19
z-score of 3.45 was necessary to establish significance at the .01 level, and a z-
score of 4.54 was necessary for the .001 level. For the regression on positive
affect, a z score of 2.89 was necessary to establish significance at the .05 level, a z
score of 3.36 was necessary for the .01 level, and a z score of 4.47 was necessary
for the .001 level. Because a different set of variables was included in each of the
two regressions, results are reported in separate tables.
Table One shows the results for negative affect. Five variables:
aggression, the outcome of violence, the outcome of disaster not related to
violence, negative emotional displays, and unusual juxtapositions significantly
predicted an increase in negative affect among respondents.
(Insert Table One Here)
Table Two shows the results for positive affect. Aggression, unusual
juxtapositions, negative emotional display, and seeing the third subject from the
side all significantly and negatively predicted positive affect. Positive emotional
display, seeing the second subject head-on and degree of closeness among
subjects predicted positive affect significantly and had a direct relationship with
the dependent variable.
(Insert Table Two Here)
Discussion and Conclusions
Although only a limited number of variables remained significant after
being subjected to the conservative tests used here, it is important to note what is
similar about those variables. In general, the features that had the most robust
impact on affective response were aspects of the images’ content. Formal features
Convention