Definition of situation refers to the state that follows after an influencing agent provides information in a deliberate or unintentional attempt to induce behaviour (Kelman, 1974). Deliberate induction of behaviour takes the form of direct commands, such as threats, requirements, orders or persuasion. Unintentional induction can be an indirect expression of expectations, thoughts and norms (Kelman, 1979). Induced behaviour is discrepant to some extent from the behaviour and attitudes of the individual upon whom influence is exerted, which typically entails some degree of the individual’s resistance to change. This is because people are inherently inclined to cognitive stability, which reduces personal receptivity to new information. If receptivity to new information is weak, the influence of the new information will probably be neutralised by the recipient (Pelinka & Suedfeld, 2017). Therefore, the chances are that the definition of situation will lead to individuals’ willingness to expose themselves to the influence, depending on the degree to which the corresponding psychological prerequisite (an expected goal of induced behaviour) and the situational condition (importance of induced behaviour) are met. To minimise resistance to social influence, the information about induced behaviour should convince an individual that the adoption of induced behaviour, although challenging an existing system of beliefs and attitudes, is conducive to the achievement of an individual’s goals (Kelman, 1974; Kelman, 1979). The definition of situation activates the willingness to expose oneself to the induced behaviour if its consequences are perceived as relevant for goal achievement. Consequently, the strength of motivation to respond to the influence depends on the degree of importance of the induced behaviour to one’s own goals (Kelman, 1974). Presentation of agent is the stage when a person evaluates the characteristics of an influencing agent, which may be social status, expertise, prestige, values, available resources and other attributes (Kelman, 1953; Kelman, 1974). The relationship between this stage of stimulus evaluation with the corresponding psychological prerequisite means that the influencing agent is evaluated in terms of how powerful the agent is in creating the required conditions for goal achievement. By deriving meaning from the attributes and communication of influencing agents, individuals can assess the relevance of the behaviour of the influencing agents to the achievement of personal goals (DeShields, Kara & Kaynak, 1996; Kelman, 1958; Kelman, 1974). The awareness of the characteristics of the influencer that are salient for a person reduces the psychological distance and resistance to the elicited information (Kelman, 1979). Characteristics that resonate with the individual signal favourable conditions for goal achievement, ensure a positive orientation toward the influencing agent and enhance predisposition towards the adoption of the induced behaviour. Hence, the factor amplifying the likelihood that agent evaluation will translate into positive orientation towards the agent is the strength of their perceived power (Kelman, 1974). Specification of response is the process preceding the adoption of behaviour, whereby an individual forms an understanding of a precise course of action for the behaviour to be implemented (Kelman, 1974). The induction of behaviour is conditioned by its prepotency. Prepotency is the extent to which induced behaviour is considered the most clearly relevant that has been activated in the influence situation. In other words, the stronger and the more relevant the induced behaviour compared to other available alternatives, the more chances are that behaviour will be induced (Kelman, 1953; Kelman, 1974).

The difference in the importance of induced behaviour, the power sources of an influencing agent, and the manner in which the induced behaviour has become prepotent contribute to the variance in behavioural responses to the social influence (Pelinka & Suedfeld, 2017; O'Keefe, 2016). The resulting changes in behaviour may be overt or covert in the form of new attitudes and beliefs (Kelman, 1979). Overt and covert behaviour change (i.e. influence acceptance) is manifested through either of the three routes, i.e. compliance with the influence, identification with an influencing agent or internalisation of the induced behaviour. The induction of the behaviour through the compliance, identification and internalisation routes is associated with different emotions and individuals’ expectations of the embeddedness of induced behaviour in the social system (Table 1).  Specifically, compliance is the acceptance of influence for individuals to receive rewards and avoid punishments for non-compliant behaviour (Kelman, 1958; Pelinka & Suedfeld, 2017). Compliance is a response to external demands motivated by concerns about the social ramifications of that behaviour (Pelinka & Suedfeld, 2017). Examples of such behaviour include abiding by laws and the prescriptions of a doctor. Such behaviour helps reduce the discrepancy in rules and norms cultivated and imposed by the external social environment, and, consequently, eliminates the feelings of social fear and embarrassment that noncompliance may cause (Kelman, 1974). The induced behaviour is likely to be performed under the surveillance of the influencing agent (Pelinka & Suedfeld, 2017). In the patient-doctor dynamics, compliance is higher when doctors have close interaction with patients and the latter tend to seek doctors’ approval (Davis, 1971). Similarly, the enforcement of laws is merely possible due to the existence of formal institutions controlling compliance with laws in different spheres of life (Jackson et al., 2012). Therefore, compliance is likely to occur if an influencing agent has the means of control. The agent would exert power and authority in relation to the recipients of communication, while the latter would have restricted behavioural choices to act otherwise (Bagozzi & Lee, 2002; Cialdini & Goldstein, 2004).

