Correlates of Consumer Trust in Online Health Information: Findings From the Health Information National Trends Survey

Abstract

The past few decades have witnessed a dramatic increase in consumers seeking health information online. However, the quality of such information remains questionable, and the trustworthiness of online health information has become a hot topic, whereas little attention has been paid to how consumers evaluate online health information credibility. This study builds on theoretical perspectives of trust such as personal-capital-based, social-capital-based, and transfer-based, and it examines various correlates of consumer trust in online health information. The author analyzed the 2007 Health Information National Trends Survey data (N = 7,674). Results showed that consumer trust in online health information did not correlate with personal capital such as income, education, and health status. Social capital indicated by visiting social networking Web sites was not associated with trust in online health information either. Nevertheless, trust in online health information transferred from traditional mass media and government health agencies to the Internet, and it varied by such information features as easiness to locate and to understand. Age appeared to be a key factor in understanding the correlates of trust in online health information. Theoretical and empirical implications of the results are discussed.

With the explosion of online health information, many people have sought and used Web-based health information (Pew Research Center, 2000, 2005; Rozmovits & Ziebland, 2004). Despite its positive influences on consumer health knowledge, health management (Baker, Wagner, Singer, & Bundorf, 2003; Wald, Dube, & Anthony, 2007; Wathen & Burkell, 2002), and social support experienced by patients (for a review, see Eysenbach, Powell, Englesakis, Rizo, & Stern, 2004), research has shown that Web-based health information suffers low credibility (Metzger, 2007). Eysenbach, Powell, Kuss, and Sa (2002) systematically reviewed the studies that evaluated the accuracy, completeness, readability, design, disclosures, and references of 5,941 health Web sites and 1,329 health Web pages and concluded that the quality of online health information was a problem, which is supported by later studies (Corcoran, Haigh, Seabrook, & Schug, 2009; Khazaal, Chatton, Cochand, & Zullino, 2008; Ream, Blows, Scanlon, & Richardson, 2009). It is therefore no surprise that the trustworthiness of online health information is rapidly becoming a hot topic, and consumers have limited trust in online health information (Bernhardt, Lariscy, Parrott, Silk, & Felter, 2002; Huntington et al., 2004).

Trust has been studied in many disciplines (Doney, Cannon, & Mullen, 1998; Hosmer, 1995) and is defined as an individual's expectation of others' behavior that will not do harm to him- or herself (Sztompka, 1999). Consumer trust in online health information is a product of the interaction among source, message, channel, and receiver characteristics (for a review, see Wathen & Burkell, 2002), including source authority, information currency, easiness to read, inclusion of scientific references, privacy policy statement, third-party endorsement, professional site design (Dutta-Bergman, 2004; Eysenbach & Köhler, 2002; Rains & Karmikel, 2009; Turner, Petrochilos, Nelson, Allen, & Liddy, 2009), presence of advertising (Walther, Wang, & Loh, 2004), and consumer characteristics (Freeman & Spyridakis, 2004; Huh, DeLorme, & Reid, 2005; Morahan-Martin, 2004; Sillence, Briggs, Harris, & Fishwick, 2007; Wang, Walther, Pingree, & Hawkins, 2008).

Although such research has offered valuable information on consumer evaluation of online health information credibility, it has been inadequate as compared to studies assessing health Web site quality from a researcher or medical professional perspective. Accordingly, in the present study, I conducted a secondary data analysis of the National Cancer Institute's 2007 Health Information National Trends Survey and examined several correlates of trust in online health information identified through a few theoretical perspectives of trust. The factors include receiver characteristics (personal capital, social capital, and trust in offline sources of health information) and message characteristics (information understandability and easiness to find).

Personal Capital and Trust

Sztompka (1999) proposed that personal capital (i.e., what people have) is more important than personality traits (i.e., what people are) in making people capable and willing to trust others; namely, the various resources one commands such as money, education, health, and good looks affect their readiness to trust others. One explanation is that personal resourcefulness fosters self-confidence and willingness to take risks associated with trusting others (Luhmann, 1979) or reduces subjective vulnerability to trust violation (Sztompka). Another explanation is that the command of ample resources enhances self-concept so that the individual is more likely to be optimistic and compassionate, which translates into more trust in others (Giddens, 1991).

