This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Online health information-seeking behavior (OHISB) is currently a widespread and common behavior that has been described as an important prerequisite of empowerment and health literacy. Although demographic factors such as socioeconomic status (SES), age, and gender have been identified as important determinants of OHISB, research is limited regarding the gender-specific motivational determinants of OHISB and differences between women and men in the use of online resources for health information purposes.
The aim of this study was to identify gender-specific determinants and patterns of OHISB by analyzing data from a representative German sample of adults (N=1728) with special attention to access and frequency of use as well as topics and sources of OHISB.
We employed a 2-step analysis, that is, after exploring differences between users and nonusers of online health information using logistic regression models, we highlighted gender-specific determinants of the frequency of OHISB by applying zero-truncated negative binomial models.
Age (odds ratio, OR for females=0.97, 95% CI 0.96-0.99) and degree of satisfaction with one’s general practitioner (GP) (OR for males=0.73, 95% CI 0.57-0.92) were gender-specific determinants of access to OHISB. Regarding the frequency of OHISB, daily Internet use (incidence rate ratio, IRR=1.67, 95% CI 1.19-2.33) and a strong interest in health topics (IRR=1.45, 95% CI 1.19-1.77) were revealed to be more important predictors than SES (IRR for high SES=1.25, 95% CI 0.91-1.73).
Users indicate that the Internet seems to be capable of providing a valuable source of informational support and patient empowerment. Increasing the potential value of the Internet as a source for health literacy and patient empowerment requires need-oriented and gender-specific health communication efforts, media, and information strategies.
Patients today are increasingly challenged to make informed choices regarding their health care and to play an active role in health-related decisions [
The Internet represents an increasingly important source of health information [
In light of this clear need—and in parallel to the discourse on the digital divide [
It has been argued that gender differences in OHISB might be concealed by differing motives for seeking health information: Whereas women are more interested in health issues and emotional support, men are more interested in informational support [
Our aim was therefore to understand gender-specific determinants and patterns of OHISB. This understanding will allow us to gain insight into gender-specific preferences regarding content and sources, and to draw conclusions regarding gender-specific targeting strategies for the development of health-related online media. To date, no representative data on gender-specific OHISB for Germany has been analyzed using multivariate statistics [
The models that are frequently used to explain health information seeking—such as the theory of planned behavior (TPB [
Although research has shown that females are more likely to conduct HISB than males, integration of this finding into theory is still lacking [
To date, there has been only minor exploration as to if and how these differences in motives and channel usage are also relevant for online health information seeking. Gaining a better understanding of gender differences in HISB would help health communication scholars to develop gender-specific health communication interventions. Regarding such considerations, our analysis may contribute to the iterative junction of theoretical approaches on HISB and gender differences.
Today, the vast majority of the population across Europe and North America has access to the Internet [
Regarding gender differences, findings are somewhat inconsistent, that is, no significant differences in general Internet use have been detected in the United States [
The Internet’s already-prominent role in HISB continues to increase, that is, in the United States, 59% of the adult population (ie, more than 72% of adult Internet users) seeks online information concerning health topics [
Gender differences have not only been reported for general Internet use, but also for general health-related behaviors and outcomes, with men having higher mortality and morbidity rates, engaging in more risky behaviors (eg, smoking, alcohol abuse), and taking part in fewer health-promoting behaviors than women [
With regard to gender-specific HISB, many studies show that women are more engaged in health information seeking in general, as well as on the Internet, specifically. Being female is among the strongest predictors of conducting OHISB [
Our main goal was to analyze gender-specific determinants and patterns of OHISB. Our first question was which sociodemographic—including gender—and health-related user characteristics explain general utilization of health information on the Internet (RQ1). The second question was whether—among those who use the Internet for health-related purposes—the same factors determine the frequency of OHISB; to address this question, we ask which sociodemographic and health-related user characteristics—in relation to gender—explain the frequency of OHISB (RQ2).
Among the health-onliners, we are also interested in the gender-specific health-related topics they are most interested in and the online media they prefer to use as sources of health information (RQ3).