Identification is a social influence with the purpose of achieving self-definite goals (Kelman, 1958; Pelinka & Suedfeld, 2017). Like compliance, the adoption of behaviour following identification does not manifest individuals’ personal values, but it is intrinsically satisfying (Cialdini & Goldstein, 2004; Kelman, 1958; Pelinka & Suedfeld, 2017). The induction of behaviour through identification is based on the expectation that the behaviour helps fulfil particular social roles. It is important for people because it can promote one’s own position in society. The induced behaviour becomes dominant when demands in social roles are clearly delineated (Kelman, 1958; Pelinka & Suedfeld, 2017). The social roles become apparent if individuals affiliate themselves with a particular social category (Kelman, 1974). For example, individuals can be nudged into the consumption of sustainable products and voting for a democratic political party in the United States of America under the expectation that identification with green consumerism and social equality would help sustain membership in respective social and political groups (Bagozzi & Lee, 2002). Identification is more likely to take place, if the relationship to an influencing agent is salient, while the source inducing the behaviour should be attractive enough to foster the desire to affiliate with it (Kelman, 1958; Pelinka & Suedfeld, 2017). Therefore, refusal to identify oneself with a social category with a dominating narrative in society can create feelings of guilt and shame (Kelman, 1974).

Influence internalisation is an act of adopting behaviour because it is construed as being congruent with personal values and views, such as deeply entrenched beliefs and attitudes about social conduct, norms and idealised images (Kelman, 1958; Pelinka & Suedfeld, 2017; O'Keefe, 2016). Such beliefs, values and attitudes usually work as self-guiding principles nurtured early on in people’s lives under the influence of the social environment the person grew up in (Bagozzi & Lee, 2002). The denial of social influence internalisation that goes against deeply rooted values could lead to emotional reactions, such as regret and self-disappointment (Kelman, 1974). For the behaviour to be perceived as congruent, it has to be relevant and easily relatable to personal goals (Kelman, 1979; Pelinka & Suedfeld, 2017). The perception of behavioural congruency and intrinsic value for the individual is facilitated when an influencing agent is perceived as credible (Kelman, 1958). The internalisation of behaviour aligned with the intrinsic needs of a person is expected to result in the restructuring of the existing value system and means-end framework for goal achievement (Kelman, 1958).

The three cognitive modes of influence acceptance can result in different behavioural responses (Bagozzi & Lee, 2002). Out of the three acceptance processes, internalisation can be considered to be the strongest predictor of long-term commitment, as it does not depend on a social and situational context (Kelman, 2006). In the case of internalisation, individuals’ behaviour is consistent with norms and beliefs, which are already internalised and deep-rooted in the existing system of values. As the influence is not dissonant with existing cognitions, the induced behaviour will more likely prevail in any social context and under any circumstances (Kelman, 1958). For example, individuals are willing to engage in pro-environmental behaviour in the long term if pro-environmental values had been deeply entrenched in the individual before influence elicitation took place. In contrast, the commitment to behaviour accepted through identification depends on the context, where the induced behaviour is relevant for maintaining desirable social identity and reciprocal social relationships (Stryker & Serpe, 1982). For example, commitment to organisations based on an employee’s corporate identification will carry on until the employee changes the workplace. The effect of compliance is short-term, as the behaviour is constrained to the context where surveillance is possible and other behavioural alternatives are limited (Kelman, 1958). In other words, the induction of behaviour is compelled by the need to create a favourable short-term impression in specific social conditions (Friedlander & Schwartz, 1985). For example, to provide services, police and healthcare workers require people to enact certain behaviours, which may not be sustained once the service or treatment is complete.

The development of Social Influence Theory marked a turning point in social influence research, addressing a gap in prior studies, which predominantly focused on instances of conformity (Asch, 1961; Sherif, 1935).  SIT advanced the literature in two ways. Firstly, it offered a nuanced and qualitative analysis of social influence situations. It made it possible to delineate between individuals' responses to persuasion, ranging from superficial conformity to genuine changes in beliefs, in both dual relationships and group dynamics. Secondly, SIT complemented the functionalist approach to attitude (Katz, 1960; Smith, Bruner & White, 1956). In addition to interpreting induced behaviour in functional terms, the theory also provided classifications of the different meanings of situations in which influences occur and the implications of these situations for attitude and behaviour change (Kelman, 1979).

