Impact of short-video health information on older adults’ health adoption intentions: A mixed-methods study of a dual-path model

Abstract

Background

With the rapid spread of short health videos, older adults frequently encounter commercialized health appeals. However, there is a lack of research on the mechanisms that drive cognitive and emotional responses and adoption decisions in such situations.

Methods

Based on the Health Belief Model (HBM) and the Heuristic–Systematic Model (HSM), this study develops a dual-path model to explain how intrinsic health needs and external persuasion cues jointly shape older adults’ emotions and behavioral intentions. Study 1 used a cross-sectional survey of older adults (aged >60 years) to test the structural relationships using the partial least squares structural equation model (PLS-SEM). Study 2 established causal evidence through a 2 × 2 between-subjects randomized experiment among active older short-video users.

Results

Study 1’s results show a positive correlation between chronic health anxiety and adoption intention through expectation trust (95% CI: [0.100, 0.191]). Conversely, Study 2 reveals that experimentally induced acute anxiety reduces expectation trust, thereby weakening adoption intention (95% CI: [-0.212, -0.071]). The heuristic cues positively affect adoption intention through source trust (95% CI: [0.101, 0.182]). In addition, media literacy attenuates the impact of heuristic cues on source trust (95% CI: [-0.315, -0.179]), and adult children’s intervention weakens the relationship between trust and adoption intention (95% CI: [-0.289, -0.149]).

Conclusion

Digital health adoption intention among older adults is a complex process driven jointly by internal health vulnerabilities and external heuristic triggers. Safeguarding this population requires a collaborative strategy: integrating strict algorithmic supervision, implementing precision literacy education to build cognitive shields, and actively empowering intergenerational family networks as decision gatekeepers.

Keywords

short video older adults health anxiety heuristic cues adoption intention media literacy health information

1. Introduction

As global society experiences significant aging, the simultaneous rise of digital technology is reshaping all aspects of social life, leading to the formation of a growing group of older Internet users. Among these platforms, short-video mobile social media applications like TikTok, YouTube, and Douyin have become important channels for older adults to obtain health information and seek social connections.¹ According to data from Quest Mobile,² users over the age of 50 in China account for more than one-third of short video platforms, and health and wellness content has long occupied the top of their interest views. Similarly, the Pew Research Center³ and the OECD⁴ pointed out in their reports that more older adults around the world are using social media platforms to obtain health information to meet their health needs. The dissemination of health information through mobile short videos has become the mainstream platform and channel for health communication in the world today.

However, in this context of abundant health information, digital technology simultaneously empowers people and introduces associated risks. On the one hand, older adults are concerned about health and information needs; on the other hand, their digital health literacy is relatively lagging.^5,6 This literacy deficit exposes them to a complex information environment filled with exaggerated propaganda, pseudoscientific packaging, and precise algorithmic pushing, making them vulnerable victims of disinformation and predatory marketing.^7–9 Especially in the usage of mobile short video health products, merchants combine emotional narratives, authoritative image packaging, and urgent sales rhetoric to break through the psychological defense line of older adults,^8,10 resulting in their irrational health decisions, which not only cause economic losses but may also endanger their lives and health due to delays in scientific diagnosis and treatment.

In the face of this real-world challenge, existing research has focused on media use and health anxiety in older adults,^5,11 false health information,^12,13 digital trust mechanisms,¹⁴ and cognitive heuristics.¹⁵ However, previous studies frequently examine these factors in isolation. They either focus on external algorithmic manipulation without accounting for older adults’ cognitive vulnerabilities or investigate internal health needs while neglecting the rapid evolution of persuasive video cues. Consequently, a critical gap remains in understanding the simultaneous, interactive effects of internal vulnerabilities and external triggers on digital health decision-making.

Based on the above background, the primary innovation of this study lies in its theoretical integration of the HSM and the HBM to construct a moderated parallel mediation model. Unlike prior studies, this novel framework systematically reveals how health anxiety and heuristic cues affect the adoption intention of older adults through the dual mediation of expectation trust and source trust in the short video environment and deeply explore the moderating role of media literacy and adult children’s intervention in it.

To this end, this study will focus on answering the following core questions:

RQ1: How do health anxiety and heuristic cues independently affect expectation trust and source trust among older adults?

RQ2: Do expectation trust and source trust mediate the relationship between these antecedents (health anxiety and heuristics) and adoption intention?

RQ3: Can media literacy, as a cognitive protection resource, significantly attenuate the positive impact of heuristic cues on source trust?

RQ4: Does adult children’s intervention play a key moderating role in the final decision-making phase, specifically in the transition from trust to adoption intention?

This study employed a mixed-methods design, including a cross-sectional survey (Study 1) and a contextualized randomized experiment (Study 2). While traditional RCTs are important criteria for assessing long-term physiological outcomes, they often lack the effectiveness needed to capture instantaneous cognitive processing and behavioral intentions in dynamic short video environments. Therefore, the combination of the two methods is more suitable for the current research scenario than a single randomized trial. It can not only reveal the complex structural relationships between variables but also effectively verify the causal mechanism. This comprehensive design provides a more comprehensive interpretive perspective for understanding digital health consumption in the elderly population.

2. Literature review and research hypotheses

2.1. Health Belief Model (HBM)

As a classic theory for understanding population health behavior in different contexts, the health belief model has been widely used in research in health and related interdisciplinary fields.^16,17 HBM believes that whether an individual adopts a certain health behavior depends on his perception of the severity of the health threat and the correlation between the benefits of recommended health behaviors and perceived barriers.¹⁶ While traditionally applied to clinical behaviors (e.g., vaccination, cancer screening),^18–22 HBM has recently demonstrated robust predictive power in digital health communication, effectively explaining users’ health motivations on social media.^23–25 For older adults, the objective decline in physical function and the prevalence of chronic diseases exacerbate their perceived vulnerability and severity.^26,27 Cognitive and emotional levels intertwine this high-intensity threat perception, forming health anxiety.²⁸ Therefore, to effectively alleviate health anxiety, older adults actively seek solutions that bring perceived benefits.

As a new medium, precise health advertisements serve as key action cues, effectively stimulating users’ health awareness.²⁹ These ads directly influence older adults’ evaluation of product efficacy through personalized narratives. When the promised effect of the advertisement matches the user’s internal anxiety, older adults believe that the product is effective in reducing health threats, thus forming expected trust. Therefore, in the context of health adoption in this study, expected trust is seen as a direct operationalized manifestation of perceived benefits.³⁰ Although HBM powerfully explains the intrinsic motivation of older adults to seek health products, it struggles to explain how users process this complex information in a short video environment full of marketing tricks and competitive information.³¹ This reveals the need to introduce the heuristic-systematic model (HSM) to complement cognitive processing perspectives.

2.2. Heuristic-Systematic Model (HSM)

The Heuristic-Systematic Model (HSM) explains attitude change through dual mechanisms: high-cognitive-input systematic processing and low-input heuristic processing.^32–34 In the context of short-video health advertising, heuristic processing overwhelmingly prevails. Short video platforms present content as a waterfall of information, with rapid switching speed and high-intensity sensory stimulation systematically diminishing the motivation and capacity of the older adults to engage in deep thinking.³⁵

In an environment where there is structural encouragement for shallow processing, heuristic processing becomes the default mode for older adults. In health advertising, advertisers will use heuristic cues like pictures of trusted experts, emotional stories, and social identity to induce older people to trust the source and adopt health solutions.^8,36,37 This suggestion of source trust is not due to the older adults carefully considering the information; rather, it stems from heuristic cues. Therefore, HSM provides core theoretical support for further explaining the relationship between heuristic cues and source trust in short video health advertisements for older adults.

2.3. Media literacy and adult children’s intervention

A robust model of consumer behavior must articulate the boundary conditions under which decision-making mechanisms take effect. This study included media literacy and adult children’s intervention as moderating variables at the cognitive and social levels, respectively. As an individual’s core ability to criticize, evaluate, and analyze information, media literacy plays an important moderating role between the heuristic cues and the source trust path.³⁸ High media literacy can make individuals actively question the manipulative intention of heuristic cues, thereby inhibiting heuristic processing and weakening their suggestion of trust in the source.³⁹ Similarly, intergenerational interactions play a significant role for older adults in family-centric East Asian cultures.⁴⁰ The advice of adult children constitutes a key role in the decision-making of older adults.⁴¹ Even if older adults generate trust during their behavior, negative advice from their adult children may block the conversion path from trust to adoption intention in older adults.