Consistent with the proposition regarding the relation between personal capital and trust in others, studies have shown that low-income people develop mistrust in health care providers (Sheppard, Zambrana, & O'Malley, 2004). People with more education tend to trust others more (Alberto Alesina & La Ferrara, 2002; Franzini, 2008). Self-rated health is significantly associated with trust in others (Franzini; Nummela, Sulander, Rahkonen, & Uutela, 2009), trust in mass media and trust in the health care system (Tokuda, Fujii, Jimba, & Inoguchi, 2009), and trust in drug information (Menon, Deshpande, Zinkhan, & Perri, 2004). However, other studies have reported contradictory findings. Relatively high trust in health care providers exists among low-income people (Benkert, Peters, Tate, & Dinardo, 2008; Sheppard et al.). Income, education, or health are not always predictive of trust in health information (Benotsch, Kalichman, & Weinhardt, 2004; Huh et al., 2005; Kalichman et al., 2006; Shon, Marshall, & Musen, 2000). Because of the discrepancy, this study tested how several personal capital indicators are correlated with trust in online health information:

• Research Question 1a: Is education correlated with trust in online health information?

• Research Question 1b: Is income correlated with trust in online health information?

• Research Question 1c: Is health correlated with trust in online health information?

Social Capital and Trust

Research suggests that a person's social capital predicts trust in others. Social capital is defined as network ties in a person's social relationship that promote advantageous outcomes to the person involved (J. S. Coleman, 1988; Portes, 1998). Putnam (1993) posited that trust is one main element of social capital; he defined social capital as characteristics of social organizations including networks, norms, and trust that facilitate action and collaboration for reciprocal benefit. Similarly, Portes maintained that enforceable trust is one source of social capital because it gives people opportunities to collect benefits from their social networks. Sztompka (1999) stated that interaction with individuals in a social network is often permeated with returned trust and mutual propensity to trust. In short, the breadth and depth of one's social network is a good indicator of his or her propensity to trust others. This study therefore proposes that virtual social network activity such as visiting social networking Web sites is positively related to trust in online health information.

The proposition also arises from empirical evidence. Participation in social networking Web sites is positively related to social capital and trust in others (Ellison, Steinfield, & Lampe, 2007; Steinfield, Ellison, & Lampe, 2008). To extend the research on social capital and trust, the present study examined the linkage between use of social networking Web sites and trust in online health information:

• Hypothesis 1: People who visit a social networking Web site such as MySpace or Second Life trust online health information more than those who do not.

The Transfer Perspective of Trust

Trust transference, a process whereby trust is transferred from a trusted third party (proof source) to another entity with which the trustor has little or no knowledge (Doney et al., 1998), has been observed mostly in such disciplines as management and e-commerce. Stewart (2003) reported that trust transferred across hypertext links because of the perceived similarity between the linked corporations, and it transferred from an offline store to the online store of the same corporation.

Regarding trust transfer between information sources, Napoli (2001) proposed that because of the hybrid nature of online health information, the credibility of online health information approximates that of the combination of various health information sources, including traditional mass media, authoritative interpersonal sources, health organizations, and informal sources such as family and friends. Consistent with this proposition, Huh et al. (2005) reported that trust in direct-to-consumer advertising in traditional media positively predicted trust in online drug information. Nevertheless, Rains (2007) argued that trust in health information in traditional media is negatively related to trust in online health information, because the distrust in health information provided by traditional media motivates individuals to search for health information online. Respondents who used online medical information rated information from radio and television news less useful than did their counterparts (Diaz et al., 2002). To advance the understanding of trust transference between different health information sources, I proposed the following research questions:

• Research Question 2a: Does trust in health information from doctors correlate with trust in online health information?

• Research Question 2b: Does trust in health information from family and friends correlate with trust in online health information?

• Research Question 2c: Does trust in health information from traditional media including radio, newspapers/magazines, and television correlate with trust in online health information?

• Research Question 2d: Does trust in health information from government health agencies correlate with trust in online health information?

