Six important theories in information behaviour research: a systematic review and future directions

Introduction

Human information behaviour research has evolved rapidly in the last decades, investigating why and how humans need, seek, process and use information for work, school and everyday life (Ford, 2015). Relatively complete theories have been established in information seeking (Ellis, 1989; Savolainen, 1995), but the scope of information behaviour is far more than that.

Information behaviour could be seen as "the totality of human behaviour in relation to sources and channels of information including both active and passive information seeking and information use" (Wilson, 2000, pp. 49). Given this broader view, we focus on the interaction process between people and information (including information products, service and environment) (Case and Given, 2016), instead of investigating a specific type of information behaviour. This process perspective leads us to be more concerned with how information behaviour occurs (e.g., how information behaviour decisions are made and how information products and services are designed), and why some information behaviour occurs (such as media use preference, collective information behaviour, consumer satisfaction and online user participation, etc.). Investigating various kinds of information behaviour in interaction, related information behaviour researchers have accumulated a large body of empirical research, exploring a broad array of motives, contexts and goals, usually based on theories from disciplines such as psychology, sociology and philosophy (Webster and Watson, 2002).

However, fragmentation, specialisation and appropriation are still manifested in some empirical outcomes, indicating that the lack of theories is a chronic problem (Wilson and Maceviciute, 2013). Furthermore, with the expanded connotation of information behaviour and overwhelming various theories, it is time-consuming and exhausting for empirical researchers to find appropriate theoretical foundations to investigate antecedents, processes and mechanisms of information behaviour. Hence, an immediate need appears to know more about these theories borrowed from other disciplines, that is what this paper is dedicated to accomplishing.

There are abundant theories involved in information behaviour research, and it is unrealistic to incorporate all the theories in one review. Based on empirical studies accumulated in the past 30 years, six theories (i.e., ethnographic decision tree theory, means-end chain theory, media richness theory, collective action theory, service encounter theory and motivation crowding theory) relatively important in information behaviour research but rarely reviewed are selected. We outlined these theories, by illustrating their development and application orientations as well as suggesting potential theoretical directions for future information behavioural researchers. Both theories that are widely applied and those with a small number of outcomes, are included.

The rest of this paper proceeds as follows. First, a comprehensive framework combining information behaviour and six theories is presented, followed by illustrating our methodology for literature search and inclusion. Next, we reviewed these six theories in the next six sections respectively: summarise the core idea and theoretical development for respective theories, identify research agenda by synthesising and analysing extant findings using these theories, and propose significant directions for future research. Finally, we concluded our contributions.

Comprehensive framework

Various information behaviour studies have widened the application boundaries of the six theories mentioned above, as well as providing solid empirical support. In the extant management information system studies, however, limited attempts have been made to systematically review the theories which are diverted from other disciplines to information behaviour literature. Hence, it is crucial to form a clear picture that will be useful for applications of the six theories in information behaviour research, in order to support a better understanding of concepts or theories and promote future research (Senyo et al., 2019).

A comprehensive framework of six theories is constructed (Table 1), based on the diversity and generality of information behaviour empirical studies, to further address the fragmentation of theories. All the theories come from disciplines more or less related to the field of information systems. As for theory application, we deconstruct the general questions about information behaviour research and further divide the research paradigm into three topics. Given that we aim to identify the research agenda of existing theories rather than providing the basic information of relevant studies, we mainly focus on the certain aspects of information behaviour and their research orientations.

Specifically, we pay attention to three aspects in information behaviour research: how behaviour happens and will evolve, why behaviour happens and why things go off track. Ethnographic decision tree theory and means-end chain theory, with their own methodologies, could be used to investigate the process of behavioural decision making and mechanisms of information product and service design, respectively; media richness theory, collective action theory and service encounter theory all perform well in investigating antecedents and factors for some kinds of information behaviour; motivation crowding theory may be helpful to explain some undesired and unexpected outcomes.

Literature search and inclusion

We searched the database Web of Science Core Collection for relevant studies on information behaviour research using the six theories. For theory, we used their names and abbreviations; for information behaviour research, we mainly locate at research area Information science & library science. Other fields, like computer science and communication, are also taken into consideration to find the most relevant articles, since information science is an interdisciplinary field straddling other disciplines and researchers can often find similar examples in related areas (Webster and Watson, 2002). Accepting the search answers provided by Web of Science (sorted by relevance), titles and abstracts of first 100~200 articles were reviewed to identify empirical research. We also obtained relevant papers through a forward and backward search.

This review is topic-centric. Our purpose is to help researchers understand main research orientations which these theories could apply to in the field of information behaviour. Hence, the focus to collect articles is to form some common topics for each theory. We do not have and do not need to integrate all the knowledge elements provided by extensive literature into an overall logic (Rowe, 2014).

Ethnographic decision tree theory

Theoretical development

Ethnographic decision tree theory, also called ethnographic decision tree modelling (Gladwin, 1989), is an inductive theory and method used to analyse the ethnographic data to determine and predict patterns of decision logic of individuals in a group who perform a certain activity (Beck, 2005).