Data were taken from the Bertelsmann Health Care Monitor 2015, a representative national German health survey (available as open access files) conducted by the Bertelsmann Foundation in cooperation with the Barmer GEK, a statutory health insurance (see [
The excluded Internet nonusers showed statistically significant differences for several demographic criteria: They were much older (mean 64.7 years, SD 12.1) than the Internet users (mean 46.0, SD 15.0) with a higher proportion of female respondents (59.9% [299/499] vs 50.04% [610/1219]) and lower SES (31.2% [150/481] vs 15.86% [177/1116]). These findings confirmed prior research concerning demographic determinants of general Internet use [
Our main analyses were based on participant responses to the question “How many times did you use the Internet for seeking health information within the last 12 months?” Answers ranged from 0 to 130 with a mean of 4.37 (SD 9.44); answers were strongly right skewed (skewness=5.85, SE 0.07). To address our first analytical goal of uncovering the gender-specific determinants of utilization of OHISB, we created a dummy variable to separate health-offliners (OHISB=0) from health-onliners (OHISB≥1). To meet our second goal of assessing gender-specific determinants of the frequency of OHISB, we left the responses on their original scale but excluded the health-offliners from the analysis, as they showed no variance in their HISB frequency. This resulted in a final sample of 643 health-onliners.
The third objective—assessing gender differences in health-related topics and information sources—was achieved by analyzing the frequencies of the health-related topics and websites the health-onliners used. Respondents were asked to select the topics on which they searched for or received information from a list of 14 items. These items ranged from very specific (eg, “drugs and their pharmacological interactions”) to more general (eg, “fitness, well-being”) topics. These items were then grouped into three categories by content: “disease and health care,” “health care policy and health care system,” and “health and well-being.”
Respondents were then asked to select the sources they used when conducting OHISB, that is, they were given the 10 items to choose from popular sources (eg, “online dictionary”) and more specific sources (eg, “websites of noncommercial health organizations”).
Participants were asked to provide their age in years and gender (female or male), whereas SES was assessed by summing up participants’ responses on their education, occupation, and income (weighted by household size) to a score ranging between 3 and 27 following the standard procedure for the Health Care Monitor [
The frequency of general Internet usage was measured using a 3-point ordinal scale of “at least sometimes per month,” “several times per week,” or “daily.”
Patient status was measured using a 4-point scale ranging from 1 (“currently not affected”) to 4 (“chronically ill”). We classified the responses from “mildly or not affected” to “severely or chronically ill,” because OHISB patterns of healthy and mildly affected respondents should be quite similar, whereas severely or chronically affected people were expected to show fundamentally different patterns. Participants’ perceived relevance of understanding somatic processes, their health-consciousness, and the satisfaction with their GP were all measured using 5-point scales, that is, to measure health-consciousness, participants were asked how much attention they generally paid to their health, which they rated from 1 (“Generally, I don’t take care of my health”) to 5 (“Generally, I take good care of my health”). Satisfaction with their GP was scored from 1 (“very dissatisfied”) to 5 (“very satisfied”), their perceived relevance of understanding somatic processes was assessed by their degree of agreement—from 1 (“totally disagree”) to 5 (“totally agree”)—concerning the statement that “patients diagnosed with an illness should understand exactly what is going on.”
The extent to which respondents reported being interested in information concerning health topics in general was originally measured on a 3-point scale indicating weak, medium, or strong interest. We transferred these answers into a dummy variable, contrasting “low or medium level of interest” with “high level of interest” to create reasonably equal group sizes (n=739 and n=450, respectively). Looking at health-onliners only, their motivations to conduct HISB were assessed using 12 dummy indicators covering a broad range of potential goals (eg, “to find general health information about health risks and diseases” or “determining the best treatment options”). On the basis of social support theory [
Items are given in
To answer RQ1, a logistic regression model was conducted to analyze the influence of sociodemographic, motivational, and health-related factors on differences between health-onliners and health-offliners. Regarding RQ2, Poisson regression models are traditionally used to model data like the frequency of OHISB, as such models are suited to fulfilling the technical needs of an outcome consisting of positive integers. However, the application of Poisson models requires a data structure that is seldom found in reality, that is, the mean is equal to the variance [
Among all of the 1219 participants who used the Internet, 643 (52.75%;
Sample characteristics of health-onliners and health-offliners.