Information Quality and Trust

The previous research questions and hypotheses examine primarily how consumer characteristics are related to trust in online health information. Trust is correlated not only with receiver characteristics but also with information characteristics (for a review, see Wathen & Burkell, 2002). Although few published studies have examined the contribution of information features to online health information credibility, research has suggested that regardless of Internet experience and issue involvement, message quality predicts Web site credibility (Hong, 2006) and is the most important factor of consumer trust in online health information in the long term (Zahedi & Song, 2008). For example, research shows the importance of information relevance, accuracy, presentation, and format in consumer perception of online information credibility (Fritch & Cromwell, 2001; Iding, Crosby, Auernheimer, & Klemm, 2009). Given both the predictive role of information characteristics in trust in online health information and limited research in this area, I tested the following hypotheses:

• Hypothesis 2a: The more effort to locate health information, the less trust in online health information.

• Hypothesis 2b: The harder to understand health information, the less trust in online health information.

Methods

Sample

The National Cancer Institute's 2007 Health Information National Trends Survey surveyed a national representative sample of 7,674 adults either through the mail (n = 3,582) or through a telephone using random digit dial (n = 4,092). The response rates for the two samples were 31.0% and 24.2%, respectively. Table 1 presents the participants' demographics. Overall, the entire sample underrepresented young, male minorities and those with low education (Cantor & McBride, 2009).

Table 1. Respondent demographics (N = 7,674)

Variable/Group	n	%
Gender
Male	2,969	38.7
Female	4,696	61.2
Unknown	9	0.1
Age (years)
18–34	1,113	14.5
35–49	1,831	23.9
50–64	2,451	31.9
65 or older	2,199	28.7
Unknown	80	1.0
Race
Caucasian	5,445	70.9
African American	687	9.0
Hispanic	622	8.1
Asian	202	2.6
American Indian or Alaska Native	61	0.8
Native Hawaiian or other Pacific Islander	15	0.2
Multiple races mentioned	146	1.9
Unknown	496	6.5
Income
$0–$9,999	373	4.8
$10,000–$14,999	409	5.3
$15,000–$19,999	360	4.7
$20,000–$34,999	1,056	13.8
$35,000–$49,999	873	11.4
$50,000–$74,999	1,203	15.7
$75,000–$99,999	775	10.1
$100,000–$199,999	992	12.9
$200,000 or more	274	3.6
Unknown	1,359	17.7
Education
Less than high school	683	8.9
High school graduate	1,804	23.5
Some college	2,192	28.6
College graduate	2,637	34.4
Unknown	358	4.7

Measures

Trust in Online Health Information

Respondents were asked to answer the following question on a 4-point Likert-type scale ranging from 1 (a lot) to 4 (not at all): “In general, how much would you trust information about health or medical topics from the Internet?” Reverse coding was applied so that higher numbers denoted higher trust.

Personal Capital (Research Questions 1a–c)

Respondents were asked to indicate the highest grade or level of schooling they completed, and responses were collapsed into four groups: “less than high school,” “high school graduate,” “some college,” and “college graduate.”

One question assessed the respondents' annual family income: “Thinking about members of your family living in this household, what is your combined annual income, meaning the total pre-tax income from all sources earned in the past year?”

Respondents were also asked to rate their health status on a 5-point Likert-type scale ranging from 1 (excellent) to 5 (poor). Reverse coding was used.

Social Capital Indicator (Hypothesis 1)

Respondents were asked to indicate whether they had visited a social networking Web site such as MySpace or Second Life in the past 12 months.

Trust in Sources of Health Information (Research Questions 2a–d)

On a 4-point Likert-type scale ranging from 1 (a lot) to 4 (not at all), respondents rated their trust in health information from different sources, including a doctor, family or friends, newspapers or magazines, radio, television, and government health agencies. Reverse coding was applied.

Information Features (Hypotheses 2a and 2b)

Respondents rated their general health information searching experience on two items, each rated on a 4-point Likert-type scale ranging from 1 (strongly agree) to 4 (strongly disagree): “It took a lot of effort to get the information you needed” and “The information you found was hard to understand.” Reverse coding was used.