Based on abundant ethnographic studies in 20th century and his ethnographic experience (Gladwin, 1983; Gladwin, 1976), Gladwin (1989, pp. 1-10) found that most decision-making can be regarded as a do or not do dichotomous issue, corresponding to the binary characteristics of decision trees, and elaborated the idea of ethnographic decision tree theory in his book Ethnographic decision tree modeling. Gladwin refers to natural decision-makers as experts using jargons in their own choice-making endeavours and comfort zones (Mishra, 2014). He seeks to reproduce their decision-making processes in real-life situations through the perspectives of the target group members themselves, instead of exploring how decisions should be made (March, 2009). This theory uses ethnographic fieldwork techniques to elicit decision criteria from decision makers, which are then translated into a hierarchical decision tree with a series of if-then rules (Gladwin, 1989).

The process of conducting ethnographic decision tree modelling, described in detail by Gladwin (1989) and improved by other scholars (e.g., Ryan and Bernard, 2006; Mwangi and Brown, 2015), entails two phases: preliminary model creation through exploratory data collection and model testing through quantitative verification. A general outline of the two phases is pictorially depicted in Figure 1.

Research agenda

Ethnographic decision tree theory provides a both predictive and descriptive framework to understand the process of group decision-making deeply, increasing the authenticity and explanatory power of results. In addition to its merits in investigating behavioural decision processes of some specific groups, recent research implies that it may help to detect factors of user churn.

Different behavioural decisions about college students have been examined, such as why students keep a blog or not in informal learning contexts (Andergassen et al., 2009) and why they play online role-playing games or not (Chang and Yang, 2013). Besides, this theory is also associated with the use of technology. Bailey and Ngwenyama (2013) developed an ethnographic decision tree model to illustrate the process through which a community member decides to use the tele-centre to support economic livelihood, with key factors identified such as social ties, opportunity recognition and support from the telecentres. Mwangi and Brown (2015) unearthed how target small enterprise users make decisions regarding mobile banking services by looking at the adopters’ and non-adopters’ decision criteria. Davenport et al. (2012) constructed a preliminary decision tree model of how older adults make smart technology decisions, investigating their perceived smart technology needs.

On the other hand, ethnographic decision tree theory plays a role in investigating the decision about why a group starts and stops doing something. Reasons are found to vary significantly, with intrinsic motivation factors keeping individuals to continue, whereas stopping is mostly attributable to external factors (Andergassen et al., 2009; Fanget al., 2009). Since ethnographic decision trees could present decision processes of decision makers (Gladwin, 1989), knowledge and information of their needs could be explored for further analysis (Jacko and Sears, 2003), thus making tailored policy interventions to prevent user churn (Gladwin, 1989; Gladwin et al., 2002).

Future direction

Gladwin (1989) highlights conducting sufficient ethnographic research with various efforts, usually including ethnographic interviews (Charmaz, 2006; Spradley, 1979) in the model building phase. However, most studies fail to do enough when asking questions and having conversations, which may lead to low quality of decision criteria elicited and low explanatory power of the model built for real-life problems. Hence, future scholars are encouraged to collect more sufficient exploratory data to understand what decision makers think about their decisions. One efficient way is to make use of the latest experience and historical decisions to prompt participants, encouraging them to search more in their memory (Breslin, 2000). In addition, combining interviews with sociological and psychological methods during the ethnographic process may help to improve its quality. For instance, field notes add details about constructs, qualitative content analysis helps to uncover subtle differences in semantics of decision criteria (Graneheim and Lundman, 2004) and the psychology of personal construct by Kelly (1995) is likely to decrease mistaken ethnocentric ideas from researchers.

After collecting sufficient data and information, future research may investigate more on the structure and dimensions of obtained tree models. Grounded theory methodology is useful to extend the logic relationships of decision trees, since it is highly sensitive to theories and uses constant comparison to ensure the best fit between concepts and data (Wiesche et al., 2017). The parallel constructs and categories used in grounded theory, could be organically associated with the if-then rules of ethnographic decision tree theory, thus expanding model scale. Besides, a role model for decision makers could be further developed in future research, since sociological symbolic interaction theory indicates that attributes of the role one played could be translated into actual behaviour (Blumer, 1969; Johnson and Williams, 1993).

In the model testing phase, future goals might be to eliminate the subjectivity of ethnographic decision tree modelling to the greatest extent, particularly, to multiply quantitative testing methods besides counting the results of questionnaires. For instance, verify those data using statistics of probability concerning the decisions, which links research outcomes with real-life situations (Gladwin, 1983). Running regressions using results data may also help to identify key factors and their correlations in the decision-making process (Gladwin et al., 2002). Except for improving quantitative verification, future researchers are supposed to strengthen the explanation and analysis of motivations and reasons about decision question, such as adding and comparing details extracted from interview notes and associating background information and key events of participants with their choices, which all provide scholars clues and information about why decision makers make the final decision. Various theories, especially those in cognitive science, are also recommended to be mix-used (Murray-Prior, 1998; Zachary et al., 2013).

To further develop the potential of ethnographic decision tree theory in the field of information behaviour, the application contexts are supposed to be expanded and discovered. Due to the proven efficiency of this theory in investigating behaviour of specific groups and factors of user churn, we encourage more studies using it especially in human-computer interaction, user behaviour on online platforms and other similar contexts. Considering the diversity of behaviour, multi-stage models with sequential decision trees may help to let things out and know more about the starting, continuing and stopping reasons for some complex actions (Gladwin, 1983).

