The online environment provides adolescents with unprecedented access to a wide range of information, including content related to health [1]. As frequent users of digital media, adolescents increasingly turn to online sources to inform their health-related decisions [2]. While this access can be empowering, it also carries risks. For instance, adolescents may encounter health information that is false, misleading, or even harmful, with potentially serious consequences [3]. Much of this content is now encountered passively through algorithm-curated feeds on platforms such as TikTok (ByteDance Ltd), Instagram (Meta Platforms, Inc), or YouTube (Alphabet Inc), which prioritize engagement over accuracy [4]. Influencer-driven health messaging—often lacking scientific grounding—can go viral and shape adolescent attitudes and behaviors in ways that are not always beneficial [5].

This risk was further amplified during the COVID-19 pandemic, which intensified young people’s reliance on digital media for health-related information while simultaneously exposing them to a surge of misinformation [6]. Studies show that adolescents with lower eHealth literacy were more vulnerable to misinformation and more likely to adopt maladaptive health behaviors during this time [7-9]. Recent data from the World Health Organization [10] indicate that many European adolescents face difficulties both in finding online health information and in evaluating its reliability, underscoring persistent gaps in essential eHealth literacy skills. These developments highlight the urgent need to equip adolescents with the skills to navigate digital health information critically and responsibly.

In this context, eHealth literacy—defined as the knowledge and skills needed to locate, understand, evaluate, and apply online health information to support health and prevent illness [11,12]—becomes a critical competency. Prior research has shown that individuals with higher levels of eHealth literacy are more likely to actively seek health information [13], engage in healthier lifestyle behaviors [14], and attend preventive medical appointments, such as screenings [15]. Beyond individual health outcomes, adolescent eHealth literacy also contributes to broader public health goals by enhancing collective resilience against misinformation and supporting informed community health decisions [6,16]. Given these benefits, fostering eHealth literacy in adolescence is essential for supporting informed and health-conscious decision-making from an early age.

Parental mediation—defined as the parenting practice that aims to enhance children’s understanding of media content [17]—is one way to increase eHealth literacy among adolescents. However, our understanding of the contributions parents make to the development of eHealth literacy in adolescents remains limited. Research indicates that parental involvement and guidance significantly shape children’s media use and health-related behaviors [18,19]. Nevertheless, there is a shortage of longitudinal studies that verify the actual impact of parental mediation.

At the same time, it is recognized that parents’ literacy and education—particularly in the field of health—are associated with their ability to convey accurate information [20]. More educated parents may therefore be more effective in helping their adolescent children acquire relevant skills. This may be due not only to their stronger health knowledge, but also to greater digital confidence [21], a tendency to model critical thinking in everyday interactions [22], and more frequent engagement in reflective conversations about online content [23]. These factors suggest that parental education may facilitate more active and effective mediation of adolescents’ encounters with digital health information. However, empirical evidence remains limited, and it is unclear whether this principle extends to parental eHealth literacy mediation (hereafter referred to as eHealth literacy mediation). Moreover, adolescents’ eHealth literacy may, in turn, influence eHealth literacy mediation. When deciding on mediation strategies, parents may consider their child’s level of literacy and adjust their efforts accordingly. However, most current research is cross-sectional, capturing primarily between-person effects, and there is a lack of longitudinal research examining within-person effects in eHealth literacy mediation. The expectation of reciprocal effects is further supported by the Self-Determination Theory [24], which emphasizes the dynamic nature of parent-child interactions and suggests that these relationships involve bidirectional influence. Adolescents’ media-related behaviors—such as their choices regarding online health information—may shape how parents engage in eHealth literacy mediation.

