Perspective Chapter: The Interactive Perspective in Social Media Usage Studies

Abstract

Social media usage is heavily influenced by individual interactions, as supported by empirical research. These interactions shape the cultural and social contexts of social media, resulting in diverse usage behaviors. Social media’s ability to transcend temporal and spatial barriers amplifies this dynamic. Interactions are fundamental to understanding social media as a systematic behavior occurring in both micro and macro systems, serving dual purposes of maintaining equilibrium and facilitating changes. They help systems achieve collective goals beyond individual capacities while also enabling necessary changes to adapt to the external environment. Thus, systematic research method, particularly employing social network analysis, is crucial for understanding human interactions in social media usage. For that social network analysis focuses on the formation and changes of structures formed by interactions. Integrating social network analysis into research can lead to a paradigm shift toward a more systematic perspective in social media research. This study aims to accomplish this by formulating hypotheses through an extensive literature review, aiming to inspire more empirical studies in the realm of social media usage.

1. Introduction

The development of digital technologies has facilitated the wide utilization of social media’s digitalized functions, enabling the exchange of video and radio information over the internet, and overcoming temporal and spatial barricades. The inherent convenience of social media has positioned it as the predominant communication tool, having 4.80 billion users constituting 60% of the global population [1]. Social media serves diverse purposes, with its primary function being the establishment and maintenance of interpersonal connections. A substantial 94% of social media users indicate their use of the platform for staying connected with friends, family, and acquaintances [1, 2, 3]. Furthermore, social media significantly influences users’ economic behaviors, as approximately 78% state their engagement in the platform to seek information about brands, informing their decision-making processes in product selection [2, 3].

In light of the escalating influence of social media, a burgeoning interest from diverse fields has emerged to scrutinize the dynamics of social media usage. Prevailing research has traditionally framed social media users as autonomous individuals, fully capable of engaging in behaviors of their own volition [4, 5, 6]. While this perspective adequately elucidates behaviors aligned with individuals’ normal beliefs, it fails to account for collective behaviors that manifest under the impact of other people, such as riots, voting, and innovation diffusion, which often diverge from individuals’ regular behavioral norms. Adopting a systemic perspective to examine social media usage behaviors unveils a paradigm wherein behaviors emerge as outcomes of reciprocal interactions between micro and macro systems, encompassing individuals and various scales of social groups. These interactions engender equilibriums via feedback loops and circular forces among systems, thereby giving rise to regular behavioral patterns among individuals. Conversely, deviations from regular behavioral patterns may occur when equilibriums are not attained, contingent upon the characteristics of the systems in which individuals are enmeshed.

That is to say, if we perceive social media users as connective entities in different systems, the transient and dynamic formation of various behaviors could be better explained. On the other side, the utilization of social media is primarily driven by the imperative of establishing and maintaining connectivity within one’s social networks, as has been elucidated in various reports [1, 2, 3]. The interactions within these networks serve as the foundational dynamics influencing subsequent behaviors [7, 8, 9]. The pivotal significance ascribed to connectivity in comprehending the utilization of social media emanates from the recognition that interactions with social relationships constitute the intrinsic cybernetic mechanism forming diverse human communication patterns, hence this mechanism is widely acknowledged as meta-communication [10, 11]. Meta-communication enables individuals to select appropriate communication patterns which is contingent upon their interactions within diverse social networks across various contexts. Feedback from various social networks allows individuals to accommodate cultural and behavioral norms belonging to each social network, thereby fostering adept and fluent communication with others.

Understanding the dynamics of social relationships is fundamental for discerning individuals’ behavioral patterns and establishing thresholds for various behaviors [8]. This comprehension significantly enhances the predictive precision of behaviors, surpassing the reliance on vague indicators such as click-through rates in all contexts. From a systematic viewpoint, these interactions serve two primary objectives necessary for the survival of a system amidst environmental fluctuations: maintaining equilibrium and facilitating adaptation [4, 9]. Interactions foster the emergence of regularities that contribute to the formation of a relatively stable structure against chaos, thereby achieving an equilibrium state within the system. Additionally, interactions exchange information and energy that disrupt this equilibrium, prompting adjustments to the system’s features to align with environmental changes.