Data Analysis

I used the SAS 9.2 SURVEYREG (SAS Institute Inc., Cary, NC) procedure to perform the regression of trust in online health information on the correlates under consideration, with age, gender, and race as covariates. The results were weighted for population estimates, and I used the jackknife variance estimation method to take replicate weights into account and to reduce estimation error (U.S. National Cancer Institute, 2007).

In addition, I performed a subgroup regression analysis to examine how the relations between the various factors and trust in online health information differ among age groups, because previous research has shown age differences in online information seeking (Utz, 2009; Valenzuela, Park, & Kee, 2009), and preliminary analyses of this study found age differences. I used the SUDAAN 10.0 program (RTI International, Research Triangle Park, NC) for this analysis to incorporate the Jackknife variance estimation method.

Results

Descriptive Statistics

Table 1 presents the participants' education level. Forty percent of the respondents' family income was less than $50,000 (see Table 1). The mean of self-reported health status was 3.39 (SD = 0.97). For visiting social networking Web sites, 1,159 (15.1%) people answered yes, whereas 3,887 (50.7%) answered no; the remaining respondents did not provide a response.

The mean of trust in online health information was 2.78 (SD = 0.91). The means of trust in various offline health information sources were 3.63 (SD = 0.61) for doctors, 2.70 (SD = 0.79) for family or friends, 2.48 (SD = 0.80) for newspapers or magazines, 2.11 (SD = 0.83) for radio, 2.29 (SD = 0.85) for TV, and 2.96 (SD = 0.89) for government health agencies.

The means regarding whether health information was hard to retrieve and to understand were 2.17 (SD = 0.96) and 1.90 (SD = 0.92), respectively.

Regression Analysis Results

Table 2 summarizes the main regression analysis, and the following details the results of the analysis.

Table 2. Summary of regression analysis for correlates of trust in online health information

Variable	β	SE	F^a
1. Income	.01	.01	2.89
2. Education	.02	.02	0.80
3. Subjective health status	–.01	.02	0.57
4. Visiting social networking websites	–.01	.04	0.09
5. Trust in health information from a doctor	.04	.03	1.70
6. Trust in health information from family and friends	.04	.02	2.37
7. Trust in health information from newspapers/magazines	.12	.03	16.86***
8. Trust in health information from TV	.19	.02	68.84***
9. Trust in health information from radio	–.02	.02	0.88
10. Trust in health information from government health agencies	.08	.02	13.82***
11. Easiness to locate health information	–.06	.02	8.25**
12. Hard to understand health information	–.07	.02	11.85**

Regarding Research Questions 1a–c, results showed that income, education, and subjective health status were not significant: F(1, 100) = 2.89, p > .05; F(1, 100) = .80, p > .05; and F(1, 100) = .57, p > .05, respectively.

Regarding Hypothesis 1, results showed that visiting social networking Web sites was not significant, F(1, 100) = .09, p > .05. Hypothesis 1 was not supported.

Regarding Research Questions 1a–d, trust in health information from doctors was not related to trust in online health information, F(1, 100) = 1.70, p > .05. Trust in health information from family and friends was not significant, F(1, 100) = 2.37, p > .05. Trust in health information from traditional media including newspapers/magazines and TV was correlated with trust in online health information: F(1, 100) = 16.86, p < .001, β = .12 (SE = .03) for newspapers or magazines, and F(1, 100) = 68.84, p < .001, β = .19 (SE = .02) for TV. Trust in radio health information was not significantly correlated with trust in online health information, F(1, 100) = .88, p > .05. Trust in health information from government health agencies was positively related to trust in online health information, F(1, 100) = 13.82, p < .001, β = .08 (SE = .02).

Concerning Hypotheses 2a and 2b, the harder it was to find health information, the less trust there was in online health information, F(1, 100) = 8.25, p < .01, β = −.06 (SE = .02); and the harder the health information was to understand, the less trust there was in online health information, F(1, 100) = 11.85, p < .01; β = −.07 (SE = .02). Hypotheses 2a and 2b were supported.

In addition, results revealed that among the covariates, only age was significant, F(3, 100) = 4.51, p < .01. Accordingly, I conducted a subgroup analysis to see how the correlations investigated in this study vary by age groups.

