Professional Documents
Culture Documents
1520 | Research and Practice | Peer Reviewed | Weaver et al. American Journal of Public Health | August 2010, Vol 100, No. 8
RESEARCH AND PRACTICE
a panel of more than 60000 participants main- accounted for 62% of the variance. Varimax measured by the BRFSS questions37 ‘‘Do you
tained by e-Rewards Marketing Research rotation to simple structures revealed that the now smoke cigarettes every day, some days, or
(www.erewardsresearch.com).31 This panel was first factor, labeled illness information–seeking not at all?’’ and ‘‘During the past month, other
chosen, in part, because it was constructed and is behavior (eigenvalue = 2.50), was defined by than your regular job, did you participate in any
maintained with a ‘‘by-invitation-only’’ recruit- high factor loadings (correlation coefficients physical activities or exercises such as running,
ment methodology designed to facilitate effective between the variables and the direct factor calisthenics, golf, gardening, or walking for exer-
demographic normalization while reducing the indices) on the illness or disease (0.79), medi- cise?’’ Dichotomous variables were computed
potential for self-selection bias.32 cations (0.86), and treatments (0.78) items. for each, with the ‘‘not at all’’ or ‘‘no’’ responses
A sample of panel members (n = 3140) were Factor 2, labeled wellness information–seeking set as 0 and all other responses as 1. Respondents
solicited via e-mail by e-Rewards to complete behavior (eigenvalue =1.22), was defined by were also asked ‘‘Do you purchase prescription
a consented survey, linked via a separate secure high factor loadings on the exercise (0.88), medications?’’ as a proximal measure of any
server, about their self-perceptions and media diet (0.82), and parenting (0.40) items. ongoing medical condition requiring pharma-
use. e-Rewards managed a stratified, multiwave Dichotomous factor indices were created by ceutical therapy. Responses were coded as
invitation procedure with oversampling of less summing the variables defining each factor and ‘‘yes’’ (1) or ‘‘no’’ (0).
responsive demographic groups designed to then coding all sums greater than zero as in-
maximize sample representativeness. This pro- dicative of the illness or wellness information– Covariates
cess yielded an American Association for Public seeking behavior and all those equal to zero as Building from previous research,21,39,40 we
Opinion Research adjusted response rate33 of not. An HISB measure with 4 discrete levels was examined 4 measures as covariates: respon-
23.69%, which is comparable to those of other created, and respondents were classified as no dent’s age, household income, perceptions of
single-invitation online surveys,34,35 and met the use (50.6%), illness only (14.1%), wellness only health information importance, and overall
survey quota in approximately 48 hours. (15.2%), and illness and wellness combined Internet use. Respondent’s age (mean = 44.7
Responses were analyzed from 559 adults, (20.0%) for subsequent analyses. years; SD =13.1) was assessed with the ques-
most of whom were between 35 and 54 years tion, ‘‘What is the YEAR of your birth?’’