Variable | Total sample | Health-onliners | Health-offliners | Difference onliners versus offliners | |||
n=1219 | n=643 | n=576 | Degree of freedom | ||||
Range | 18-79 | 18-79 | 18-78 | ||||
Mean (SDc) | 46.02 (14.96) | 45.35 (14.55) | 46.78 (15.38) | 2.812a | 1 | .09 | |
Female | 610 (50.04) | 338 (52.6) | 272 (47.2) | 3.5b | 1 | .06 | |
Male | 609 (49.96) | 305 (47.4) | 304 (52.8) | ||||
Low | 177 (15.86) | 79 (13.4) | 98 (18.6) | ||||
Medium | 658 (58.96) | 339 (57.5) | 319 (60.6) | 13.1b | 2 | .001 | |
High | 281 (25.18) | 172 (29.2) | 109 (20.7) | ||||
At least sometimes per month | 189 (15.50) | 68 (10.6) | 121 (21) | ||||
Several times per week | 319 (26.17) | 155 (24.1) | 164 (28.5) | 34.9b | 2 | <.001 | |
Daily | 711 (58.33) | 420 (65.3) | 291 (50.5) | ||||
Chronically or severely ill | 243 (20.05) | 141 (22.1) | 102 (17.8) | 3.4b | .06 | ||
Mildly or not affected | 969 (79.95) | 498 (77.9) | 471 (82.2) | 1 | |||
Perceived relevance of understanding somatic processesd, mean (SD) | 4.25 (0.86) | 4.34 (0.83) | 4.16 (0.90) | 13.012a | 1 | <.001 | |
Low or medium | 739 (62.15) | 336 (53.8) | 403 (71.5) | 39.5b | 1 | <.001 | |
Strong | 450 (37.85) | 289 (46.2) | 161 (28.5) | ||||
Esteem support | 0.24 (0.26) | 0.24 (0.26) | - | N/Ag | |||
Emotional support | 0.06 (0.20) | 0.06 (0.20) | - | N/A | |||
Informational support | 0.37 (0.23) | 0.37 (0.23) | - | N/A | |||
Health-consciousnessh, mean (SD) | 3.47 (0.77) | 3.59 (0.71) | 3.34 (0.81) | 33.431a | 1 | <.001 | |
Satisfaction with general practitioneri, mean (SD) | 4.07 (0.86) | 4.02 (0.86) | 4.13 (0.86) | 4.492 | 1 | .03 |
a
bChi-square values derived from chi-square test for shares.
cSD: standard deviation.
dScale ranges from 1 (“strongly disagree”) to 5 (“strongly agree”).
eHISB: health information-seeking behavior.
fScale ranges from 0 (“no” for all items of the scale) to 1 (“yes” for all items of the scale).
gN/A: not applicable.
hScale ranges from 1 (“Generally, I don’t take care of my health”) to 5 (“Generally, I take good care of my health”).
iScale ranges from 1 (“very unsatisfied”) to 5 (“very satisfied”).
The results of the logistic regression models are depicted in
A higher frequency of general Internet use was associated with a nearly triple-increase in the odds of being a health-onliner (OR for “daily” use=2.91, 95% CI 1.92-4.41), with the slightly stronger effects for female (OR 3.23, 95% CI 1.86-5.59) than for male respondents (OR 2.50, 95% CI 1.30-4.78).
Persons who were chronically ill or severely affected by health problems were significantly more likely to be health-onliners, but only if they were women (OR 2.12, 95% CI 1.28-3.53). A similar relationship was found between perceived relevance of understanding somatic processes and HISB, that is, for women, a one-point increase in the perceived importance of health literacy was associated with an OR 1.39 (95% CI 1.09-1.78) of being a health-onliner, whereas men had only a moderately heightened OR that did not reach significance. Although both male and female respondents appeared to be significantly influenced by having general interest in information on health topics, this impact was much stronger among female participants (ORwomen 2.07, 95% CI 1.36-3.14; ORmen 1.70, 95% CI 1.09-2.63, respectively).
Degree of health-consciousness was associated with significantly increased OR for men (OR 1.46, 95% CI 1.10-1.94) and for the combined model (OR 1.33, 95% CI 1.10-1.61), but not for women alone. Higher satisfaction with one’s GP had a negative effect on the odds that men would seek health information online, that is, be health-onliners (OR 0.73, 95% CI 0.57-0.92).