For people between the ages of 18 and 34 years, trust in health information from TV and government health agencies was significantly related to trust in online health information: F(1, 100) = 18.43, p < .001, β = .21 (SE = .05); and F(1, 100) = 4.50, p < .05, β = .11 (SE = .05), respectively. In addition, the harder health information was to find, the less trust there was in online health information, F(1, 100) = 3.96, p < .05, β = −.09 (SE = .04).

For people between the ages of 35 and 49 years, trust in health information from both newspapers/magazines and TV was significantly related to trust in online health information, F(1, 100) = 13.74, p < .001, β = .17 (SE = .05); and F(1, 100) = 26.93, p < .001, β = .19 (SE = .04), respectively. Further, the harder it was to find health information, the less trust there was in online health information, F(1, 100) = 5.18, p < .05, β = −.06 (SE = .03); and the harder it was to understand health information, the less trust there was in online health information, F(1, 100) = 4.44, p < .05, β = −.07 (SE = .03).

For people between the ages of 50 and 64 years, results were consistent with those based on the entire sample. Trust in health information from newspapers/magazines, TV, or government health agencies was positively related to trust in online health information: F(1, 100) = 9.89, p < .01, β = .13 (SE = .04); F(1, 100) = 23.63, p < .001, β = .13 (SE = .03); and F(1, 100) = 7.66, p < .01, β = .09 (SE = .03), respectively. Whether health information was hard to locate and to understand were related to trust in online health information: F(1, 100) = 6.05, p < .05, β = −.07 (SE = .03), and F(1, 100) = 7.02, p < .01, β = −.09 (SE = .03), respectively.

For people age 65 years or older, none of the correlates was significant.

Discussion and Conclusions

This study explored various correlates of consumer trust in online health information from several theoretical frameworks of trust. The factors ranged from consumer characteristics, such as personal capital, social capital, and past experience with offline health information sources, to message features such as easiness to locate and to understand. Overall, this study demonstrates trust in online health information varies by certain consumer characteristics and information features. These results have important theoretical and managerial implications.

Regarding personal capital and trust, this study shows that education, income, and subjective health status do not correlate with trust in online health information. This is in accordance with Huh et al.'s (2005) study, but it contradicts other studies (Benotsch et al., 2004; Kalichman et al., 2006; Menon et al., 2004) and does not support the proposition that high personal capital enables people to trust others more (Sztompka, 1999), at least regarding Web information sources. It is possible that personal capital does not necessarily increase consumer self-confidence in dealing with a variety of online health information. Shon et al. (2000) reported that even highly educated people had trouble identifying misleading health information. Consequently, individuals with higher education and those with lower education may have difficulty judging online health information credibility and homogeneously trust online health information. A post hoc analysis of the data revealed that 30.4% of those with at least some college education reported that it was “hard” or “very hard” to understand medical statistics and that 49.5% of the respondents with high school education or less reported the same. Therefore, organizations responsible for health Web sites need to ensure that information can be readily located and presented in ways that can be easily understood by the general public. Another explanation may reside in the risks associated with trust. Trust is always accompanied by risks and is a solution for risks (Luhmann, 1988). Hence, it is conceivable that when it comes to health information, people are highly motivated to scrutinize it, and the judgment of online health information credibility is associated less with personal health status but more with information characteristics. Although this contradicts Menon and colleagues' (2004) argument that better health makes people more willing to trust online drug information, it supports Huh and colleagues' (2005) finding on the nonsignificant correlation between health status and trust in online drug information.

Regarding social capital and trust, this study demonstrates that visiting social networking Web sites is not significantly related to trust in online health information. One explanation is that in a social networking Web site, users not only communicate with people from their offline social network, but they also meet new people online (Boyd & Ellison, 2007). Hence, a social networking Web site usually encompasses a group of people with diverse backgrounds and interests so that ties among members may be weak, which is less conducive to trust. Although Internet users tend to have larger and richer social relationships, the virtual bonds are not sufficiently strong, substantial, or sustaining (for a review, see Hlebec, Lozar, & Vehovar, 2006). Donath and Boyd (2004) observed that participation in social networking Web sites often nurtured the formation and maintenance of weak ties among users and promoted bridging social capital (i.e., the social relationship among heterogeneous groups of people; Ellison et al., 2007) rather than bonding social capital (i.e., the networks among homogeneous groups of people wherein close and strong network ties exist; Gittell & Vidal, 1998; Pfeil, Arjan, & Zaphiris, 2009).

