Page 93 - Media Effects Advances in Theory and Research

82 SHRUM

Moreover, the judged applicability of the construct is a function of the
overlap between its attended features and the features of the judgment.
In terms of the cultivation effect, the recalled construct would presum-
ably be a television example. However, it is counterintuitive that people
would perceive a television example (e.g., doctor, lawyer) as applicable to
a judgment about its real-world prevalence. If they do not perceive the
example as relevant, alternative information would be retrieved and used
as a basis for judgment (Higgins, 1996; Higgins & Brendl, 1995; Shapiro &
Lang, 1991).
One way in which a television example could be perceived as relevant to
a real-world judgment is if people generally do not consider the source of
the example they retrieve in the course of judgment construction. Note that
perceived applicability is a function of the overlap between the attended fea-
tures of the recalled construct and the features of the judgment. It may well
be that the source characteristics of the retrieved construct may not be a
salient feature that is attended to, particularly when judgments are made
automatically and with little effort. This may be a function of either lack of
motivation to attend to source features (consistent with low involvement
processing; Petty & Cacioppo, 1986, 1990) or lack of ability to recall source
information (consistent with research on errors in source monitoring; John-
son, Hashtroudi, & Lindsay, 1993; Mares, 1996; Shrum, 1997).
To test Proposition 3, Shrum, Wyer, and O’Guinn (1998) conducted two
experiments in which source characteristics were primed prior to judg-
ments. In the first experiment, the priming events consisted of a source-
priming condition, in which participants provided information regarding
their television viewing habits prior to providing prevalence and likeli-
hood judgments of crime and occupations, and a relation-priming condi-
tion, in which participants were told that the constructs they would be
estimating appeared more often on television than in real life. In a third,
no-priming condition, participants provided their estimates prior to pro-
viding television viewing information. Analyses revealed that when par-
ticipants provided estimates under no-priming conditions, a cultivation
effect was noted, but when they provided estimates under either source-
or relation-priming conditions, the cultivation effect was eliminated.
Follow-up analyses indicated that the estimates of light viewers did not
differ as a function of priming conditions, but the priming conditions
served to bring the estimates of heavy viewers more in line with those of
3
light viewers. This pattern of results can be seen in Fig. 4.1. A second
study replicated this pattern of results, and further suggested that the

3 The graph shown in Fig. 4.1 is merely a general representation of the effects across depen-
dent variables and is for illustration purposes only. For details of the actual effects for each
dependent variable, see Shrum et al. (1998).