Our study focuses on the reciprocal dynamics between eHealth literacy mediation and adolescents’ eHealth literacy, while also considering the moderating role of parental education. We use dyadic data from Czech parents and their adolescent children, collecting parent-reported data on eHealth literacy mediation and adolescent-reported data on eHealth literacy. This design allows us to examine whether changes in parental mediation practices are associated with subsequent changes in adolescents’ eHealth literacy, and conversely, whether adolescents’ literacy levels influence the extent to which parents engage in mediation. By adopting a longitudinal perspective, the study aims to clarify how these processes unfold over time within families and whether their strength varies according to parental educational background.

In this study, we examine the reciprocal dynamics between eHealth literacy mediation (as reported by parents) and adolescents’ eHealth literacy (as reported by the adolescents themselves) using a longitudinal research design. We consider the moderating role of parental education and apply a within-person approach, which enables the analysis of individual-level changes and variations in both eHealth literacy mediation and adolescents’ eHealth literacy over time within the same participants. Specifically, we examine whether increases in eHealth literacy mediation lead to subsequent increases in adolescents’ eHealth literacy, and conversely, whether increases in adolescents’ eHealth literacy influence the extent of parental eHealth literacy mediation. By focusing on within-person dynamics, we aim to better disentangle the complexities of these associations and clarify the potential causal mechanisms underlying changes in adolescents’ eHealth literacy. Additionally, we examine the moderating role of parental education to test whether higher education strengthens the effectiveness of eHealth literacy mediation. The model also controls for adolescents’ age and gender.

The data for this study came from a survey evaluating various aspects of adolescents’ use of ICTs and their impact on well-being. Because of the COVID-19 pandemic, we selected an online panel for participant recruitment and used computer-assisted web interviewing for data collection. An external research agency in the Czech Republic managed the sampling and data collection. Eligible participants were Czech households with at least one adolescent aged 11-16 years and a caregiver, enabling data collection from adolescent-parent dyads within the same household.

We utilized quotas based on region (Nomenclature of Territorial Units for Statistics level 3 [NUTS3]), municipality size, and household socioeconomic status (proxied by parents’ education) to ensure proportional representation of Czech households with children. Additionally, quotas for adolescents’ age and gender were established to achieve a balanced distribution. Data were collected across 3 waves, spaced 6 months apart, beginning in June 2021 (time 1 [T1]) and concluding in June 2022 (T3).

The full sample for this study comprised 2500 adolescent-parent dyads at T1. At T1, of the 2500 adolescents, 1250 (50%) were girls (mean_age 13.4 years, SD 1.7 years), and 1592 out of 2500 (63.68%) parents were women (mean_age 42.7 years, SD 7.1 years). At T2, the sample included 1654 dyads (adolescents: girls, 800/1654, 48.37%, mean_age 13.4 years; parents: women, 1004/1654, 60.70%, mean_age 43.3 years), and at T3, 1102 dyads (adolescents: girls, 532/1102, 48.28%, mean_age 13.4 years; parents: women, 661/1102, 59.98%, mean_age 43.3 years). Retention rates were over 66% from T1 to T2 (1654/2500, 66.16%) and from T2 to T3 (1102/1654, 66.63%). See the “Analysis” section for details on attrition.

The study protocol was approved by the Research Ethics Committee at Masaryk University (EKV-2018-068). Before participation, respondents were informed about the nature and purpose of the survey, their right to decline involvement, and their ability to skip any questions by selecting the “I prefer not to say” option available for all items. Informed consent was obtained from both parents and adolescents. Participants were rewarded with reward points redeemable for approximately US $4.

eHealth literacy mediation was measured with the following: “Children can find/encounter various health-related information on the internet. In the past 6 months, how often have you talked to your child about...” followed by 4 statements (eg, How can we tell that health-related information on the internet is true or false?; How should we evaluate the quality of such health-related information on the internet?). These items were previously used in the study by Gulec et al [28] to capture parental advice regarding the critical evaluation and recognition of online health information. Parents responded using a 6-point scale ranging from 1 (never) to 6 (several times a day). The internal consistency was excellent, with ω=0.95 at T1, ω=0.96 at T2, and ω=0.96 at T3. Descriptive statistics for the scale are presented in Table 1. The results supported longitudinal metric invariance, as indicated by a nonsignificant difference between the configural and metric models (Δχ²₆=2.812, P=.83).