Investigating the mechanisms of interactions with social relationships on social media benefits from employing social network analysis as an ideal research tool. This method treats a single social relationship as an analysis unit, with various combinations of these units formed by interactions illustrated by network graphs. Structural features of established networks and trends of change over time in these network graphs can be quantified and calculated using statistical models. The attributes of these points and lines, including density, degree, and homogeneity, assume significance as crucial parameters for statistical models, facilitating a nuanced assessment of the underlying structures of interest. The precision of these models in predicting dynamic processes within social networks increases with the amount of information about the features of social networks gathered. In this case, social media platforms could serve as valuable sources for collecting digital data on social network features due to the digital trails recorded by these platforms, such as the frequency and direction of interactions. The ability to accumulate vast digital information on social media not only facilitates data collection but also allows researchers to explore various features of social networks across different sizes globally without the need for on-site efforts.

Hence, this study employs social network analysis to scrutinize two fundamental systematic attributes associated with social media usage: maintaining equilibrium and adapting to changing environments. Consequently, two research questions emerge: Research Question 1 addresses the characteristics of social networks shaped by interactions on social media, while Research Question 2 delves into the temporal evolution of social networks through interactions. The subsequent Section 2 delves into prior research on social media usage, which emphasizes the interactive features of social media, whereas Section 3 elucidates the application of social network analysis in exploring the formation and evolution of social networks over time, thereby addressing the aforementioned research questions. Section 4 outlines potential hypotheses based on the findings of Section 3, serving as an exploratory research framework for future inquiries. Finally, the concluding segment of Section 5 underscores the interpretative prowess of an interactive perspective in uncovering the intricate dynamics underlying social media utilization, thus enhancing comprehension of the complex social processes intertwined with social media usage.

2. Literature review of social media usage

In contemporary society, the pervasive use of social media has become prevalent across diverse facets of social life. Social media refers to computer-based technology that enables the creation, sharing, and exchange of information, ideas, and content through virtual networks and communities. It encompasses various online platforms and applications designed to facilitate interaction, including Facebook, Instagram, TikTok, WhatsApp, WeChat, and X. These platforms typically enable users to create profiles, connect with other users, share text, images, videos, links, and other forms of content, engage in discussions, and interact with content posted by others. Social media has become a pervasive aspect of modern digital communication, profoundly influencing social interactions, information dissemination, marketing, business, entertainment, and various other aspects of society and culture. A noteworthy 78% of social media users actively seek brand information through these platforms, with 31% of individuals belonging to Generation Z (born between 1995 and 2009) utilizing social media as their primary channel for receiving commercial advertisements. This discernible trend indicates a heightened inclination among younger demographics to engage in economic activities through social media platforms. Specifically, 31% of Generation Z individuals acquire commercial information via social media, surpassing the corresponding figures for millennials (30%), generation X (27%), and the baby boomers (21%) [2].

Hence, social media emerges as a crucial vantage point for scrutinizing the economic behaviors of young individuals. The accessibility provided by social media proves advantageous for those seeking insights into the usage patterns of the younger demographic. Despite its empowerment as a versatile tool for various activities and the considerable autonomy granted to users in content creation, recent reports indicate a decline in trust in social media ads since 2022 [3]. Factors contributing to this erosion of trust encompass concerns related to privacy breaches, an overload of ad content, and the perceived inefficiency of time spent scrolling through advertisements. Primarily, dissatisfaction with the information presented stands out as the key factor influencing trust loss. As highlighted by scholar Castells [4], the commercial algorithm underpinning social media relies heavily on the click-through rate, potentially overlooking genuine behavioral intentions and the nuanced dynamics of social media usage.

The click-through rate represents merely one facet of social media, focusing solely on the direct interaction with advertisements. Often, the motivations behind such interactions are oversimplified or attributed solely to user interests. However, it is crucial to recognize that the rationale for clicking on ads can vary significantly. Instances where individuals click by mistake or other factors may come into play are frequently overlooked [4, 5]. In such scenarios, the information conveyed through ads may not effectively cater to the diverse needs of social media users. Acknowledging the complexity of human behavior, it becomes evident that a singular or a limited set of factors cannot comprehensively account for the behavioral variations observed. This assertion is substantiated by numerous empirical studies investigating patterns in social media usage.