old (48.5%), female (51.9%), White (82.2%), Health Indicators Household income (mean = $86 829.84;
married (62.9%), college graduates (63.7%), Six questions adopted from the 2006 Be- SD = $58194.34) was measured by asking
and reporting a household income of $75 000 havioral Risk Factor Surveillance Survey respondents to ‘‘please indicate your gross
or more (56.5%). These characteristics (BRFSS) instrument37 assessed respondents’ family income.’’ The perceived importance of
(Table 1) were consistent with those samples perceptions of their health. These included health information was assessed through re-
derived via random-digit-dialing sampling in health status (‘‘Would you say that in general spondents’ ratings of ‘‘how important’’ they
the Seattle–Tacoma DMA,30 with 2 exceptions: your health is excellent [5], very good [4], good considered ‘‘health information,’’ with re-
respondents reported both higher household [3], fair [2], or poor [1]?’’), diminished physical sponse options including ‘‘I don’t use’’ (0) and
income and educational achievement. These health (‘‘Now thinking about your physical a scale ranging from ‘‘not at all important’’
differences were anticipated, however, on the health, which includes physical illness and injury, (1) to ‘‘extremely important’’ (9). All respon-
basis of the demographic characteristics of for how many days during the past 30 days was dents attributed some importance to health
Internet users.36 your physical health not good?’’), diminished information, with responses ranging from 2
mental health (‘‘Now thinking about your mental to 9 (mean = 5.67; SD =1.96). Respondents
Health Information–Seeking Behaviors health, which includes stress, depression, and were also asked how much time they spent
HISBs were assessed by asking respondents problems with emotions, for how many days during a typical week ‘‘surfing the Internet’’
‘‘about how much time’’ they spent ‘‘during during the past 30 days was your mental health outside of work or school responsibilities to
a typical week’’ obtaining information online not good?’’), and poor quality of life (‘‘During assess overall Internet use (mean = 7.39 h/wk;
related to 6 content domains: diet, exercise, the past 30 days, for about how many days did SD = 8.77). The variables for household in-
illness or disease, medications, parenting, and poor physical or mental health keep you from come and Internet use were transformed with
treatments. The 6 items assessing information doing your usual activities, such as self-care, a base-10 logarithm to improve normality.
content domains were selected from a pool of work, or recreation?’’). Respondents were asked
several potential items following factor analysis their height and weight, and body mass index Statistical Analyses
of formative pilot-test data drawn from mem- was calculated by a standard formula.38 The The primary predictor variable for all
bers of the sampling frame. For each item, variables for diminished physical and mental analyses was HISB (no use, illness only,
responses were computed to reflect usage per health and poor quality of life were transformed wellness only, and illness and wellness com-
week in minutes and transformed with a base- with a base-10 logarithm to improve normality. bined). The 5 health indicators and 3 health
10 logarithm to improve normality. risk factors were the outcome variables. Bi-
Principal component factor analysis, Health Risk Factors variate analyses (Table 1) revealed that only 1
employed to isolate the primary HISB content Three health risk factors were assessed. demographic characteristic—respondent’s
dimensions, yielded a 2-factor solution that Smoking and physical activity behaviors were sex—varied significantly as a function of
August 2010, Vol 100, No. 8 | American Journal of Public Health Weaver et al. | Peer Reviewed | Research and Practice | 1521
RESEARCH AND PRACTICE
TABLE 1—Participant Characteristics as a Function of Online Health Information–Seeking Behavior (HISB): Seattle–Tacoma,
Washington, 2006
HISB
% of Estimated No Use Illness Wellness Combined
Seattle–Tacoma % of Estimated (n = 283), (n = 79), % or (n = 85), % or (n = 112), % or
Characteristic Population Sample % or OR (95% CI) OR (95% CI) OR (95% CI) OR (95% CI) c2 (df)
Note. Population estimates were compiled from the 2007 Scarborough Research Report, Release 2 Seattle–Bellevue–Everett Metropolitan Statistical Area.30
a
The population and sample distributions differed significantly across education (c22 = 128.49; P < .05) and household income (c23 = 15.57; P < .05).
b
For the ‘‘No Use’’ HISB category, n = 282. For the ‘‘Illness’’ HISB category, n = 78.
c
For the ‘‘Wellness’’ HISB category, n = 84.
d
Across race/ethnicity, small sample sizes among the African American, Asian American, and Other groups within the HISB cohorts threatened the validity of the c2 test. However, a follow-up test
with these 3 groups collapsed yielded the same statistical decision (c23 = 6.85; not statistically significant).
e
For the ‘‘No Use’’ HISB category, n = 280.
f
Referent categories of dummy-coded variables.
*P < .05, using Sidak’s adjustment.