Theory Extension

A model of interpersonal influence characteristics

SIT has inspired numerous studies in social psychology, with scholars building upon and developing new frameworks. One of the notable modifications of SIT is the development of a new Model of Interpersonal Influence Characteristics, by Levy et al. (Levy, Collins & Nail, 1998). Levy et al. (Levy, Collins & Nail, 1998) reconceptualised the social influence literature produced in 1950, largely dominated by the research of Kelman (1953; 1958; 1974) and newer social influence taxonomies (Raven, 1992). The new conceptual taxonomy was aimed at addressing the challenge in the existing theories on social influence, namely vagueness in the concepts within the sphere of social influence, and the challenge of differentiating social influence from other topics in social psychology.

The model of Levy et al. (1998) was different from traditional approaches in that it did not attempt to offer a mutually exclusive and discrete categorisation of social influence instances for predictive purposes. Instead, it offered a parsimonious model for analysing interpersonal influence cases. The model consists of four fundamental elements: (1) the level of cognitive processing, (2) the intentionality of the influence, (3) social status, and (4) the direction of change. The level of cognitive processing refers to the degree of an individual’s awareness of being exposed to the influence of another person, which is represented as the conscious-unconscious dichotomy (Levy, Collins & Nail, 1998). This dichotomy corresponds to the views of other cognitive psychologists, suggesting that people tend to process information in an automatic or deliberate manner (Petty & Cacioppo, 1986). Automatic and deliberate cognitive processes contrast in the degree to which they are controlled by an individual and the level of attentional resources they require (Wyer & Srull, 1994; Levy, Collins & Nail, 1998). For example, conscious cognitive processing is prompted by an individual’s awareness of the attempt to induce behaviour. The influencee puts mental effort into considering whether a behavioural or attitudinal response should align with or oppose the induced behaviour. Responses could be the decision to abide by enforced legal regulation or rebel against the imposition of rules by legitimate authorities. Unconscious cognitive processing occurs outside of the influencee’s awareness and control. It happens without cognitive guidance and results in such responses as intuitive alignment to commonly accepted social norms, instinctive social responses (e.g. equity and reciprocity), and spontaneous reactions in the form of attraction or repulsion (Levy, Collins & Nail, 1998).

The intentionality of the influence refers to whether an influencing agent has the intention to exert power on an individual. It is considered only in cases of the conscious processing of influences (Levy, Collins & Nail, 1998). The notion of unintentionality can be traced back to Social Influence Theory, emphasising that identification can result from an unintentional attempt to provoke actions (e.g. self-categorisation to achieve a desired social status) (Kelman, 1958). The degree of intentionality takes the form of a three-level category: intentional, unintentional and irrelevant influences. Intentional influence is the one which is exerted deliberately (e.g. rules of conduct) and inflicts either compliant or non-compliant responses (Levy, Collins & Nail, 1998). Unintentional influence stipulates imitation or anti-imitation, manifested as the increased or decreased similarity in behavioural, cognitive or affective characteristics between individuals and influencers. Examples of imitation could be acting in line with group norms and the adoption of the traits and characteristics of the influencer. Consequently, the instances of anti-imitation could be acting in opposition to social norms and the imitation of traits and characteristics of the opposite social group (Levy, Collins & Nail, 1998; Levy & Nail, 1993). The irrelevant category encompasses the concepts that are theoretically irrelevant, where the intentionality of the influence is impossible to trace. Under such circumstances, the acceptance of influence can hinge upon individuals’ personal experience, observations or exposure to persuasive or impactful communication (Levy, Collins & Nail, 1998).

Social status refers to the relative difference of an individual from an influencing agent (lower vs peer, vs higher vs irrelevant), which can correspond to different forms of social influence acceptance (Levy, Collins & Nail, 1998). Specifically, a higher social status of an influencing agent bestows upon them the power to make individuals obedient or, conversely, can trigger rebellious reactions against influencers (Levy, Collins & Nail, 1998). In the condition of influencers’ lower social status, individuals can also succumb to social pressure if they have responsibility over the influencers. Such a form of compliance is referred to as power-of-the-powerless (Raven, 1993). Negative power dynamics for influencees can also provide them with the grounds to oppose the influence—an act referred to as counter power-of-the-powerless (Levy, Collins & Nail, 1998). Intentional influence from peers is postulated to result in either direct conformity or non-conformity with social expectations and norms. Similarly, an unintentional influence attempt from peers is poised to result in either imitation or anti-imitation of the behaviour and attitudes of a social group (Levy, Collins & Nail, 1998).