In these empirical studies, scholars have endeavored to elucidate social media usage through diverse lenses, encompassing personal, social, and mass communication theories. Nevertheless, these perspectives have yielded outcomes that are not only mixed but also disparate, occasionally giving rise to conflicting results. An illustrative case emerges in the examination of attitudes, where its impact on behaviors exhibits variance contingent upon contextual shifts. For instance, in the investigation of behavioral intention regarding food safety, subjective norms prove more influential than attitudes [6]. Conversely, when scrutinizing moral behavior, attitudes emerge as the most potent predictor of behaviors [7]. The fluctuating predictability observed in individual and social factors predominantly stems from the dynamic and transitory nature of the social environment. Individuals may exhibit behaviors incongruent with their conventional beliefs, especially during engagement in social activities [8].

The regulation of human behaviors appears remarkably complex, posing a challenge for scholars to discern the elusive effects of individual and social factors. A cybernetic perspective provides valuable insight by recognizing that individuals are not isolated entities; rather, their behaviors unfold within specific systems, wherein interactions with other system parts exert influence. Consequently, behaviors manifest as outcomes of the dynamic equilibrium maintained through interactions among system elements, rather than being solely determined by specific sets of factors. Within these systems, interactions serve two primary purposes: maintaining existence through balance and responding to environmental fluctuations through change. The pursuit of balance ensures order and relative stability within a system, allowing its constituent elements to sustain their existence, while adaptation to environmental variations necessitates change [9].

Hence, interaction emerges as the fundamental dynamic across systems, spanning from the micro to the macro level. This interactive process generates a circular force facilitating the exchange of information and energy among various system components. At the individual level, humans engage in interactions to acquire essential information and food, thereby supporting life and personal development through engagements with others and their environment. Similarly, at the social and national levels, within a social or a nation, interactions contribute to the establishment of social relationships, fostering cultural formation that maintains equilibrium among diverse groups. Between societies and nations, interactions serve as conduits for the exchange of information and resources, fostering collective development [9, 10]. In this systematic perspective, from individuals to nations, interaction, and its circular force emerge as the cybernetic mechanism governing all activities of organism systems.

Regarding social media communication, the cybernetic mechanism is also interaction. Through different strengths of interactions, diverse forms of social relationships are formed. The interdependence of interaction and social relationships establishes a circular dynamic, wherein interactions establish social relationships, and social relationships can either facilitate or impede behaviors reciprocally. Serving as a crucial control mechanism for communication behaviors, social relationships define the contexts and interpretation of information shared in communication processes. Scholar Bateson [11] aptly posits that during communication, the exchanged information includes two distinct levels: the factual meaning and the social relationships level. The factual level encapsulates the original meaning of information, such as the statement that there is a dog, indicating the existence of a four-legged canine mammal, a dog. However, on the social relationships level of information, the same statement, when exchanged between individuals engaged in a game, may function as a coded instruction for another player. The interpretation of this statement is potentially signifying an order to attend to the dog if the communicators share familial relationships. Consequently, social relationships are often referred to as meta-communication, elucidating their role as an underlying control mechanism [11, 12].

From systematic and cybernetic perspectives, the inclination of individuals to engage in communication within their familiar social relationships on social media becomes evident, the report has shown that the strongest motive to use social media is to connect with their family members and close social relationships [1]. When individuals are conceptualized as systems, the imperative to maintain control over themselves arises to uphold balance in both cognitive and physical realms [13]. As highlighted earlier, the fundamental prerequisites for the existence of systems involve the maintenance of balance and adaptability to environmental changes. These conditions serve as safeguards against the onset of chaos and the potential loss of control, which could detrimentally impact their existence. Consequently, individuals opt to communicate with their established social relationships on social media, recognizing the interactions with these social relationships as crucial control mechanisms that contribute to the preservation of stable and consistent cognitive frameworks for both their ego identities and surroundings. Despite the expansive digital communication capabilities that transcend temporal and spatial barriers, individuals exhibit a propensity to communicate locally and within specific contexts on social media, rather than globally and universally [6, 14, 15].

The dynamic mechanism underpinning social media usage has significantly broadened its influence on individuals due to the robust communication functions it offers. Primarily, social media has played a pivotal role in facilitating connections between individuals and their established social relationships, thereby substantially reducing communication costs. Notably, these interactions can be systematically recorded on social media platforms in the form of digital data, providing scholars with a valuable means to analyze the dynamics of social media usage. In the examination of social media interactions, researchers can leverage the diverse functions provided by these platforms as analytical tools to scrutinize various types of connections. As emphasized by researcher Burke [16], the multifaceted utilization of social media functions can give rise to a spectrum of connections, underscoring the nuanced nature of the impact and potential outcomes associated with different uses of these platforms.

