Sage Journals: Discover world-class research

Abstract

WeChat tweets have become an important way to spread knowledge of the importance of personal information security and to enhance older Chinese adults’ awareness of information security. Based on the knowledge adoption model and the elaboration likelihood model, this study identifies two types of tweets’ source features, that is, publisher type (official or unofficial) and author type (expert or non-expert). In addition, based on cognitive fit theory, this study explores the effects of the consistency and inconsistency between the two types of source features on the knowledge adoption intentions of older Chinese adults. The proposed research model is validated by carrying out a 2 × 2 between-subjects experiment, which also takes into account the three control variables of: “amount of readership,” “gender,” and “the perceived credibility of WeChat.” The results show that, compared with non-official tweets or non-expert tweets, official tweets or expert tweets are more conducive to enhancing older adults’ perception of the credibility and usefulness of the tweets and thus improve their knowledge-adoption intentions. However, the inconsistent combinations of unofficial and expert tweets and official and non-expert tweets can weaken the positive impact of official or expert sources on perceived credibility and usefulness. In addition, as the older adults’ perception of the relevance to themselves of personal information security increases and as their ability to process tweets strengthens, the influence of source features on perceived credibility and usefulness diminishes.

Keywords

older adults knowledge adoption intention personal information security WeChat tweet source features information consistency

Introduction

With the comprehensive implementation of various policies, such as strengthening the research and development of age-friendly information technology equipment and providing community training to promote older adults’ information skills, the barriers that previously stopped older adults from accessing the internet have been significantly reduced (Zhang et al., 2023). Widespread penetration of the internet by the older population is driving their growing demand for internet connectivity, browsing, and utilization (AARP Research, 2022). Nevertheless, the significant gap in information technology ownership, usage, and proficiency between older adults and other age groups means the awareness and practice of information security within the older demographic significantly lag behind their internet engagement (Bayne et al., 2023). The prevalence of internet scams has seen older adults falling prey to fake short videos or getting trapped in fraudulent schemes. This continues to undermine their sense of security in cyberspace (Moore & Hancock, 2020). Consequently, addressing the issue of how to enhance information security education is imperative (Bayne et al., 2023; Moore & Hancock, 2020). Steps must be taken to foster greater awareness among older adults and to facilitate their optimal utilization of the internet’s many advantages.

In 2019, China implemented the national strategy of “Strengthening Cybersecurity Awareness and Skills for All People.” Now, social media platforms such as WeChat (the predominant social media platform in China) have emerged as prominent and original channels through which personal information security knowledge can be disseminated (National Computer Network Emergency Response Coordination Center [CNCERT/CC], 2021). Via tweets, authoritative institutions and self-publishers utilize their WeChat public accounts to share with the public valuable knowledge on information security. Such knowledge can enhance older adults’ information literacy (Kim & Lee, 2023), transform their mindsets, and influence their behaviors regarding information security, ultimately contributing to their overall security posture. As a result, WeChat tweets have become the most efficient, cost-effective, and impactful means of promoting public information security education. WeChat serves as the primary gateway for older Chinese adults to access the internet (Shang et al., 2020), and therefore, WeChat tweets have also naturally become the main source of information security education for older adults. However, existing studies have emphasized that knowledge dissemination performance through social media is highly dependent on the recipients’ willingness to adopt knowledge (Hajli, 2016; Sussman & Siegal, 2003). This means that social media operators and tweet content creators need to fully understand the factors that influence recipients’ knowledge adoption intentions. This is also the critical path to effectively creating content and improving the performance of personal information security knowledge transfer. However, existing studies have only focused on exploring the factors that influence the adoption of social media–based health knowledge among older adults (Huo, Ma, et al., 2018; Huo, Zhang, & Ma, 2018; Shang et al., 2020). Consequently, there is an urgent need to systematically explore the influencing factors and the mechanism associated with older adults’ adoption intentions regarding social media–based personal information security knowledge.

Existing research based on the knowledge adoption model has emphasized the critical role of recipients’ perceptions of the credibility and usefulness of the information in shaping recipients’ knowledge adoption intentions (Hajli, 2016; Sussman & Siegal, 2003). Basing their studies on the elaboration likelihood model (ELM), some researchers have further proposed that, except for argument quality (which acts on the central route), source credibility (which acts on the peripheral route) is also an important factor in improving the recipient’s perceptions of information’s credibility and usefulness (Shahab et al., 2021). These studies have summarized the heuristic cues that are beneficial for improving source credibility (Sundar, 2008). The cues are manifested as various source features, such as reviewers’ ratings, reviewers’ reliability, and expertise (Hlee et al., 2019). Notably, older adults exhibit a higher level of trust (compared to other age groups) in information sources such as experts and official sources (Fung et al., 2023). Consequently, this study examines the influence of source credibility, characterized by a tweet’s publisher type (official or unofficial) and author type (expert or non-expert) on older adults’ perceived credibility, perceived usefulness, and knowledge adoption intention with regard to the published information. Inconsistency of information may undoubtedly interfere with recipients’ information decision-making (Mohammad et al., 2017). In spite of this fact, only a limited number of studies have paid attention to the potential effects of inconsistency between cues on perceived usefulness and perceived credibility (Li et al., 2024). In this vein, this study fills the abovementioned research gap by exploring the interaction effect of these two source features on perceived usefulness and perceived credibility.

Furthermore, existing researchers have found that the recipients’ information processing capability (Arendt et al., 2018; Shahab et al., 2021) and information involvement (Tang, 2016; Tang & Wu, 2022) can potentially diminish the positive effect of peripheral cues (such as source credibility) on knowledge adoption. However, due to age-related cognitive limitations, older adults generally possess lower levels of information processing capability than other age groups (Fung et al., 2023). In addition, due to the prevalence of cyber fraud and scams that target older adults, the elderly often consider themselves as marginal groups in cyberspace (Bayne et al., 2023; Van den Broeck et al., 2015). This results in the elderly’s generally low awareness of and limited attention to information security (Ebner et al., 2020), which may create obstacles to their involvement in learning about personal information security. Therefore, this study further investigates whether these two factors can continue to weaken the facilitating effect of peripheral cues on older adults’ knowledge adoption intentions. The results provide more empirical evidence that helps to deepen the understanding of the influence mechanism of peripheral cues on older adults’ adoption intentions toward social media–based personal information security knowledge.

Theoretical Foundation and Literature Review

Knowledge Adoption Model

Knowledge transfer encompasses activities that facilitate the flow of knowledge from one person, group, or organization to another (Sussman & Siegal, 2003). At the individual level, knowledge transfer focuses on how individuals apply the knowledge acquired in one context to another context (Sussman & Siegal, 2003). Knowledge adoption, a crucial process in the knowledge management cycle, refers to the internalization phase of knowledge transfer (Chou et al., 2015). During this phase, individuals transform explicit information into internalized knowledge and meaning (Hardy et al., 2018). This transformation reflects the extent to which individuals utilize (adopt) the transferred knowledge (Huo, Zhang, & Ma, 2018). Therefore, in the context of online platforms, knowledge adoption is defined as the willingness of recipients to accept, implement, or utilize the knowledge they acquired online (Chou et al., 2015).

The knowledge adoption model emphasizes the recipient’s perception of information usefulness and credibility as the primary factors influencing the recipient’s knowledge adoption (Sussman & Siegal, 2003). This model has been extensively applied in the study of knowledge adoption across various domains. Examples include information adoption behavior in online virtual communities (Chou et al., 2015), social media platforms (Erkan & Evans, 2016; Huo, Zhang, & Ma, 2018), and e-commerce reviews (Tang & Wu, 2022). The widespread use of this model confirms its universal relevance in understanding various knowledge adoption phenomena. Studies based on this model have further proposed that, for a given piece of information, apart from the quality of the information itself, both the perceived usefulness and the trustworthiness of that information are primarily influenced by the source credibility of the information (Sussman & Siegal, 2003). “Source credibility” refers to the extent to which users believe that a source can provide objective and relevant information or knowledge, and the degree of that belief is based on the source’s expertise and authenticity (Chou et al., 2015). To be specific, “expertise” refers to the level of knowledge and experience that the source possesses regarding the topic related to the user’s information needs (Sussman & Siegal, 2003). For instance, information published by experts plays a vital role in promoting users’ comprehension of scientific texts (Tzu-Jung et al., 2014) and positively influences users’ evaluation of online health information (Huo, Ma, et al., 2018; Kostagiolas et al., 2014), which in turn facilitates knowledge adoption in these contexts. In addition, “authenticity” emphasizes the integrity and honesty of the information source (Sussman & Siegal, 2003). For example, e-commerce websites that acquire third-party authentication enhance consumers’ perception of the credibility of the platform’s information sources (Filieri et al., 2018). Similarly, information published by official hospital publications (as opposed to health care business-like publications) is more likely to contribute to the public’s perception of source credibility, thereby facilitating knowledge adoption (Huo, Zhang, & Ma, 2018).