Adolescents’ eHealth literacy was measured with 5 items from the Norman and Skinner [34] eHealth literacy scale and the following question: “One can find various health-related information on the internet. How true are the following statements for you?” Adolescents reported how strongly they agreed or disagreed with 5 statements (eg, I know where to find helpful health resources on the internet; I have the skills I need to evaluate the health resources I find on the internet) on a scale that ranged from 1 (strongly disagree) to 5 (strongly agree). The internal consistency was ω=0.92 for T1, ω=0.93 for T2, and ω=0.93 for T3. The results supported the longitudinal metric invariance (difference between the configural and the metric model: Δχ²₆=9.334, P=.16).

Parent’s education was assessed with the following question: “What is your highest level of completed education?” Parents were asked to respond on a scale that ranged from 1 (unfinished primary) to 5 (university, including higher vocational school). To examine the role of parental education, participants were categorized based on the parent’s education level reported at T1 into 3 groups: without a maturity diploma (735/2500, 29.40%), with a maturity diploma (1118/2500, 44.72%), and with a university degree (647/2500, 25.88%). Within the Czech educational system, obtaining a maturity diploma and pursuing university studies represent key milestones that signify different levels of educational attainment.

The age and gender of the adolescent were indicated at T1.

We analyzed the longitudinal data using structural equation modeling in lavaan 0.6-18 [92], applying the maximum likelihood estimator with robust standard errors. All data analyzed during this study are included in this published article and Multimedia Appendices 1 and 2. Structural equation modeling enables simultaneous estimation of multiple relationships between observed and latent variables while accounting for measurement error. To test our hypotheses, we used a random intercept cross-lagged panel model (RI-CLPM). RI-CLPM is particularly suitable for longitudinal research questions, as it decomposes observed variance into within-person (ie, autoregressive and lagged effects) and between-person levels (ie, random intercepts), enabling a clearer understanding of temporal dynamics and individual changes over time [93]. Because of this decomposition, autoregressive effects are usually weaker in RI-CLPM and, rather than indicating stability, reflect within-person carry-over effects.

The RI-CLPM was recursive and included 2500 observations and 38 free parameters. We also included the child’s age and gender at T1 as time-invariant covariates. All variables were specified as manifest. For the attrition analysis, we conducted a multivariate analysis of variance. Participants who completed all waves differed overall from dropouts in demographic variables and the variables of interest (Pillai V=0.09, F_8,4628=2.51, P=.01). The results of a separate univariate ANOVA revealed a significant difference only in eHealth literacy mediation (F_2,2316=4.56, P=.01). Further analysis showed that this difference was driven by participants who took part only in T1 and T2 (mean_T12 2.47), who had the highest mean score in eHealth literacy mediation compared with those who participated only in T1 (mean_T1 2.30, P=.04), and those who participated in all waves (mean_T123 2.33, P=.02). To address this, we used full information maximum likelihood to handle missing data.

To test H3, we used a multigroup version of the RI-CLPM described above. We tested for differences in path estimates across groups using a likelihood-ratio test, which compares the fit of nested models to determine whether constraining specific paths significantly worsens model fit.