In summary, this study constructs a dual-path parallel intermediary model integrating HBM and HSM. The model argues that older adults’ adoption intention is driven by two pathways: an active demand path driven by health anxiety (mediated by Expectation trust) and a passive persuasion path driven by marketing cues (mediated by source trust). Media literacy and adult children’s intervention play a key moderating role in the front end of information processing and the back end of decision-making, respectively.

2.4. Research Hypothesis

2.4.1. Health anxiety and expectation trust

For older people, perceived susceptibility and perceived severity constitute the intrinsic basis of their health anxiety.¹⁶ This anxiety gives rise to intense motivational reasoning that seeks solutions to reduce and alleviate their restlessness.⁴² When the promised content of a short video health advertisement accurately matches the health anxiety of older users, they tend to believe that the solutions provided in the advertisement are effective and trust in the efficacy of the product to alleviate their anxiety. Therefore, the following hypotheses are proposed:

H1: The level of health anxiety in older adults has a significant positive impact on their expectation trust.

2.4.2. Heuristic cues and source trust

Short video health advertising limits the possibility of in-depth systematic processing of information by older adults with the characteristics of fast-paced and fragmented information flow, so that heuristic processing becomes the dominant path of cognitive processing.³⁴ In this scenario, advertisers quickly build persuasiveness by using heuristics such as authority symbols, social identity, and emotional narratives. Especially when older groups lack the motivation or ability to conduct in-depth content analysis, they rely on these external cues to assess the credibility of information.³³ Based on this, the following hypotheses are proposed:

H2: Heuristics in short video health advertisements have a significant positive impact on older adults’ source trust.

2.4.3. Trust and adoption intention

Trust is an important psychosocial tool that makes transactions less risky and uncertain. In the context of virtual information, trust is recognized as a crucial antecedent variable for predicting user adoption intention, due to physical distance and information asymmetry.^43–45 Previous studies have demonstrated a positive relationship between consumer trust and adoption intention, including merchants, platforms, and particular products.^46–48 This study elucidates the transformation mechanism by distinguishing between two distinct yet complementary forms of trust.⁴¹ Expectation trust serves as a cognitive hub for the active demand pathway, transforming older adults’ intrinsic health anxiety into adoption motivation based on product efficacy assurance. Source trust, on the other hand, serves as a bridge to the passive persuasion path, directly transferring the reputational assessment of external marketing leads to a willingness to adopt the product through authority and social effects. Based on this, the following hypotheses are proposed:

H3: Expectation trust has a positive impact on the adoption intention of older adults.

H3a: Expectation trust plays a significant mediating role between health anxiety and adoption intention.

H4: Source trust has a positive impact on the adoption intention of older adults.

H4a: Source trust plays a significant mediating role between heuristic cues and adoption intention.

2.4.4. The moderating role of media literacy

Media literacy is defined as an individual’s ability to critically evaluate and reflect on information.³⁸ It is resistant to heuristic processing paths in HSMs. A high level of media literacy indicates that individuals possess strong cognitive abilities, can evaluate information effectively, and are capable of actively engaging their own systematic processing skills.⁴⁹ Therefore, when an older person with high media literacy is exposed to heuristic cue information content, they may question its authenticity rather than automatically accepting the information and generating trust. Media literacy plays the role of cognitive judgment in this process, which interrupts the automatic path from heuristic cues to generating trust. Therefore, we infer that the effect of heuristic cues is not the same for all older adults, and it is moderated by individual media literacy levels.

H5: Media literacy plays a negative moderating role between heuristic cues and source trust.

2.4.5. The moderating role of adult children’s intervention

Social networks and significant others heavily influence individual health decisions, especially in the context of family-oriented East Asian cultures.⁴⁰ The opinion of adult children is one of the most important subjective norms for older adults’ parents.⁴¹ Even if older adults have independently formed a high source trust and expectation trust by watching advertisements, the final step from the internal attitude of trust to the explicit behavior of adoption is usually constrained by social factors. An explicit objection from a child or a well-intentioned reminder can create a strong perceptual barrier and potentially lead to family conflict. This consideration of social costs will inhibit or even reject the original impulse to adopt. Based on this, the following hypotheses are proposed:

H6a: Adult children’s intervention plays a negative moderating role in the relationship between trust (expectation trust) and adoption intention.

H6b: Adult children’s intervention plays a negative moderating role in the relationship between trust (source trust) and adoption intention.

Based on the health belief model and the Heuristic–Systematic Model, a manipulation and demand dual-path model was constructed to analyze the causal relationship between why older adults are willing to accept short video health advertisements (Figure 1).

Figure 1.

Research model.

2.5. Overview of mixed-methods study design

To test the proposed two-path model and boundary conditions through empirical studies (Figure 1), a sequential mixed-method design was adopted.

Study 1 (Cross-sectional Survey): A Partial Least Squares Structural Equation Model (PLS-SEM) was employed to delineate structural relationships and validate the mediating pathways of health anxiety and heuristic cues in the natural environment.

Study 2 (randomized experiment): A 2×2 control experiment was designed to build upon study 1. This step isolates the manipulated variables to establish rigorous causal evidence and tests the moderating effects of media literacy and adult children’s interventions.

Together, these two integrated studies comprehensively answer the research question from both structural and causal perspectives.

3. Study 1: Structural validation via cross-sectional survey

3.1. Sample selection and data collection

To assess the model proposed in this study, after receiving approval from the School of Animation and Digital Arts, Communication University of Shanxi review board (No. 20250917), we conducted a questionnaire survey of older adults aged over 60 in second- and third-tier cities in China (Taiyuan, Shenyang, Nanchang etc.). These cities are currently the primary hubs of urban development and population mobility in China,⁵⁰ which helps ensure the sample’s universality and mitigates issues arising from regional disparities in information and media literacy. A structured questionnaire was developed through the Wenjuan Xing platform (China), containing personal information and data (gender, age, education, etc.), as well as questions related to the research objectives, which were distributed via mobile phone and data recovered. This method was chosen because sampling is geographically independent, inexpensive, and responsive.⁵¹

To increase the willingness and trust of the respondents to participate, they were informed of the purpose and significance of the study before filling out the questionnaire, and the implementation began after obtaining their informed consent. After answering the questions carefully, participants could randomly receive a monetary incentive reward of 2-5 RMB. After the survey, a total of 538 questionnaires were collected. To ensure sample validity, the dataset was strictly screened based on the following criteria: Inclusion Criteria: (1) Individuals aged 60 years and above; (2) Individuals possessing basic cognitive and digital skills to complete online questionnaires independently via mobile phones. Exclusion Criteria: (1) Questionnaires with a completion time of less than 5 minutes; (2) Questionnaires exhibiting obvious logical contradictions or patterned responses (e.g., selecting the same option for all items). A total of 46 questionnaires were eliminated, and finally, 492 valid questionnaires were retained, with an effective recovery rate of 91.4%. Because the online survey platform was configured to require responses for all mandatory items, there was no missing data in the final retained dataset. Sample attrition was strictly limited to the predefined exclusion criteria. The demographic characteristics of specific survey respondents are presented in Table 1.

Table 1.

Demographic profile of respondents (study 1).

Measure	Items	Frequency	Percent
Gender	Male	223	45.3%
Gender	Female	269	54.7%
Age	60-74	342	69.5%
Age	75-89	150	30.5%
Education	Middle school and below	248	50.4%
	High school	198	40.2%
	Undergraduate	46	9.4%
Living Environment	Living alone	131	26.6%
	With spouse	245	49.8%
	With children	116	23.6%
Income	Below 1000 RMB	103	21.0%
	1001-3000 RMB	212	43.0%
	3001-6000 RMB	88	17.8%
	Above 6000 RMB	89	18.2%

Note. RMB for the income variable.

