ABSTRACT
We proposed a conceptual model to predict health information-seeking behaviors (HISBs) from three different sources (family, the Internet, doctors). To test the model, a structural equation modeling (SEM) analysis was conducted using data from the 2012 Annenberg National Health Communication Survey (ANHCS) (N = 3,285). Findings suggest higher social support from family predicts higher trust in health information from family members (abbreviated as trust in this article). Trust is positively related to HISBs from all three sources, with the path linking trust to HISB from family being the strongest. The effect of social support on HISB from family is partially mediated by trust, while effect of social support on HISBs from the Internet/doctors is fully mediated by trust. Implications of the study are discussed.
Acknowledgments
The authors thank the Annenberg Schools at the University of Pennsylvania (UPenn) and the University of Southern California (USC) for conducting the Annenberg National Health Communication Survey (ANHCS) and making the data available online.
Notes
1 First, a panel of participants was recruited by Knowledge Networks (KN) from a list-assisted, random digit dial (RDD) population of all landline telephone numbers in the United States to provide a probability-based starting sample. Then, an address-based sampling (ABS) method was used by providing the individuals who agreed to be included in the panel with free hardware and/or Web access. The provision of identical hardware to all participants allowed for a consistent delivery of the survey content and did not bias the sample in favor of computer owners or Internet users.
2 We used generalized questions as a proxy to measure perceived social support, because previous research suggests that perceived social support is primarily based on personal experiences (Albrecht & Adelman, Citation1987; Vinokur et al., Citation1987). This proxy measure of perceived social support has been applied in previous studies (e.g., Minnis, Doherty, Cheng, Otero-Sabogal, & Padian, Citation2010).
3 Values smaller than .05 for the standardized root mean square residual (SRMR) and root mean square of approximation (RMSEA) suggest close model fit, while values between .05 and .08 show reasonable fit (Browne & Cudeck, Citation1993; Hu & Bentler, Citation1998). Values greater than .90 for comparative fit index (CFI) and Tucker–Lewis Index (TLI) indicate reasonable fit, while values greater than .95 indicate excellent fit (Bentler & Bonett, Citation1980; Hu & Bentler, Citation1998).