The research questions behind decision criteria also need to be explored. Some studies have misidentified key decision criteria as critical success factors (Woodside and Baxter, 2013; Mwangi and Brown, 2015), which means researchers can use ethnographic decision tree modelling as exploratory steps in behavioural research, by using interview transcripts and model results to provide support for developing empirical propositions or hypotheses about information behaviour (Davenport et al., 2012).

Media richness theory

Theoretical development

In the early eighties, the means of communication began to diversify due to the development of computer communication technology. With different work tasks and more media channels, some questions about choices and the efficiency of media arise. Given that, American organizational theorists, Daft and Lengel (1987) explicitly introduced media richness theory. The core idea is that different media are classified into richness levels depending on information carrying capacity, and the degree of match between media richness and task complexity determines the communication effectiveness of users.

There are four basic elements of media richness theory (Daft and Lengel, 1987). (1) The definition of media richness (also called information richness), refers to the potential information carrying capacity of a medium. High wealth means that a medium could articulate issues clearly to attain consensus. (2) Four dimensions to discern the richness, including timeliness of feedback, multi-channel communication cues, linguistic diversity and personal attention. (3) The reduction of uncertainty and ambiguity. Organizational communication has two basic information purposes, which are to satisfy the need for information quantity and to reduce the degree of ambiguity and uncertainty (Shannon, 1948). (4) The degree of match between task complexity and media wealth, if the two are not matched, there will be insufficient information supply to support decision-making or information overload, increasing task complexity. According to the theory, communication media can be classified into five categories in descending order of richness: face-to-face communication; telecommunication; personal written documents, such as letters and memos; formal written documents, such as announcements and official documents; and digital documents (Daft and Lengel, 1987; Trevino et al., 1990).

Initially, the theory was used to describe and evaluate the communication media within an organization, but with the widespread adoption of various evolving media represented by social media, theoretical connotation and extension have been enlarged and supplemented accordingly. The hierarchical classification of media is not static, but evolves and is refined according to different contexts (Trevino et al., 1990). Traditional media richness theory considers human media adoption as rational behaviour, highlighting the objectivity of media and task characteristics; while subsequent theoretical extension studies point out this kind of behaviour is also related to socialisation conditions (Lee, 1994), personal preferences (El-Shinnawy and Markus, 1997) and even media-use skills (Rice and Shook, 1990). The concept of media wealth, therefore, is supposed to be investigated based on specific situations and research objects.

Research agenda

Individual performance in different contexts

At first, media richness theory was applied to solving organizational communication problems, such as demonstrating the efficiency of virtual teams dispersed in different times and spaces. At the beginning of organizing a virtual team, information and communications technology may restrict the interaction between members, but this limitation will gradually diminish over time, especially through skilful training (Warkentin and Beranek, 1999), and eventually will show a positive impact on communication performance (Chidambaram, 1996). There may be no significant difference between face-to-face communication and virtual communicating environment, which both convey useful information and help to achieve goals (Tan et al., 2012).

Using information systems to enhance knowledge exchange among members and subsequently boost the output performance of development collaboration is also a concern of some organizations. When performing complex, multi-party collaboration and high-speed information exchange tasks, the use of higher-rich media, such as collaborative product commerce, facilitates the transfer of difficult or skilful knowledge (Banker et al., 2006). The e-mail, as a higher-level medium than face-to-face communication, has positive impacts on the innovation and development of new products significantly (Ganesan et al., 2005).

In online learning, computer-mediated communication is thought to facilitate constructive learning and personalised learning (Muir-Herzig, 2004). Sun and Cheng (2007) found that the use of higher media-rich learning systems such as video and animation significantly improves learning efficiency especially when the content is difficult. For instructionally designed videos, the visual cues and their better logical connection in the storyline seem to improve the media effects, even triggering behavioural intention (Alamäki, 2019). Knowledge presented in different media varies in terms of attractiveness, expressiveness and ease of acceptance for learners, and will exert significant impacts on learners' concentration (Liu et al., 2009), motivation (Lan et al., 2011) and sense of stress (Rau et al., 2009).

Online fraud and user trust

Given the virtual and anonymous nature of the Internet, the medium of communication, the participants involved and the presentation of the information cues may give rise to problems such as online fraud. Kahai and Cooper (2003) indicated that richer media could foster social perception and enhance an individual's ability to detect deceptive behaviour, thus reducing its occurrence. Interestingly, both deceivers and receivers seem to perceive themselves as better able to perform in higher levels of highly synchronous media to detect deception, and deceivers' confidence in detecting deceptive behaviour is enhanced in high interaction-rich media (Carlson and George, 2004).

User trust is closely related to the amount of information carried in interaction and the ability of media to resolve uncertainty (Shannon, 1948). Aljukhadar et al. (2010) elucidated the moderating role of agent-retailer trust between media richness and users' purchase intentions in e-commerce. Similarly, highly media-rich online reviews may enhance users' perceptions of usefulness, credibility and persuasiveness, all three of which strengthen their willingness to consume online (Xu et al., 2015). A game experiment on three environments with different media richness, including physical commerce, virtual experiential commerce and traditional e-commerce, showed that the virtual experiential commerce environment has the highest investment rate and return rate, since it could shape the strongest perceived trust between the communicating actors (Chesney, 2017).

