Seeking Health Information in the Information Age: The Role of Internet Self-Efficacy

Abstract

The study reported here explored Internet self-efficacy in the process of acquiring health information on the World Wide Web. Internet self-efficacy was examined as a partial mediator of exogenous variables reflecting an individual's motivation to take action to manage his or her health and experience using the Web, and endogenous variables representing information-seeking behaviors and outcomes. The results show that Internet self-efficacy partially mediated the relationship between respondents' experience using the Web and their attitude about the quality of health information available online as well as the relationship between respondents' desire for informational involvement and their attitude about information quality. Internet self-efficacy completely mediated the relationships between respondents' Web experience and the perceived success of a recent information search as well as between respondents' desire for informational involvement and perceived search success.

Keywords:

• E-health
• Health Communication
• Information Seeking
• Internet
• Self-efficacy

Notes

Note. IHLOC = internal health locus of control. Ratings for all variables were made on a seven-point scale, with one exception. Ratings of intention to use the Web in the future to seek health information were made on an 11-point scale. Larger values indicate a greater amount of a variable.

∗p < .05.

Note. ISE = Internet self-efficacy. IHLOC = internal health locus of control. Unstandardized indirect effects were computed by multiplying the unstandardized values for the exogenous variable-mediator and mediator-endogenous variable paths. Evidence for mediation is present when the confidence interval does not include zero.

It should be noted that this study does not represent a formal test of the CMIS. Instead, the CMIS is used to inform the exogenous and endogenous variables examined in this study. Additionally, because the CMIS focuses on predicting media use in the context of one specific health condition, the variables used for this study were adapted to reflect a more general approach to health information-seeking behavior that is not tied to one specific health condition.

It should also be noted that the Web-based survey tool made it possible to examine the time at which a respondent accessed the questionnaire, the time the questionnaire was completed, and the respondent's IP address. Each of these features was reviewed in an attempt to attempt to identify any suspicious questionnaires; none were found.

Factor loadings for items in the health locus of control measure and behavioral involvement sub-scale were lower than desirable (< .40). However, these items were retained because both measures are well established and have been used fairly frequently in research related to health communication. Further, both of the measures fit the sample data adequately.

Although Holbert and Stephenson (2003) recommend using the distribution of products analysis to test for mediation in SEM, the asymmetric distribution of products test was used in this study because it performs better in those situations where the two paths corresponding to the mediating variable (exogenous variable-mediator, mediator-endogenous variable) are of different magnitudes (large, zero; small, large; etc.) (MacKinnon et al., 2002). In these instances, the differences between the Type I error rates for the asymmetric distribution of products test and the distribution of products test are quite substantial; the distribution of products test results in inflated Type I error rates. Given that the many of the paths in this study from the exogenous variables to the mediator and from the mediator to the endogenous variables are not of the same magnitude, the asymmetric distribution of products test provides a more conservative estimate of the mediating effects of Internet self-efficacy.

Upper and lower confidence limits were constructed using the following formulae from MacKinnon et al. (2004): Upper confidence limit = α∗β + Meeker Upper Limit ∗σ_αβ Lower confidence limit = α∗β + Meeker Lower Limit ∗σ_αβ α refers to the exogenous-mediator path, β refers to the mediator-endogenous path, and