The Hosmer-Lemeshow test inform on the proper classification of all cases included and gives a chi-square value of 11.5 (df=8;
Results of the logistic regression models predicting online health information-seeking behavior.
Determinants | Total (n=950)a | Male (n=463) | Female (n=487) | ||||
ORb (95% CI) | OR (95% CI) | OR (95% CI) | |||||
Age | 0.99 (0.98-1.00) | .01 | 0.99 (0.98-1.01) | .46 | 0.97 (0.96-0.99) | .002 | |
Gender |
1.21 (0.90-1.61) | .21 | -c | - | - | - | |
.01 | .02 | .57 | |||||
Medium | 1.13 (0.77-1.66) | .54 | 1.07 (0.62-1.86) | .80 | 1.16 (0.67-2.01) | .59 | |
High | 1.82 (1.15-2.88) | .01 | 1.97 (1.06-3.68) | .03 | 1.46 (0.72-2.93) | .29 | |
<.001 | .005 | <.001 | |||||
Several times per week | 1.57 (1.02-2.41) | .04 | 1.46 (0.72-2.94) | .29 | 1.70 (0.99-2.94) | .06 | |
Daily | 2.91 (1.92-4.41) | <.001 | 2.50 (1.30-4.78) | .006 | 3.23 (1.86-5.59) | <.001 | |
Patient status: chronically or severely ill |
1.56 (1.11-2.19) | .01 | 1.22 (0.76-1.95) | .42 | 2.12 (1.28-3.53) | .004 | |
Perceived relevance of understanding somatic processesd | 1.27 (1.08-1.50) | .005 | 1.22 (0.97-1.53) | .10 | 1.39 (1.09-1.78) | .008 | |
Strongly interested in information concerning health topics |
1.89 (1.40-2.54) | <.001 | 1.70 (1.09-2.63) | .02 | 2.07 (1.36-3.14) | .001 | |
Health-consciousnesse | 1.33 (1.10-1.61) | .004 | 1.46 (1.10-1.94) | .008 | 1.24 (0.95-1.62) | .11 | |
Satisfaction with general practitionerf | 0.82 (0.70-0.96) | .02 | 0.73 (0.57-0.92) | .008 | 0.91 (0.72-1.14) | .40 | |
Constant | 0.31 | .000 | 0.35 | .006 | 0.35 | .002 | |
Hosmer-Lemeshow test (chi-square, df; |
11.5, 8; .17 | 12.2, 8; .14 | 4.7, 8; .79 | ||||
Goodness of fitg (chi-square, df; |
116.3, 11; <.001 | 50.2, 10; <.001 | 75.5, 10; <.001 | ||||
Nagelkerke |
.154 | .137 | .192 |
aThe difference between the number of total cases included in the descriptive section and in the logit models is due to the listwise exclusion of missing cases.
bOR: odds ratio.
cThe dash indicates the absence of the variable “gender” in both gender-specific models.
d1 (“strongly disagree”) to 5 (“strongly agree”).
e1 (Generally, I don’t take care of my health” to 5 (“Generally, I take good care of my health”).
f1 (“very unsatisfied”) to 5 (“very satisfied”).
g (−2 Log L compared with −2 Log L of the empty model).
The results of the zero-truncated negative binomial regression models are shown as incidence rate ratio (IRR) in
Although the effects of higher frequencies of general Internet use are similar in size and
Being a patient with a chronic or severe disease was a positive predictor of OHISB frequency (IRR 1.57, 95% CI 1.26-1.95) regardless of gender, as the estimated IRRs do not differ substantially between men and women. Respondents who reported being strongly interested in information concerning health topics were much more likely to seek out information online more frequently (IRR 1.45, 95% CI 1.19-1.77), and gender played a much weaker role than health status.
Self-reported health-consciousness in the zero-truncated negative binomial regression models was—as compared with the findings from the logit models predicting the utilization of the Internet for health information purposes—not associated with significant effects. In contrast, perceived relevance of understanding somatic processes had the opposite effect on OHISB frequency of that predicted by the logit models: In the zero-truncated negative binomial regression models, belief in health literacy became a significant negative predictor, but for men only (IRR 0.81, 95% CI 0.68-0.96). Satisfaction with one’s GP changed from being a significant factor only for males to being a significant factor only for females, the latter now with a strong negative effect (IRR 0.75, 95% CI 0.65-0.88), whereas men’s frequency of OHISB seems to be statistically unrelated to their degree of satisfaction.