This study explored the reciprocal dynamics between eHealth literacy mediation and adolescents’ eHealth literacy, considering the moderating role of parental education. We found a significant between-person association—adolescents with higher eHealth literacy tended to have parents who engaged more frequently in eHealth literacy mediation—which aligns with findings from existing cross-sectional studies [50,64]. However, contrary to our expectations, we found no within-person associations, meaning that changes in eHealth literacy mediation were not followed by the anticipated increases in adolescents’ eHealth literacy, nor vice versa. This contrast suggests that although some families maintain consistently higher levels of both eHealth literacy mediation and adolescent eHealth literacy compared with others (between-person level), these differences appear relatively stable and are not easily influenced over time (within-person level). While we did not observe significant cross-lagged effects between eHealth literacy mediation and adolescents’ eHealth literacy, our results revealed a significant autoregressive effect for eHealth literacy mediation between T2 and T3. This suggests that short-term fluctuations in parental mediation tend to persist over time within individuals, indicating some intraindividual stability in how parents engage with their children around online health information. Taken together, these findings imply that simply increasing parental mediation efforts at a single point in time may not necessarily lead to improvements in adolescent eHealth literacy, particularly among older adolescents. From a policy and intervention perspective, this underscores the importance of sustained, developmentally appropriate strategies that extend beyond eHealth literacy mediation alone.

One possible reason for the lack of an apparent impact of parental efforts on adolescents’ skills is that, based on previous findings [53], adolescents in this age group are already quite autonomous and may not notice or acknowledge parental mediation efforts. This can be understood through the lens of cognitive development: as adolescents acquire more advanced reasoning skills, they tend to question authority and assert their independence [94]. Such autonomy needs may conflict with parental mediation efforts, especially when these efforts are perceived as controlling. This notion is further supported by studies comparing parents’ and adolescents’ perspectives, which reveal substantial discrepancies in their reports. A meta-analysis of parent-adolescent discordance in parenting variables found only a moderate correlation between their reports (pooled r=0.28) [95]. For instance, in a study on parental mediation of video games [96], parents reported a higher frequency for all 3 examined mediation strategies—active, restrictive, and coplaying. Similarly, Beyens and Valkenburg [97] found that parents reported greater use of autonomy-supportive strategies but lower use of controlling or inconsistent approaches. These findings suggest that the extent of parenting efforts, as perceived by parents, is not always recognized by their children, which may limit their effectiveness. Moreover, our study relies solely on parent-reported data. Prior research has shown that parents and adolescents often differ in their perceptions of mediation, with parents frequently reporting higher levels of engagement than their children perceive [97]. Future studies should therefore adopt a dual-informant approach to capture both perspectives and better reflect the mediation process as experienced by adolescents. Additionally, the relatively low levels of online parental mediation in the Czech Republic compared with the rest of Europe [25] could further reduce the potential impact of mediation in this context. Future research may thus consider whether similar patterns emerge in other cultural contexts with different digital parenting norms.

It is also possible that adolescents’ eHealth literacy is influenced more by socialization agents other than parents, such as teachers or peers. Prior research suggests that parental mediation is most impactful during early childhood, particularly up to 12 years of age [98,99], whereas our sample includes adolescents aged 11-17 years, a stage when peer influence tends to surpass that of parents [100]. In the context of active mediation, Shin and Lwin [101] found that peer and teacher guidance, rather than parental mediation, was linked to adolescents’ experiences with online risks. Similarly, in the health domain, advice from peers, but not from family members, was positively associated with physical activity among adolescents with overweight [102]. These findings suggest that peer eHealth literacy mediation may play an increasingly important role during adolescence and should be considered in future research. Additionally, the timing of our data collection during the COVID-19 pandemic, a period marked by heightened health anxiety and widespread health information exposure [103,104], may have amplified adolescents’ access to diverse sources of health guidance, potentially diminishing the relative influence of parental mediation.

Our study did not find support for the reverse effect, meaning that changes in adolescents’ eHealth literacy were not followed by corresponding changes in parental eHealth literacy mediation. One possible explanation is that parents may assume their children are already receiving adequate guidance from peers and teachers and thus perceive no need for additional mediation [105]. In the eHealth domain, this may be further exacerbated by parents’ limited confidence or expertise [106], which might lead them to delegate responsibility to schools or other professionals. Additionally, our assumption that parents would adjust their mediation based on their child’s needs may not hold if parents struggle to recognize those needs. Qualitative studies show that parents often lack insight into what their adolescent children do online [107], and adolescents frequently withhold information about their online challenges [108]. In the Czech Republic, more than half of adolescents rarely or never discuss their online activities with their parents [25], which can hinder parents’ ability to tailor mediation strategies effectively.