3.2. Measurement

This study adopts established scales of previous studies, and the wording of the questions is appropriately adjusted according to the actual background of the research. Since the survey was conducted in China, we followed a process of combined translation and back-translation.⁵² First, the original English questionnaire was translated into Chinese. During this phase, a cultural adaptation process was conducted to ensure that the phrasing, terminology, and health-related contexts were natural, respectful, and easily comprehensible for Chinese older adults. Following this cultural adaptation, it was translated back into English by two native Chinese translators with PhDs in the English language to confirm its validity across languages. When comparing the two versions of the questionnaire, no semantic differences were found. Therefore, we used the Chinese version of the questionnaire for data collection. All items on the questionnaire were rated on a 5-point Likert scale, with 1 meaning “strongly disagree” and 5 meaning “strongly agree.” To ensure the reliability and validity of the questionnaire, we first invited 2 experts with medical backgrounds and 3 experts with communication backgrounds to evaluate and verify the validity of the content of the questionnaire, and the final objective consistency index (IOC) was about 1, indicating that the validity of the questionnaire content was fully confirmed.⁵³ We then invited 102 respondents to pretest the questionnaire, and it is worth noting that the Cronbach’s alpha coefficient exceeded 0.7, meeting the criteria for a reliability coefficient,⁵⁴ indicating acceptable internal consistency and stability. A comprehensive assessment of the final measurement model’s reliability (Cronbach’s alpha, Composite Reliability) and validity (Average Variance Extracted, HTMT) is detailed in Section 3.4.2. Table 2 shows the measurement items.

Table 2.

Constructs and measurement items.

Variable	Factor	Item	Reference
Health Anxiety (HA)	HA1	I worry about my health a lot.	Salkovskis et al.⁵⁵; Lasher & Faulkender.⁵⁶
	HA2	I seem to be more aware of my bodily sensations than others.
	HA3	When I feel a bodily discomfort, I worry that it might be a symptom of a serious illness.
	HA4	The thought of being seriously ill in the future makes me very anxious.
	HA5	I spend a lot of time thinking about my health.
Heuristic Cues (HC)	HC1	The person speaking in the video looked like a trustworthy doctor or expert.	Metzger et al.⁵⁷; Petty & Cacioppo.⁵⁸
	HC2	The video showed many positive reviews and success stories from other users.
	HC3	The video told a very touching story that resonated with me emotionally.
	HC4	The video used authoritative-sounding phrases like “scientifically certified” or “exclusive patent”.
Source Trust (ST)	ST1	I feel the source of this video is trustworthy.	Hovland et al.⁵⁹; Ohanian.⁶⁰
	ST2	I believe the source of this video is an expert in this health field.
	ST3	I believe the information provided by the source of this video is honest.
	ST4	Overall, I have a high regard for the source of this video.
Expectation Trust (ET)	ET1	I believe the product featured in the video can effectively address my health concerns.	Gefen & Straub.⁴⁵; Munuera-Aleman et al.⁶¹
	ET2	I trust that this product will perform as promised in the video.
	ET3	I believe that using this product would be beneficial for my health.
	ET4	Based on the information in the video, I am confident in this product’s efficacy.
Media Literacy (ML)	ML1	When I see a piece of health advice on a short-form video, I am confident in my ability to judge whether it is good or bad.	Norman & Skinner.⁶²; Livingstone.³⁸
	ML2	I can tell whether a “health tip” that sounds amazing on a short-form video is reliable or not.
	ML3	I understand many of the “experts” and “success stories” in short-form videos are advertisements designed to sell products.
	ML4	Before deciding to trust or try a health method from a short-form video, I will first verify it elsewhere
Adult Children’s Intervention (ACI)	ACI1	Before buying a new health product, I usually ask for my children’s opinion first.	Rau et al.⁶³
	ACI2	My children often take the initiative to remind me to be wary of online health product advertisements.
	ACI3	If my children strongly object, I will generally not purchase a certain health product.
	ACI4	Overall, my children’s advice has a significant impact on my health adoption decisions.
Adoption Intention (AI)	AI1	It is very likely that I will purchase the product recommended in the video.	Dodds et al.⁶⁴; Sussman & Siegal.⁶⁵
	AI2	I will actively consider purchasing this product in the future.
	AI3	If I had the opportunity, I would be willing to try purchasing this product.

Note. Adoption Intention items were adapted from Dodds et al. (1991). Following the Information Adoption Model (Sussman & Siegal, 2003), given the commercialized context, adoption was operationalized as the intention to purchase the recommended solution.

3.3. Data analysis

The structural equation model of the partial least squares method (PLS-SEM) was used to analyze and estimate the research model. First, the reliability and validity analysis are carried out, and secondly, the path coefficient and explanatory power of the structural model are estimated and verified, and finally, the reliability and validity of the constructed model are verified.⁶⁶ PLS-SEM was selected because it is particularly robust for exploratory models handling complex relationships and non-normal data distributions, which are common in social survey research.^67–69 This method is highly suitable for our integrated HBM and HSM framework to identify key predictive drivers of older users’ adoption intentions. Majchrzak et al.⁷⁰ recommended that the number of sample size in the model should be at least 5–10 times the maximum number of model paths, and in this study, the sample size was 492 and the maximum number of paths was 6, which met the recommended criteria. Kline⁷¹ suggested that a sample size greater than 200 is necessary to guarantee the stability of model parameter estimation; this study’s sample size also adheres to this criterion. In addition, a post hoc power analysis conducted using G*Power 3.1.9.7 software showed that the statistical power exceeded 0.95 when the sample size was 492, the significance level was 0.05, and the effect size was moderate. This confirms that the sample size is fully sufficient to validate our structural equation model. This study used the SmartPLS 4.0 software for data analysis based on the PLS-SEM. To isolate the main effects of the theoretical constructs, demographic characteristics including age, gender, education, income, and living environment were included as control variables in the PLS-SEM model to account for potential confounding effects.

3.4. Results

3.4.1. Common Method Bias (CMB) and Variance Inflation Factor (VIF)

Since the study used a self-reporting method to collect data, it is necessary to examine the potential problems caused by common method biases and variance expansion factors. The results of the Harman univariate test indicated that the single factor explained 17.69% of the total variance, which was lower than the critical standard of 40%.⁷² The multicollinearity of the factors was detected by the variance inflation factor, and the results showed that the VIF values of all latent variables were between 1.60-2.15, which was lower than the standard of 5, and there was no problem of multicollinearity for the latent variables.⁷³ The data collection process in this study appears to be free from method bias and multicollinearity issues.

3.4.2. Measurement model evaluation

Before the structural model can be evaluated, the basic requirements of the model construct must be met. First, the Cronbach’s α of each construct needs to be greater than the threshold of 0.70, the factor load needs to reach the threshold of 0.6,⁷⁴ and the combined reliability (CR) of each variable needs to meet the threshold of 0.7,⁷⁵ which indicates that the model construct has a high consistency. At the same time, the average variance extracted (AVE) for each variable exceeded 0.5, indicating that the model had excellent convergence validity⁷⁶ (Table 3). Table 4 shows that the square root of the average variance extracted for each construct is higher than its corresponding correlation coefficient with other constructs, and the heterotrait–monotrait ratio (HTMT) value is less than 0.85, all of which meet the standard.⁷⁷ This study demonstrates the reliability and effectiveness of the measurement model.

Table 3.

Reliability and validity analysis (study 1).

Variable	Factor	Factor loadings	Cronbach’s alpha	CR	AVE
Health Anxiety (HA)	HA1	0.825	0.875	0.876	0.666
	HA2	0.827
	HA3	0.823
	HA4	0.810
	HA5	0.796
Heuristic Cues (HC)	HC1	0.750	0.821	0.832	0.649
	HC2	0.816
	HC3	0.839
	HC4	0.816
Source Trust (ST)	ST1	0.857	0.862	0.863	0.707
	ST2	0.839
	ST3	0.835
	ST4	0.831
Expectation Trust (ET)	ET1	0.838	0.853	0.856	0.694
	ET2	0.823
	ET3	0.831
	ET4	0.841
Media Literacy (ML)	ML1	0.806	0.854	0.861	0.695
	ML2	0.821
	ML3	0.862
	ML4	0.842
Adult Children’s Intervention (ACI)	ACI1	0.832	0.861	0.867	0.706
	ACI2	0.844
	ACI3	0.828
	ACI4	0.856
Adoption Intention (AI)	AI1	0.893	0.831	0.840	0.747
	AI2	0.859
	AI3	0.840

Note. CR= Composite Reliability; AVE= Average Variance Extracted.