Some of the three sum indices representing different goals of OHISB showed strong explanatory potential: whereas
Results of the zero-truncated negative binomial regression models on the frequency of online health information-seeking behavior (OHISB).
Determinants | Total (n=510)a | Male (n=241) | Female (n=269) | ||||||||||
IRRb (95% CI) | IRR (95% CI) | IRR (95% CI) | |||||||||||
Age | 0.99 (0.982-0.998) | .01 | 1.00 (0.99-1.01) | .58 | 0.99 (0.975-0.996) | .009 | |||||||
Gender (Ref: male) | 0.99 (0.82-1.22) | .99 | -c | - | - | - | |||||||
Medium | 1.06 (0.79-1.41) | .71 | 1.01 (0.66-1.55) | .97 | 0.93 (0.63-1.36) | .70 | |||||||
High | 1.25 (0.91-1.73) | .17 | 1.14 (0.72-1.79) | .57 | 1.14 (0.72-1.81) | .56 | |||||||
Several times per week | 1.60 (1.12-2.27) | .009 | 1.72 (0.94-3.16) | .08 | 1.54 (1.01-2.35) | .04 | |||||||
Daily | 1.67 (1.19-2.33) | .003 | 2.49 (1.43-4.35) | .001 | 1.28 (0.84-1.96) | .25 | |||||||
Patient status: chronically or severely ill (Ref: mildly or not affected) | 1.57 (1.26-1.95) | <.001 | 1.67 (1.22-2.29) | .001 | 1.43 (1.07-1.91) | .02 | |||||||
Perceived relevance of understanding somatic processesd | 0.90 (0.80-1.01) | .06 | 0.81 (0.68-0.96) | .02 | 0.96 (0.82-1.13) | .65 | |||||||
Strongly interested in information concerning health topics (Ref: weekly or not interested) | 1.45 (1.19-1.77) | <.001 | 1.46 (1.09-1.97) | .01 | 1.42 (1.10-1.83) | .01 | |||||||
Esteem support | 1.91 (1.28-2.83) | .001 | 1.49 (0.82-2.72) | .19 | 2.22 (1.30-3.79) | .004 | |||||||
Emotional support | 0.90 (0.57-1.43) | .65 | 1.13 (0.53-2.40) | .75 | 0.83 (0.46-1.49) | .52 | |||||||
Informational support | 3.12 (1.97-4.96) | <.001 | 2.56 (1.34-4.90) | .004 | 4.03 (2.17-7.49) | <.001 | |||||||
Health-consciousnessf | 1.10 (0.97-1.26) | .14 | 1.70 (0.89-1.29) | .49 | 1.08 (0.91-1.30) | .38 | |||||||
Satisfaction with general practitionerg | 0.86 (0.77-0.96) | .007 | 1.09 (0.93-1.28) | .29 | 0.75 (0.65-0.88) | <.001 | |||||||
Constant | 0.47 | .04 | 0.32 | .36 | 0.65 | .01 |
aThe difference between the total number of cases included in the descriptive section and in the models depicted in this table is due to the listwise exclusion of missing cases.
bIRR: incidence rate ratio.
cThe - indicates the absence of the variable “gender” in both gender-specific models
dFrom 1 (“strongly disagree”) to 5 (“strongly agree”).
eOHISB: online health information-seeking behavior.
fFrom 1 (“Generally, I don’t take care of my health”) to 5 (“Generally, I take good care of my health”).
gFrom 1 (“very unsatisfied”) to 5 (“very satisfied”).
Online media offering the opportunity to share information and to interact with others, specifically, online health communities and social networking sites are not yet established as a means of OHISB in the broad public, with an overall usage of 17.0% and 9.7% (corresponding to 109/640 and 62/640 respondents), respectively. In some cases, we can detect significant differences in issue- and channel-related preferences between women and men, that is, in general, men focus more on topics concerning health care policy and systems (66.4%, 202/304 males vs 53.0%, 178/336 females;
Gender differences in online health information-seeking behavior (OHISB) concerning topics and sources of online communication. NHOs: Noncommercial health organizations (total n=640).