Additionally, our assumption relied on parents recognizing their children’s current level of eHealth literacy. We assumed that when parents perceive a need (eg, a low level of eHealth literacy), they would engage in mediation. However, accurately recognizing their children’s literacy level in this domain may not be easy. Qualitative studies show that parents often report lacking the knowledge or understanding of what their adolescent children do online [107]. In addition, adolescents often do not confide in their parents about their struggles and negative online experiences [108]; therefore, parents may not be fully aware of their child’s actual eHealth literacy levels or the specific challenges they face online, which could limit their ability to adjust mediation strategies effectively. In the Czech Republic, more than half of adolescents rarely or never talk to their parents about their online activities [25]. As a result, parents may easily miss important cues that would signal the need to enhance their eHealth literacy.

In our study, we examined the moderating effect of parental education on the relationship between eHealth literacy mediation and adolescents’ eHealth literacy. Previous research has shown that parents with higher education levels tend to be more confident in their digital skills and can engage more effectively in active mediation on various digital topics [21,50]. However, our findings did not support the moderating effect of parental education. There was no significant relationship between eHealth literacy mediation and adolescents’ eHealth literacy across any of the education groups. This suggests that higher parental education does not enhance the effectiveness of eHealth literacy mediation for this adolescent age group. One possible explanation is that although higher education is generally associated with more effective mediation, better-educated parents may overestimate their children’s digital competence, leading to less frequent or less engaged mediation. For example, Livingstone et al [21] observed that highly educated parents tend to assume their children are more digitally skilled than they actually are. Similarly, Kuldas et al [109] noted that parents with higher self-efficacy—often linked to higher education—may underestimate the frequency of their children’s online risk experiences, which can undermine the accuracy and responsiveness of their mediation strategies.

Future research could explore other potential moderators on the parental side, such as the quality of the parent-adolescent relationship. It is possible that the effectiveness of parental mediation is more closely linked to positive family relationships than to parents’ education or knowledge levels. Future research should therefore examine the complex interplay between parental awareness, adolescent disclosure, and family dynamics, as well as explore alternative socialization agents—such as peers, schools, or digital tools—that might effectively support adolescents’ eHealth literacy development.

Previous studies have primarily relied on cross-sectional research, demonstrating associations between constructs and differences among adolescents (ie, the between-person approach). Our study is innovative in that it examines individual-level changes within a longitudinal framework, presenting both within- and between-person results. Similar to previous research on adolescents’ digital skills [50], our study observed a between-person association, indicating that adolescents with higher eHealth literacy experienced more frequent eHealth literacy mediation. This finding is important because it highlights the distinction between the aforesaid 2 approaches in our study. Our study shows that in families with higher levels of eHealth literacy mediation, adolescents also have higher eHealth literacy. These 2 factors tend to co-occur. However, changes in eHealth literacy mediation do not correspond to changes in adolescents’ eHealth literacy within our sample. This finding has 2 important implications. First, this highlights the need for researchers and policy makers to exercise caution when relying solely on cross-sectional findings. Studies on parental mediation (and media effects more broadly) should increasingly use within-person methods if the field aims to generate more robust evidence to advance theory and inform evidence-based policy. Second, this finding suggests that—at least for early and middle adolescents—factors other than parental mediation play a more significant role in shaping their level of eHealth literacy.

One limitation of our study is the reliance on self-reported data from both adolescents and their parents. A potential shortcoming is that, due to social desirability bias, respondents might provide answers they believe are expected of them. Additionally, the tendency to overestimate or underestimate one’s own eHealth literacy could affect the accuracy of our findings. Similarly, the reported frequency of discussions about online health information may not fully capture the quality or effectiveness of those conversations. Future research could address these limitations by incorporating performance-based assessments, such as practical tasks or quizzes, to provide a more objective measure of eHealth literacy and minimize the biases inherent in self-reported data. Furthermore, although quota sampling was used to ensure demographic representativeness, the reliance on an online panel may have attracted participants with above-average digital engagement and skills, potentially underrepresenting individuals with limited internet access or lower digital literacy.