ELM and the Knowledge Adoption Behavior of Older Adults

The ELM provides a solid theoretical foundation for studying this phenomenon and explains how people’s attitudes change through both the central and peripheral routes (Kitchen et al., 2014; Shahab et al., 2021). The model offers valuable insights for this study since the process by which older adults decide whether to adopt knowledge, based on their needs after reading tweets, can be viewed as the process of individual attitude change. The ELM is widely recognized as one of the most influential theoretical models for explaining consumers’ information processing behavior (Kitchen et al., 2014). The ELM elucidates the two ways in which individuals process cognitive information, namely the central route and the peripheral route (Petty & Cacioppo, 1986). The central route involves individuals carefully considering and evaluating the received information, based on their previous experiences and knowledge, thereby engaging in in-depth processing of that information (Petty & Cacioppo, 1986). Conversely, the peripheral route offers a quick way for individuals to accept or reject information. Taking this route emphasizes that individuals do not actively invest significant cognitive resources or effort in processing information events. Instead, they process the information briefly, relying on heuristic cues and other shortcuts (Petty & Cacioppo, 1986). Furthermore, the ELM posits that the outcome of an individual’s attitude change is influenced by three factors: (1) content strength, which refers to the quality of the information itself; (2) peripheral cues, such as source credibility; and (3) the likelihood of elaborate processing (Kitchen et al., 2014). Thus, when the likelihood of elaborate processing is high, individuals will be more inclined to thoroughly examine the content and consider other relevant material related to the news event. In these instances, such individuals are more susceptible to factors like content quality. Conversely, when the likelihood of elaborate processing is low, individuals tend to minimize cognitive effort and become more influenced by heuristic cues (Musuva et al., 2019; Shahab et al., 2021).

In this age of information explosion, it is hugely challenging and somewhat infeasible that information users will expend significant effort in reading every argument or topic in a tweet (Özkan & Tolon, 2015). Going a step further, older adults, who belong to the information-disadvantaged group, inherently struggle with screening and judging the quality of information (Zhang et al., 2023). Consequently, when evaluating arguments, older adults rarely engage in elaborate processing and are more influenced by heuristic cues (Bayne et al., 2023; Shao et al., 2019). Therefore, this study primarily focuses on examining the mechanism through which source features impact older adults’ willingness to adopt knowledge. However, existing research based on the ELM has also suggested that an individual’s motivation to engage in elaborate information processing (Tang & Wu, 2022) and their ability to process content (Arendt et al., 2018) are crucial factors that negatively influence the effect of heuristic cues. Therefore, this study further investigates whether older adults’ level of information involvement and their information processing capability both mitigate the impact of heuristic cues (such as source features) on their willingness to adopt knowledge.

Hypotheses Development and Research Model

Source Features of Tweets and Perceived Credibility

Perceived credibility refers to the overall perception of information receivers with regard to the credibility of transferred information (Tzu-Jung et al., 2014). Source credibility, as a core component of perceived credibility, is a fundamental requirement for the formation of users’ willingness to adopt information and knowledge (Chou et al., 2015). Sussman and Siegal (2003) highlighted how individuals consider various source characteristics—such as the expertise and credibility of the information publisher—when making judgments about source credibility. These two types of source features have consistently emerged as the most influential heuristic cues in terms of shaping user credibility judgments (Kitchen et al., 2014). Similarly, older adults, being an information-disadvantaged group, face inherent challenges in screening and identifying information (Bayne et al., 2023). This leads older adults to rely more heavily on heuristic cues during the knowledge acquisition process (Huo, Ma, et al., 2018).

Compared to unofficial public accounts, such as personal self-media, official public accounts associated with government agencies and public service organizations are better positioned to signify the authority and authenticity of the information source. This is particularly relevant in the field of information security, an area that, in recent years, has experienced increased regulation in China (CNCERT/CC, 2021). The complex, comprehensive, and important nature of information security has reinforced older individuals’ perception of the professionalism of the knowledge published within this domain (AARP Research, 2022). Consequently, older adults are more inclined to perceive the content of information security tweets published by official public accounts as being more credible. Furthermore, authorship serves as a critical cue for users when assessing content quality (Saarinen & Vakkari, 2013; Tang, 2016). The author’s expertise and popularity also contribute to the perceived professionalism and authority of the information (Tang, 2016). Therefore, providing information about the author’s expertise in tweets can reduce older adults’ uncertainty about the information and enhance their trust in both the source and content and ultimately increase the perceived credibility of the information (Huo, Ma, et al., 2018; Huo, Zhang, & Ma, 2018). Based on this, the following hypotheses are proposed:

H1: Compared to unofficial public accounts, older adults perceive that the credibility is higher when the tweet related to personal information security knowledge is published by official public accounts.

H2: Compared to non-expert authors, older adults perceive that the credibility is higher when the tweet related to personal information security knowledge is published by expert authors.

Source Features and Perceived Usefulness

Perceived usefulness refers to the extent to which information recipients perceive the information as being valuable in enhancing their knowledge in a specific domain (Li et al., 2024). Source expertise, on the other hand, stems from the information recipient’s overall perception of the skills or knowledge possessed by the source in a particular domain (Sussman & Siegal, 2003). The latter serves as a crucial determinant of the information recipient’s perception of the usefulness of the information (Sussman & Siegal, 2003).

In traditional environments, individuals can evaluate the professionalism of the publisher and author of the released information based on tangible cues, such as the physical office address of the organization and the actual title of the information publisher. By comparison, assessing the source’s professionalism becomes significantly challenging in an online environment (Chou et al., 2015; Filieri et al., 2018). For example, in an online environment, information is often disseminated through anonymous relationships and predominantly takes the form of asynchronous communication activities (Filieri et al., 2018). This leads older adults to rely more on other discernible cues within tweets, such as whether the tweets originate from an official organization or have been authored by an expert in a specific field. Older adults use such information to better evaluate the source’s professionalism.

In addition, the professionalism of personal information security knowledge goes significantly beyond the cognitive scope of ordinary people’s daily work and life (Van den Broeck et al., 2015). This unknown field of knowledge can enhance older adults’ risk perception (Ebner et al., 2020) and improve their inherent dependence on the concept of information being “official.” This provides them with a greater sense of confidence in the truthfulness (Carstensen & Hershfield, 2021) and utility (Fung et al., 2023) of the information being disseminated. Such higher risk perception also encourages older adults to trust “experts” and view the associated content as professional (Huo, Ma, et al., 2018). This trust in turn can lead older adults to find that the information source is substantially more credible and reliable (Arendt et al., 2018), consequently fostering a greater belief that the information can provide valuable insights to enhance their knowledge of information security. Thus, older adults’ perceived usefulness is higher when tweets are written by experts. The following hypotheses are therefore proposed:

H3: Compared to unofficial public accounts, older adults’ perceived usefulness is higher when the tweet related to personal information security knowledge is published in official public accounts.

H4: Compared to non-expert authors, older adults’ perceived usefulness is higher when the tweet related to personal information security knowledge is published by expert authors.

Interactive Effects of Publisher Type and Author Type on Perceived Credibility and Perceived Usefulness

According to cognitive fit theory, the key to influencing individual performance decisions lies in the matching relationship between problem representation and task type (Vessey & Galletta, 1991). The term “problem representation” refers to the ways in which problems are expressed, such as symbolic representations of tables or spatial representations of graphs. Meanwhile, problem-solving tasks are categorized into symbolic tasks (such as numerical retrieval) and spatial tasks (such as trend retrieval; Vessey, 1991). Previous research has suggested that cognitive fit only occurs when the problem representation and the problem-solving task are of the same type. This leads to consistent mental representations for problem-solving and facilitates faster and more accurate decision-making (Vessey & Galletta, 1991). Similarly, whether or not older adults adopt personal information security knowledge through tweets can be viewed as an information processing problem for the recipients. The processing performance of this problem is also influenced by the matching relationship between the representation of tweets and the task type. Specifically, due to the professionalism of personal information security knowledge, when older adults receive tweets about such knowledge, this task is represented as a retrieval task for professional text information. To enhance the quality of information processing efficiency, older adults will actively seek out symbolic representation-class information that aligns with the task type. However, if internal inconsistency is found within the symbolic class of information that defines the problem, this can hinder older adults’ extraction and comprehension of such information (Li et al., 2024). Hence, the consistency in the form of issue representation, which encompasses tweet publisher type and author type, plays a crucial role in older adults’ overall processing of the source feature information. This, in turn, impacts older adults’ overall perception of the source’s credibility and subsequently influences their evaluation of the credibility and usability of the content.