The direction of change takes the form of positive, negative and irrelevant responses. The inclusion of the direction of change characteristic in the framework was rooted in the proposition that sometimes influence can be counter-effective and raise an opposite reaction (Levy, Collins & Nail, 1998). Table 2 summarises and defines all types of responses to social influence. The relationships between the four fundamental interpersonal influence characteristics making up the decision tree are presented in Figure 2.

Note: the concept appears in the order in which it is presented in Figure 2

The decision tree of interpersonal influence characteristics served as a novel conceptualisation of Kelman’s Social Influence Theory and other research in the broad domain of social influence (French, 1956; Raven, 1992). Consisting of only four types of characteristics, the framework represents a logical system for differentiating a higher number of interpersonal influence instances, compared to Social Influence Theory. Also, reformulating existing knowledge on social influence opened avenues for theoretical advances, especially around influence intentionality and unconscious processing, which had not been emphasised in prior theories. The decision tree by Levy et al. (Levy, Collins & Nail, 1998) presents distinct subcategories for unintentional influence (i.e. indirect conformity, anti-conformity, identification, disinhibitory contagion, and residual influence types) and unconscious influence processing (i.e. attraction, echo contagion, hysterical contagion, repulsion, social loafing and other influences theoretically irrelevant to the concept category). Consequently, the model not only addressed the knowledge gaps persistent in social influence research at that time, but paved the way for further empirical inquiry into the underexplored dimensions of social influence situations.

Applications

For decades, Social Influence Theory guided studies across various disciplines. Initially, its application was predominantly seen in organisational, medical, and socio-political studies. In the discipline of organisational management, the SIT was employed to discern behavioural patterns related to compliance, identification, and the internalisation processes influencing employees' behavioural development. For example, researchers delved into the impact of authority on employee compliance (Freedman, 1981), negative behaviour development among co-workers due to group affiliations (Herold & Conlon, 1981) and the changes in internalised beliefs among demographically (dis)similar individuals (Eagly, Wood & Fishbaugh, 1981). In medical studies, the theory was utilised to explore the factors inherent to social environmental influencing health-destructive and health-protective behaviours, such as resistance to smoking (Covington, 1981) and adherence to treatments advocated by healthcare practitioners (Dembroski, Lasater & Ramirez, 1978; Rodin & Janis, 1979). In the socio-political context, the theory found application in explaining the implications of large-scale interventions for specific nations' economic and political submissiveness (vs dominance) (Richardson, 1976), civil movements and socio-cultural transformations (Ewens & Ehrlich, 1973; Moscovici & Mugny, 1985; Pool, Wood & Leck, 1998), political polarisation (Kelman, 1970; MOR, 2007; Watanabe et al., 2017) and other phenomena. Equally valuable was SIT’s application in understanding social relations at the community level. For instance, researchers investigated the role of the social context in the development of cultural preferences among children (Bunton & Weissbach, 1974; Kochanska, Tjebkes & Fortnan, 1998) and the influence of values on the perception of policies of international financial institutions (Riggs, 1980). At the individual level, the theory was employed to examine various aspects of interpersonal relations, including but not limited to the perception of power and the legitimacy of actions (Bickman, 1974), stigmatisation and dynamics in interracial interactions (Page, 1997).  

In the past two decades, along with the unfaded interest in the social influence phenomenon within organisational and larger social systems (Binyamin, 2020), the application of SIT has also expanded in the domains of marketing and information system management (Baker et al., 2014; Cheung et al., 2022; Huang, 2019). As research on user interaction around information systems has burgeoned, scholars have adapted Kelman's framework to investigate shifts in attitudes toward new technology. For example, SIT was integrated with the Elaboration Likelihood Model developed by Petty and Cacioppo (1986) to understand the effectiveness of influence in the process of information systems acceptance. Modifications in SIT also concerned the integration of context-specific factors associated with influence acceptance. Studies examined the impact of system characteristics (design quality, technology quality, information quality) on identification (Cheung et al., 2022). SIT was also used for examining post-adoption behaviour, whereby the effect of identification processes on continuous usage of IT services was tested alongside the effect of the expectation-confirmation mechanism (i.e. the assessment of the degree to which expectations are confirmed by actual performance) (Huang, 2019).