Online consumption and self-disclosure

The content on media not only informs users about the product or activity, but also supports relevant decisions after browsing and generates or reinforces users' behavioural intentions. For instance, displaying complex items such as jewellery and electronics in a multimedia environment using a higher media richness may increase consumers' positive attitudes and their willingness to buy (Lynch and Dan, 2000; Simon and Peppas, 2004); the fit among product complexity, consumer involvement and system interface media richness may enhance consumers' purchase intention (Walia, 2016). Besides, timeliness of feedback, multi-channel communication cues and personal attention may promote user loyalty by enhancing functional, social and self-expression values, respectively (Tseng et al., 2017).

Some pay attention to self-disclosure on social media. Blog content with high media richness can minimise the uncertainty and ambiguity of information and thus improve the efficiency of information dissemination. Therefore, the appropriate increase in media richness of content should be considered when writing a blog (Chang and Yang, 2013). As for stickers prevalent on social media nowadays, Lee and Lin (2019) found that social media users prefer stickers to text messages in communication because of their inherent needs for self-expression and showing off, mainly influenced by the uniqueness, variety and fun of image design. Liu et al. (2018) also found that stickers on social media have eased the difficulty of understanding users' messages while lowering users' decision threshold for sharing these funny things.

Individual differences in information and communications technology use

In the current trend of refined and tailored information services, the differences between groups in their information and communications technology choices and usage patterns are increasingly gaining attention. Their choices are not influenced solely by the degree of match between media wealth and task uncertainty, but also closely related to the distinct factors from the population itself, such as cultural backgrounds, demographic characteristics (e.g., sex and age) and job/skill differences. Zakaria (2017) found that cross-cultural teams turn to show switching patterns in their choice of medium, meaning that members' choice of communication mode changes accordingly in different decision-making processes and task situations. Media with low richness such as e-mail help filter out differential understanding from cultural backgrounds, thereby reducing the negative impact of cultural differences on communication and collaboration.

When it comes to sex, Savicki and Kelley (2000) found that the sex composition of communication groups makes sense, and women in female-only groups are more able to overcome the limitations imposed by the low richness of textual forms and achieve a higher sense of presence through self-disclosure, compared to male-only groups showing lower satisfaction with computer-mediated communication. Bryant et al. (2009) found that female members perceive more social laziness than male members in virtual teamwork, which subsequently reduces their use of media with high wealth. In addition, the impact of age differences on information needs has also received attention. Jung's (2017) interview results suggested that the low media-rich information abundant on social media does not meet older adults' need for uncertainty elimination and therefore discourages their use of social media and turns them to direct face-to-face communication which could provide sufficient nonverbal cues.

Future direction

Studies using media richness theory have accumulated in the mentioned aspects. For future research, there are several aspects worth noting. (1) Consideration of social and personal and technical factors is not yet well developed. A comprehensive and general model, more significant variables and new media contexts deserve to be taken into consideration. (2) Excessive emphasis on the classification of different media. Future researchers should always examine the core idea of the theory that the appropriate media richness and task uncertainty determine the success of the system, making use of the core value of the theory. (3) Limited literature concerning the mixed use of media with different richness. Future improvements could be made to the nesting and blending effects of different rich media, as well as choices and perceptions of users in this case.

Means-end chain theory

Theoretical development

Means-end chain theory explains the means and purpose of products or services through the personal perception of users. It suggests that users usually view the attributes of a product or service as means to their ultimate ends through the benefits generated by these attributes. Based on previous research on user value (Rokeach, 1968; Rokeach, 1973) as well as means and purposes of user behaviour (Howard, 1977; Vinson et al., 1977; Young and Feigin, 1975), Gutman (1982) first proposed the means-end chain theoretical model, which contains three levels of attributes, consequences and values. Attributes are the characteristics of the object or the activity in which people are involved; consequences can be defined as any result, either physiological or psychological, that accumulates directly or indirectly from the consumers' behaviour (sooner or later); based on attributes and consequences, value is defined here as desirable end-states of existence, such as pleasure, security, fulfilment and so on. The model attempts to explain the process of how product or service choices enhance the desired state of purpose through users’ personal perceptions, during which the value plays a dominant role in guiding choice patterns.

Building on Gutman et al., (1983) further subdivided attributes, consequences and values: concrete and abstract attributes of choice collections, functional and psychosocial consequences of such attributes and instrumental and terminal values which users want to achieve further.

Research agenda

Thanks to the development of information technology, the application boundaries of means-end chain theory have extended from early application mainly in marketing to a series of new information products and services nowadays. Most studies mainly focus on the conceptual design, attribute and quality of information products and services.

Attributes of information products and services

Attributes are the most fundamental part of means-end chain theory. Exploring the attributes could accumulate understanding of product features, which helps to analyse the key functions playing greater impact on users, thus enabling targeted enhancements of products and services. Sun et al. (2009) found that in the e-learning system, instruction presentation and student learning management are successful attributes. For online games, simulated reality and self-created game content may be the key aspects triggering learning results of players, determined by soft laddering approach and means-end chain model (Lin and Lin, 2017). However, studies investigating attributes pay little attention to the attributes of environments and contexts, which are emphasised in classic means-end chain theory (Gutman, 1982).

Conceptual design of information products and services

Means-end chain theory combines attributes and user values, helping to design product or service characteristics tailored to user needs, helpful for building conceptual design models. Wang and Chen (2011) used the ladder method to mine users' demand information and build hierarchical value maps through content analysis, building a multi-professional collaborative design model for enabling product concept design. Building on elements of a community and specific characteristics of the mobile channel, Prykop and Heitmann (2006) applied qualitative laddering interviews and means-end chains to design mobile brand communities according to perceived consumer value.