Specifically, the combination of official publishing bodies and expert authors, characterized by authority, formality, and professionalism, best exemplifies consistency in the representation of an issue (Shahab et al., 2021; Sussman & Siegal, 2003). Therefore, compared to the other three source feature combinations, the combination of these two types of source features best reflects the consistency of professionalism and authenticity connotations of problem characterization information. This combination also facilitates the cognitive fit between the issue representation and the task of processing tweet information (Guo & Li, 2022; Vessey, 1991; Vessey & Galletta, 1991), thereby enhancing older adults’ perception of the tweet content’s credibility and usability (Rafique et al., 2024). Based on this, the following hypothesis is proposed:

H5: Among the four combinations of source features, tweets published by an official public source and authored by an expert have the greatest impact on enhancing older adults’ perceived credibility (a) and perceived usefulness (b).

The current combination of non-professionals and non-experts also ensures a high level of consistency in the representational information that tweets lack in terms of expertise and authenticity. This consistency helps to reduce the cognitive effort required for older adults to evaluate the source of any given tweet (Li et al., 2024; Rafique et al., 2024). This consistency will further reduce the professional attributes of the content, enhance the tweets’ experiential attributes (Ebner et al., 2020), and make older people more likely to perceive the tweet as a casual conversation about personal information security, rather than a professional domain article on related knowledge (Kelton et al., 2010). All of this will serve to reduce older adults’ perception of the usefulness and credibility of the tweet (Rafique et al., 2024). Hence, the following hypothesis is proposed:

H6: Tweets posted by unofficial publishers and written by non-experts enhance the negative impact of existing source features (unofficial publisher or non-expert authors) on the perceived credibility (a) and usefulness (b) among older adults.

However, the combinations of official publishers and non-expert authors on the one hand and unofficial publishers and expert authors on the other hand are ambivalent representations of the two dimensions of publishing type and author type in terms of expertise (Li et al., 2024; Mohammad et al., 2017). These combinations increase the cognitive effort required for older adults to process the overall information from the source (Özkan & Tolon, 2015; Vessey & Galletta, 1991), causing them to question the authenticity of the publishing organization’s public account. They may even doubt the plausibility and professionalism of the experts. The use of such combinations may even lead to a sense of aversion toward the experts’ information. Therefore, for older adults, the combination of source features that exhibit inconsistent problem representation information adds to the cognitive burden when processing problem representation information (Guo & Li, 2022). Such combinations of source features interfere with the realization process of cognitive matching and consequently diminish the influence of each of these features in enhancing older adults’ perceived trustworthiness and usability. Hence, the following hypothesis is proposed:

H7: The combinations of unofficial publishers and expert authors and official publishers and non-expert authors diminish the positive effects of existing source features (official publisher or expert authors) on perceived credibility (a) and perceived usefulness (b).

Perceived Credibility, Perceived Usefulness, and Knowledge Adoption Intention

Previous studies have demonstrated that the features of credible sources (such as experts and authorities) play a crucial role in guiding information receivers’ decision-making processes through a peripheral path rather than complex cognitive processes (Sussman & Siegal, 2003). Therefore, the higher the perceived credibility of personal information security knowledge tweets among older adults is, the more motivated those older adults will be to actively adopt the knowledge conveyed in the tweets. Furthermore, information recipients perceive that content that helps them familiarize themselves with the information and evaluate its quality before making decisions encourages them to be more willing to adopt the content (Filieri et al., 2018). This suggests that older adults’ willingness to adopt knowledge is stronger when they perceive that the content in the tweets is beneficial to enhancing their awareness of information security. Based on these observations, the following hypotheses are proposed:

H8: Perceived credibility positively influences older adults’ adoption intentions of personal information security knowledge from tweets.

H9: Perceived usefulness positively influences older adults’ adoption intentions of personal information security knowledge from tweets.

Moderating Effect of Information Processing Capability on Source Features, Perceived Credibility, and Perceived Usefulness

The term “information processing capability” refers to an individual’s capacity to engage in mental processing tasks, such as discovering, categorizing, organizing, and ultimately making decisions about information (Özkan & Tolon, 2015). However, when individuals are exposed to large amounts of information—both in terms of quantity and quality—than their information processing capacity allows, they may experience information overload. This will lead to suboptimal decision-making and non-exhaustive evaluation processes (Özkan & Tolon, 2015). In other words, individuals with weaker information processing capabilities are more likely to encounter information overload when exposed to the same amount of information as individuals with stronger abilities. This will result in there being insufficient time to evaluate all the information and potentially missing critical components (Gao et al., 2012). This limited capacity to process, respond to, and comprehend information increases the likelihood that, when there is insufficient information, individuals will rely more heavily on heuristic cues to assess the overall content (Arendt et al., 2018; Gao et al., 2012). What is evident is that the weaker an older person’s information processing skills are, the more difficult it is for them to handle the content of tweets. This, in turn, makes it more difficult for them to assess the credibility and usefulness of the content of tweets using the central pathway. Consequently, they are more likely to rely on the peripheral pathway, where heuristic cues (such as the publication type, topic, and author’s expertise) play a more prominent role in influencing their evaluations. Hence, the following hypotheses are proposed:

H10: The weaker the information processing capability of older adults, the stronger the effect of publisher type on perceived credibility (a) and perceived usefulness (b).

H11: The weaker the information processing capability of older adults, the stronger the effect of author type on perceived credibility (a) and perceived usefulness (b).

Moderating Effect of Information Involvement on Source Features With Perceived Credibility and Perceived Usefulness

The term “information involvement” refers to the degree to which individuals perceive the content of information as relevant to their own interests and needs (Tang, 2016). Previous research has demonstrated that, when individuals perceive low involvement or low relevance to themselves, they engage in less elaborate information processing. Conversely, when individuals perceive high involvement or high relevance to themselves, they are highly motivated to process the information and exert cognitive efforts to thoroughly analyze that information (Sussman & Siegal, 2003). Consequently, individuals who deem themselves to be highly involved in the information are more likely to utilize the central route of processing (Tang & Wu, 2022). Conversely, in their evaluation processes, individuals with lower involvement are more inclined to rely on peripheral cues, such as the credibility of the information source (Kitchen et al., 2014). Similarly, although older adults often perceive themselves as being marginalized in the digital society, the increasing instances of older adults falling victim to online fraud have heightened their risk perceptions in the realm of cyberspace (Bayne et al., 2023). As a result, more older adults have developed a heightened sense of involvement in information security tweets. The greater the level of information involvement is, the more likely older adults are to perceive the content of tweets as personally relevant. This increased motivation to process the information leads them to rely more heavily on the quality of the content and arguments, rather than the source characteristics, when assessing the credibility and usefulness of the information. Based on these observations, the following hypotheses are proposed:

H12: The stronger the information involvement of older adults, the weaker the effect of publisher type on perceived credibility (a) and perceived usefulness (b).

H13: The stronger the information involvement of older adults, the weaker the effect of author type on perceived credibility (a) and perceived usefulness (b).

The research framework is shown in Figure 1.

Figure 1.

The research framework.

Research Methodology

Experimental Design

In this study, the scenario experiment method is employed to validate the research model. The experimental design consists of a between-group experiment with a 2 (publisher type: official, unofficial) × 2 (author type: expert, non-expert) framework. Preventing elderly people from being scammed online is a key focus of information security education in China (CNCERT/CC, 2021). Therefore, the tweet is themed around authentic materials pertaining to the improvement of personal financial information security for older adults.

The manipulation design of specific variables is illustrated in Figure A1 (in Appendix A). For information sourced from official public accounts, a prompt was displayed to participants, stating: “The following is a tweet related to personal information security knowledge, which was published by the Public Account ‘Network Information Security Education’. This public account is registered by the Level Evaluation Center of Information Security Protection Education. The center is a third-party testing and evaluation organization approved by the Personnel Department of the Ministry of Education of China. The center is also affiliated with the Education Management Information Center of the Ministry of Education of China, which provides a range of security services, such as information system security level evaluation, risk assessment, confidential evaluation, off-site disaster recovery services, information security protection training and education, and so on.” In contrast, for information originating from an unofficial public account, the prompt displayed was: “The following is a tweet related to personal information security education, which was issued by the public account ‘Information Security DaWenjing.’ The public account ‘Information Security Dawenjing’ focuses on popularizing and promoting security and provides the public with the latest information on information security and review articles.” After carefully reading all the experiment materials, each participant was required to fill in the questionnaire.