HISB. In light of this finding and previous computed with the t test (linear models) or the substantial sex differences evident across the
research,4,11,21,39 sex was incorporated as Wald c2 test (logistic models). 4 HISB categories (Table 1).
a moderator of HISB in subsequent analyses. Table 2 displays the test statistics resulting
Consequently, the focal relationships between RESULTS from linear and logistic models examining the
HISB and the 8 health outcome measures were sex (female, male) by HISB (no use, illness only,
tested with multivariate linear (for the 5 health The sample comprised 559 adults, 49.4% of wellness only, and illness and wellness com-
indicators) and logistic (for the 3 health-risk whom reported HISB during a typical week. bined) factorial model, adjusted for the cova-
factors) regression models parameterized in More women (60.5%) than men (39.5%) riates. The HISB main effects were significant
a sex-by-HISB (2 · 4) factorial design with cova- reported HISB (odds ratio [OR] =1.99; 95% (P < .05) for 5 of the 8 health measures: health
riate adjustment. Post hoc comparisons were confidence interval [CI] =1.42, 2.79), with status, diminished physical health, poor quality
1522 | Research and Practice | Peer Reviewed | Weaver et al. American Journal of Public Health | August 2010, Vol 100, No. 8
RESEARCH AND PRACTICE
TABLE 3—Differences in Respondents’ Health Indicators and Health Risk Factors as Types of Health Information Sought
This report is one of the first to explore
a Function of Online Health Information–Seeking Behavior (HISB): Seattle–Tacoma,
differences in health perceptions and behaviors
Washington, 2006
among individuals seeking different types of
HISB online health information. We discovered
No Use (n = 283) Illness (n = 79) Wellness (n = 85) Combined (n = 112) considerable variability in health indicators
and health risk factors across the 4 levels of
Health indicator, mean (SE)
HISB, with prominent differences between in-
Health status 3.81y (0.53) 3.52x (0.10) 4.05z (0.10) 3.65x,y (0.09)
dividuals exclusively seeking wellness infor-
Diminished physical health 0.22x,y (0.02) 0.31y (0.05) 0.14x (0.05) 0.31y (0.04)
mation and those in the other 3 HISB groups.
Diminished mental health 0.26 (0.03) 0.33 (0.05) 0.28 (0.05) 0.34 (0.04)
Overall, wellness-information seekers reported
Poor quality of life 0.15y (0.02) 0.24z (0.04) 0.11x (0.04) 0.20y,z (0.03)
the most positive health indicators and the
Body mass index 27.15 (0.36) 27.19 (0.65) 27.04 (0.64) 27.77 (0.57)
lowest levels of health risks. The opposite
Health risk factors, %
pattern was evident for illness-information
Physical activitiesa 78.8x 82.3x 94.1y 86.6x,y
x y x seekers, who tended to report the most nega-
Prescription medications 63.6 83.5 57.7 80.4y
tive health indicators and more health risks.