Despite the fact that men and women reported equal access to online health information, our data indicate that OHISB should be explained using gender-specific models, to account for several significant gender differences among health-onliners. Dissatisfaction with primary care seems to more often trigger women to seek patient esteem support through online health information seeking; OHISB might therefore serve a compensatory function. These and additional results—particularly regarding gender differences, implications, and methodical limitations—are discussed and compared with international data.
Our results indicate that SES and age remain relevant barriers to general access to health information on the Internet, but only for specific genders. We found increasing age to be significantly associated with both access to and frequency of OHISB for women only, thus enhancing understanding of the gender-specificity of the well-established negative correlation between age and OHISB [
Higher frequencies of general Internet use revealed to be consistently associated with more frequent OHISB [
The effect of being severely or chronically ill affected OHISB differently for different groups. Only severely ill women, not men, were significantly more likely to be health-onliners, consistent with findings from a French study [
The association between OHISB and related online activities and interests (eg, buying drugs and other health-related products online) that indicate a high level of interest in health information is neither surprising nor new, as this has been reported in both an analysis of cross-sectional data from 7 European countries [
We found that whereas women are inclined to engage in more frequent OHISB in light of their goals reflecting needs for esteem support and informational support, men tend to be driven more by purely informational motives. This is consistent with another recent finding that women were more likely than men to conduct OHISB for social motives and enjoyment [
Our results further indicate that using the Internet can serve a compensatory function, but in different ways for women and men. Whereas a lower satisfaction with one’s GP motivates men to turn to the Internet for health-related purposes (raw usage, independent of the frequency), a lower satisfaction with one’s GP is associated with an increased frequency of OHISB reported by women. These findings are in line with another study reporting that dissatisfied cancer patients seek health information from sources other than their physicians [
Our findings regarding sources employed for OHISB are partly consistent with a similar study: Females from our sample reported using health content-related websites significantly more often than males, which may reflect the stronger social supportive patterns detected among women [
The first limitation is that the cross-sectional data used in our analysis do not allow for any causal attributions, even in cases that seem straightforward, such as the effects of age or health status on OHISB.
The second limitation is that outcome operationalization was somewhat explorative, as a well-established, validated scale for assessing access to and frequency of OHISB does not yet exist. Development of a validated measure to assess OHISB is the central precondition of conducting internationally comparable research on this behavior. Such a measure would also complement the valid and reliable measure for assessing eHealth literacy (ie, the ability to seek, find, understand, and appraise health information from electronic sources and apply it to addressing or solving a health problem) that has already been developed [
Further limitations include, third, that no differentiation is made between people who are searching for information for themselves and those who are searching for others (“surrogate seekers”). Finally, we only used a binary categorization of men and women, which does not cover all facets of such a complex construct [
Our results provide promising and innovative insights into OHISB and indicate that a deeper understanding of OHISB requires differentiating between access to online health information (ie, differentiating between health-onliners and health-offliners) and the frequency of OHISB. This deeper understanding would be particularly valuable for the analysis of what are often subtle gender-based differences. Furthermore, sociodemographic, health-related, and motivational determinants of OHISB should be taken into account when explaining such complex behavior. This recommendation also applies to the associations between skills-related (ie, eHealth literacy) and behavior-related (OHISB) concepts, whose interrelations have yet to be analyzed sufficiently [
Overall, although users indicate that the Internet is capable of providing a valuable source of informational support and esteem support, gender-specific, user-oriented sources and empowerment-strategies should be developed to increase the benefits of OHISB. This may include enlisting the support of health care providers to supply patients with health information sources that offer evidence-based, transparent, and credible information. To close the gap in OHISB due to age and SES, such resources might, for example, reduce the complexity of the language and enhance the understandability of the health information offered. Gender-specific determinants and patterns in information-seeking behavior should also be taken into account in theories of health information seeking and in the provision of online health information by offering information in accordance with male and female preferences regarding goals, sources, and topics. For example, men’s technical affinity might be used as a pathway in health communication to raise their interest in health content about diseases and well-being [
Items.
general practitioner
Health Information National Trends Survey
health information-seeking behavior
incidence rate ratio
online health information-seeking behavior
odds ratio
socioeconomic status
theory of planned behavior
The authors thank the Bertelsmann Foundation for their cooperation and for providing the final dataset.
None declared.