984) neo-associationistic model the affective aggression
model. This model incorporates affect and arousal into the network frame-
work and introduces a three-stage process by which situations influence
aggressive behavior and affect. In the first stage, situational variables such
as pain, frustration, or depictions of violence prime aggressive cognitions
(e.g., hostile thoughts and memories) and affect (e.g., hostility, anger),
which results in increased arousal. In the second stage, the primed cogni-
tions and affect, in conjunction with the increased arousal, influence pri-
mary appraisal. Primary appraisal involves the automatic interpretation of
the situation (Fazio & Williams, 1986; Houston & Fazio, 1989) and of one’s
arousal in that situation (Fazio, Zanna, & Cooper, 1979; Schachter & Singer,