Another explanation may reside in the target of trust. The theoretical perspective on social capital and trust centers on interpersonal trust, whereas in the present study I investigated trust in online health information. Visiting a social network Web site is positively related to social capital and trust in others (Ellison et al., 2007; Steinfield et al., 2008), but it may not translate to trust in the Internet as a source of health information. For interpersonal trust, the trustee acknowledges and accepts the credit of trust and therefore feels compelled to have a moral obligation to meet it; in contrast, the Internet is a technology-based medium, and there is no moral obligation involved. Hence, the risk accompanied by trust in online health information is greater than that associated with placing trust in the actions of others, and people are less likely to place trust in the Internet as a source of health information. This results in a weaker correlation between visiting social network Web sites and trust in online health information than between visiting social network Web sites and trust in others.

Regarding trust transfer, this study shows that trust in online health information is associated with trust in health information from both traditional mass media and government health agencies. These results are consistent with previous studies on trust in direct-to-consumer advertising (Huh et al., 2005) and add to the literature by demonstrating a positive association between trust in health information from government health agencies and trust in online health information. The finding that trust in health information from interpersonal sources does not predict trust in online health information also merits discussion. This is probably due to the lack of similarity between interpersonal communication channels and the Internet. Trust transferability depends partly on similarity between trust targets (Kim, 2008; Stewart, 2003). Limited connection exists among such interpersonal sources as doctors, family and friends, and the Internet, because the former are related to interpersonal communication and the latter is associated with media communication. By contrast, the trust transfer from traditional mass media to the Internet occurs because of similarities between the two. Health information from traditional mass media is often closely related to that offered on the Web, because most mass media cover health issues and make that information available on their Web sites. Just like e-commerce consumers adopt a bank's online service because of the bank's offline reputation (Lee, Kang, & McKnight, 2007), health consumers also tend to trust health information on a government health agency's Web site because of the established reputation and credibility of the agency (Eysenbach, 2001).

Meanwhile, this study shows generational effects in trust transfer. First, none of the transfers was observed for the age group 65 years or older. This may reflect that elderly adults have growing health concerns and their trust in online health information has more to do with such factors as health beliefs and health-information orientation than with the factors investigated in the present study. Among people age 55 years or older, both computer use and health attitudes influenced their evaluation of online health information (McMillan & Macias, 2008). Dutta-Bergman (2003) reported the significance of health orientation in trust in a particular online source. Second, trust transference from newspapers/magazines to the Internet did not occur for the age group 18 to 34 years. This may be explained by less frequent use of print media among young people (R. Coleman & McCombs, 2007). Last, the transfer from government health agencies to the Internet did not occur for the age group 35 to 49 years. This age group searches for information online more frequently than do their older and younger adult peers (Ybarra & Suman, 2006); and because of more experience with online health information, they may be less likely to perceive similarity between government health agencies and general health Web sites, and the transfer between the two is less likely to happen.

Regarding information features and trust in online health information, both easiness to locate and understandability of health information are positively related to trust in online health information. These results enrich the limited literature on the effect of information characteristics on online health message trustworthiness (Rains & Karmikel, 2009; Walther et al., 2004; Wang et al., 2008). The ability to retrieve relevant health information (U.S. Department of Health & Human Services, 2000; U.S. Institute of Medicine, 2004) is one of the health literacy skills that is critical for health communication and management tasks. Hence, online health information providers should improve information retrieval with consumer-friendly tools and should also aim for health information that is easy to understand.

