Moderators of Priming Effects: A Theory and Preliminary Evidence from an Experiment on Swiss European Policy
| Author | Lionel Marquis |
| Published date | 01 March 2007 |
| Date | 01 March 2007 |
| DOI | http://doi.org/10.1177/0192512107075405 |
| Subject Matter | Articles |
Marquis: Moderators of Priming Effects 185
International Political Science Review (2007), Vol. 28, No. 2, 185–224
DOI: 10.1177/0192512107075405 © 2007 International Political Science Association
Sage Publications (Los Angeles, London, New Delhi, and Singapore)
Moderators of Priming Effects:
A Theory and Preliminary Evidence from
an Experiment on Swiss European Policy
Lionel Marquis
Abstract. This article proposes an extended model of how “media priming
effects” come about, distinguishing between several mediators and
moderators of priming. The model is tested using data from an experiment
on Swiss foreign policy in which undergraduate students were provided
with different types of biased information about the relationships between
Switzerland and the European Union. The empirical analysis suggests that
cross-sectional and cross-temporal effects are facilitated by quite differ-
ent moderators, and that very few variables moderate priming according
to both perspectives. In that respect, political knowledge, the level of recall
of the experimental message, and awareness of the persuasive intent of
the message appear central to the priming mechanism.
Keywords: • Experiment • Foreign policy • Moderators • Priming effects
• Switzerland
Introduction
The study of political information and of its effects on voters has been on the
scholarly agenda for several decades now. However, especially since the 1970s, there
has been an accelerating development of research methods in communication
science and a striking expansion in the scope and variety of mass media effects
which have progressively been taken into account (see Kinder, 1998). In this
article, I shall deal with just one aspect of political information, namely, the so-
called priming effects of the mass media. The priming theory argues that the media
provide voters with the “issues of the day,” and thus infl uence the criteria by which
they will judge the performance, personality, capacities, and other attributes of
candidates. In other words, “by calling attention to some matters while ignoring
others, [political news] infl uences the standards by which governments, presidents,
and candidates for public offi ce are judged” (Iyengar and Kinder, 1987: 63). Any
type of information can, under proper circumstances, permeate the considerations
186 International Political Science Review 28(2)
that citizens bring to bear on their attitudes and, sometimes, on their behavior.
Likewise, although the priming theory was developed primarily to study political
attitudes and behavior in the context of elections, it can be equally applied to the
context of direct democratic issues (for example, De Vreese, 2004).
My task in this article will be to explore whether experimental stimuli have the
potency to prime evaluations of the possibility of Switzerland joining the European
Union (EU). More broadly, my purpose is threefold. First, by investigating the
basic priming hypothesis using experimental data, I wish to perform a stringent
test of some central, but largely unexamined, assumptions of the priming theory.
For example, does the assumed “hydraulic pattern” of priming (whereby stressed
issues gain prominence to the detriment of ignored ones) hold when more than
two issues are considered? A second purpose of my study is to determine whether
priming is a mediated process. More specifi cally, I shall attempt to determine
which individual-level variables matter in the priming process. I thus want to identify
which “fi lter variables,” or moderators, assume a role in enhancing or inhibiting
priming effects. Third, I wish to clarify some methodological issues associated
with the study of priming effects. Most notably, it may be instructive to pit the
cross-temporal method against the cross-sectional method, and to ascertain
whether both methods can be combined into a single design which may prove
more effi cient for the detection of presumably modest effects. This article will
address these and related questions using experimental data collected at fi ve
Swiss universities in April and May 2005.
The second section of this article is devoted to the presentation of my theoretical
model. I shall distinguish between the various mechanisms (or “mediators”)
implied in the priming process, as well as between the main independent variables
(or “moderators”) which have been shown or can be thought to come into play in
that process. In Section 3, I familiarize the reader with the empirical data; issues
related to the research design, to data collection, and to variable measurement will
be addressed in detail. Section 4 then takes up the task of applying my model to
the data at hand, showing that priming indeed occurred, but only under the most
“favorable” conditions. Next, a concluding section discusses some ambiguous or
unexplored aspects of my analysis, and sums up the main results of the article.