The utilization of social media can be broadly classified into three discernible types, as outlined by researcher Burke [16]: direct interaction with targeted individuals, passive consumption, and broadcasting. The term “direct interaction” denotes engagements with specific users on social media platforms aimed at either sustaining existing relationships or establishing direct connections between users. Common features facilitating direct communication on social media encompass the like button, in-line comments, messages, synchronized chat, and photo tagging. Conversely, interactions involving an unspecified number of contacts are characterized as indirect communication. This category encompasses both passive consumption and broadcasting, involving the consumption of aggregated streams of news, status updates, links, and profiles on social media platforms. It is noteworthy that direct communication demands a higher degree of concentration and attention compared to indirect communication. The maintenance of relationships with specific individuals necessitates supportive and attentive feedback, contributing to the potential for the creation of robust social bonds. In contrast, indirect communication presents a greater likelihood of fostering weak ties or bridging ties, allowing individuals to connect with a broader, less specific network of contacts. Consequently, direct communication tends to foster strong interpersonal bonds, while indirect communication offers opportunities for the formation of weaker or bridging ties, facilitating connections with a more diverse array of individuals.

Nevertheless, the general application of these functions does not adequately capture the intricate dynamics inherent in social interactions. The meanings associated with interactions are frequently contingent upon the specific social relationships within which individuals are situated, as previously discussed. The contextual nature of social relationships dictates the interpretation of interactions, challenging the assumption that strong ties invariably indicate deep bonds or weak ties exclusively denote bridging effects. For example, within intimate relationships, individuals who endorse the appropriateness of public displays of affection may cultivate weak ties through indirect connections, such as wall posting, status updates, and profile links, as these are perceived as effective means to foster intimacy [16, 17]. Conversely, those who prioritize privacy in their intimate relationships on social media may opt for more direct communication features, like messages and synchronized chats. The selection of social media functions is contingent upon a shared understanding of behaviors defining intimate relationships, leading individuals to adopt diverse usage patterns on social media based on their preferences and contextual norms.

Therefore, there research methods that offer a more nuanced portrayal of the dynamic processes inherent in social media usage? The affirmative response is evident, and a pertinent approach in this regard is the application of social network analysis, particularly in the exploration of human relationships.

3. The approach of social network analysis

A social network typically denotes a social structure comprising a set of social actors engaged in regular interactions, wherein these structures may form various social groups such as familial units, friendship circles, working affiliations, and the like. In the context of this study, social networks pertain to a collective of individuals engaging in regular online interactions, unfettered by temporal or spatial constraints inherent in offline interactions. In this virtual realm, individuals possess greater freedom in selecting their associates compared to the physical world, thus rendering social network formation more fluid. However, as documented in the literature, individuals often maintain connections with preexisting social relationships established offline [4, 8], with homophilous individuals, sharing similar social and cultural backgrounds, displaying a propensity to form connections, thereby perpetuating equilibrium within systems.

Thus, the research methodology employed to study social media usage needs to adeptly capture the dynamic interplay between social media activities and social contexts, given that human behaviors, including social media usage, are inherently influenced by variations in social contexts [18, 19, 20]. An apt approach for elucidating these intricate interactions is through the application of social network analysis (SNA). Defined as the systematic examination of interaction structures, SNA involves designating interacting entities, such as individuals or other entities, as nodes, and the connections between them as lines or edges. Employing this methodology allows for the creation of network graphs comprising nodes and lines, facilitating the visual representation of structural patterns. As elucidated earlier, the fundamental objectives of systems involve maintaining balance and adapting to environmental changes. In this regard, SNA emerges as a valuable tool from a systematic perspective, providing useful insights into the structural dynamics of social interactions within the context of social media usage.

3.1 The balance of social networks

Achieving system equilibrium is contingent on the regularity of interactions. Frequent and repetitive interactions give rise to relatively stable structures within social networks, acting as a stabilizing force against the random actions of individual components within the system and introducing order to counteract chaos. The analysis of the presence and persistence of these structures serves as an indicator of the overall balance within the social network. Employing statistical methods rooted in the entropy laws of physics provides a systematic approach to gauging the equilibrium of social networks and understanding the extent to which they exhibit stability and order [18, 21, 22, 23].

