seven point issue scales, assessments of the candidates’ character traits, retrospective evaluations
of the Ford presidency, and familiar demographic measures. In addition, it contains the
questions necessary to develop measures of three different types of attitude strength (uncertainty,
ambivalence, and importance). In short, it has everything necessary to test the theory.
The Presidential Campaign Impact on Voters survey asked respondents the same basic set
of survey questions in five waves and supplemented these with two shorter surveys—one after
the presidential debates and one after the election. The waves occurred in February, April, June,
August, and October. The Debate survey was in late October and the post election survey was
completed mostly in November. The survey designers included the vote choice question in each
wave, but only included the measures of most of the independent variables in the main pre-
election waves.
Measurement
Dependent variable: The dependent variable in these analyses is the respondent’s vote
choice. This is assessed using two different questions. Before Carter is the presumptive
Democratic nominee and Ford defeated Reagan, the survey asked respondents “Please think
ahead to this fall’s presidential election. Do you think you will probably vote for the Democratic
or Republican candidate for president this fall?” Every survey after the fourth wave adds the
names of the candidates to the questions. This is, obviously, not ideal. It would be much better
if the surveyors were clairvoyant and knew who the nominees would be. They could not. So I
am faced with the choice of either throwing out the first three waves of the panel or treating the
two questions as if they were synonymous. The former strategy makes the analyses worthless—
the campaign had clearly begun by the August wave and makes the comparison to how voters
would have voted absent the campaign impossible.
9
Convention