The questionnaire used in this study incorporated various items from relevant literature. Minor modifications were incorporated to align with the themes of the scenarios. To assess perceived usefulness, three items were adapted from the work of Sussman and Siegal (2003). These items include: “I believe that adopting this tweet’s content can enhance personal awareness of financial information security protection.” “I believe that adopting this tweet’s content can improve the effectiveness of financial information security protection for myself.” “I believe that adopting this tweet’s content is beneficial for my own financial information security protection.” Perceived credibility was evaluated using three items from the study by Sussman and Siegal (2003). These items were: “I believe the tweet’s content is true.” “I believe the tweet’s content is accurate.” “I believe the tweet’s content is trustworthy.”

Furthermore, information processing capability was measured using six items from Özkan and Tolon (2015). The items include: “It is difficult for me to comprehend the content of this tweet.” “It is challenging for me to internalize the content of this tweet.” “Sometimes, I find it difficult to concentrate on the content of this tweet.” “I feel anxious when evaluating the content of this tweet.” “I feel overwhelmed when evaluating the content of this tweet.” “I do not feel in control when evaluating the content of this tweet.”

Information involvement was measured by using two items from the study by Sussman and Siegal (2003), including “The topic of this tweet is very relevant to me” and “Much of this tweet is something I have thought about recently.” Knowledge adoption intention was measured using three items from the study by Sussman and Siegal (2003), including: “I will focus on increasing my awareness of information security in finance.” “I will take steps to prevent information security incidents in finance.” “I will actively take steps to protect the security of my personal information in finance.”

A manipulation check was adopted to assess the correctness of the experimental design. Table B1 (in Appendix B) shows the detail regarding the manipulation of the test questions for each variable. Finally, participants’ evaluations of the authenticity of the experimental scenarios were tested with the item, “I think the tweet-reading scenarios described in this questionnaire are authentic.” All of the above items were measured using a seven-point Likert-type scale.

Experimental Procedure

Pre-Experiment

In accordance with the World Health Organization’s definition of older adults (i.e., individuals aged 60 years and above), a pre-experiment was conducted to ensure the validity and accuracy of the experimental design. Due to the fact that the amount of readership often affects a reader’s first impression of the quality and dissemination of tweets, which may interfere with the effectiveness of source features, the amount of readership was considered as a control variable. Thus, a 12-group experimental scenario was formed with a 2 (publisher type: official or unofficial) × 2 (author type: expert or non-expert) × 3 (information about the amount of readership: none, high, or low) design.

From May 20 to May 27, 2022, a total of 120 older adults were invited to participate in this pre-experiment. A professional questionnaire collection platform (wenjuanxing.com) was employed to distribute the questionnaires. Participants who successfully completed the survey were compensated with US $2 per questionnaire. The 120 participants were randomly assigned to 12 groups of 10 individuals each. To ensure the questionnaire’s validity, participants were required to spend a minimum of 120 seconds completing it.

The findings from the pre-experiment indicate that the questionnaire has high reliability. The Cronbach’s α values range from .799 to .823 for each variable, suggesting a high level of reliability. In addition, as shown in Table B1 in Appendix B, the manipulation test results confirm the validity of the different scenario designs concerning the publisher type, author type, and the amount of readership. However, the mean score for participants’ perceived authenticity of the scenarios in the “no information about the amount of readership” group was observed to be significantly lower (M = 2.386) than the median score of 4. This finding indicates that showing the information about the amount of readership plays a crucial role in tweet design. As a result, this group was excluded from the formal experiment. Furthermore, two questions related to information processing capability exhibited factor loadings below 0.50, namely “I feel overwhelmed when evaluating the content of this tweet,” and “I do not feel in control when evaluating the content of this tweet.” Consequently, these questions were eliminated from the analysis.

Formal Experiment

Based on the conclusions of the pre-experiment, a formal experiment with 2 (publisher type: official or unofficial) × 2 (author type: expert or non-expert) × 2 (information about the amount of readership: high or low) design was conducted from June 10 to June 20, 2022. Due to the readers’ perceived credibility of WeChat also affecting their perception of the quality and trustworthiness of the tweets published on this platform, this factor was considered as another control variable. One item “I believe that WeChat is trustworthy” with a seven-point Likert-type scale was adopted to measure this variable. In all, 281 participants were recruited through the previously-mentioned professional questionnaire collection platform (wenjuanxing.com). To ensure the validity of the data, 22 questionnaires that were completed in less than 120 seconds and 19 questionnaires that consistently provided the same answer to all questions were deemed invalid and were therefore excluded from the analysis. Ultimately, 240 valid questionnaires were obtained. The participants were randomly assigned to eight groups, with each group consisting of 30 individuals. Table 1 provides a summary of the sample information.

Table 1.

Basic Information About the Samples.

Items		Number	Frequency (%)
Gender	Female	113	47.1%
Gender	Male	127	52.9%
Age	60–65 years old	126	52.5%
	66–70 years old	75	31.3%
	71–75 years old	27	11.3%
	76 years old and above	12	5.0%
Education background	Primary school	57	23.8%
	Junior high school	91	37.9%
	High school	47	19.6%
	Bachelor’s degree or above	45	18.8%
Daily time of browsing WeChat public account	1 hour or less	22	9.2%
	Between 1–3 hours	125	52.1%
	3 hours and above	93	38.8%

Experimental Result

Assessing the Measurement Model

The measurement model was rigorously evaluated using structural equation modeling with AMOS 22.0 software. The results of the confirmatory factor analysis yielded the following statistical values: χ² = 173.252, d.f. = 88, GFI (Goodness of Fit Index) = 0.925, AGFI (Adjusted Goodness of Fit Index) = 0.906, RFI (Relative Fit Index) = 0.904, NFI (Normed-fit Index) = 0.912, TLI (Tucker-Lewis Index) = 0.934, IFI (Incremental Fit Index) = 0.945, RMR (The Root Mean Square Residual) = 0.040, and RMSEA (Root Mean Square Error of Approximation) = 0.050. These outcomes indicate a favorable fit between the data and the proposed model. Furthermore, as illustrated in Table 2, the standardized loadings of all factors exceed 0.7, with p < .001. This finding signifies the existence of a strong relationship between the variables and the underlying constructs, which reinforces the adequacy of the model for this study (Hu & Bentler, 1999). In addition, the Cronbach’s α values for each of the five variables all exceeded .8, indicating the trustworthiness of the measurements and their suitability for subsequent research (Nunnally & New, 1994). The average variance extracted (AVE) values for each variable were greater than 0.5, and the combined reliability (CR) exceeded 0.7. Both these findings affirm the scales’ strong convergent validity (Hair et al., 2009). In addition, the square root of the AVE for each variable was found to be greater than the correlation coefficient between that variable and the other variables. This finding indicates the high discriminant validity of the scale (Hair et al., 2009).

Table 2.

The Questionnaire’s Reliability and Validity Test Results.

Constructs	Items	Loadings	CR	AVE	α	M	SD	PC	PU	KAI	IPC	II
PC	PC1	0.820	0.813	0.595	0.805	5.389	0.852	0.771
	PC2	0.645
	PC3	0.834
PU	PU1	0.751	0.810	0.587	0.802	5.307	0.917	0.30.97	0.776
	PU2	0.749
	PU3	0.797
KAI	KAI1	0.706	0.809	0.585	0.801	5.490	0.921	0.191	0.331	0.765
	KAI2	0.792
	KAI3	0.794
IPA	IPA1	0.807	0.871	0.628	0.867	4.678	0.915	0.220	0.354	0.374	0.792
	IPA2	0.805
	IPA3	0.786
	IPA4	0.770
II	II1	0.721	0.739	0.586	0.733	5.011	0.923	0.231	0.267	0.352	0.289	0.766
II	II2	0.808

Note. In the table, the values marked in bold font are the square roots of the AVE values of each variable. PC = perceived credibility; PU = perceived usefulness; KAI = knowledge adoption intentions; IPA = information processing capability; II = information involvement.

Manipulation Test

Table B1 in Appendix B shows the results of the manipulation test, which indicate that the experimental scenario design is aligned with this study’s expectations. In addition, the results of an independent samples t-test reveal that the mean value of all groups’ responses regarding the authenticity of the scenarios exceeded the median value of the scale (4) (M = 5.508, t = 87.889, p < .001). This finding indicates that the experimental scenarios employed in this study are valid and that they accurately reflect the real scenario for older adult individuals when reading tweets about personal information security education through the WeChat public account.