Smoking 13.8 15.2 4.7 12.5
These differences suggest that distinct motiva-
Note. For the health indicators, least square means having different superscripts (x–z) differ at P < .05 by the 2-tailed t test. tions may underlie HISBs,27,40 a phenomenon
For health risk factors, percentages having different superscripts (x–z) differ at P < .05 by the Wald c2 test. Missing data on frequently overlooked in previous research.11
body mass index excluded 16 respondents (total respondents, n = 543; no use, n = 272; illness, n = 77; wellness, n = 84;
illness and wellness combined, n = 110). One key conclusion of this study, illustrated
a
Participation in ‘‘any physical activities or exercises’’ was assessed. by these trends, is that a substantial portion of
health information seekers (30.8%) perceive
August 2010, Vol 100, No. 8 | American Journal of Public Health Weaver et al. | Peer Reviewed | Research and Practice | 1523
RESEARCH AND PRACTICE
themselves as well and healthy and are using seeks health-related information on the linkages between the type of health informa-
online health information about wellness per- Internet’’41(p209)—or ‘‘incidental health infor- tion sought (e.g., wellness, illness, and so on)
haps, as others have suggested, ‘‘in a proactive mation use’’ (e.g., inadvertent or nonpurposive and several health indicators and health risk
manner for health promotion.’’23(p187) This use of health information).42 Instead, the data factors. As reported in previous research,
suggests that, with thoughtful formative research at hand suggest that further exploration of in- much HISB seems to be disease or illness
and effective targeting, the Internet could pro- dividual-level antecedents of HISB,27,40 particu- oriented. Actively seeking wellness informa-
vide an efficient channel for primary health larly the cognitive and affective components of tion, on the other hand, emerges as a unique
promotion and disease prevention activities, health anxiety,40,43–47 could prove a fruitful behavior among many health-conscious in-
encouraging many individuals engaged in HISB avenue for future research. dividuals. These findings demonstrate ‘‘that
to maintain a healthy lifestyle.8 Equally important, the fact that most consumers who are active health information
By contrast, the findings also show that (50.6%) of this Seattle–Tacoma sample seekers are not a monolithic group, and
many who engage in HISB pursue content reported no HISB raises critical questions should not be treated as one.’’27(p42) The
about medical concerns (e.g., illness or disease, about the viability of the Internet as a channel findings suggest that integration of a basic com-
medications, or treatments) and that these for health promotion and disease prevention munication-tailoring strategy, targeted toward
illness information seekers (28.6%) report less that require further consideration. The consis- specific health indicators and health risks, into
positive health indicators and more health tency of this observation across numerous Internet health communication endeavors
risks. This would suggest that, consistent with studies1,5,6,9,15,24 highlights the need for research could substantially improve the efficacy of the
the bulk of previous research, seeking health to extend our understanding of why so many Internet as a channel for primary, secondary,
information may be a common behavioral people apparently ignore easily accessible in- and tertiary prevention. j
response among those facing health challenges formation resources that could aid in maximizing
and that online health information may effec- their health. The current results offer few in-
tively bolster secondary and tertiary preven- sights, since several potential determinants (e.g.,
tion interventions. poor health literacy and media literacy, demo- About the Authors
James B. Weaver, III, Darren Mays, Stephanie Sargent
At the same time, the observation that graphics) seem unlikely contributors given the Weaver, Doğan Eroğlu, and Jay M. Bernhardt are with the
many health information seekers sought both characteristics of the sample (i.e., Internet users National Center for Health Marketing, Centers for Disease
illness and wellness information (40.6%; with elevated income and education levels). Control and Prevention, Atlanta, GA. Gary L. Hopkins is
with the Center for Media Impact Research, Institute for the
combined HISB group) was somewhat un- Clearly, if our goal is to convey health-promoting Prevention of Addictions, Andrews University, Berrien
expected and needs further investigation. One content to the broadest audience, further re- Springs, MI.
can quickly envision a number of circum- search examining the barriers and motives Correspondence should be sent to James B. Weaver,
III, Centers for Disease Control and Prevention, 1600
stances in which there may be coexisting influencing HISB is necessary to facilitate en- Clifton Road MS-E21, Atlanta, GA 30333 (e-mail:
needs for different types of health informa- gagement with those who are not actively seek- jim.weaver@cdc.gov). Reprints can be ordered at
tion. Therapies for prevalent chronic diseases ing health information. http://www.ajph.org by clicking on the ‘‘Reprints/
Eprints’’ link.
