20
responsiveness was significant, neither the main effects of the source nor the source x
responsiveness interaction term was significant: the source, F (1, 126) = .47, p > .50; the
responsiveness, F (1, 126) = 14.74, p < .001; the interaction term, F (1, 126) = .08, p > .78.
Therefore, Hypothesis 1c was not supported.
Moderating Effects of Responsiveness (H2a to H2c)
Hypotheses 2a through 2c predicted that when the Web site is little responsive, a
government source will be more effective in enhancing favorable attitude toward health
information (H2a), health behavior intentions (H2b) and attitude toward the Web site (H2c) than
a commercial source; when the Web site is highly responsive, however, effects of a government
source and a commercial source will be similar. As the above tests for Hypotheses 1a through 1c
have shown, moderating effects of responsiveness on the relationship between the source and the
dependent variables were not found in attitude toward health information and attitude toward the
Web site. Therefore, Hypotheses 2a and 2c were not supported.
Since the moderating effects of responsiveness was found in two behavior intention
measures, preventive behavior intention and information seeking intention, planned contrasts
were performed to test if the direction of the moderating effects is found as predicted. The
results showed that when the Web site was little responsive, a commercial source (M = 3.76, SD
= .97) was significantly more influential in promoting preventive behavior intentions than a
government source (M = 3.21, SD = .96) (p < .05). When the Web site was highly responsive,
effects of commercial source (M = 3.23, SD = .96) and government source (M = 3.63, SD = .97)
did not differ (p > .11). Similarly, when the Web site was little responsive, a commercial source
(M = 4.34, SD = 1.23) was significantly more influential in promoting information seeking
intentions than a government source (M = 3.61, SD = 1.22) (p < .05). When the Web site was
highly responsive, effects of commercial source (M = 3.75, SD = 1.22) and government source
(M = 3.97, SD = 1.22) did not differ (p > .48). Therefore, Hypothesis 2c was partially supported.