Hypothesis Testing

To test Hypotheses H1–H7, MANCOVA (multivariate analysis of covariance) was conducted using SPSS 23.0 software. The Pearson correlation coefficient of perceived credibility and usefulness is 0.493 (p < .001). The five essential assumptions were checked and verified before the MANCOVA was conducted. The data met the linearity and normal distribution assumptions (the p-values for each level combination were greater than .05 in the Shapiro-Wilk test), and there were no outliers. The result of the Durbin-Watson test is 1.821, which is between 0 and 4, proving that the data meet the requirement of statistical independence of errors. No significant interactions were found between the perceived credibility of WeChat (covariate variable, PCW) and two independent variables for perceived credibility (F_{PCW*publisher type} = 1.648, p = .113; F_{PCW*author type} = 1.456, p = .166) and usefulness (F_{PCW*publisher type} = 0.837, p = .571; F_{PCW*author type} (1, 63) = 1.076, p = .382). These findings indicate that the assumption of homogeneity of regression was upheld. The results of the Levene’s test were also non-significant, indicating that the homogeneity of variance criteria was satisfied. The results of Box’s test (F_{(9,638265.415)} = 1.77, p = .068 > .05) verify the equivalence of covariance matrices.

The results of the MANCOVA analysis reveal a significant covariate of perceived credibility of WeChat (M_PCW = 5.216, Wilks’ λ = 0.941, F = 7.397, p = .001 < .01), a significant main effect of both author type (Wilks’ λ = 0.819, F = 25.929, p < .001) and publisher type (Wilks’ λ = 0.823, F = 25.231, p < .001), and significant author type by publisher type interaction (Wilks’ λ = 0.771, F = 34.651, p < .001). The results of between-subjects effects reveal that, compared to unofficial information, “official publisher” information has a positive impact on older adults’ perceived credibility (M_official = 5.825, M_unofficial = 4.922, F = 50.668, ηp² = 0.177, p < .001) and perceived usefulness (M_unofficial = 4.889, M_official = 5.736, F = 33.540, ηp² = 0.125, p < .001). Moreover, compared to tweets written by non-expert authors, tweets authored by experts were found to have a stronger influence on older adults’ perceived credibility (M_non-expert = 4.928, M_expert = 5.819, F = 50.752, ηp² = 0.178, p < .001) and perceived usefulness (M_non-expert = 4.848, M_expert = 5.778, F = 42.348, and ηp² = 0.153, p < .001). Consequently, H1–H4 are confirmed.

The interaction effects of both publisher type and author type on perceived credibility (F = 58.727, p < .001, and ηp² = 0.226) and perceived usefulness (F = 53.965, p < .001, and ηp² = 0.187) were also significant. Figures 2 and 3 present the results of the interaction test between publisher type and author type effects on perceived credibility and perceived usefulness, respectively.

Figure 2.

The interactive impact of publisher type and author type on perceived credibility.

Figure 3.

The interactive impact of publisher type and author type on perceived usefulness.

The results of the simple effect analysis (see Tables 3 and 4) further reveal the interaction effects. The findings show that when inconsistencies occur between source features, the readers’ perceptions of the credibility (M_{official*expert} = 6.795, M_{official*non-expert} = 4.856, and M_{unofficial*expert} = 4.844) and usefulness (M_{official*expert} = 6.734, M_{official*non-expert} = 4.739, and M_{unofficial*expert} = 4.822) of the tweet are weakened. In addition, compared to the two inconsistent combinations, perceived credibility (M_{unofficial*non-expert} = 5.000) and perceived usefulness (M_{unofficial*non-expert} = 4.956) were slightly higher when the publication was unofficial and authored by a non-expert.

Table 3.

The Simple Effect Analysis Results of Publisher Type and Author Type on Perceived Credibility.

	SS	DF	MS	F	p
publisher type × author type	65.81	1	65.81	70.13	***
author type × publisher type (official)	112.77	1	112.77	98.84	***
author type × publisher type (unofficial)	0.73	1	0.73	0.64	0.425
publisher type × author type (expert)	114.13	1	114.13	100.50	***
publisher type × author type (non-expert)	0.62	1	0.62	0.55	0.459

***

p < .001.

Table 4.

The Simple Effect Analysis Results of Publisher Type and Author Type on Perceived Usefulness.

	SS	DF	MS	F	p
publisher type × author type	67.97	1	67.79	54.95	***
author type × publisher type (official)	119.32	1	119.32	84.41	***
author type × publisher type (unofficial)	0.54	1	0.54	0.38	0.537
publisher type × author type (expert)	109.63	1	109.63	75.57	***
publisher type × author type (non-expert)	1.41	1	1.41	0.97	0.325

p < .05; **p < .01; ***p < .001.

However, the influence of publisher type on perceived credibility was significant only when the information was written by an expert (F_{publisher type*expert} = 100.05 and p < .001; F_{publisher type*non-expert} = 0.55 and p = .459 > .05); the influence of author type on perceived credibility was significant only when the information was written by an official institution (F_{author type*official} = 98.84, and p < .001; F_{publisher type*unofficial} = 0.64, and p = .425 > .05). In addition, the influence of author type on perceived usefulness was solely driven by official sources (F_{author type*official} = 84.41 and p < .001; F_{author*unofficial} = 0.38 and p = .537 > .05). The impact of publisher type on perceived usefulness was primarily influenced by expert sources (F_{publisher type*expert} = 75.57 and p < .001; F_{publisher type*non-expert} = 0.97 and p = .325 > .05). This indicates that only when existing features are used to enhance the credibility and usefulness of information will the negative effect of inconsistency between features occur. Thus, H6 is not supported, while H5 and H7 are supported.

A MANCOVA was also adopted to evaluate the effects of the other two control variables (i.e., the amount of readership and gender). As shown in Table B2 (Appendix B), the interaction effects of these two control variables with the two independent variables were not significant. This indicates that the influence of these two control variables was controlled, thus further validating the robustness of the model.

Hypotheses H8 and H9 were tested using Amos 22.0 software. The results indicate that a positive relationship exists between the perceived credibility of information among older adults and their knowledge adoption intentions (β = .398, t = 3.934, SD = 0.084, p = .003 < .001). Thus, H8 is validated. In addition, the results reveal that the perceived usefulness of information among older adults positively influences their knowledge adoption intentions (β = .301, t = 2.979, SD = .078, p = .003 < .01). Thus, H9 is also validated.

Test of Moderating Effects

To verify Hypotheses H10 and H11, this study first categorized the publisher type as either official (coded as “1”) or unofficial (coded as “0”); the author type was categorized as either experts (coded as “1”) or non-experts (coded as “0”). Then, the product terms were calculated by multiplying these dummy variables with information processing capability. The moderating effect of information processing capability was tested using hierarchical regression analysis. The results are presented in Tables 5 –8. The product term of publisher type and information processing capability show a significant effect on both perceived credibility (β = −1.175, t = −8.628, and p < .001) and perceived usefulness (β = −1.138, t = −8.018, and p < .001). This finding suggests that the information processing capability has a negative impact on the relationship between publisher type and perceived credibility and perceived usefulness. Thus, H10a and H10b are confirmed. Furthermore, the product terms of author type and information processing capability also had a significant effect on both perceived credibility (β = −1.207, t = −8.868, and p < .001) and perceived usefulness (β = −1.623, t = −8.650, and p < .001). This finding indicates that information processing capability negatively influences the relationship between author type and perceived credibility and author type and perceived usefulness. As a result, H11a and H11b are confirmed.

Table 5.

The Moderating Effect of Information Processing Capability on the Relationship Between Publisher Type and Perceived Credibility.

Steps	Independent variable	β	t	p	ΔR²
Step 1	publisher type	0.847	7.233	***	0.180***
Step 2	publisher type	0.959	7.436	***	0.014*
Step 2	information processing capability	0130	2.014	.045*
Step 3	publisher type	1.242	10.579	***	0.193***
	information processing capability	1.046	8.700	***
	publisher type × information processing capability	−1.175	−8.628	***

Note. The dependent variable is perceived credibility.

p < .05; **p < .01; ***p < .001.

Table 6.

The Moderating Effect of Information Processing Capability on the Relationship Between Publisher Type and Perceived Usefulness.

Steps	Independent variable	β	t	p	ΔR²
Step 1	publisher type	0.741	6.172	***	0.138***
Step 2	publisher type	0.857	6.481	***	0.015*
Step 2	information processing capability	0.135	2.033	.043*
Step 3	publisher type	1.131	9.242	***	0.181***
	information processing capability	1.022	8.152	***
	publisher type × information processing capability	−1.138	−8.018	***

Note. The dependent variable is perceived usefulness.

p < .05; **p < .01; ***p < .001.