Quality of information products and services

Zeithaml et al. (1988) might be the first to introduce means-end chain theory into quality research of traditional products and services, whose theoretical model suggesting that the level of quality perceived is the result of customers' perceptions of the internal and external attributes, thus affecting their value evaluation and behaviour. Drawing on this, Parasuraman et al. (2005) built theoretical models of e-service quality, elaborating the antecedents of e-service quality in detail.

The improvement of the quality of information products and services may be the constant focus of enterprises conducting user analysis using means-end chain theory, which could identify key factors concerning quality by analysing or comparing users, products and service quality in comprehensive ways. For instance, calculating the utility score of each attribute level of the written content on tourism web sites could uncover the customers' preferred attribute level portfolio and their preferred tour package (Lin and Liao, 2010); different hierarchical value map diagrams built from advertising messages of credit card websites in different cultural context suggested that the effectiveness of online advertising should be distinct in different countries (Fu and Wu, 2010).

Future direction

For theoretical application, there are several aspects deserving attention in future research. (1) The diversity of research methods. Some studies only use laddering methods and hierarchical value maps, likely to be influenced by subjectivity. Future researchers are suggested to combine quantitative methods such as data mining to uncover hidden correlations or use grounded theory methodology to further increase the depth of interviews. (2) The improvement of integrated research or theoretical frameworks. Most empirical studies focus on the application of means-end chain theory in certain contexts, while there are limited outcomes about comprehensive frameworks. Theoretical issues such as effect measurement are supposed to be taken into consideration further. (3) Enrichment and development of theoretical connotation in the era of new information products and services. There are few breakthroughs in the application of means-end chain theory, although with some extensions of attributes, consequences and values. Information products and services differ significantly from traditional ones in the new technology era, hence there may be more complex relationships between the attributes and values of users.

For application contexts, current research engages more on e-services such as e-commerce, games and online education, while future research can explore artificial intelligence and virtual reality, which focus on cognitive processes of users and are in line with the internal logic of means-end chain theory. In addition, other research priorities and frontier issues in management information system can be investigated, such as design science in human-computer interaction (Vanden Abeele et al., 2012.

Using momentary compensation and price incentives to motivate individuals may exert unexpected detrimental effects (e.g., Deci and Ryan, 1985; Titmuss, 1970). To explain it, Frey and Oberholzer-Gee (1997) advocated the motivation crowding theory. It tends to expound the relationship between extrinsic and intrinsic motivations, with the central idea being that, depending on the perceived controllability of extrinsic motivation, individuals' intrinsic motivation to perform a task would be either weakened or strengthened by their extrinsic motivation, and are likely to be destroyed especially when price incentives are introduced.

Building on self-determination theory (Deci and Ryan, 1985) explaining extrinsic and intrinsic motivations and cognitive evaluation theory illustrating the impacts of extrinsic motivation on intrinsic ones (Deci and Ryan, 2000), Frey and Jegen (2001) found two psychological processes about how external interventions influence intrinsic motivations: impaired self-determination and impaired self-esteem. When individuals perceive external intervention to reduce their self-determination or debase their involvement and competence, they reduce their intrinsic motivation and effort level. They further developed two psychological conditions based on these processes: if the affected individuals perceive themselves as being in control, the external interventions crowd-out (diminish) intrinsic motivations, with both self-determination and self-esteem suffering; conversely, if individuals perceive the external intervention as supportive, it crowd-ins (strengthens) intrinsic motivation, with self-esteem fostered and self-determination enlarged.

Though contradictory to the commonly held opinion that external interventions such as monetary compensation may have a positive motivational effect, the existence of crowding-out effect has been confirmed by several meta-analysis studies on related topics (Deci et al., 1999; Tang and Hall, 1995; Wiersma, 1992). Motivation crowding theory has been widely applied in various fields, such as public sector management (Georgellis et al., 2011), environmental protection (Huang et al., 2014), organizational management (Gneezy and Rustichini, 2000) and information systems (Tong et al., 2007).

Research agenda

User adoption and engagement

The call on user engagement has led community managers to administer explicit incentives to increase members’ adoption and involvement. Most communities, however, still suffer from weak contributions caused by crowding-out effects. In new online communities with limited size, both direct and indirect financial incentives help to promote user adoption without the crowding-out effect, but direct rewards have little influence on increasing usage (Becker et al., 2010). In large public online communities, monetary incentives may help to increase the active adoption and willingness of members to participate in the short term, but can reduce their participation motivation in the long run, which should be taken as the hidden costs of rewards (Garnefeld, 2012).

The dilemma in online project participation and online survey response could be explained by motivation crowding theory. In online open-source projects, developers' motivations are linked in a complex way, where the external motivation of gaining status and reputation (Nov et al., 2014) enhances the internal motivation. Individuals feel more autonomous when engaged in open-source projects than in private ones, with their autonomy and self-esteem not impaired (Osterloh and Rota, 2007). Unlike open-source projects, online surveys are often associated with more direct extrinsic rewards, such as donation incentives and small compensation, which are perceived by respondents as a control tactic (Pedersen and Nielsen, 2016) and uninteresting signals (Sauermann and Roach, 2013), thus crowding out their intrinsic motivations and leading to lower survey response rates.

