ABSTRACT
The influence of presumed influence hypothesis (IPI hypothesis) explains that people have biased perceptions of media influence and they change their behavior based on such perceptions. This study explicated the mechanisms of influence of presumed influence in health communication by integrating the theoretical explanations of the IPI hypothesis with theories of normative influence. The causal chains of the IPI hypothesis were examined using an experimental methodology with a HIV prevention, PrEP (pre-exposure prophylaxis). The results supported the expectations. Presumed exposure to health messages about PrEP shaped presumed influence of the messages on others, which in turn affected one’s own intentions for information seeking and prosocial behaviors about PrEP. The findings also show that descriptive norms and injunctive norms interact with presumed influence differently. This study discusses the potential benefits of the IPI hypothesis in health communication.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1. First, model chi-square statistic (χ2) was inspected in comparison with the degrees of freedom. The relative chi-square in which a chi-square statistic divided by its degrees of freedom (χ2/df) was used, because chi-square statistic tends to over-reject the model fit when the model is estimated with large sample size (Bollen, Citation1989). A model with the value of relative chi-square less than 2 is considered as acceptable fit (Mueller, Citation1996). Other model fit indices were used with cutoff criteria as follows: comparative fit index (CFI) and Tucker–Lewis index (TLI) at or above .95 (Hu & Bentler, Citation1999), root mean square error of approximation (RMSEA) at or below .05, and standardized root-mean-square residual (SRMR) at or below .08 indicate a good fit (Browne & Cudek, Citation1993).
2. Unstandardized error (δx) = VAR(X)(1 – ρ), where VAR(X) is the sample variance of presumed exposure and ρ is the reliability estimate of presumed exposure. The ideal reliability estimate should be drawn from previous studies. As there is no known reliability estimate of presumed exposure, as a rule of thumb I took .85 as a reliability estimate, which indicates that a total 15% of total variance of this measure is estimated to be error (Brown, Citation2006). The sample variance of presumed exposure is 5.851. Thus, unstandardized error of perceived other exposure is fixed to .88 in the model.