Table 7.

The Moderating Effect of Information Processing Capability on the Relationship Between Author Type and Perceived Credibility.

Steps	Independent variable	β	t	p	ΔR²
Step 1	author type	0.685	5.635	***	0.118***
Step 2	author type	0.787	5.729	***	0.009
Step 2	information processing capability	0.109	1.577	.116
Step 3	author type	1.059	8.605	***	0.218***
	information processing capability	1.002	8.555	***
	author type × information processing capability	−1.207	−8.868	***

Note. The dependent variable is perceived credibility.

p < .05; **p < .01; ***p < .001.

Table 8.

The Moderating Effect of Information Processing Capability on the Relationship Between Author Type and Perceived Usefulness.

Steps	Independent variable	β	t	p	ΔR²
Step 1	author type	0.942	5.635	***	0.118***
Step 2	author type	1.126	5.986	***	0.016*
Step 2	information processing capability	0.196	2.081	.039
Step 3	author type	1.492	8.796	***	0.209***
	information processing capability	1.397	8.565	***
	author type × information processing capability	−1.623	−8.650	***

Note. The dependent variable is perceived usefulness.

p < .05; **p < .01; ***p < .001.

Similarly, this study employed hierarchical regression analysis to examine the moderating effect of information involvement. In Tables 9 –12, the results indicate that the interaction between the publisher type and information involvement significantly impacts perceived credibility (β = -.192, t = −2.014, and p < .05) and perceived usefulness (β = -.180, t = −2.008, and p < .05). These findings suggest the existence of a negative influence of information involvement on the relationship between publisher type and perceived credibility, as well as perceived usefulness. Thus, H12a and H12b are confirmed. Moreover, the interaction between author type and information involvement also demonstrates a significant effect on perceived credibility (β = -.269, t = −2.924, and p < .01) and perceived usefulness (β = -.271, t = −3.231, and p < .01). This finding indicates the existence of a negative moderating effect of information involvement on the relationship between author type and perceived credibility, as well as on perceived usefulness. Thus, H13a and H13b are also confirmed.

Table 9.

The Moderating Effect of Information Involvement on the Relationship Between Publisher Type and Perceived Credibility.

Steps	Independent variable	β	t	p	ΔR²
Step 1	publisher type	1.072	7.233	***	0.180***
Step 2	publisher type	0.426	4.512	***	0.525***
Step 2	information involvement	0.972	20.536	***
Step 3	publisher type	0.436	4.635	***	0.005*
	information involvement	1.054	16.965	***
	publisher type × information involvement	−0.192	−2.014	.045*

Note. The dependent variable is perceived credibility.

p < .05; **p < .01; ***p < .001.

Table 10.

The Moderating Effect of Information Involvement on the Relationship Between Publisher Type and Perceived Usefulness.

Steps	Independent variable	β	t	p	ΔR²
Step 1	publisher type	1.019	6.172	***	0.138***
Step 2	publisher type	0.244	2.745	.007**	0.641***
Step 2	information involvement	1.167	26.209	***
Step 3	publisher type	0.253	2.858	.005**	0.004*
	information involvement	1.244	21.283	***
	publisher type × information involvement	−0.180	−2.008	.046*

Note. The dependent variable is perceived usefulness.

p < .05; **p < .01; ***p < .001.

Table 11.

The Moderating Effect of Information Involvement on the Relationship Between Author Type and Perceived Credibility.

Steps	Independent variable	β	t	p	ΔR²
Step 1	author type	0.867	5.635	***	0.118***
Step 2	author type	0.353	3.778	***	0.580***
Step 2	information involvement	0.998	21.335	***
Step 3	author type	0.357	3.855	***	0.011**
	information involvement	1.129	17.534	***
	author type × information involvement	−0.269	−2.924	.004**

Note. The dependent variable is perceived credibility.

p < .05; **p < .01; ***p < .001.

Table 12.

The Moderating Effect of Information Involvement on the Relationship Between Author Type and Perceived Usefulness.

Steps	Independent variable	β	t	p	ΔR²
Step 1	author type	0.942	5.635	***	0.118***
Step 2	author type	0342	4.011	***	0.669***
Step 2	information involvement	1.164	27.235	***
Step 3	author type	0.344	4.110	***	0.009**
	information involvement	1.296	22.120	***
	author type × information involvement	−0.271	−3.231	.001**

Note. The dependent variable is perceived usefulness.

p < .05; **p < .01; ***p < .001.

Discussion

The results of this study reveal that the two kinds of source features of WeChat tweets about personal information security knowledge (i.e., publisher type and author type) affect older adults’ perception of the credibility and usefulness of the tweet content. This in turn affects their knowledge adoption intention. Compared to unofficial tweets and/or tweets written by non-experts, official tweets and/or tweets written by experts are perceived by older adults as being more credible and useful. This finding indicates that older adults rely on heuristic cues and tend to process tweet information through the peripheral path. Several factors contribute to this preference for heuristic cues. First, older adults may lack confidence in their ability to discern accurate information. This is evidenced by lower self-report scores and lower information literacy, compared to younger groups (Kim & Lee, 2023; Van den Broeck et al., 2015). Consequently, older adults are more inclined to avoid detailed processing of content from the central pathway (Musuva et al., 2019). Furthermore, older adults tend to place blind trust in authoritative sources (Ebner et al., 2020). This is especially due to the highly professional and novel nature of personal information security knowledge. These factors reinforce Chinese older adults’ feeling of unfamiliarity with this type of knowledge, as well as their perception of a gap between this type of knowledge and their own existing body of knowledge (Zhang et al., 2023). Therefore, information published by officials or written by experts is more likely to be perceived by older adults as being both credible and useful, subsequently increasing their knowledge adoption intention.

However, the impact of source features on older adults’ perception of the credibility and usefulness of WeChat tweets on personal information security knowledge is influenced by the two source features’ internal consistency. This consistency is reflected in whether these source features equally transmit credible and available information to the older adults. When WeChat tweets are both officially published and written by experts, older adults perceive the tweets to be highly credible and useful. However, when tweets are unofficially published but written by experts, or when the tweets are officially published but written by non-experts, both the perceived credibility and perceived usefulness diminish among older adults. This finding emphasizes the importance of cognitive matching in older adults’ information processing of WeChat tweets. When inconsistent information is presented by two different types of sources (specifically in terms of authority and expertise), this hampers the ability of older adults to judge the credibility of the information. This inconsistency increases the cognitive load on the representation of the problem, which in turn makes matching the problem type with its representation challenging (Vessey, 1991; Vessey & Galletta, 1991). Consequently, the older adults’ perception of the credibility and usefulness of the overall content is affected. Furthermore, inconsistent information about sources weakens the positive impact of authoritative and expert information on perceived credibility and usefulness. This may be due to the fact that information that may appear to have contradictory meanings raises doubts about previously-acquired information, leading to self-skepticism (Li et al., 2024; Mohammad et al., 2017). As a result, not only does this self-doubt reduce the evaluation of the credibility of the source; self-doubt also causes older adults to focus their attention on inconsistent information features or to shift to the central path of information processing. Consequently, the positive effect of prior information on perceived credibility and usefulness is diminished, possibly to a level even lower than that of unofficial WeChat tweets written by non-experts. This finding further explains the positive impact of internal consistency in information presentation on the information-task fit, which is favored as a way to reduce cognitive load (Guo & Li, 2022). Thus, this study fully echoes calls from previous research to pay attention to the influence of the presentation of diverse information on cognitive fit in task-specific contexts (Kelton et al., 2010).

In addition, in the context of aging and decreasing cognitive ability, older adults generally have weaker information processing abilities (Bayne et al., 2023). Despite this fact, this study still finds that the stronger the information processing capability of older adults is, the less they are influenced by source features. This ability means the older adults are more likely to shift their processing path from the peripheral path to the central path. They are also less likely to be swayed by the type of publisher and author when it comes to their perceptions of tweets’ credibility and perceived usefulness. Conversely, older adults with weaker information processing abilities tend to experience information overload in information-rich cyberspace (Özkan & Tolon, 2015). This causes such adults to rely more on heuristic cues to organize and judge the credibility and usefulness of tweets.

This study also delves deeper into the impact of information involvement on the relationship between source features and the way individuals perceive the credibility and usefulness of content. The findings reveal that, when older adults perceive information security knowledge in WeChat tweets as being important and relevant to their lives, their high level of information involvement weakens the influence of source features on the older adults’ perceptions of the credibility and usefulness of those tweets. This finding suggests the existence of selective cognition in the process of knowledge adoption, wherein individuals are inclined to pay attention to information that aligns with their existing interests, habits, and needs (Carstensen & Hershfield, 2021). Consequently, under the influence of selective cognition, older adults’ information processing shifts from a broad overview to a more detailed analysis. This, in turn, diminishes the role of heuristic cues in promoting knowledge adoption.