Another potential limitation is the 6-month interval between data collection waves for eHealth literacy mediation and adolescents’ eHealth literacy, which may have been too long. A shorter interval might have been more effective in capturing the potential impact of parental mediation. A follow-up study could also explore other factors that contribute to adolescents’ eHealth literacy. For instance, examining parental attitudes and behaviors toward digital technologies—alongside their own eHealth literacy levels and digital skills—may offer deeper insights into how eHealth literacy is transmitted from parents to children. In our study, we assessed parental education using a simplified classification (no diploma, diploma, and university). However, we acknowledge that this approach may not fully capture important aspects such as the parents’ field of study or alternative forms of education. Future research could improve upon this by more clearly distinguishing between different types of educational experiences, which may contribute differently to digital and eHealth literacy. Moreover, as educational background does not always equate to digital competence, examining digital skills as a distinct factor could offer a more nuanced understanding of the mediation process. Additionally, investigating broader familial dynamics, such as the quality of the parent-adolescent relationship, communication patterns, and parental role modeling of eHealth behaviors, could provide valuable insights into how the family context shapes adolescents’ engagement with online health information and the development of their eHealth literacy.

Future research could also benefit from distinguishing between different types of online health information that adolescents engage with, such as fitness, mental health, sexual health, or nutrition. These domains may not only attract varying levels of adolescent eHealth literacy and trust but may also elicit different parental mediation strategies. Disaggregating eHealth literacy by topic may, therefore, offer a more detailed understanding of how adolescents navigate specific eHealth domains and how parents support or influence these processes. Regarding age, future studies could also explore whether the relationship between parental mediation and adolescent eHealth literacy follows a nonlinear pattern. Testing for curvilinear associations could provide more nuanced insights into how mediation functions across developmental stages.

Lastly, we address the issue of sample attrition and the associated potential loss of statistical power. This may have reduced the likelihood of detecting small within-person effects (β=.10), resulting in less certain estimates. However, most of the observed effects were weaker than this threshold, indicating negligible effect sizes. Further, RI-CLPM has the advantage of decomposing variance into between- and within-person components [93], which reduces the likelihood of detecting spurious effects and generally provides less biased estimates. As most of the existing literature is based on cross-sectional data, which inherently cannot distinguish between these sources of variance, we consider our results to be an important contribution. We also acknowledge the use of manifest variables instead of multiple-indicator latent variables in our model. This choice limits the precision of our estimates, as measurement error was not controlled. However, because these variables did not represent multidimensional constructs and were primarily used to establish directional relationships, we still consider our results a valuable contribution. Nonetheless, we recommend that future research apply RI-CLPM with multiple indicators whenever possible to enhance measurement accuracy.

This study explored the reciprocal dynamics between eHealth literacy mediation and adolescents’ eHealth literacy using a longitudinal research design. While no within-person reciprocal associations were found, noteworthy between-person associations emerged. This means that adolescents in families with higher eHealth mediation tend to have higher eHealth literacy; however, changes in eHealth literacy mediation do not appear to affect changes in adolescents’ eHealth literacy. Contrary to expectations, the study did not confirm a moderating role of parental education level. Our findings suggest that other factors may influence adolescents’ eHealth literacy, highlighting the need for further research. These results challenge the widely held theoretical assumption—based largely on cross-sectional studies—that parental mediation causally affects adolescents’ digital literacy and skills. The study’s results suggest that such associations may overestimate the true impact of parental mediation over time and underscore the need for more nuanced models that clarify when, how, and for whom parental mediation is effective.