Four Perspectives on . . .
17
premessage fear, acceleration, velocity, and deceleration respectively. Although all four are
nonsignificant at p < .05, the significance test alone is not a convincing means of assessing a no-
difference hypothesis. Rather, a power analysis is needed. With an N of 361 and two-tailed alpha
of .05, power is .46 for small correlations (i.e., .10) and greater than .99 for moderate (i.e., .30)
and large correlations (i.e., .50) (Cohen & Cohen, 1983). Thus, while we cannot rule out a small
relationship, H3 is clearly supported for moderate to large associations.
H4-H6: Acceleration, Velocity, and Deceleration Perspectives
H4-H6 predicted positive associations between persuasion and acceleration, velocity, and
deceleration respectively. And, as Table 2 shows, the simple correlations support all three
predictions. They are .18, p < .01, for acceleration, .18, p < .01, for velocity, and .15, p < .01, for
deceleration. However, a method was needed that allowed more stringent comparisons between
perspectives. Consequently, we turned to regression analysis. For each analysis, the change in
likelihood of obtaining a flu vaccination served as the dependent variable. The independent
variables were entered in two blocks. The first block contained one of the three fear variables,
whereas the second block contained a different fear variable that was being contrasted with first.
Due to collinearity among the fear indices, only two tests per equation were conducted. We ran
each equation twice, in each case reversing the order of entry of the two variables that were to be
contrasted. In this way, each predictor was given the opportunity to consume as much variance in
the dependent variable as possible prior to its competitor. Comparing the results obtained from
different entry orders helped to clarify the contribution of each perspective.
Table 3 presents the results of the regression analyses in three parts.
2
In the upper portion
of the table, the acceleration and velocity perspectives are compared. Inspection of the R
2
)
values
Convention