22
DISCUSSION
This study investigated if the effects of source that were found in the traditional source
and source credibility literature were also found in an online environment and if the effects of
source were moderated by responsiveness of the information channel, the Web site. In general,
findings of this study did not confirm the previous findings in the source and source credibility
literature. Perhaps the most contradicting finding is the direction of moderating effects of
responsiveness on source effects. Contrary to our prediction, when responsiveness was low, a
commercial source, presumably a low credible source, was more effective than a government
source, a high credible source. When responsiveness was high, there was no statistically
significant difference between the two sources, although a comparison of means indicated that a
government source was more effective than a commercial source for both preventive behavior
intention and information seeking intention. Since both commercial source and government
source did not differ in perceived credibility in this study, we are not able to determine why a
commercial source was more effective than a government source under the low responsiveness
condition. While there may be a number of explanations, expectancy violation theory (Burgoon,
1978) would reason that participants paid more attention to the information from a commercial
source in the low responsiveness condition, because provision of non-commercial, health
information from a commercial source “violated” their expectation. When participants were told
that the commercial source recently released an allergy medication, they might have expected to
see commercially oriented allergy information. However, when they found there was no
commercial message in the information content, their expectation was not confirmed and, as a
result, they paid more attention and “bought” the message. Expectancy violation theory would
also explain participants’ response in the high responsiveness condition, but in a different way.
Considering that Web sites sponsored by governmental or non-profit organizations are generally
less interactive and flashy than those by commercial organizations (Stout, Villegas, and Kim,
2001), the highly responsive Web site by a government source violated participants’ expectation
and thus participants paid more attention to the information by the government source.
Convention