Entropy stands as a fundamental concept employed to quantify the degree of uncertainty or randomness associated with a random variable [10, 24]. It offers a quantitative measure of the potential states or forms that a factor may assume. Greater variability in the potential statuses of this factor signifies heightened randomness, hindering the attainment of balance. The explicit formula articulating entropy is presented below:

$$\text{H}\left(\text{x}\right)=-\Sigma \text{P}{\left(\text{x}\right)}^{\ast }{\log }_2\text{P}\left(\text{x}\right)$$E1

In the realm of Formula (1), the introduction of entropy involves the consideration of H(x), representing the entropy of the random variable X. Here, P(x) stands for the probability associated with a particular outcome or behavior exhibited by X, highlighting the probability of observing said outcome or behavior. The symbol Σ signifies the summation of all potential outcomes or behaviors of X. The logarithm base 2, denoted as log₂ P(x), logarithmically scales the probabilities and functions as a metric of information, measured in binary bits. By multiplying P(x) by log₂ P(x), one can quantify the information required to describe the specific outcome or behavior of P(x). This information quantity serves as an indicator of the certainty of factor X. Consequently, the negation of the sum of these values (−Σ P(x) * log₂ P(x)) indicates that entropy is expressed as the opposite of certainty, signifying the degree of randomness. Thus, more information about X corresponds to increased certainty and decreased entropy, reinforcing the correlation between information, certainty, and entropy in the context of Formula (1).

Entropy emerges as a crucial concept in assessing the likelihood of self-regulation and control systems arising from states of chaos and randomness. Probability assumes a central role in capturing the intricate interactions among factors within a designated system, serving as an indicator for the formation of recurring patterns indicative of stability. These recurring patterns result from the interactions among the elements comprising the system. Researchers have observed that individual particles, when isolated, exhibit stochastic and unpredictable behaviors. However, when multiple particles engage in interactions, their collective behaviors give rise to discernible and recurrent patterns. This enables the estimation of associated probabilities for each pattern, facilitating the prediction of these recurrent behaviors.

In the case of isolated individual particles, stochastic and unpredictable behaviors are evident, akin to the wave-particle duality where each status holds a 50% probability. The unpredictability arises because these particles, as singular entities, demonstrate equal probabilities for all conceivable behavioral patterns. Predicting the status of a single particle becomes nearly impossible, given the equal likelihood of each status emerging at the same rate. Consequently, the status of such particles remains unstable. However, the scenario changes when particles interact with one another or with external entities. In these interactions, specific patterns begin to emerge, leading to the stabilization or collapse of probabilities [10, 25].

Likewise, individuals acquire prioritized and stabilized behavioral patterns through interactions with other entities. The dynamic interplays between individuals and their surroundings are readily observable on social media, facilitated by digitization [24, 26, 27]. Digitalization encodes information into digital bits, rendering interactive patterns accessible to electrical devices. Next, social network analysis employs graphic structures comprising nodes and edges to code these interactions and relationships. Transforming individuals and entities into geometric units, represented as nodes, and delineating relationships between them through lines, social network analysis could analyze the probabilities of interactive structures, spanning from micro to macro systems.

At the core of human behaviors are interactions that can coalesce into intricate structures at both micro and macro levels. To assess the stability of such social network structures, social network analysis utilizes exponential random graph models, estimating the probabilities associated with the fundamental units of interactions and structures that emerge within a specific system, as articulated by Eq. (2) [28, 29, 30, 31, 32, 33]. This analytical approach provides a comprehensive framework for understanding the complexities inherent in social interactions and their manifestation in network structures on various scales.

$$\mathrm{Pr}\left(Y=y\right)=\left(\frac{1}{k}\right)\exp \left\{\sum _A{\eta }_A{g}_A\left(y\right)\right\}$$E2

In Eq. (2), the variable Y encompasses the entire set of potential ties within a certain social network comprising n actors. For instance, in a social network with 5 actors, the sum of all possible ties among them amounts to 10, determined by the combination formula C52 = 5! /(2!*(5–2)!) = 120/12 = 10. While the variable y represents the actual number of ties observed within this social network. A represents the configurations, or structures, formed by interactions in this social network. Interactions could form structures such as edges, two-stars, three-stars, and triangles presented in Table 1, the number of observed configuration A is denoted as g_A(y). Here, the term ${\eta }_A$ represents the parameter corresponding to the configuration A, and K is a normalizing quantity that ensures the model has a proper probability distribution, Σ_A means the summation of all configurations. In Table 1, the calculated results based on Eq. (2) have been presented:

Table 1.