The Model
Mediators of Priming
In part, political attitudes are structurally determined. Swiss attitudes toward the
question of EU accession are no exception in this regard (for example, Christin
and Trechsel, 2002; Sciarini and Listhaug, 1997). Yet, however important this
structural component may be, a substantial part of the variance in evaluations
of Swiss EU membership has contingent causes, depending on which issues are
tackled by the mass media, by referendum campaigns on European matters, or by
friends and relatives, and depending on how deeply these issues are highlighted.
In that respect, the priming theory points to how issues and arguments leave an
imprint on citizens’ minds, and how they can determine preferences toward EU
membership. As we shall see, the theory is extremely fertile, but it is also complex
to operationalize, not least because the very concept of “priming” is polysemic
and there exist several understandings of what priming really means. Originally,
the concept was introduced in biology, medicine, and engineering, from where
Marquis: Moderators of Priming Effects 187
it was borrowed by social and cognitive psychology. The common idea of most
psychological defi nitions is that mental constructs once activated (or “primes”)
remain temporarily accessible and applicable to the interpretation of stimuli
encountered in the present (“the priming effect”).1 Solicitation of a primed category
renders this category immediately available for making sense of a new piece of
information, which is then encoded in memory together or closely associated with
the prime. Thus, information will be evaluated in close proximity with the prime
(for example, positively or negatively, depending on the affective value of the
prime) at some later time (for a review, see Fiske and Taylor, 1991: 257–66; Wyer
and Srull, 1989: Ch. 6).
Later on, the concept was adopted by communication and political scientists,
who have made somewhat simplifi ed use of it. In their conception, priming is
essentially a “retrieval bias,” whereby frequently or recently activated memories
tend to outweigh less salient or less easy-to-use information in the making of judg-
ments about political objects; the assimilation and biased-encoding features of
the priming process are generally neglected. However, because of its focus on the
retrieval step of the process, this priming theory is not suffi cient, in my opinion,
to account for the full effect of those variables mediating the encoding step. At the
same time, introducing new variables sets out new concerns, because some of them
yield opposite effects from one step of the process to the next (Iyengar and Kinder,
1987: 95–6; Miller and Krosnick, 2000: 303–4). On occasion, these countervailing
effects may cancel one another, producing a seemingly absent or trivial net impact
on priming.
What we need, then, is a more encompassing picture of the successive steps
a piece of information must take in order to become an ingredient of some
political judgment. My reasoning is that the whole process is roughly similar to
the stochastic chain of psychological mechanisms outlined by specialists in the
fi eld of attitude-change research, but with differential emphasis on some mediating
steps. Briefl y described, “information-processing” models (for example, McGuire,
1985; Zaller, 1992) posit that the path to attitude change is punctuated by a
series of “mediators,” each of which is conditional on the fulfi llment of that preceding it.
Accordingly, extensive efforts to persuade individuals to accept or reject EU
membership can be ruined by the failure to overcome a single step, because a
possessive information “is like a chain. It cannot be stronger than its weakest link”
(Alcalay and Bell, 2000: 18).
Elsewhere (Marquis and Gilland Lutz, 2004), we have developed a fuller account
of the causal chain of mediators, distinguishing seven main steps from exposure to
evaluation. Now, from the standpoint of priming theory, the importance of some
mediators is reduced in comparison with more standard persuasive settings,
while others deserve more careful consideration. To begin with, I share McGuire’s
(1985: 286) concern that attitude research has focused too exclusively on the
yielding mediator. There is reason to believe that a person can retain pieces of
information which she takes to be incompatible with her own values, predispositions,
or knowledge (for example, Eagly et al., 2000). Moreover, although getting
such “unconvincing” information is hardly relevant for the modifi cation of core
attitudes toward an object, it does contribute to the activation of pre-existing
memories and thus to “refreshing” one’s attitudes. This occurs because of the
assumed associative structure of human memory (for example, Anderson, 1983) and
because “spreading activation” between related concepts in the memory network
reinforces the utilized associative pathways (McGuire, 1960).2
To continue reading
Request your trial