In the present study, I observed age differences in the relations between the two information features and trust in online health information. Both information features under consideration are significantly related to trust in online health information for the two age groups 35–49 years and 50–64 years. However, people age 18 to 34 years trust in online health information regardless of whether or not the information is hard to understand, and people age 65 years or older trust in online health information no matter how hard the information is to locate and to understand. This may be because young adults are generally healthier than other age groups, perceive less vulnerability to low-quality health information, and consequently, they are more willing to trust in online health information regardless of information quality. In contrast, elderly adults have more health concerns than the younger generations, and their evaluation of online health information, as mentioned previously, tends to correlate more with factors such as health orientation and health beliefs than with information features. This study has several limitations. First, the present study was based on a cross-sectional survey and could not assess the directionality of the various relations under consideration. Second, despite statistical procedures being used to adjust the underrepresentation of people who were young, male, and a minority, and those who had low education, the underrepresentation is still a limitation and has implications for the results. For example, only about 15% of the sample were young adults and, in part as a result of this, less than 20% of the participants visited social networking Web sites. Social networking Web sites are used predominantly by adolescents and young adults (Utz, 2009; Valenzuela et al., 2009). Hence, the underrepresentation may pose difficulty in finding a significant correlation between visiting social networking Web sites and trust in online health information. One related issue is that 30% of the responses to the question regarding visiting social networking Web sites were missing, despite sampling weights being used to adjust for nonresponse and to minimize estimation bias (Rao, Sigurdson, Doody, & Graubard, 2005).

Third, the response rates were low, which is not surprising, given that the response rates of most major American national surveys have been decreasing over decades (Baruch & Holtom, 2008; Krosnick, 1999). Although low response rates do not necessarily compromise results and are less of a concern for studies that test relations between multiple variables or aim for theory development (Keeter, Miller, Kohut, Groves, & Presser, 2000; Krosnick), any degree of nonresponse could bias results (Burkell, 2003). Therefore, the results reported here should be interpreted taking into consideration the low response rates.

Fourth, the number of Health Information National Trends Survey questions available for some of the domains studied in this study—particularly social capital and information characteristics—was limited. For example, in the present study, I used only one indicator of social capital (i.e., visiting social networking Web sites) to test the relation between social capital and trust in online health information.

Last, the measurement of trust in online health information merits discussion. The Health Information National Trends Survey focused on a broad category of trust (B. Hesse, personal communication, September 25, 2009) and did not differentiate various health Web sites and pages with varying levels of credibility. This is a limitation. Nevertheless, although the credibility of health Web sites and pages may depend on site genres (Flanagin & Metzger, 2007; Iding, Crosby, Auernheimer, & Klemm, 2009), factors such as site design can counteract the effect of Web site genre on site credibility (Flanagin & Metzger). Meanwhile, a number of studies also measured consumers' global trust in various sources of health information, such as doctors, television, newspapers and magazines, relatives and friends, and the Internet (Khoo, Bolt, Babl, Jury, & Goldman, 2008; Musa, Schulz, Harris, Silverman, & Thomas, 2009; Narhi, 2007). This practice appears to be acceptable: Several Health Information National Trends Survey studies that use such measures have been published (Hesse et al., 2005; Hong, 2008; Ling, Klein, & Dang, 2006; McQueen, Vernon, Meissner, Klabunde, & Rakowski, 2006; Rains, 2007; Roach et al., 2009; Rutten, Augustson, Doran, Moser, & Hesse, 2009), and some of them have been highly cited (Hesse et al.; McQueen et al.; Ramanadhan & Viswanath, 2006).

In conclusion, this study expands current research on consumer trust in health information and finds that trust in online health information is correlated more with information features than with receiver characteristics, which is consistent with previous research (Zahedi & Song, 2008). Further, it reveals that trust transfers from both traditional mass media and government health agencies to the Internet. Last, it shows that age is a key factor in understanding the correlates of trust in online health information. More research is needed in order to explore correlates of trust in online health information among seniors, because this study shows that none of the factors investigated is significant. This study also calls for more research to test the relation between social capital and trust in online health information to advance the theory on social capital and trust. Meanwhile, research in this area should obtain a sample with more young adults to investigate how social capital indicators such as visiting social networking Web sites and participation in online support groups are related to trust in online health information.

Notes

^a All degrees of freedom are 1.

*p < .05. **p < .01. ***p < .001.