Structural features in social networks.

Table 1 illustrates an exemplary social network consisting of 28 students [34, 35, 36]. The right column delineates the estimated parameters associated with each configuration. The derivation of these parameters involves employing a sequence of Monte Carlo estimation algorithms. This process entails iterative calculations utilizing gathered social network data, facilitating the simulation and stabilization of the normalizing constant (K) and the pertinent parameters linked to each configuration. The main objective of this methodology is to converge toward an approximation of the observed social network, demonstrating the efficacy of the Monte Carlo estimation algorithms in capturing the regularities of these structures within the network.

In examining a social network, various configurations may exert distinct influences on the network. The parameter associated with edges serves as an indicator of social network density, reflecting the average degree of connectivity. Notably, the presence of two-star structures exhibits a heightened inclination toward forming closed triangles, wherein the latter signifies the propensity of a social network to fragment into smaller cliques [37, 38, 39]. Conversely, the features associated with three-star configurations may signify transitive relationships among individuals and other segments of the social network, fostering connections between individuals and these segments. Analyzing the results presented in Table 1, the positive estimations suggest a heightened likelihood of the emergence of two-star and triangle structures, indicating a tendency for individuals to form small cliques rather than fostering unity with all members.

Derived from real social network data, these structures may manifest collectively, exhibiting either independence or dependence on one another. If these configurations operate independently, their distribution probabilities can be estimated separately using Eq. (1), as illustrated in the sample provided in Table 1. Occasionally, these configurations are interdependent, particularly in scenarios such as intimate relationships. For instance, if a triangular tie or relationship forms within a couple, the reciprocal edge connecting the same couple dissolves, leading to their separation. In such instances, the summation of configurations in Eq. (1) encompasses both reciprocal edges and triangle configurations rather than just one category of configurations. Contrary to separate estimations, these configurations do not remain distinct but instead converge into one equation, incorporating their respective parameters.

Moreover, attributes about nodes, such as age, gender, and socioeconomic position, can serve as indicative factors for distinct categories of configurations. In a specific social network, individuals sharing similar features exhibit an increased likelihood of forming connections, for instance, with ties between individuals of the same age exhibiting higher probabilities of occurrence. Therefore, it becomes pertinent to scrutinize the assumption of dependence between ties with similar attributes and the observed configurations within this social network [25, 28]. The precision of social network analysis in estimating the probabilities of various configurations present in social networks is contingent upon the granularity of the collected data. As data becomes more detailed and comprehensive, the analytical capacity of social network analysis improves, enabling a more accurate estimation of the prevalence of diverse configurations within the social network landscape.

3.2 The change in social networks

Aligned with the main purpose inherent in all systems, the adaptive capacity of social networks to undergo transformation or evolution within their environment is realized through interactions across diverse societal domains. In the preceding section, we delved into the assessment of structures stabilized through interactive processes, elucidating the regularities emerging from randomness. While these structures encapsulate patterns resulting from frequent interactions, they are not perpetually stable, constituting dynamic equilibriums. Any alterations in the frequency and strength of interactions prompt corresponding changes in the structures. This study introduces a few research endeavors examining the dynamics of change within social networks. By elucidating these studies, a nuanced comprehension of the systematic nature of behaviors within social networks is facilitated, shedding light on the investigation of social media usage.

The first is the Threshold Model of collective behaviors, which was proposed by Granovetter in 1978. In this study, Granovetter [8] believed that human behavior does not always conform to institutionalized and agreed-upon norms. People with the same norms and values may behave differently in different social networks when engaging in certain collective behaviors, such as innovation, rumor diffusion, voting, and strikes. This is largely dependent upon the behavioral threshold of these social networks. For example, when few people are participating in collective behavior, other individuals are more likely to stay away from it; on the other hand, when more people are participating in events, the movement is more likely to spread. Thus, exploring the general norms, preferences, motives, and beliefs of individuals only provides a necessary but not sufficient explanation of behavioral outcomes [8].