Table 4.

Discriminant validity (HTMT) and correlations (study 1).

	ACI	ET	HA	HC	ML	AI	ST
ACI	0.840
ET	-0.024 (0.039)	0.833
HA	-0.069 (0.081)	0.525 (0.605)	0.816
HC	-0.013 (0.048)	0.050 (0.080)	0.017 (0.055)	0.806
ML	0.009 (0.063)	-0.077 (0.087)	-0.073 (0.089)	0.048 (0.079)	0.833
AI	0.163 (0.190)	0.260 (0.304)	0.144 (0.166)	0.132 (0.158)	-0.043 (0.057)	0.864
ST	0.014 (0.049)	-0.001 (0.041)	-0.053 (0.068)	0.415 (0.486)	0.207 (0.239)	0.340 (0.399)	0.841

Note. Square root of AVE is showed in diagonal; value within()is the value of HTMT ratio; HTMT= Heterotrait-Monotrait ratio. AI= Adoption Intention; ACI= Adult Children’s Intervention; ET= Expectation Trust; HA= Health Anxiety; HC= Heuristic Cues; ML= Media Literacy; ST= Source Trust.

3.4.3. Measure structural models

To measure the validity of the model, the model fit index standard (SRMR<0.08) proposed by Henseler et al.⁷⁸ was referenced. The results of this study indicated that the SRMR was 0.045. As an important index to evaluate model fitting, the R² values of this study were 0.271, 0.276, and 0.301, respectively.⁷⁹ Similarly, the Q2 value is also an important supplement to the model fitting, and a Q2 value greater than 0 indicates predictive ability,⁷⁹ and the Q2 value in this study is between 0.413-0.502, both of which are greater than 0. In summary, the model in this study has a good fit.

3.4.4. Hypothesis test

In direct alignment with our enumerated research questions (RQ1 to RQ4), we sequentially tested the main effects, mediation mechanisms, and moderation effects. The pathway analysis revealed that older adults’ health anxiety significantly and positively impacted expectation trust (HA→ET, p < 0.001), supporting H1. Similarly, heuristic cues positively affected source trust (HC→ST, p < 0.001), supporting H2. Furthermore, both expectation trust (ET→AI, p < 0.001) and source trust (ST→AI, p < 0.001) positively influenced adoption intentions, confirming H3 and H4 (Table 5).

Table 5.

Path coefficients of the structural model (study 1).

Hypothesis	Path-coefficient	t values	p values	Bootstrap95% CI	Result
H1: HA→ET	0.525	15.746	0.001	[0.460, 0.591]	Supported
H2: HC→ST	0.401	11.328	0.001	[0.334, 0.472]	Supported
H3: ET→AI	0.273	7.062	0.001	[0.196, 0.347]	Supported
H4: ST→AI	0.346	8.993	0.001	[0.270, 0.421]	Supported

Note. AI= Adoption Intention; ET= Expectation Trust; HA= Health Anxiety; HC= Heuristic Cues; ST= Source Trust.

3.4.5. Mediation effect

This study proposes that expectation trust significantly influences the relationship between health anxiety and the adoption intention. In contrast, source trust is a significant mediator in the relationship between heuristic cues and the adoption intention. To empirically test these hypotheses, the bootstrap method, which includes bias-corrected confidence estimation, was employed to evaluate the mediating effects. More specifically, 95% confidence intervals for the mediating effects were calculated using 5000 bootstrap samples. The findings indicated that expectation trust significantly mediated the relationship between health anxiety and adoption intention (HA→ET→AI, p<0.001), and that source trust significantly mediated the relationship between heuristic cues and adoption intention (HC→ST→AI, p<0.001). The confidence intervals of the Bootstrap 95% CI values of the two mediation pathways did not contain 0, indicating that the significant mediating effect was confirmed. H3a and H4a are supported (Table 6).

Table 6.

Mediation analysis results (study 1).

Hypothesis	Path-coefficient	t values	p values	Bootstrap95% CI	Result
H3a:HA→ET→AI	0.144	6.202	0.001	[0.100, 0.191]	Supported
H4a:HC→ST→AI	0.139	6.745	0.001	[0.101, 0.182]	Supported

Note. AI= Adoption Intention; ET= Expectation Trust; HA= Health Anxiety; HC= Heuristic Cues; ST= Source Trust; Number of bootstrap samples (5000).

3.4.6. Moderation effect

Based on the research model, two variables, media literacy and adult children’s intervention, are introduced as moderators of the model. The results indicated that the interaction terms of media literacy and heuristic cues significantly negatively impacted source trust (HC×ML→ST, p<0.001), supporting hypothesis H5. Similarly, the interaction terms of adult children’s intervention with expectation trust and source trust significantly negatively affected adoption intention (ST×ACI→AI, p<0.001; ET×ACI→AI, p<0.001), which supports hypotheses H6a and H6b (Table 7). Media literacy weakens the positive effect of heuristic cues on source trust, and adult children’s intervention can also play a role in the transformation of trust into adoption intention.

Table 7.

Moderation analysis results (study 1).

Hypothesis	Path-coefficient	t values	p values	Bootstrap95% CI	Result
H5:HC×ML→ST	-0.245	-6.993	0.001	[-0.315, -0.179]	Supported
H6a:ST×ACI→AI	-0.219	-6.109	0.001	[-0.289, -0.149]	Supported
H6b:ET×ACI→AI	-0.199	-5.421	0.001	[-0.269, -0.126]	Supported

Note. AI= Adoption Intention; ET= Expectation Trust; HA= Health Anxiety; HC= Heuristic Cues; ML= Media Literacy; ACI= Adult Children’s Intervention; ST= Source Trust. Number of bootstrap samples (5000).

In conclusion, the results of study 1 support the dual-path model. However, cross-sectional data are difficult to confirm the strict causal effects of variables such as heuristic cues. To detect causality between variables in a more rigorous manner, it will be validated by experimental manipulation in the study 2.

4. Study 2: Causal establishment via randomized experiment

4.1. Sample selection and data collection

To effectively overcome the limitations of causal inference in Study 1, the causal relationship of the model is verified by designing a randomized experiment to further validate the research hypothesis. To avoid the learning and testing effects of previous studies and maintain the independence of the sample, this study re-recruited 262 older users over 60 years old in Taiyuan and Jinzhong, Shanxi Province. For this study, the criteria were specifically tailored to accommodate the experimental design: Inclusion Criteria: (1) Adults aged 60 years and above; (2) Individuals who have used mobile short video platforms to watch health information or advertisements within the past six months. Exclusion Criteria: (1) Participants who failed to pass the embedded directed-response item (attention checks) during the experiment; (2) Participants who submitted incomplete questionnaires.

4.2. Experimental materials

This study adopts a 2 (health anxiety: high vs. low) × 2 (heuristic cue: strong vs. weak) between-subjects design. Four 30-second vertical videos (9:16 ratio) were professionally produced, maintaining identical duration, background visual elements, and audio volumes across all conditions. The speakers in the video are played by professionally trained actors who follow a standardized script to ensure strict consistency in tone, speech speed, and facial expressions under all experimental conditions while ensuring consistent exposure time to effectively isolate manipulative variables. During the experiment, no on-site intervention executor was set up because the intervention was implemented entirely through pre-recorded online videos. Health anxiety manipulation: (1) The high health anxiety version emphasizes the severity and irreversible consequences of the disease, such as “if left untreated, the disease can be life-threatening” and other serious statements. (2) The low health anxiety version adopts a more neutral description, for example: “If left untreated, some people will experience symptoms such as dizziness and nausea.” Heuristic cue manipulation: (1) Strong cue condition description: the speaker in the short video is dressed in a doctor’s costume, the background shows the hospital’s logo, and medical terminology is used. (2) Weak cue condition description: the same speaker is dressed in casual clothes and uses everyday language in daily life. Simultaneously, 40 older participants were invited for pre-testing, and the results showed no significant differences in video clarity, interest, information comprehensibility, or situational realism ratings.