Online content contribution

For knowledge contribution in communities, a Web-based questionnaire (Jeppesen and Frederiksen, 2006) showed that the innovative contribution of members is associated with the desire of getting recognition from their organizations, but external factors may crowd-out this kind of motivation. Furthermore, the intrinsic motivations of members to exchange knowledge and trust are likely to be undermined by rewards, with the pursuit of rewards becoming the priority for many in the community rather than valuable knowledge contribution (Fahey, 2007). This may be due to a mismatch between the external reward and the specific behaviour that needs to be motivated (Bock et al., 2005).

Kuang et al. (2019) found the initial financial incentives on knowledge sharing platform increase their social engagement, but they spill over to users’ desirable non-incentivised online engagement behaviour, such as knowledge seeking. Similarly, for social network users, financial incentives may dilute the impact of intrinsic incentives through a crowding-out effect, shifting their motives for sharing e-commerce content from enjoyment to discounts (Vilnai-Yavetz and Levina, 2018).

Online word-of-mouth communication

In more extensive online contexts, such as social media, word-of-mouth communication is also likely to be influenced by crowding-out effects. Rehnen et al. (2017) found that offered rewards may undermine customers' autonomy and thus negatively affect their intention to engage in word-of-mouth communication on social media. There exists a contradiction between self-reporting, which suggests the dominance of internal motivation, and experimental manipulation, which shows that external incentives (i.e., monetary compensation) trigger a greater willingness to share (Vilnai-Yavetz et al., 2018). Users may initially be internally motivated to share e-commerce experience on social networks, but the introduction of monetary incentives weakens their intrinsic motivation, switching them gradually to external motivation. If a reviewer becomes driven mostly by status recognition and reciprocal obligation, their initial intrinsic enjoyment may suffer a crowding-out effect (Wu, 2019).

Future direction

The crowding-out effect has been observed and confirmed in various contexts, such as online adoption and engagement, content contribution and word-of-mouth communication. Different scholars, however, still disagree on whether motivational crowding-out effects exist in certain contexts and whether external intervention has a crowding-out or crowding-in effect on intrinsic motivation (Becker et al., 2010; Garnefeld, 2012; Rehnen et al., 2017). There are no clear and specific boundary conditions for the emergence of crowding out and crowding in, combined with types of online platforms, cultural backgrounds and the ways of collecting and measuring data.

In response to such a dilemma, the following improvements are encouraged in future research. (1) Conduct meta-analysis with divergent findings, taking significant moderating variables into consideration. This may help to explore reasons for ambivalent outcomes and conditions under which crowding-out effects occur, thus finding general patterns with higher external validity when applying the theory; (2) Use mixed-methods to collect data and detailed information. The process of how crowding-out or crowding-in affects change under the influence of incentives could be observed by longitudinal field study, context experiments and in-depth interviews, thus increasing the explanatory power of the results. (3) Subdivide types of behaviour and extend application contexts. For instance, motivation crowding of contributing different types of knowledge may vary, and possible distinctions between positive and negative word-of-mouth might be an interesting topic. Research contexts other than the above agenda need to be further developed.

Collective action theory

Theoretical development

Collective behaviour in real-life situations often does not match the idealised state of common pursuit of collective interests described by scholars of pluralist theory (Truman, 1981). It is common to see conflicts between collective and individual interests, such as the free-rider problem and other collective action dilemmas. To explain these phenomena, American economist and sociologist Olson (1965), building on the rational person assumption, individualistic methodology and public goods theory, formally introduced the collective action theory in his book The logic of collective action.

The core idea is, if the collective is large enough and there is no internal coercion or incentive to motivate individuals to act for collective benefits, rational social actors will give priority to their own interests, instead of taking collective interests as a main mission. Rational members tend to enjoy increased collective benefit without cost, meanwhile lacking motivation to fight for it. To avoid conflict, Olson advocated a method of giving selective incentive to individual members.

Since its inception, collective action theory has been developed and refined by many scholars. Oliver (1993, pp. 271) claimed that there are many types of collective action that cannot all be captured with the same model, and models of dynamic interactions require further development. Medina (2013, pp. 279) reviewed the evolution of collective action models and argued that Olson's model, a single-equilibrium public-goods game, is a special case. When multiple equilibrium game models are used, group members are likely to adopt cooperative strategies to achieve collective interests jointly. With collective action becoming more dynamic and iterative, actors tend to change initial self-interested choices and work more cooperatively (Axelrod and Hamilton, 1988). In addition to the continuous process of collective action, choice behaviour of individuals could also be influenced by their bounded rationality (Finkel et al., 1989) and social selective incentive (Sandell and Charlotta, 1998).

Research on the relationship between types of public goods and collective action continues to intensify. The public goods studied by Olson are limited to zero-summer public goods, which may lead to collective action dilemmas. Other kinds of public goods, however, such as pure public goods (Chamberlin, 1974) and non-zero-summer ones (Marwell and Oliver, 1993), are likely to provide individuals with more benefits as the number of group actors increases. Besides, countermeasures for collective dilemmas continue to improve, with more priority given to trigger intrinsic motivation. Fireman and Gamson (1977) criticised Olson for mainly using external rewards and financial punishments and ignoring the influence of internal selective incentives such as loyalty, morality and friendship in group members. Olsson (2009) proposed the idea of relational rewards, which motivates members to achieve collective goals by enhancing communication and sense of community among members.