To find the threshold of various social networks, Granovetter [8] proposed a model to delineate the variations inside a social network:

$$r\left(t+1\right)=F\left[r\left(t\right)\right]$$E3

Within the given equation, r(t) represents the proportion of individuals within a social network engaging in an action at time t. F[r(t)] signifies the cumulative distribution function capable of depicting the developmental trajectory of this proportion. The subsequent term, r(t + 1), designates the proportion of individuals at time t + 1. As time progresses, if the observed proportion continues to grow, an equilibrium point, denoted by r(t + 1) = r(t), can consistently be identified.

Upon reaching the threshold of an individual within a social network, that individual becomes inclined to participate in collective actions. Consider a scenario in a social network comprising 100 individuals, each with thresholds distributed from 0 to 99. For instance, when the person with a threshold of 0 initiates action, a cascading effect occurs, leading the person with a threshold of 1 to join, and this domino effect ceases upon reaching the equilibrium point of 100. In alternate scenarios, alterations in the standard deviation of the threshold distribution among these 100 individuals influence the equilibrium point, as dictated by the cumulative distribution function.

Within the threshold model, collective behaviors unfold within social networks through interdependent influences among their members. Over time, the distribution of thresholds changes, exerting an impact on behavioral performances. Therefore, a systematic exploration of social media usage necessitates an appreciation of the evolving nature of social networks. To examine changes with time, the application of differential equations proves valuable. In a parallel example, differential equations serve as a tool to scrutinize the temporal transformations in ties within social networks. Adopting a systematic perspective that incorporates these dynamic considerations enhances the comprehension of how social networks evolve over time and underscores the relevance of employing differential equations in such analyses.

In the study by van Dijk and van Winden [39], the reciprocal dynamics between alterations in social ties and the provision of local public goods were examined [40]. The authors introduced a differential equation as a conceptual tool to assess the evolution of social ties. The specific equation is articulated as follows:

$$d_{\partial ij}/d_t=f_i\left(G_{ij}\text{,}{\partial }_{ij}\right)$$E4

In this equation, the formation of a tie needs the weighted relative contributions (G_ij) from individual i to individual j, and the sentiment state (α_ij) of individual i toward individual j. The boundaries of the sentiment state are from -1 to 1, which stands for negative and positive feelings. Thus, f_i is a S-shaped function of G_ij, where f_i → 1-α_ij (−1 − α_ij), when G_ij → ∞(−∞). When this slope grows with time, it approaches the maximum value of 1, and the diminishing marginal effect of Gij occurs, which means the growth of G_ij over time brings about the decline of the sentiment of individual i, and just like threshold theory, this equation also has an equilibrium point where α_ij = f_i (G_ij). This depiction of tie evolution suggests that the establishment of social ties is inherently dynamic and subject to temporal variations. Any alterations in the equilibrium point will consequently induce changes in ties, leading to either a decline or an increase in their prevalence. Consequently, the corresponding structures within social networks transform in response to these dynamic shifts.

In summary, when viewed through the lens of social network analysis, the fundamental unit of social behaviors—interactions—can give rise to diverse social relationships and simple or intricate structures within social networks. These established social relationships and structures, in turn, exert an influence on behavioral outcomes. With accessible data, this reciprocal relationship can be systematically analyzed, unveiling the underlying patterns in social media usage as an integral component of a broader systematic process.

4. Possible hypotheses regarding structural features of social media usage

Social network analysis, as discussed in this paper, provides a means to investigate the balances and changes that occur in social networks, thus the influences on individual behaviors. The dynamic analysis could track the changes in interaction processes and thence could make a good simulation of the real systematic processes [41, 42, 43]. With the development of digital and internet techniques, interactions on social media could be recorded accurately, which could benefit scholars of social media research to find out the systematic dynamics to better understand the various uses of social media. The possible hypotheses could be brought up as follows:

H1: The configurations of edges/2-stars/3-stars/triangles are more likely to occur within social networks on social media.

H2: The configurations and social ties in given social networks could change over time on social media.

H3: The formation/changes of configurations and social ties in given social networks could positively/negatively impact social media usage.

Hypothesis 1 (H1) aims to elucidate the underlying regularities inherent in social networks, just as the first part of Section 3. Specifically, it posits that if interactions within a given social network yield an abundance of 2-star configurations and triangles, individuals are inclined to form stable cliques rather than expand their social circles. Consequently, individuals’ patterns of social media usage are likely to be influenced by their close-knit relationship groups. Consequently, tailored information catering to the diverse interests of these cliques is expected to garner greater acceptance. Conversely, when 3-star configurations predominate within the social network, it suggests that individuals exhibit a propensity to reach out to a wider array of contacts. This inclination signifies a proclivity toward seeking novel information beyond their immediate social milieu. Consequently, information spanning diverse fields is anticipated to be well-received under such circumstances.

