A University of Toronto Ph.D. thesis. Copyright 1995 by Alejandra Rojo

Chapter 2: LITERATURE REVIEW

In this chapter I present the study theoretical framework. First, I review models that explain media use and adoption. Second, I review participation in collaborative media as a multilevel phenomenon. Third, I review the approaches shaping the framework, public goods- based theories and the users' perspective approach.

Models to explain media use and adoption

Four recent approaches to explain use and adoption of media in general were reviewed: 1) the social influence approach (Fulk, Schmitz and Steinfield, 1990); 2) the reciprocal influence approach (Contractor and Eisenberg, 1990); 3) the public goods-based approaches (Markus, 1990; Connolly and Thorn, 1990) and; 4) the users' perspective approach (Dervin, 1989). In the 'social influence', the 'reciprocal influence' and the 'user's perspective' approaches there is an effort to overcome rationalistic, conduit, and objectivity-oriented models of communication. The social influence and reciprocal influence models characterize the communication processes as embedded in social process. The public goods-based approaches (Critical mass and Discretionary database theories) contribute to conceptualizing the phenomena in ways that address adequately collective phenomena; its explanations are based in a special incentive structure called "dilemma of public goods". In the user's perspective model there is an intent for looking at communication processes as influenced by users' perceptions and needs.

2.1.1 The social influence model

Attitudes and behaviors of others in the shared environment are influencial in how media are perceived and used; media use behavior is shaped by others' overt statements about media and, by vicarious learning. Social norms contain cues regarding appropriate use of media. Media behavior sense-making is also shaped by the social environment (Fulk et al., 1990). They criticize traditional media use research by pointing to some faulty assumptions: a) each medium has fixed characteristics and these objective characteristics are salient to users; b) individuals choose media through a rational cognitive matching process: assessing the requirements of the task at hand and selecting a medium with communication capabilities that match these requirements; c) individuals make independent choices, the interpersonal context does not play any role in the decision process; d) behavior is efficiently motivated. At difference with traditional media use research, Fulk's et al., (1990) central premise is that social context plays an important role in media-use behavior. They affirm that media perceptions are in part "subjectively and socially constructed" (Fulk et al, 1990; p.121). In this model, social context has been operationalized as processes of social influence.

2.1.2 The reciprocal influence model

Contractor and Eisenberg's (1990) recursive model of reciprocal influences between media use and social structure, highlights the interplay among interaction patterns and media practices by operationalizing social context in terms of communication network theory . They specify how individuals' perception and behaviors with new media are shaped, and how in turn, their patterns of media use affect social network participation and network structure. For example, 'key communicators' are important shapers of new media perceptions. Having access to a broad range of contacts, they are exposed to novel information about new media and being prominent in a given communication network, they are well-positioned to introduce these ideas in it. In turn, access to media influences the prominence of individuals in a network and can influence the range and diversity of an individual's communication network. Media use can change the structure of networks by changing existing boundaries.

The approaches selected as the theoretical framework for this study, public goods-based theories and the users' perspective approach are examined in the next section. Public goods-based approaches were selected because they provide a conceptualization that allows the capture of essential characteristics of the participation phenomenon in collaborative mass media, and the users' perspective approach was selected because it provides an advantageous complementary entry point into participation.

Participation in SEFs as a multilevel phenomenon

I decided to give prominence to a group or community level of analysis, because it is more adequate to examine a phenomenon that distinguishes interactive media in general and collaborative mass media in particular--the reciprocal interdependence among users. Reciprocal interdependence means that individual users of interactive media cannot achieve the potential benefits offered by the medium on their own. Markus (1990) has discussed the issue of reciprocal interdependence among users in relation to the adoption process. Prospective adopters of interactive media need to know that they will have sufficient and relevant communication partners; if these conditions are absent there is a risk that medium use will not only fail to spread but also will become extinguished. Markus (1990) explains that there is a difference between types of innovation in terms of influence among early and late adopters. For most innovations later adopters are influenced by early adopters but not vice versa. For example, someone considering starting to use a computer for word processing will be influenced by peers who have already adopted word processing but does not need to worry about the number of future adopters of word processing. In interactive media influence goes both ways: early adopters can be influenced by later adopters and vice versa. If early adopters do not have their messages reciprocated or if only a few follow them in adopting the medium, they can leave, thus initiating a process of use extinction. On the other hand, if early users' communications are reciprocated and new people adopt the medium, early users are stimulated to stay and to increase their use, thus generating benefits and attracting new users and a cycle of increased adoption is initiated.