Research agenda

Antecedents of knowledge sharing in online community

Knowledge in online communities could be viewed as public goods in nature (non-competitive and non-exclusive), online members, in some cases, are prone to be free-riders without efforts. Studies based on collective action theory have investigated the role of intrinsic motivation in solving knowledge sharing dilemmas in online communities. For instance, Wasko et al. (2009) found that reputation-enhancing mechanisms can motivate members who want to enhance their personal reputation to be more proactive in creating and sharing public knowledge. Similarly, Spaeth et al. (2008) proved that members who made outstanding contributions during the development of open-source projects would gain a good reputation, which in turn increases their motivation to further participate. Presenting appreciation for contributors on websites and displaying the ranking of each member's contributions would motivate individuals to provide better solutions to questions posted on the webpage (Cheshire and Antin, 2008).

Collective information and communications technology actions within groups

Information and communications technology are considered as an effective means to solve collective action dilemmas. Lupia and Sin (2003) pointed out that their use reduces the cost of information exchange and management within organizations, allowing for higher noticeability among members and thus reducing the likelihood of collective action dilemmas. Such technologies also make it easier for members to communicate with each other and therefore strengthen their social ties, blurring the boundaries between collective and private interests to a certain extent (Bimber et al., 2005). The improved social networks could facilitate collaborative problem solving among members in the same organization (Kang et al., 2010).

For real communities, the use of the Internet helps to reduce the cost of communication among community members, enables them to quickly learn about the community, thus solving public problems in the community by enhancing the community's ability to mobilise its members (Kavanaugh et al., 2005), such as creating online discussion groups for community residents to consider issues such as road paving and greenery coverage (Hampton and Wellman, 2003).

Inter-organizational information systems development

Collective action theory could explain factors diminishing cross-organizational efforts to accomplish common goals. Weiss and Cargill (1992) found that any company, since the spin-off of AT&T, is free to use information system standards regardless of contribution. This has resulted in a lack of motivation for some to devote resources and efforts to information systems standards development and reduce inter-organizational cooperation. The excessive number of organizations engaged in standard development would also make it difficult for the organizational association to effectively identify contributions of members (Foray, 1994). This kind of free-riding behaviour could be overcome by forming clubs with the same interests and limiting members.

Distinct interests between organizations also impact their cooperation in achieving collective benefits. For instance, the different benefits of user organizations and IT vendors slow down the process of cooperation in industry vertical information systems standardisation projects (Markus, 2006). This is a dilemma not only in the standard development stage and standard promotion stage, but also in the stage of initiating cooperation among organizations, during which members appear to care for their own interest rather than overall benefits (Klein and Schellhammer, 2011). In addition, the different strategic goals and the degree of interdependence among organizations can influence them to conduct collective actions (de Reuver et al., 2015).

Future direction

The collective action theory focuses on investigating negative factors affecting collective efforts. However, the performance of members in dynamic and repetitive collective action has rarely been examined, which needs systematic measures to facilitate joint action and attention to the impacts of punitive measures on members' behaviour. Besides, findings obtained mainly from subjective reporting methods such as observations and interviews may be subjective to some degree.

To remedy these possible shortcomings, the following four issues should be noted in future research. (1) Investigate factors causing collective action dilemmas in a more dynamic and repetitive collective action process (e.g., in the development process of upgrading and iterating information systems). (2) Analyse the internal logic of factors influencing collective action and develop common measures to promote group activity accordingly. (3) Explore whether punitive measures can make members participate in collective action, such as limiting the number of times that non-sharers get access to knowledge and information. (4) Combine qualitative and quantitative methods (e.g., designing contextual experiments and quantifying relevant indicators of members' willingness to participate in collective action) to gain results with more practical significance. On the other hand, future research is also supposed to expand the application contexts such as network public opinion (Ward and Ostrom, 2006), knowledge sharing among corporate alliances and consumer reviews in e-commerce websites.

Service encounter theory

Theoretical development

In 1985, Shostack formally introduced service encounter theory in his study of service quality management in enterprises. The core idea is that the service contact between customers and the service system is the key factor that affects customer's perceived service quality.

Considering the important role of service contact for service enterprise management and service marketing, scholars have been paying much attention to the development and breakthrough of theory itself. Research on the meaning of service contact has deepened. Initial theory focused on the binary interactions between customers and service providers in every moment of truth (Solomon et al., 1985). Shostack (1985) proposed a definition of multiple service contact, suggesting that service contact should also include contact between customers and service facilities with other tangible objects. Further with the development of information and communications technology, a broader theory began to incorporate technological factors based on multiple interactions and commonly adopted all kinds of touch-points, such as technical, physical, interpersonal and invisible touch-points (Froehle and Roth, 2004; Massad et al., 2006).

Models concerning service contact are improving. Grove et al. (1992) developed a service performance club consisting of actors, audience and setting, using a drama metaphor to explore the services marketing mix. Bateson (1985), considering the role of organization in the process of service contact, proposed a triadic model including service organization, service personnel and customers. The two models are both based on interpersonal interaction, without taking technical factors and service environments into consideration. Building on previous outcomes, Gronroos (1990) proposed a more detailed service contact system model, consisting of three phases: service operation systems (driving factors, including the corporate mission, service concept and core technology on the strategic level), service delivery systems (key and direct factors affecting customer satisfaction, consisting of internal and external facilities, service personnel and equipment) and other touch-points (indirect factors such as advertisements and word-of-mouth).