H2 examines the changes in configurations within specific social networks. As elaborated in the latter part of Section 3, these changes can be systematically quantified and analyzed using differential equations, particularly in the context of longitudinal studies. Previous scholarly works have highlighted variations in the distribution of thresholds concerning behavioral performance [8], the establishment of social connections, and shifts in configurations [29, 39]. A nuanced understanding of changes within specific social networks enables a more comprehensive exploration of the dynamic processes caused by interactions, thus facilitating investigations into the dynamics influencing behavioral performance.

H3 posits that the formation and changes of social networks significantly influence patterns of social media usage behavior. Prior research indicates that social structures, including social networks, not only facilitate the establishment of social relationships but also contribute to the formation of common behavioral beliefs and norms, thereby shaping social and cultural contexts. Within these contexts, individuals adhere to prevailing beliefs and norms to access social support and resources [16, 17]. Consequently, social structures formed by interactions, such as social networks, could influence the behavioral patterns of their participants. Existing literature has already explored the impacts of social networks on various behaviors, including economic behaviors and intimate relationships, as well as other collective behaviors [8, 14, 39]. Therefore, it is logical to hypothesize that social networks similarly impact patterns of social media usage.

These hypotheses are formulated to scrutinize the behavior of social media usage through a systematic lens, acknowledging that achieving equilibrium and adaptation to change are central objectives for the systems of social networks. Additional specific hypotheses are to be tailored according to the research objectives outlined by scholars. The precision of probability distributions can be enhanced with more detailed information obtained from social networks concerning social media usage. Empirical investigations could further facilitate the inclusion of interactions on different levels, spanning from the micro-level of individual interactions to the macro-level involving social groups and other entities coming from various domains of societies.

5. Conclusions

This review paper delves into the dynamics underpinning behaviors, which is interaction. Interactions could facilitate connections between systems to attain equilibriums and bring changes in systems to adapt to environmental fluctuations. Adopting a systematic perspective offers a comprehensive and adaptable approach to scrutinizing the intricacies of human behaviors, including those related to social media usage. In practical contexts, conventional institutionalized beliefs and norms governing behavioral patterns often fall short of elucidating collective behaviors such as strikes, rumor propagation, and riots, which frequently diverge from established norms and beliefs. This divergence underscores the inadequacy of solely relying on normal beliefs and norms to explain human behaviors across diverse social and cultural contexts. However, with an interactive and systematic perspective, interactions are viewed as the basic dynamic and control mechanism of systems, hence behaviors adhering to established norms and beliefs can be viewed as outcomes of equilibriums shaped by interactions among systems. Conversely, stochastic and collective behaviors that deviate from established norms are regarded as manifestations of system evolution through interactions. Within this framework, social network analysis emerges as a pivotal methodology for studying the dynamics of social media usage. It could examine the structural outcomes of interactions, which manifest as various forms of social networks, by utilizing the interaction between two actors as the fundamental unit of analysis.

Social media, functioning as an ideal platform for observing dynamic interactions at both micro and macro levels, facilitates the exploration of digital records through interactions as analytical units. This approach allows social network analysis to employ mathematical techniques for the comprehensive study of dynamic interactions on social media. Structural features are recognized as crucial elements influencing behaviors, yet historically, dominant theories have often overlooked their interpretive capabilities for stochastic and dynamic interactions. The limitations in tracking and analyzing interactions across various scales have contributed to this oversight. However, the growing prevalence of the Internet and advancements in digitalization techniques reveal the potential of social network analysis to analyze dynamic interactions, emphasizing the need for additional empirical research for validation.

The paradigm of social media studies is suggested to evolve through the adoption of relational and interactive perspectives. Utilizing mathematical techniques such as maximum likelihood estimation and p models introduced in this study, social network analysis becomes instrumental in studying stochastic and dynamic events, unraveling the structural features of social media usage. These techniques hold the promise of unveiling the underlying mechanisms driving social media usage, urging scholars in the realm of social media studies to dedicate attention to these methodologies in future research endeavors.

Conflict of interest

The authors declare no conflict of interest.