(e.g., cardiovascular disease, diabetes), for in- Although this study reveals new insights re- This article was accepted December 8, 2009.
stance, often involve both medically super- garding HISB, there are several limitations to
vised treatments and healthy-lifestyle behav- acknowledge. First, this is a cross-sectional study Contributors
ioral changes. Notably, however, there was and attribution of causality is not warranted. In James B. Weaver, III originated the study and supervised
essentially no variability in health indicators, addition, the sample characteristics (e.g., concen- all aspects of the preparation of the article. Darren Mays
and Stephanie Sargent Weaver assisted with data anal-
health risk factors, perceived importance of trated in western Washington State, drawn from ysis and interpretation and assumed a primary role in
health information, overall Internet use, and an Internet-based panel) limit generalizability. writing the article. All other authors conceptualized ideas,
demographic characteristics between the Finally, although informed by formative research interpreted results, and reviewed drafts of the article.
illness-only and combined groups. and more detailed than most previous research,11
the 6 health information–seeking domains Acknowledgments
This research was supported in part by a grant from the
Possible Motives for Health Information- examined here were not comprehensive and left Center for Media Impact Research in the Institute for
Seeking Behavior considerable room for respondents’ interpreta- Prevention of Addictions at Andrews University and by
Taken together, the direct linkages between tion. Future studies of HISB can substantially appointments of D. Mays and S. S. Weaver to the Re-
search Participation Program at the Centers for Disease
health assessments and the type of health advance our understanding by refining measures Control and Prevention (CDC) administered by the Oak
information sought online observed in this to capture in greater detail the characteristics Ridge Institute for Science and Education through an
study suggests that different health-related of respondents’ choices and preferences regard- interagency agreement between the US Department of
Energy and the CDC.
motives may underlie each type of HISB. This ing health information. We are indebted to Richard E. Dixon, Wendi
observation appears inconsistent with expla- In conclusion, the findings reveal distinct Kannenberg, Duane C. McBride, John V. Stevens Jr,
nations for Internet HISB built on concepts patterns in the health information content and the anonymous reviewers for their significant con-
tributions to this project.
such as the ‘‘cyberchondriac’’1—a word sought by adult Internet users in western Note. The findings and conclusions in this article are
frequently ‘‘used to describe anyone who Washington State and highlight unique those of the authors and do not necessarily represent the
1524 | Research and Practice | Peer Reviewed | Weaver et al. American Journal of Public Health | August 2010, Vol 100, No. 8
RESEARCH AND PRACTICE
views of the CDC or the US Department of Health and 15. Baker L, Wagner TH, Singer S, Bundorf MK. Use of e-rewardsresearch.com/eri_m_methRecruit.htm.
Human Services. the Internet and e-mail for health care information: Accessed August 25, 2009.
results from a national survey. JAMA. 2003;289(18): 33. Standard Definitions: Final Dispositions of Case Codes
2400–2406. and Outcome Rates for Surveys. 5th ed. Lenexa, KS:
Human Participant Protection
The institutional review boards of Andrews University, 16. Bundorf MK, Singer S, Wagner TH, Baker L. American Association for Public Opinion Research;
Emory University, Loma Linda University, and Virginia Consumers’ use of the Internet for health insurance. Am 2008. Available at: http://www.aapor.org/uploads/
Polytechnic Institute and State University approved the J Manag Care. 2004;10(9):609–616. Standard_Definitions_04_08_Final.pdf. Accessed
protocol for this study. August 10, 2009.
17. Goldner M. Using the Internet and email for health
purposes: the impact of health status. Soc Sci Q. 2006; 34. Ekman A, Litton JE. New times, new needs; e-
87(3):690–710. epidemiology. Eur J Epidemiol. 2007;22(5):285–292.
References
1. Harris Interactive Inc. Harris Poll shows number of 18. Goldner M. How health status impacts the types of 35. Gosling SD, Vazire S, Srivastava S, John OP. Should
information consumers seek online. Inf Commun Soc. we trust Web-based studies? A comparative analysis of
‘‘cyberchondriacs’’—adults who have ever gone online for
2006;9(6):693–713. six preconceptions about Internet questionnaires. Am
health information—increases to an estimated 160 mil-
Psychol. 2004;59(2):93–104.