With the enrichment of theoretical content and models, contexts of service contact theory have expanded from service marketing to service design. It provides new ideas for service designers to explore customer needs from touch-point perspective. Some scholars have developed critical incident techniques to identify touch-points (Clatworthy, 2011), service blueprints (Shostack, 1984) and customer journey maps (Stein and Ramaseshan, 2016) to visualise touch-points.

Research agenda

E-service satisfaction

The development of information and communications technology gives rise to new ways of service contact, and service contact based on interpersonal interaction is gradually replaced by electronic service contact realised by advanced Web technologies. Therefore, many scholars have begun to study the role played by technology in intervening in service contact. Verhagen et al. (2014) focused on virtual customer service agents and designed an experiment on changing mobile service plansResults showed that friendliness and professional knowledge of virtual customer service agents could stimulate a sense of social presence and personalisation, indirectly and significantly influencing customer satisfaction during e-service process. Inbar et al. (2012, pp. 245) also argued that increasing customer interaction with electronic displays during service can increase information sharing and customers’ sense of control over system, which in turn leads to customer satisfaction through the mediating effect of trust and perceived system effectiveness. Self-service technology exposure has a positive impact on customer satisfaction (Beatson, 2007).

Lin’s (2007) fuzzy neural networks showed that interpersonal service contact has a greater influence on customer satisfaction than technical service contact, especially in information technology and tourism industries, with both contacts indirectly affecting customer satisfaction by improving perceived service quality of customers. The critical incident technique (Massad et al. 2006) also found that interpersonal service contacts (including the characteristics and behaviour of service personnel) are primary factors directly affecting customer satisfaction in e-retail service. Park et al. (2012) explored the direct impact of service touch-points before and after online shopping on customer satisfaction based on a processual perspective of service contacts, and confirmed that the direct impact of post-purchase service touch points on customer satisfaction is greater.

Customer retention

Service contact theory indicates that the touch-point is the key element affecting customer experience, and the effect of experience will be directly expressed through customers’ behavioural willingness in service process. Ieva et al. (2018) found a significant positive relationship between the frequency of customer contact with traditional retail touch-points (e.g., physical stores, brochures, promotions), mobile touch-points (e.g., mobile apps, mobile messaging), social media touch-points (e.g., online advertising, online word-of-mouth) and customer loyalty intentions (measured in relationship commitment, self-representation and positive word-of-mouth). Considering time dimensions, Scherer et al. (2015) conducted a survival analysis using panel data on roadside assistance services, showing that the frequency of contact with online self-service and telephone customer service dynamically affects customer retention in a U-shaped manner, which indicates that technology service contact has a significant long-term effect on customer retention.

Future direction

For future research, the application of service contact theory requires attention in the following aspects. (1) Contexts of most studies are limited to retail e-commerce. Expanding service contact theory to new areas and contexts is needed, such as e-government, online education and even virtual reality environment. Besides, future research, based on service failure and service remediation perspectives, can also provide reverse insight into touch-points that lead to e-service failures and user complaint behaviour. (2) Research on the relationship between service contact and customer satisfaction/loyalty has not yet reached a consensus. Future scholars could explore more, investigate reasons for differences, as well as the interaction effects between service touch-points and moderating factors such as contact frequency and customer's emotional response. (3) Most studies use cross-sectional data. Future concern could be perceptions of users collected from data mining and dynamic effects of touch-points from time dimension.

Discussion

In this paper, we conducted a review of six theories used in information behaviour research, including ethnographic decision tree theory, means-end chain theory, media richness theory, collective action theory, service encounter theory and motivation crowding theory. After introducing their core principles and development history, we identified several research agenda based on existing findings as well as proposing potential directions for future research respectively (summarised in Table 2).

Conclusion

Empirical studies in information behavioural research borrowed some theories from other disciplines, such as psychology, sociology and philosophy (Webster and Watson, 2002). The fragmentation of fundamental theory, however, has increasingly affected the development of information behavioural research. Hence, it is crucial to form a clear picture of concepts or theories and promote future research (Senyo et al., 2019). It is unrealistic to review all theories used, but we try hard to review six theories relatively useful but less summarised in information behavioural research.

We made several contributions. Firstly, we illustrated the general development of each theory, such as how its concepts and principles evolve, how its models advance and in what contexts its application has been extended. Detailed introductions about theories are expected to help researchers understand the connotation and extension more clearly, encouraging rigour and integrity in empirical research. Secondly, the research contexts, research questions, and research objects to which theories can be applied have been demonstrated, beneficial for researchers to grasp the application boundaries and find appropriate theoretical support for their own information behaviour research, thus shifting from ambiguous description to prediction and explanation (Cole, 2013). Finally, we proposed possible directions, in both theoretical and methodological aspects. Information behaviour researchers are encouraged to utilise various methods and extend application contexts to investigate some unsolved issues and uncharted territories.

Acknowledgements

We would like to thank the "Focus Project" Program of School of Information Management at Nanjing University.

About the author

Xinyue Yang is a PhD student at the School of Information Management at Nanjing University, Nanjing, Jiangsu Province, P.R. China. She can be contacted at ruby22yang@outlook.com.
Qinjian Yuan is a Professor at the School of Information Management at Nanjing University, Nanjing, Jiangsu Province, P.R. China. He can be contacted at yuanqj@nju.edu.cn.

How to cite this paper