4.3. Measurement

To achieve fairness in the evaluation process of the two studies, this study continued to use the scale used in study 1, and all items were evaluated on a 5-point Likert scale (1=strongly disagree; 5=strongly agree). The experimental procedure followed a standardized online protocol to simulate real-world browsing: (1) First, participants were presented with the informed consent form and detailed instructions on their mobile screens. (2) Second, the online system utilized a randomized algorithm to assign each participant to view only one of the four 30-second intervention videos. (3) Third, participants were instructed to watch the video on their own mobile devices in their natural, everyday environment (e.g., at home). This naturalistic viewing environment was deliberately chosen to maximize ecological validity. (4) Finally, immediately after the video concluded, the system automatically redirected them to the post-test measurement scales. Where they completed two five and four-items operational validation scales (health anxiety and heuristic cues) and core construct scales (expectation trust, source trust, media literacy, adult children’s intervention, and adoption intention). A directive response item is presented in random order, distracted responses are detected, and demographic questions are presented. Twenty-two questionnaires were excluded from attention testing, and 240 valid samples were obtained, meeting the requirement for sample size saturation. Like Study 1, the mandatory response setting ensured no missing data among the retained samples. Furthermore, a priori power analysis using G*Power 3.1.9.7 for an ANOVA with four groups (2×2 design) showed that to detect a medium effect size (f = 0.25) with an alpha of 0.05 and a power of 0.95, a total sample size of 210 was required. Our final valid sample size of 240 exceeded this requirement, ensuring robust statistical power for causal inference. The detailed demographic characteristics of the Study 2 participants are presented in (Table 8).

Table 8.

The detailed demographic characteristics in study 2.

Measure	Items	Frequency	Percent
Gender	Male	115	48.0%
Gender	Female	125	52.0%

Note. Average age= 71.3 (SD=4.3).

4.4. Data analysis

Before hypothesis testing, chi-square tests and one-way ANOVA were conducted to confirm that demographic variables (e.g., age, gender, education) were evenly distributed across the four experimental groups (p>0.05), thereby effectively controlling for baseline confounders. The data were statistically analyzed using SPSS 29 and PROCESS 4.0.⁸⁰ First, internal consistency,⁵⁴ normality, homogeneity of variance, and multiple collinearity tests were performed for each variable.⁸¹ Prior to conducting the ANOVA, statistical assumptions were rigorously examined. The Shapiro-Wilk test indicated an acceptable normal distribution of the residuals, and Levene’s test confirmed the homogeneity of variance across the experimental groups (p>0.05). The manipulative effect of variables was verified by an independent sample t-test for high/low health anxiety and strong/weak heuristic cues,⁸² and the main and interactive effects of health anxiety and heuristic cues on expectation trust and source trust were tested by 2×2 analysis of variance. To test the moderating effects of media literacy and adult children’s intervention, they were included in the study as moderating variables for analysis. Using PROCESS Models (1) and (4), 5000 bootstraps were used for mediation and conditioning inspections.

4.5. Results

4.5.1. Operational verification

The Cronbach’s α coefficient for all items ranged from 0.851 to 0.903, reaching a minimum standard value of 0.70.⁷⁴ The subjective score of the high health anxiety group was significantly higher than that of the low health anxiety group, t (238) =13.227, p<0.001, and the effect size Cohen’s $d$ =1.849. The perception of the strong heuristic cue group was also significantly higher than that of the weak-cue group, t (238) =11.802, p<0.001, and the effect size Cohen’s $d$ =1.710; both groups had a large effect.⁸² Therefore, both sets of manipulations met expectations, and subsequent experiments could be continued.

The analysis of variance (ANOVA) results for expectation trust showed significant main effects for both health anxiety (p = 0.004) and heuristic cues (p < 0.001), as well as a significant interaction between them (p = 0.046) (Table 9 for full details including F-values and effect sizes η²p). The simple effect test showed that in the low health anxiety group, strong heuristic cues significantly increased expectation trust (p < 0.001), but in the high health anxiety group, this effect was not significant (p > 0.05).

Table 9.

Two-way ANOVA results (study 2).

Dependent variable	Independent variable	F	p values	η²
Expectation Trust	Health Anxiety	8.524	0.004	0.035
	Heuristic Cues	15.271	0.001	0.060
	Health Anxiety × Heuristic Cues	4.012	0.046	0.017
Source Trust	Health Anxiety	7.385	0.007	0.030
	Heuristic Cues	18.214	0.001	0.071
	Health Anxiety × Heuristic Cues	5.127	0.025	0.021

Similarly, the source trust analysis indicated significant main effects for both heuristic cues (p < 0.001) and health anxiety (p = 0.007), alongside a significant interaction effect (p = 0.025) (Table 9). The simple effect analysis revealed that for those with low health anxiety, heuristic cues significantly increased source trust (p < 0.001), while no significant difference was observed for those with high health anxiety (p > 0.05).

4.5.2. Mediation and moderation

The mediation results indicated that expectation trust played a significant mediating role between health anxiety and adoption intention (HA→ET→AI, p<0.001), and source trust played a significant mediating role between heuristic cues and adoption intention (HC→ST→AI, p<0.001) (Table 10). The positive effect of heuristic cues on source trust was significantly negatively moderated by media literacy (HC×ML→ST, p=0.002). The positive effect of expectation trust on adoption intention was significantly negatively moderated by the adult children’s intervention (ET×ACI→AI, p<0.001). The positive effect of source trust on adoption intention was also significantly negatively moderated by adult children’s intervention (ST×ACI→AI, p<0.001) (Table 11).

Table 10.

Mediation analysis results (study 2).

Hypothesis	Path-coefficient	t values	p values	Bootstrap95% CI	Result
H3a:HA→ET→AI	-0.136	-3.820	0.001	[-0.212, -0.071]	Supported
H4a:HC→ST→AI	0.121	3.760	0.001	[0.062, 0.197]	Supported

Note. AI= Adoption Intention; ET= Expectation Trust; HA= Health Anxiety; HC= Heuristic Cues; ST= Source Trust; Number of bootstrap samples (5000).

Table 11.

Moderation analysis results (study 2).

Hypothesis	Path-coefficient	t values	p values	Bootstrap95% CI	Result
H5:HC×ML→ST	-0.286	-3.160	0.002	[-0.466, -0.107]	Supported
H6a:ET×ACI→AI	-0.777	-3.964	0.001	[-1.161, -0.389]	Supported
H6b:ST×ACI→AI	-0.925	-4.812	0.001	[-1.297, -0.551]	Supported

5. Discussion

This study, by combining the Health Belief Model (HBM) and the Heuristic-Systematic Model (HSM), provides empirical evidence for the dual pathways that influence older adults’ intentions to adopt health behaviors presented in short videos. The results suggest that this process is shaped by both internal health-related motivations and external persuasive elements. Both studies confirm that expectation trust and source trust serve as critical mediators between these stimulus factors and behavioral intentions.

The results of this study are consistent with the reliability of the source and the theory of HSM.^32,59 The two-path structure broadens the traditional health persuasion framework, implying that both cognitive emotional needs and environmental cues can shape trust. Two studies revealed the two-sided effects of health anxiety in different contexts. The results of study 1 showed that high levels of health anxiety were positively correlated with expectation trust and adoption intention, indicating that chronic or idiosyncratic health anxiety in older adults prompted them to seek and rely on trusted health information sources. They actively seek and trust information and products that solve problems, in line with HBM’s predictions. However, in study 2, the effect was reversed when health anxiety was acutely induced through fear-inducing information, suggesting that excessive anxiety in older adults may trigger defensive processing or psychological reactance. This suggests that the impact of health anxiety on the persuasion effect may have a nonlinear or threshold effect; moderate anxiety promotes trust and acceptance, while excessive exogenous anxiety can trigger fear control in older adults and weaken trust and lead to resistance behavior. This result is consistent with previous persuasion studies that suggest that high-threat information can be counterproductive by triggering avoidance or distrust.^83,84

In addition, our findings provide important extensions and contrasts with previous studies. Previous research suggested that health anxiety generally leads to irrational spending behavior or higher levels of gullibility in older adults. However, previous literature has often identified aging-related anxiety as a common vulnerability. Our dual study design provides strong, data-based empirical support for a situational dependence mechanism. Specifically, we went beyond theoretical assumptions, and quantitative findings empirically suggested that long-term intrinsic anxiety (Study 1) prompted an aggressive, trust-seeking approach to health information (95% CI: [0.100, 0.191]), while acute, extrinsically induced fear (Study 2) triggered critical defensive processing and significantly reduced trust (95% CI: [-0.212, -0.071]). This empirical contrast underscores the importance of avoiding universal assumptions about cognitive vulnerability, showing that the source and nature of anxiety significantly influence information processing in older adults.