The adaptability of SIT to various scenarios of interpersonal interactions has led to implications for various sub-streams in marketing and public relations research. In the traditional marketing context, for instance, researchers have delved into consumption behaviour under the influence of communication from referent groups and the alignment of influence with personal goals and values (Yang, Tseng & Lee, 2021). In the business-to-business environment, the focus has extended to examining the impact of brand communications on the behaviour and performance of customer service representatives (Baker et al., 2014). Recently, with more companies venturing into sustainable production, there is a growing body of research utilising SIT for understanding the underpinnings of pro-environmental behaviour. Studies have examined the influence of sustainability advocacy groups and personal norms on sustainable consumption choices (Confetto et al., 2023; Elgaaied-Gambier, Monnot & Reniou, 2018). The applications of Social Influence Theory have been instrumental in designing persuasive messages and campaigns that align with different company objectives, whether aiming for momentary compliance or the enduring internalisation of influence. In the digital marketing context, SIT has been used to explain user behaviour on social media platforms (Oliveira, Garcia & Vivacqua, 2021; Sánchez-Fernández & Jiménez-Castillo, 2021; Santiago, Magueta & Dias, 2020) and the attitudinal and behavioural implications of online influencer marketing (Fan & Chan, 2023; Tafesse & Wood, 2021). In particular, studies found that social media influencers’ personality characteristics (e.g. closeness, interactivity, homophily and originality) drive imitation and the adoption of induced behaviour ((Chloe) Ki, Park & Kim, 2022; Li & Peng, 2021). Researchers have applied the theory loosely, however, primarily focusing on the characteristics of an influencing agent. This emphasis often occurs without due consideration to the relation of the examined characteristics to a particular route of influence acceptance.

Limitations

Social Influence Theory garnered critique around its lack of acknowledgement of contextual factors, its inability to differentiate the intentionality of influences and associated behavioural responses, and insufficient attention to the role of power dynamics in social influence situations and cognitive processes of persuasion (Latané, 1981; Levy, Collins & Nail, 1998; Lisha et al., 2017; Petty & Cacioppo, 1986; Wang et al., 2022). In particular, researchers emphasised the need to explore the contextual differences of situations affecting variance in social influence effects. To that end, some critics argued that Kelman's theory may not be adequate to address the role of situational factors in social influence processes (Dong et al., 2021). It was found that temporal-spatial factors (i.e. time difference and the crowdedness of the environment where influence is elicited) affect the evaluation of the characteristics of an influencing agent and the persuasiveness of communication consequently (Dong et al., 2021). Other scholars contended that the theory does not account for the variations in social influence processes across different cultures (Lisha et al., 2017; Wang et al., 2022). Evidence from cross-cultural research has suggested that social influence dynamics can be influenced by cultural norms and values (Wang et al., 2022). Therefore, cultural differences should be considered to eliminate flawed generalisations. Second, the notion of intentionality was implicit in Social Influence Theory, suggesting that identification did not require an influencing agent to deliberately exert power upon an individual to stimulate the induction of behaviour to secure membership in a particular social group (Kelman, 1953; Levy, Collins & Nail, 1998). The lack of explicit distinctions between intentional and unintentional influences limited the understanding of the types of potential behavioural responses that unintentional influences may entail (Levy, Collins & Nail, 1998). Drawing on this limitation of SIT, Levy et al. (Levy, Collins & Nail, 1998) developed the Model of Interpersonal Influence Characteristics and provided an explicit categorisation and scenarios under which unintentional influence can be exerted and the responses it entails. Third, it has also been highlighted that SIT did not pay necessary consideration to power dynamics in social influence situations. In that regard, the Social Impact Theory by Latané (1981) emphasises the role of power and its distribution. Social Impact Theory postulates that the strength of the influencing group of people, the immediacy (proximity) of the agent to the individual and the number of people in the group determine how people respond to social influence. Finally, it has been argued that, in an attempt to cover the broader scope of responses to social influences, SIT provides an insufficient explanation of the cognitive processes underscoring communication persuasion (Petty & Cacioppo, 1986).  To address that limitation, the Elaboration Likelihood Model developed by Petty and Cacioppo (1986) delves into the motivational and ability factors impacting information processing and persuasion.