Implications and Limitations

Theoretical Implications

This study makes significant contributions by constructing a comprehensive research model based on the knowledge adoption model and the ELM. The research model focuses on the dissemination of personal information security knowledge through WeChat public account tweets. In addition, the model systematically examines the features of tweet sources that enhance the older adults’ knowledge adoption intentions. The contributions can be divided into three main parts.

First, this study explains the influence of two types of source features—the publisher type and the author type—on the knowledge adoption intention of older adults. The study reveals that older adults tend to trust authoritative and professional information sources, such as official public accounts and expert authors. Consequently, older adults perceive the content of such tweets to be credible and usable, which in turn enhances their intention to adopt knowledge. This study therefore expands the application of the knowledge adoption model and the ELM in explaining the information processing behavior of older adults. In addition, this study also addresses the gap in existing research that has predominantly focused on the health-information-seeking behavior of older adults, while neglecting their information security knowledge learning behavior.

Second, some researchers have found that the consistency between the heuristic cues and the argument itself is essential for reducing the difficulty of individual information processing. However, this study focuses on examining the impact of the inherent consistency of source features, such as authority and expertise, in interacting with perceived credibility and perceived usefulness among older adults. The results lead to an enhancement of the understanding of how heuristic cues influence information processing. In addition, this study reveals the negative effect of inconsistency on trustworthiness and usefulness features. These findings provide new explanatory variables for comprehending the cognitive pathways of information processing in older adults. This is because, compared to other age groups, processing WeChat tweets on personal information security knowledge can lead to higher cognitive pressure for older adults.

Furthermore, this study introduces new variables to help elucidate the role of heuristic cues in the information processing process of older adults. This is done by discovering how both information processing capability and information involvement moderate the relationship between tweet source features and older adults’ perceptions of the credibility and usefulness of the tweets. This highlights the importance of considering individual differences in information processing abilities when examining the influence of source characteristics. By understanding these dynamics, relevant parties can better tailor information-dissemination strategies to meet the needs of older adults with different processing abilities. In addition, the findings of this study underscore the importance of considering the role of information involvement, particularly among older adults, when designing effective communication strategies. By recognizing the selective nature of information processing, communicators can tailor their messages to align with the specific interests and needs of their target audience, thereby enhancing the likelihood of knowledge adoption.

Managerial Implications

The findings of this study offer valuable insights into how personal information security education for older adults can be enhanced through the use of WeChat public accounts. First, considering that older adults tend to place greater trust in tweets from professional and authoritative sources, this study recommends that official organizations leverage this inclination by establishing official public accounts specifically dedicated to personal information security. These public accounts should focus on disseminating authoritative, scientific, and reliable information. By doing so, these accounts can effectively address the information needs of the elderly population. Also crucial is ensuring that tweet authors are aligned with the official background and that they maintain consistency in their expertise. Industry experts and institutional professionals should be encouraged to actively participate in writing tweets on personal information security knowledge, based on their own professional characteristics. This could contribute greatly to the dissemination of scientifically sound and practical information to older adults through official public accounts. Also important is that relevant organizations prioritize the professional verifications of tweet authors. This type of scrutiny is necessary to avoid any inconsistencies between an author’s expertise and their official background. Such inconsistencies not only undermine the credibility and usability of the content and knowledge evaluated by older adults, but also damage the official and professional image of actual officials and experts.

In addition, the impact of source characteristics on older adults’ willingness to accept tweets highlights the need for collaboration between WeChat public account management organizations and relevant departments. This is necessary to enhance the management of WeChat public accounts and their tweets and will be crucial in preventing fake public accounts from exploiting older adults’ inclination to trust officials and experts. This step would also help prevent the dissemination of false information on information security. Furthermore, strengthening personal information security education for older adults and increasing their awareness in this domain are imperative. The operators of WeChat public accounts should also focus on developing content that is suitable for older adults, avoiding the complexity of current personal information security education tweets. By doing so, the cognitive load associated with processing the content will be reduced, thereby enabling older adults to more effectively process the information. This approach will help diminish the influence of heuristic cues on older adults’ information processing, encouraging them to utilize the central path for processing tweets. By avoiding excessive reliance on peripheral paths, which may result in shallow or superficial knowledge adoption or even blind acceptance, older adults can make more informed decisions regarding the information they encounter.

Research Limitations

This study has three primary research limitations that need to be acknowledged. First, this study only focuses on the elderly population in China and the information security education of China’s largest social media platform as the background. Therefore, future research can apply the research model of this study to other countries and social media platforms, to verify the universality of this study’s research conclusions. Second, it is worth noting that previous studies have suggested dividing the older population into two groups: younger senior adults (aged 60–75 years) and older senior adults (aged 75 years and above). This division would allow for a more in-depth exploration of the impact of physiological aging on the information processing behavior of older adults. Subsequent studies could therefore expand on this aspect to gain a more comprehensive understanding of the subject matter. Finally, due to the weaker information-screening ability of older adults (compared to other age groups), the experimental design of this study only considers tweet materials with authentic and reliable content. Consequently, tweets containing false information are not taken into account. To complement the role of authenticity in tweeted information on the perceived credibility and usefulness among older adults, future research could conduct a comparative study.

Footnotes

Appendix A

Appendix B

Table B2.

Results of Tests of Between-Subjects Effects (Considering the Other Two Control Variables).

DV	IV	F	p	ηp ²
Perceived credibility	Author type	26.039	***	0.109
	Publisher type	16.206	***	0.071
	Readership	1.605	0.207	0.008
	Gender	0.506	0.679	0.007
	Readership × Author type	0.152	0.697	0.001
	Readership × Publisher type	1.778	0.184	0.008
	Gender × Author type	0.252	0.860	0.004
	Gender × Publisher type	0.763	0.516	0.011
Perceived usefulness	Author type	22.319	***	0.095
	Publisher type	13.894	***	0.062
	Readership	1.197	0.275	0.006
	Gender	0.729	0.536	0.010
	Readership × Author type	0.533	0.466	0.003
	Readership × Publisher type	0.036	0.849	0.000
	Gender × Author type	0.785	0.504	0.011
	Gender × Publisher type	0.196	0.899	0.003

Note. DV = dependent variable; IV = independent variables.

Declaration of Conflicting Interests

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

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors thank the support from the Natural Science Foundation Project of China (no. 72302029), the Science and Technology Research Program of Chongqing Municipal Education Commission (no. KJZD-K202400611), and Chongqing Spoken and Written Language Research Project (no. yyk24103).

ORCID iD

Yu Zhang

Author biographies

Yu Zhang (PhD Chongqing University of Posts and Telecommunications) is Associate Professor in the School of Modern Posts at the Chongqing University of Posts and Telecommunications, Chongqing, China. Her research interests include in AI service failure and recovery, elderly information behavior, and transformation service research.

Qian Zhang (MA East China University of Science and Technology) is a graduate student in the Institute of Science and Technology Information at the East China University of Science and Technology, Shanghai, China. Her research interests include elderly information behavior and AI-enabled service management.

References

AARP Research. (2022). 2022 tech trends fact sheet: Online privacy remains a key concern among the 50-plus.https://www.aarp.org/content/dam/aarp/research/surveys_statistics/technology/2022/2022-technology-trends-older-americans-online-privacy-factsheet.doi.10.26419-2Fres.00493.010.pdf

Arendt

Bräunlein

Koleva

Mergen

Schmid

Tratner

(2018). Effects of gain- and loss framed quit messages on smokers: Test of the ability to process the health message as a moderator. Journal of Health Communication, 23(8), 800–806.

Bayne

Mumford

E. A.

Lancaster

Sheridan Johnson

(2023). Technology-facilitated abuse among Americans age 50 and older: A latent class analysis. Journal of Elder Abuse & Neglect, 35(1), 65–87.

Carstensen

L. L.

Hershfield

H. E.

(2021). Beyond stereotypes: Using socioemotional selectivity theory to improve messaging to older adults. Current Directions in Psychological Science, 30(4), 327–334.

Chou

C. H.

Wang

Y. S.

Tang

T. I.

(2015). Exploring the determinants of knowledge adoption in virtual communities: A social influence perspective. International Journal of Information Management, 35(3), 364–376.

Ebner

N. C.

Ellis

D. M.

Lin

Rocha

H. A.

Yang

Dommaraju

Soliman

Woodard

D. L.

Turner

G. R.

Spreng

R. N.

Oliveira

D. S.

(2020). Uncovering susceptibility risk to online deception in aging. Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 75(3), 522–533.