The results further show that media literacy as a critical defense mechanism for older adults has been verified in both studies, especially in study 2, where media literacy can effectively weaken the causal relationship of heuristic cues. Highly literate individuals can independently recognize and judge information content, thereby reducing the manipulation of video content cues, while low-literacy individuals are completely affected by their operation. This finding confirms that the core value of media literacy lies in improving critical evaluation skills and effectively inhibiting automated heuristic processing.^38,39

Adult children’s intervention, as an important factor affecting the adoption intention of health products in older adults, has also been confirmed, which significantly attenuates the translation of both expectation trust and source trust into final adoption intentions. The results of study 2 further confirm that adult children’s intervention can directly block the conversion from trust to adoption intention in older adults. The evidence indicates that when adult children participate in the decision-making regarding healthy product adoption among older adults, their impulsive decision-making diminishes, regardless of whether trust is derived from personal convictions or external influence. In the family-centric East Asian cultural milieu, the supervisory norms of adult offspring serve a pivotal veto function,⁴¹ particularly in intergenerational communication as a social filter. Moreover, children’s recommendations significantly influence their decision-making and can effectively mitigate potential digital risks.⁶³

5.1. Theoretical implications

This study provides several theoretical contributions. Firstly, HBM and HSM are innovatively integrated to construct a two-path trust model, and two persuasion mechanisms are explained at the same time: internal (demand-driven) and extrinsic (cue-driven). This integration deepens our understanding of how health-related emotions and cognitive heuristics collectively influence digital health behavior, providing a more comprehensive framework for persuasion processes in the digital age.⁶ Second, it deepens the research on trust in digital contexts. Through the empirical distinction between source trust and expectation trust, it reveals their entirely different formation paths and deepens the understanding of the multi-dimensional construct of trust. This echoes the multidimensional model of trust proposed by scholars such as McKnight et al.⁸⁵ and embodies it in the context of health advertising. This suggests that future online trust research must move beyond single-dimensional measurement to examine the various facets of trust and its differentiated antecedents. Third, it expands the cultural and causal boundaries of moderating variables. This study not only verifies the cognitive moderating effect of media literacy but also confirms the causal moderating power of children’s intervention, a variable with Chinese cultural characteristics, in decision-making. This study provides a new empirical example of the subjective normative construct in TPB theory in the context of East Asian culture and confirms the need to incorporate the micro perspective of family decision-making when studying older adults’ population. Finally, the combination of PLS-SEM and experimental verification provides a rigorous path for clarifying correlation and causality in digital persuasion research. This hybrid design provides stronger empirical support for causal inference and improves the internal validity of the study.

5.2. Practical implications

This study provides multi-dimensional enlightenment for intervening and guiding the digital health behavior of older adults. First, for older users, improving their own media literacy is the key. Especially in today’s Internet age, media literacy has transcended tool attributes and has become an enabler of social integration, risk resistance, and improvement of quality of life. Therefore, the government and society should carry out targeted training through universities for older adults, community older adults service centers, and other channels to help older adults learn to identify business intentions and verify sources to build the ability to resist the risk of incorrect information. Second, for adult children, well-intentioned intervention is crucial and effective. Children not only support older adults in daily life but also need to play the role of guardians and should play the role of information consultants more actively and patiently to help older adults identify the information content in complex information environments. Finally, for policymakers, health managers, and clinical practitioners, interventions must go beyond the broader scope of digital governance. Policymakers should mandate the establishment of a digital health green channel for short video platforms, prioritizing algorithm-driven exposure for videos verified by certified clinical professionals. In addition, health managers and clinicians should incorporate home-based digital literacy counseling into routine geriatric care, formally recognizing and empowering adult children as key information gatekeepers in the clinical intervention and decision-making process of older adults.

5.3. Limitations and future research

Although this study has been rigorous through multiple study designs, there are still limitations.

First, experimental operations rely on short video stimulation materials, which may not fully reflect the diversity of real-world media exposure, and future studies can use longitudinal or field experiments to enhance ecological validity.

Second, this study primarily draws its sample from online platforms, indicating that the respondents are members of the digitally active older population. For older adults who are significantly affected by the digital divide and lack basic network skills, their literacy levels may be lower, and they may face a higher risk of manipulation; therefore, the findings of this study should be generalized with caution when applied to older adult groups.

Third, in study 2, the adult children’s intervention was examined as a measurement variable rather than a manipulative variable. Although the results are statistically significant, to obtain more rigorous causal evidence, future studies can use scenario simulation experiments to directly manipulate children’s feedback types to further verify their causal moderation.

Fourth, this study mainly measures self-reported adoption intentions rather than actual purchase behavior. Although intention is a strong predictor of behavior, it can still be influenced by social expectation bias or intention-behavior gaps. Fifth, although Study 2 established a short-term causal relationship in a controlled experimental environment, the long-term cumulative effect of short video exposure on sustained health decision-making in older adults remains to be explored. Therefore, further validation of the results of extrapolating findings to the real world is needed.

Finally, both personal medical history and current physical condition can significantly influence the popularity of digital health applications. Although we have successfully controlled for basic demographic variables (e.g., age, gender) to ensure equal baseline levels between the experimental groups, our study currently does not fully document participants’ previous chronic diseases. Future studies must consider detailed medical history and baseline health assessment as key covariates or moderators.

6. Conclusion

This study deepens the understanding of the vulnerability and agency of older adults in the digital health information ecosystem. Breaking through the traditional single-perspective research framework, the research findings show that the digital health information consumption behavior of older adults is a complex process formed by the combination of internal health vulnerabilities (chronic anxiety) and external environmental triggers (heuristic marketing signals). The results emphasize that although digital media literacy is a key cognitive shield, intergenerational family intervention is the decisive gatekeeper mechanism. Conversely, considering that this investigation utilized self-reported intentions and simulated video stimuli, these theoretical understandings should be interpreted as an initial explanatory structure, not as conclusive findings. To comprehensively confirm these mechanisms within authentic purchasing contexts, subsequent longitudinal research will be crucial. Ultimately, protecting the rights and interests of older adults in the short video era requires a collaborative strategy that integrates algorithmic supervision, precision literacy education, and the active empowerment of family networks.

Footnotes

Acknowledgments

The authors would like to thank the communities and their residents for their kind cooperation and for facilitating the development of this study.

ORCID iDs

Wenxia Xuan

Kun Tian

Lijie Hao

Ethical considerations

This study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of the School of Animation and Digital Arts, Communication University of Shanxi (Approval Number: 20250917).

Consent to participate

Informed consent was obtained from all individual participants prior to their enrollment in the study. Before beginning the online questionnaire, participants were presented with the study’s purpose, data usage policies, and voluntary nature. All of participants were assured that their responses would remain confidential.

