empirically demonstrate which type of attitude strength is responsible for the changes. I test the
mechanism underpinning campaigns by testing how the campaign changes attitude strength
(ambivalence, importance, or certainty) and how these change voters’ opinions (Miller and
Peterson 2004). In particular, I demonstrate that the heresthetic change that occurs during a
campaign occurs because the campaign changes the uncertainty voters have about candidates.
The other potential mediators of campaign effects studied here (ambivalence and importance) do
not have any significant impact on how voters use their attitudes.
The paper proceeds as follows. In the next section I develop a theory of campaign effects
focusing on the roles of changes in the content of citizens’ attitudes and the changing strength of
these attitudes. I then move to a discussion of the data, the Presidential Campaign Impact on
Voters: 1976 Panel developed by Patterson (1976).
In particular, I address the nature of this
panel survey and why this is the best available data for testing this question. The next section of
the paper presents the method for analyzing these data, a non-linear growth curve. I then present
the results and conclude with what these results mean for the study of campaigns and how these
findings need to be extended.
Campaign effects
In a recent review of the campaign effect literature, Bartels (2006) suggests that
campaign effects (by which he means changes in a voter’s decision making) stem from two
distinct processes: persuasion and priming. Persuasion is the change in the content of an
individual’s attitude. The voter becomes more liberal or more Republican. Priming occurs when
the attitude does not change, but the weight that a voter applies to the attitude changes. Voter’s
1
The data were originally collected by Thomas E. Patterson and made available by the Inter-university
Consortium for Political and Social Research. Neither Patterson nor ICPSR bear any responsibility for the analyses
or interpretation of the data presented here. See Patterson (1980) for a larger description of the data.
4
Convention