Erkan

Evans

(2016). The influence of eWOM in social media on consumers’ purchase intentions: An extended approach to information adoption. Computers in Human Behavior, 61, 47–55.

Filieri

Mcleay

Tsui

Lin

(2018). Consumer perceptions of information helpfulness and determinants of purchase intention in online consumer reviews of services. Information & Management, 55(8), 956–970.

Fung

N. L. K.

Fung

H. H.

Chu

Gong

X. M.

(2023). Facial trustworthiness influences age differences in visual attention toward credible versus non-credible messages. Innovation in Aging, 7(5), igad051.

10.

Gao

Zhang

Wang

(2012). Understanding online purchase decision making: The effects of unconscious thought, information quality, and information quantity. Decision Support Systems, 53(4), 772–781.

11.

Guo

H. Y.

(2022). Can the amount of information and information presentation reduce choice overload? An empirical study of online hotel booking. Journal of Travel & Tourism Marketing, 39(1), 87–108.

12.

Hair

J. F.

Black

W. C.

Babin

B. J.

Anderson

R. E.

Tatham

R. L.

(2009). Multivariate data analysis. Pearson Prentice Hall.

13.

Hajli

(2016). Ethical environment in the online communities by information credibility: A social media perspective. Journal of Business Ethics, 149(4), 1–12.

14.

Hardy

Vorobjovas-Pinta

Eccleston

(2018). Enhancing knowledge transfer in tourism: An elaboration likelihood model approach. Journal of Hospitality and Tourism Management, 37, 33–41.

15.

Hlee

Lee

Yang

S. B.

Koo

(2019). The moderating effect of restaurant type on hedonic versus utilitarian review evaluations. International Journal of Hospitality Management, 77, 195–206.

16.

L.-t.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 1–55.

17.

Huo

C. G.

F. C.

Qiu

Y. F.

Wang

Y. C.

(2018). Exploring the determinants of health knowledge adoption in social media: An intention-behavior-gap perspective. Information Development, 34(4), 346–363.

18.

Huo

C. G.

Zhang

F. C.

(2018). Factors influencing people’s health knowledge adoption in social media: The mediating effect of trust and the moderating effect of health threat. Library in TECH, 36(1), 129–151.

19.

Kelton

A. S.

Pennington

R. R.

Tuttle

B. M.

(2010). The effects of information presentation format on judgment and decision making: A review of the information systems research. Journal of Information Systems, 24(2), 79–105.

20.

Kim

Lee

(2023). Does social media use mitigate or exacerbate loneliness among Korean older adults? Focusing on the moderating role of media literacy. Social Media + Society, 9(2), 20563051231177959.

21.

Kitchen

P. J.

Kerr

Schultz

D. E.

McColl

Pals

(2014). The elaboration likelihood model: Review, critique and research agenda. European Journal of Marketing, 48(11/12), 2033–2050.

22.

Kostagiolas

Korfiatis

Kourouthanasis

Alexias

(2014). Work-related factors influencing doctors search behaviors and trust toward medical information resources. International Journal of Information Management, 34(2), 80–88.

23.

Q. L.

Park

Kim

(2024). Impact of information consistency in online reviews on consumer behavior in the e-commerce industry: A text mining approach. Data Technologies and Applications, 58(1), 132–149.

24.

Mohammad

A. S.

Jaradat

Mahmoud

A. A.

Jararweh

(2017). Paraphrase identification and semantic text similarity analysis in Arabic news tweets using lexical, syntactic, and semantic features. Information Processing & Management, 53(3), 640–652.

25.

Moore

R. C.

Hancock

J. T.

(2020). Older adults, social technologies, and the coronavirus pandemic: Challenges, strengths, and strategies for support. Social Media + Society, 6(3), 2056305120948162.

26.

Musuva

P. M. W.

Getao

K. W.

Chepken

C. K.

(2019). A new approach to modelling the effects of cognitive processing and threat detection on phishing susceptibility. Computers in Human Behavior, 94, 154–175.

27.

National Computer Network Emergency Response Coordination Center. (2021). China internet network security report of 2020. http://www.xinhuanet.com/info/20220105/0c729b36ebed4253a0a4331cf211255d/c.html

28.

Nunnally

New

M. G.

(1994). Psychometric theory (3rd ed.). McGraw Hill.

29.

Özkan

Tolon

(2015). The effects of information overload on consumer confusion: An examination on user generated content. Bogazici Journal Review of Social Economic & Administrative Studies, 29(1), 27–51.

30.

Petty

R. E.

Cacioppo

J. T.

(1986). Communication and persuasion: Central and peripheral routes to attitude change. Springer

31.

Rafique

Ul Islam

Shamim

(2024). Acceptance of e-learning technology by government school teachers: Application of extended technology acceptance model. Interactive Learning Environments, 32(2), 2970–2988.

32.

Saarinen

Vakkari

(2013). A sign of a good book: Readers’ methods of accessing fiction in the public library. Journal of Documentation, 69, 736–754.

33.

Shahab

M. H.

Ghazali

Mohtar

(2021). The role of elaboration likelihood model in consumer behavior research and its extension to new technologies: A review and future research agenda. International Journal of Consumer Studies, 45(4), 664–689.

34.

Shang

L. L.

Zhou

J. J.

Zuo

M. Y.

(2020). Understanding older adults’ intention to share health information on social media: The role of health belief and information processing. Internet Research, 31(1), 100–122.

35.

Shao

Lin

Lei

(2019). Credulity rather than general trust may increase vulnerability to fraud in older adults: A moderated mediation model. Journal of Elder Abuse & Neglect, 31(2), 146–162.

36.

Sundar

S. S.

(2008). The MAIN model: A heuristic approach to understanding technology effects on credibility. In Metzger

M. J.

Flanagin

A. J.

(Eds.), Digital media, youth, and credibility (pp. 73–100). MIT Press.

37.

Sussman

S. W.

Siegal

W. S.

(2003). Informational influence in organizations: An Integrated Approach to Knowledge Adoption. Information Systems Research, 14(1), 47–65.

38.

Tang

M. C.

(2016). The effect of online word of mouth on fantasy readers’ preference stability: The moderating role of involvement and knowledge. Library & Information Science Research, 38(4), 292–300.

39.

Tang

M. C.

P. M.

(2022). Reconciling the effects of positive and negative electronic word of mouth: Roles of confirmation bias and involvement. Online Information Review, 46(1), 114–133.

40.

Tzu-Jung

Horng

R. Y.

Anderson

R. C.

(2014). Effects of argument scaffolding and source credibility on science text comprehension. Journal of Experimental Education, 82(2), 264–282.

41.

Van den Broeck

Poels

Walrave

. (2015). Older and wiser? Facebook use, privacy concern, and privacy protection in the life stages of emerging, young, and middle adulthood. Social Media + Society, 1(2), 2056305115616140.

42.

Vessey

(1991). Cognitive fit: A theory-based analysis of the graphs versus tables literature. Decision Sciences, 22(2), 219–240.

43.

Vessey

Galletta

(1991). Cognitive fit: An empirical study of information acquisition. Information Systems Research, 2(1), 63–84.

44.

Zhang

Deng

Yuan

(2023). Reducing technology-mediated service exclusion by providing human assistant support for senior citizens: Evidence from China. Information Technology for Development, 29(4), 692–715.

The Effects of Consistency and Inconsistency in the Source Features of WeChat Tweets on Older Chinese Adults’ Intentions to Adopt Personal Information Security Knowledge

Abstract

Keywords

Introduction

Theoretical Foundation and Literature Review

Knowledge Adoption Model

ELM and the Knowledge Adoption Behavior of Older Adults

Hypotheses Development and Research Model

Source Features of Tweets and Perceived Credibility

Source Features and Perceived Usefulness

Interactive Effects of Publisher Type and Author Type on Perceived Credibility and Perceived Usefulness

Perceived Credibility, Perceived Usefulness, and Knowledge Adoption Intention

Moderating Effect of Information Processing Capability on Source Features, Perceived Credibility, and Perceived Usefulness

Moderating Effect of Information Involvement on Source Features With Perceived Credibility and Perceived Usefulness

Research Methodology

Experimental Design

Experimental Procedure

Pre-Experiment

Formal Experiment

Experimental Result

Assessing the Measurement Model

Manipulation Test

Hypothesis Testing

Test of Moderating Effects

Discussion

Implications and Limitations

Theoretical Implications

Managerial Implications

Research Limitations

Footnotes

Appendix A

Appendix B

Declaration of Conflicting Interests

Funding

ORCID iD

Author biographies

References