lion nationwide. July 31, 2007. Available at: http:// 19. Houston TK, Allison JJ. Users of Internet health
www.harrisinteractive.com/vault/Harris-Interactive-Poll- information: difference by health status. J Am Med Inform 36. Pew Internet and American Life Project. February
Research-Cyberchondriacs-2007-07.pdf. Accessed Assoc. 2002;4:e7. 15–March 7, 2007 tracking survey. 2008. Available
August 4, 2008. at: http://www.pewinternet.org/;/media/Files/
20. Renahy E, Raizot I, Chaivin P. Health information
Questionnaire/Old/PIAL%20Feb07%20Tracking
2. Online Activities: Total. Washington, DC: Pew Internet seeking on the Internet: a double divide? Results from
%20FINAL%20Topline%20Excerpt.pdf. Accessed
and American Life Project; 2008. Available at: http:// a representative survey in the Paris metropolitan area,
June 27, 2008.
www.pewinternet.org/Trend-Data/Online-Activites-Total. France, 2005–2006. BMC Public Health. 2008;8:69.
aspx. Accessed October 9, 2008. 37. 2006 Behavioral Risk Factor Surveillance System
21. Rice RE. Influences, usage, and outcomes of Internet
Questionnaire. Atlanta, GA: Centers for Disease Control
3. Anderson JG. Consumers of e-health: patterns of health information searching: multivariate results from
and Prevention; 2006.
use and barriers. Soc Sci Comput Rev. 2004;22(2):242– the Pew surveys. Int J Med Inform. 2006;75(1):8–28.
248. 38. BMI-Body Mass Index: How Is BMI Calculated and
22. Lambert SD, Loiselle CG. Health information seek- Interpreted? Atlanta, GA: Centers for Disease Control and
4. Elkin N. How America Searches: Health and Wellness. ing behavior. Qual Health Res. 2007;17(8):1006–1019. Prevention; 2008. Available at: http://www.cdc.gov/
Scottsdale, AZ: iCrossing; 2008. Available at: http://
23. Pandey SK, Hart JJ, Tiwary S. Women’s health and healthyweight/assessing/bmi/adult_bmi/index.html.
www.icrossing.com/research/how-america-searches-
the Internet: understanding emerging trends and impli- Accessed June 27, 2008.
health-and-wellness.php. Accessed August 1, 2008.
cations. Soc Sci Med. 2003;56(1):179–191. 39. Drentea P, Goldner M, Cotton S, Hale T. The
5. Fox S. Health information online. Pew Internet and association among gender, computer use, and online
24. Dutta-Bergman MJ. Health attitudes, health cogni-
American Life Project, 2005. Available at: http:// health searching, and mental health. Inf Commun Soc.
tions, and health behaviors among Internet health in-
www.pewinternet.org/Reports/2005/Health- 2008;11(4):509–525.
formation seekers: population-based survey. J Med Inter-
Information-Online.aspx. Accessed June 11, 2008.
net Res. 2004;6(2):e15. 40. Lemire M, Pare G, Sicotte C, Harvey C. Determinants
6. Fox S, Fallows D. Internet health resources. Pew of Internet use as a preferred source of information
25. Dutta-Bergman MJ. Developing a profile of con-
Internet and American Life Project, 2003. Available at: on personal health. Int J Med Inform. 2008;77(11):
sumer intention to seek out additional information
http://www.pewinternet.org/Reports/2003/Internet- 723–734.
beyond a doctor: the role of communicative and moti-
Health-Resources.aspx. Accessed June 11, 2008.
vation variables. Health Commun. 2005;17(1):1–16. 41. Smith PK, Davies P, Hamidi-Manesh L. Cyberchon-
7. Tu HT, Cohen G. Striking Jump in Consumers Seeking driacs. Int J Adolesc Med Health. 2006;18(2):209–213.
26. Verheijden MW, Jans MP, Hildebrandt VH, et al.
Health Care Information. Washington, DC: Center for
Rates and determinants of repeated participation in 42. Tian Y, Robinson JD. Incidental health information
Studying Health Systems Change; 2008. Available at:
a Web-based behavior change program for healthy body use on the Internet. Health Commun. 2009;24(1):41–49.