Author contributions

Conceptualization: WX, KT; Methodology: KT; Formal analysis and investigation: KT, LH; Writing - original draft preparation: WX; Writing - review and editing: KT, LH; Funding acquisition: KT. All authors read and approved the final manuscript.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Research Project on High-Quality Development in Shanxi (SXGZL2025003).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

References

Tian

Hao

Xuan

, et al. The impact of perceived value and affection on Chinese residents’ continuous use intention of mobile health science information: an empirical study. Front Public Health 2023; 11: 1034231. https://doi.org/10.3389/fpubh.2023.1034231

QuestMobile . China short video industry report. QuestMobile, 2024. https://www.questmobile.com.cn

Pew Research Center . Social media use in 2023. Pew Research Center, 2023. https://www.pewresearch.org

OECD . Promoting health literacy for the ageing population. OECD Health Working Papers, 2022, vol 143. https://www.oecd.org/health/promoting-health-literacy-for-ageing-populations.pdf

Zhao

Song

. Online health information seeking behaviors among older adults: systematic scoping review. J Med Internet Res 2022; 24(2): e34790. https://doi.org/10.2196/34790

Wang

Luan

. Research progress on digital health literacy of older adults: a scoping review. Front Public Health 2022; 10: 906089. https://doi.org/10.3389/fpubh.2022.906089

Zhang

. An intelligent recommendation model for health culture based on short video content analysis in the mobile internet environment. J Environ Public Health 2022; 2022(1): 6009974. https://doi.org/10.1155/2022/6009974

Xuan

Phongsatha

Hao

, et al. Impact of health science popularization videos on user perceived value and continuous usage intention: based on the CAC and ECM model framework. Front Public Health 2024; 12: 1382687. https://doi.org/10.3389/fpubh.2024.1382687

Raffoul

Santoso

, et al. Diet pills and deception: a content analysis of weight-loss, muscle-building, and cleanse and detox supplements videos on TikTok. Eat Behav 2024; 55: 101911. https://doi.org/10.1016/j.eatbeh.2024.101911

10.

Mozafari

Otaghi

Paskseresht

, et al. Effect of video games on cognitive performance and problem-solving ability in the aged with cognitive dysfunction: a randomized clinical trial. Iran J Med Sci 2025; 50(2): 77–86. https://doi.org/10.30476/ijms.2024.101861.3452

11.

Kim

Jeon

Byun

, et al. Exploring e-health literacy and technology-use anxiety among older adults in Korea. Healthcare 2023; 11(11): 1556. https://doi.org/10.3390/healthcare11111556

12.

Zhou

Xiang

Xie

. Better safe than sorry: a study on older adults’ credibility judgments and spreading of health misinformation. Univers Access Inf Soc 2023; 22(3): 957–966. https://doi.org/10.1007/s10209-022-00899-3

13.

Kim

. Trust in health information websites: a systematic literature review on the antecedents of trust. Health Informatics J 2016; 22(2): 355–369. https://doi.org/10.1177/1460458214559432

14.

Alzaidi

Agag

. The role of trust and privacy concerns in using social media for e-retail services: the moderating role of COVID-19. J Retail Consum Serv 2022; 68: 103042. https://doi.org/10.1016/j.jretconser.2022.103042

15.

Peng

Meng

Issaka

. Navigating persuasive strategies in online health misinformation: an interview study with older adults on misinformation management. PLoS One 2024; 19(7): e0307771. https://doi.org/10.1371/journal.pone.0307771

16.

Rosenstock

. Historical origins of the health belief model. Health Educ Monogr 1974; 2(4): 328–335. https://doi.org/10.1177/109019817400200403

17.

Champion

Skinner

. The health belief model. In: Health behavior and health education: theory, research, and practice. 4th ed. San Francisco, CA: Jossey-Bass, 2008, pp. 45–65.

18.

Limbu

Gautam

Pham

. The health belief model applied to COVID-19 vaccine hesitancy: a systematic review. Vaccines 2022; 10(6): 973. https://doi.org/10.3390/vaccines10060973

19.

Alqahtani

Chang

, et al. Knowledge, beliefs, and practice regarding human papillomavirus (HPV) vaccination among American college students: application of the health belief model. J Am Coll Health 2023; 71(8): 2329–2338. https://doi.org/10.1080/07448481.2021.1967362

20.

Lau

Lim

Wong

, et al. The health belief model and colorectal cancer screening in the general population: a systematic review. Prev Med Rep 2020; 20: 101223. https://doi.org/10.1016/j.pmedr.2020.101223

21.

Rakhshani

Razeghi

Kashfi

, et al. The effect of educational intervention based on health belief model on colorectal cancer screening behaviors. BMC Public Health 2024; 24(1): 1640. https://doi.org/10.1186/s12889-024-19180-8

22.

Lufthiani

Bukit

Siregar

. Health belief model in the prevention of chronic disease in older adults. J Kesehat Masy 2022; 17(4): 500–508. https://doi.org/10.15294/kemas.v17i4.27296

23.

Mao

Wang

Zheng

. Health belief model and social media engagement: a cross-national study of health promotion strategies against COVID-19 in 2020. Front Public Health 2023; 11: 1093648. https://doi.org/10.3389/fpubh.2023.1093648

24.

Shang

Zhou

Zuo

. Understanding older adults’ intention to share health information on social media: the role of health belief and information processing. Internet Res 2021; 31(1): 100–122. https://doi.org/10.1108/INTR-12-2019-0512

25.

Zhang

Baker

Pember

, et al. Persuading me to eat healthy: a content analysis of YouTube public service announcements grounded in the health belief model. South Commun J 2017; 82(1): 38–51. https://doi.org/10.1080/1041794X.2016.1278259

26.

Bechard

Bergelt

Neudorf

, et al. Using the health belief model to understand age differences in perceptions and responses to the COVID-19 pandemic. Front Psychol 2021; 12: 609893. https://doi.org/10.3389/fpsyg.2021.609893

27.

Keshavarz

Karimi

Nazari

, et al. The effect of a health belief model-based educational intervention on the determinants of intention to influenza prevention behaviors among the older adults. Educ Gerontol 2022; 48(8): 381–389. https://doi.org/10.1080/03601277.2022.2043612

28.

Bergman

Bodner

. Age is not just a number: age awareness, subjective nearness-to-death, self-esteem, and depressive symptoms among older adults. Aging Ment Health 2020; 24(6): 906–913. https://doi.org/10.1024/1662-9647/a000295

29.

Vassilenko

Taba

Marcello

, et al. The effects of a short video intervention on digital health literacy skills: protocol for an online randomised controlled trial. Health Lit Commun Open 2025; 3(1): 2489723. https://doi.org/10.1080/28355245.2025.2489723

30.

Kok Wah

. Empowering healthier decisions: the impact of consumer innovativeness on linking perceived benefits with protective health behaviors through digital marketing in the health-care sector. Int J Pharm Healthc Mark 2025; 20: 18–44. https://doi.org/10.1108/IJPHM-07-2024-0073

31.

Glanz

Bishop

. The role of behavioral science theory in development and implementation of public health interventions. Annu Rev Public Health 2010; 31(1): 399–418. https://doi.org/10.1146/annurev.publhealth.012809.103604

32.

Chaiken

. Heuristic versus systematic information processing and the use of source versus message cues in persuasion. J Pers Soc Psychol 1980; 39(5): 752–766. https://doi.org/10.1037/0022-3514.39.5.752

33.

Chaiken

Trope

(eds). Dual-process theories in social psychology. Guilford Press, 1999.

34.

Chaiken

. The heuristic model of persuasion. In: Social influence. Psychology Press, 2014, pp. 3–39.

35.

Bawden

Robinson

. Information overload: an overview. Oxford encyclopedia of political decision making. Oxford University Press, 2020. https://doi.org/10.1093/acrefore/9780190228637.013.1360

36.

Sbaffi

Rowley

. Trust and credibility in web-based health information: a review and agenda for future research. J Med Internet Res 2017; 19(6): e218. https://doi.org/10.2196/jmir.7579

37.

Chen

Hay

Waters

, et al. Health literacy and use and trust in health information. J Health Commun 2018; 23(8): 724–734. https://doi.org/10.1080/10810730.2018.1511658

38.

Livingstone

. What is media literacy? Intermedia 2004; 32(3): 18–20. https://eprints.lse.ac.uk/1027/

39.

Vraga

Tully

Akin

, et al. Modifying perceptions of hostility and credibility of news coverage of an environmental controversy through media literacy. Journalism 2012; 13(7): 942–959. https://doi.org/10.1177/1464884912455906

40.

Qiao

Chen

, et al. The role of filial piety in filial tourism: an intergenerational analysis of decision-making. Asia Pac J Tour Res 2024; 29(8): 1017–1031. https://doi.org/10.1080/10941665.2024.2362199

41.

Ajzen

. The theory of planned behavior. Organ Behav Hum Decis Process 1991; 50(2): 179–211. https://doi.org/10.1016/0749-5978(91)90020-t

42.

Kunda

. The case for motivated reasoning. Psychol Bull 1990; 108(3): 480–498. https://doi.org/10.1037/0033-2909.108.3.480

43.