http://hschange.org/CONTENT/1006. Accessed August
weight and healthy lifestyle. J Med Internet Res. 2007;
21, 2008. 43. Eastin MS, Guinsler NM. Worried and wired: effects
9(1):e1.
of health anxiety on information-seeking and health care
8. Lintonen TP, Konu AI, Seedhouse D. Information
27. Navarro FH, Wilkins ST. A new perspective on utilization behaviors. Cyberpsychol Behav. 2006;9(4):
technology in health promotion. Health Educ Res. 2008;
consumer health Web use: ‘‘valuegraphic’’ profiles of 494–498.
23(3):560–566.
health information seekers. Manag Care Q. 2001;9(2):
44. Hadjistavropoulos HD, Craig KD, Hadjistavropoulos
9. Ayers SL, Kronenfeld JJ. Chronic illness and health- 35–43.
T. Cognitive and behavioral responses to illness infor-
seeking information on the Internet. Health (London).
28. Fox S, Jones S. Americans’ pursuit of health takes mation: the role of health anxiety. Behav Res Ther.
2007;11(3):327–347.
place within a widening network of both online and 1998;36(2):149–164.
10. Huang GJ, Penson DF. Internet health resources and offline sources. Pew Internet and American Life Project,
45. Lee SY, Hwang H, Hawkins R, Pingree S. Interplay of
the cancer patient. Cancer Invest. 2008;26(2):202–207. 2009. Available at: http://pewinternet.org/Reports/
negative emotion and health self-efficacy on the use of
11. Weaver JB III, Mays D, Lindner G, Eroglu D, 2009/8-The-Social-Life-of-Health-Information.aspx.
health information and its outcomes. Communic Res.
Fridinger F, Bernhardt JM. Profiling characteristics of Accessed July 31, 2009.
2008;35(3):358–381.
internet medical information users. J Am Med Inform 29. Sillence E, Briggs P, Harris PR, Fishwick L. How do 46. Sonnenberg FA. Health information on the Inter-
Assoc. 2009;16(5):714–722. patients evaluate and make use of online health infor- net—opportunities and pitfalls. Arch Intern Med. 1997;
12. Josefsson U, Murero M, Rice RE. Online information- mation? Soc Sci Med. 2007;64(9):1853–1862. 157(2):151–152.
seeking behavior. In: Murero M, Rice RE, eds. The Internet 30. Seattle Times Company. Seattle market overview. 47. Turner MM, Rimal RN, Morrison D, Kim H. The role
and Health Care: Theory, Research, and Practice. Mahwah, 2007. Available at: http://www.seattletimescompany. of anxiety in seeking and retaining risk information:
NJ: Lawrence Erlbaum Associates; 2006:127–147. com/advertise/pdfs/smo_spring07.pdf. Accessed July testing the risk perception attitude framework in two
13. Morahan-Martin JM. How Internet users find, eval- 14, 2007. studies. Hum Commun Res. 2006;32(2):130–156.
uate, and use online health information: a cross-cultural 31. e-Rewards Marketing Research. e-Rewards US
review. Cyberpsychol Behav. 2004;7(5):497–510. consumer panel. 2008. Available at: http://www.e-
14. Schwartz KL, Roe T, Northrup J, Meza J, Seifeldin R, rewardsresearch.com/downloads/US_Consumer_
Neale AV. Family medicine patients’ use of the Internet Panels.pdf. Accessed September 23, 2008.
for health information: a MetroNet study. J Am Board 32. e-Rewards Marketing Research. e-Rewards meth-
Fam Med. 2006;19(1):39–45. odology: recruitment. 2009. Available at: http://www.
August 2010, Vol 100, No. 8 | American Journal of Public Health Weaver et al. | Peer Reviewed | Research and Practice | 1525