Morgan

Hunt

. The commitment-trust theory of relationship marketing. J Mark 1994; 58(3): 20–38. https://doi.org/10.1177/002224299405800302

44.

Alnoor

Al-Abrrow

, et al. Uncovering the antecedents of trust in social commerce: an application of the non-linear artificial neural network approach. Compet Rev 2022; 32(3): 492–523. https://doi.org/10.1108/CR-04-2021-0051

45.

Gefen

Karahanna

Straub

. Trust and TAM in online shopping: an integrated model. MIS Q 2003; 27: 51–90. https://doi.org/10.2307/30036519

46.

Chen

Hsiao

. Purchase intention in social commerce: an empirical examination of perceived value and social awareness. Libr Hi Tech 2018; 36(4): 583–604. https://doi.org/10.1108/LHT-01-2018-0007

47.

Srivastava

Mukherjee

Datta

, et al. Impact of perceived value on the online purchase intention of base of the pyramid consumers. Int J Consum Stud 2023; 47(4): 1291–1314. https://doi.org/10.1111/ijcs.12907

48.

Guerra-Tamez

Kraul Flores

Serna-Mendiburu

, et al. Decoding Gen Z: AI's influence on brand trust and purchasing behavior. Front Artif Intell 2024; 7: 1323512. https://doi.org/10.3389/frai.2024.1323512

49.

Nikou

De Reuver

Mahboob Kanafi

. Workplace literacy skills—how information and digital literacy affect adoption of digital technology. J Doc 2022; 78(7): 371–391. https://doi.org/10.1108/JD-12-2021-0241

50.

Chen

, et al. Population prediction of Chinese prefecture-level cities based on multiple models. Sustainability 2022; 14(8): 4844. https://doi.org/10.3390/su14084844

51.

Tan

Teo

. Factors influencing the adoption of Internet banking. J Assoc Inf Syst 2000; 1(1): 5–44. https://doi.org/10.17705/1jais.00005

52.

Van de Vijver

Leung

. Methods and data analysis for cross-cultural research. Cambridge University Press, 2021, vol 116.

53.

Tongprasert

Rapipong

Buntragulpoontawee

. The cross-cultural adaptation of the DASH questionnaire in Thai (DASH-TH). J Hand Ther 2014; 27(1): 49–54. https://doi.org/10.1016/j.jht.2013.08.020

54.

Nunnally

Bernstein

. Psychometric theory. 3rd ed. McGraw-Hill, 1994.

55.

Salkovskis

Rimes

Warwick

HMC

, et al. The Short Health Anxiety Inventory (SHAI). Psychol Med 2002; 32(5): 843–853. https://doi.org/10.1017/s0033291702005883

56.

Lasher

Faulkender

. Measurement of aging anxiety: development of the Anxiety about Aging Scale. Int J Aging Hum Dev 1993; 37(4): 247–259. https://doi.org/10.2190/B62J-24A6-8A46-X1G2

57.

Metzger

Flanagin

Medders

. Social and heuristic approaches to credibility evaluation online. J Commun 2010; 60(3): 413–439. https://doi.org/10.1111/j.1460-2466.2010.01488.x

58.

Petty

Cacioppo

. The elaboration likelihood model of persuasion. In: Berkowitz

(ed). Advances in experimental social psychology. Academic Press, 1986, vol 19, pp. 123–205.

59.

Hovland

Janis

Kelley

. Communication and persuasion: psychological studies of opinion change. Yale University Press, 1953.

60.

Ohanian

. Construction and validation of a scale to measure celebrity endorsers’ perceived expertise, trustworthiness, and attractiveness. J Advert 1990; 19(3): 39–52. https://doi.org/10.1080/00913367.1990.10673191

61.

Munuera-Aleman

Delgado-Ballester

Yague-Guillen

. Development and validation of a brand trust scale. Int J Mark Res 2003; 45(1): 1–18. https://doi.org/10.1177/147078530304500103

62.

Norman

Skinner

. eHEALS: the eHealth literacy scale. J Med Internet Res 2006; 8(4): e27. https://doi.org/10.2196/jmir.8.4.e27

63.

Rau

PLP

. Filial piety and its mixed effects on older adults’ adoption of smart home care technology in China. Comput Hum Behav 2021; 120: 106752. https://doi.org/10.1016/j.chb.2021.106752

64.

Dodds

Monroe

Grewal

. Effects of price, brand, and store information on buyers' product evaluations. J Mark Res 1991; 28(3): 307–319. https://doi.org/10.2307/3172866

65.

Sussman

Siegal

. Informational influence in organizations: an integrated approach to knowledge adoption. Inf Syst Res 2003; 14(1): 47–65. https://doi.org/10.1287/isre.14.1.47.14767

66.

Hulland

. Use of partial least squares (PLS) in strategic management research: a review of four recent studies. Strateg Manag J 1999; 20(2): 195–204. https://doi.org/10.1002/(SICI)1097-0266(199902)20:2<195::AID-SMJ13>3.0.CO;2-7

67.

Haenlein

Kaplan

. A beginner’s guide to partial least squares analysis. Underst Stat 2004; 3(4): 283–297. https://doi.org/10.1207/s15328031us0304_4

68.

Petter

Straub

Rai

. Specifying formative constructs in information systems research. MIS Q 2007; 31: 623–656. https://doi.org/10.2307/25148814

69.

Hair

. A primer on partial least squares structural equation modeling (PLS-SEM). Sage, 2014.

70.

Majchrzak

Malhotra

John

. Perceived individual collaboration know-how development through information technology–enabled contextualization: evidence from distributed teams. Inf Syst Res 2005; 16(1): 9–27. https://doi.org/10.1287/isre.1050.0044

71.

Kline

. Principles and practice of structural equation modeling. 5th ed. Guilford Press, 2023.

72.

Podsakoff

MacKenzie

Lee

, et al. Common method biases in behavioral research: a critical review of the literature and recommended remedies. J Appl Psychol 2003; 88(5): 879–903. https://doi.org/10.1037/0021-9010.88.5.879

73.

Hair

Risher

Sarstedt

, et al. When to use and how to report the results of PLS-SEM. Eur Bus Rev 2019; 31(1): 2–24. https://doi.org/10.1108/ebr-11-2018-0203

74.

Hair

Black

Babin

, et al. Multivariate data analysis. 7th ed. Prentice Hall, 2010.

75.

Chin

. The partial least squares approach to structural equation modeling. In: Modern methods for business research. Lawrence Erlbaum Associates, 1998. https://doi.org/10.4324/9781410604385-10

76.

Fornell

Larcker

. Evaluating structural equation models with unobservable variables and measurement error. J Mark Res 1981; 18(1): 39–50. https://doi.org/10.1177/002224378101800104

77.

Gold

Malhotra

Segars

. Knowledge management: an organizational capabilities perspective. J Manag Inf Syst 2001; 18(1): 185–214. https://doi.org/10.1080/07421222.2001.11045669

78.

Henseler

Hubona

Ray

. Using PLS path modeling in new technology research: updated guidelines. Ind Manag Data Syst 2016; 116(1): 2–20. https://doi.org/10.1108/IMDS-09-2015-0382

79.

Hair

Hult

GTM

Ringle

, et al. A primer on partial least squares structural equation modeling (PLS-SEM). 3rd ed. Sage, 2022.

80.

Hayes

. Introduction to mediation, moderation, and conditional process analysis: a regression-based approach. Guilford Press, 2017.

81.

Field

. Discovering statistics using IBM SPSS statistics. Sage, 2024.

82.

Cohen

. Statistical power analysis for the behavioral sciences. Routledge, 2013.

83.

Nabi

. Emotional flow in persuasive health messages. Health Commun 2015; 30(2): 114–124. https://doi.org/10.1080/10410236.2014.974129

84.

Witte

Allen

. A meta-analysis of fear appeals: implications for effective public health campaigns. Health Educ Behav 2000; 27(5): 591–615. https://doi.org/10.1177/109019810002700506

85.

McKnight

Choudhury

Kacmar

. Developing and validating trust measures for e-commerce: an integrative typology. Inf Syst Res 2002; 13(3): 334–359. https://doi.org/10.1287/isre.13.3.334.81