In my view, reciprocal interdependence among users has another referent besides the adoption process and that is the on-line interaction process. Some researchers trying to explain participation in on-line forums posit that it is on-line interaction itself that draws people in to contribute messages. Each message has the potential of accomplishing two goals: to communicate content and to stimulate the responsiveness of interlocutors (Feenberg, 1987; Rafaeli, 1988; Rafaeli and LaRose, 1991). A more interactive on-line communication setting is likely to increase participation (Rafaeli, 1988; Rafaeli and LaRose, 1991; Rafaeli and LaRose, 1993). I selected two approaches based in public goods theory--Critical mass theory (Markus, 1990) and Discretionary database theory (Connolly and Thorn, 1990)--to inform this study's framework because they model these reciprocal interdependence processes.

On the other hand, it seemed important not to neglect the experiences of SEFs' individual users in order to capture the specificity of SEFs. I also considered it necessary to uncover SEFs members' purposes in using the medium and their patterns of usage in order to better understand SEFs participation dynamics. A theoretical approach focusing on users' views in conceptual terms as well as in methodological terms was selected. Dervin's approach (1989) brings into focus media use as a socially situated process and looks at the process from the user's viewpoint.

Next, I will examine the two public goods-based approaches and the users' perspective approach that inform this study.

2.3 Theoretical approaches

Participation in collaborative mass media as a public goods dilemma

The dilemma applied to interactive media is this: the content of the medium (and the medium itself) is available to all members of the community regardless of whether or not individuals contribute to achieve it (public radio or television are good examples for broadcasting media; bulletin boards; SEFs are good examples for interactive media). Thus, the incentive to contribute to the medium content (and by doing so increasing the likelihood of getting the whole community participating in the medium) is minimal and a 'rational' user would tend to withhold contribution and enjoy the benefit of other members' contributions without contributing him/herself. In other words, the incentive structure of interactive media stimulates 'free rides'. However, if users behave in this 'rational' way--withholding contribution--nobody will contribute and therefore nobody will get any content (and participation of the whole community in the medium will not be achieved). Both Critical mass theory and Discretionary database theory developed overlapping versions of how this particular incentive structure comes into existence for interactive media. These theories seem especially useful for examining contribution in collaborative mass media such as SEFs because the audience is the creator and the recipient of the content (Rafaeli and LaRose, 1993).

Critical mass theory

Three main concepts in this theory are: collective behavior, public goods and critical mass. Collective action is the individuals' efforts in the service of public goods. Public goods are "benefits that individuals cannot be prevented from enjoying, whether or not they have helped to secure them" (Markus, 1990; p.201). Critical mass is "a small segment of the population that chooses to make big contributions to the collective action while the majority do little or nothing" (Oliver et al., 1985, p. 524, cited by Markus, 1990; p.201).

Critical mass theory was conceptualized in relation to interactive media such as telephone and electronic mail. In these cases users' contributions correspond to efforts in reciprocating communication. New interactive media such as bulletin boards, computer conferencing, SEFs, depend on users' contribution of information to exist; there is no medium without participants' contributions. Therefore, contribution in these cases corresponds to efforts in reciprocating information.

Markus (1990) posits that the key factors determining the likelihood of an interactive medium achieving universal access when introduced into a community of interest are: 1) costs incurred by users--the extent of usage knowledge and skills requirements, communication discipline, and individual access costs, and 2) variation in the ability of individuals to obtain benefits from participating in the medium (heterogeneity of interests) and the variation in terms of the resources individuals can contribute (heterogeneity of resources).

The higher the contribution costs the lower the level of contribution to the medium. The costs that affect individual contribution are those costs borne by the users. In the case of SEFs there are three types of costs borne by the user: a) knowledge and skill requirements--general computer skills, specific communication programs knowledge; b) communication discipline--readiness to reciprocate communication and c) access costs--access to a computer connected to BITNET/Internet.

Participants' heterogeneity increases the possibility of adoption and contribution because the differential ability to derive benefits or contribute resources increases the likelihood that there will be some people willing to contribute more than others. The existence of a group of highly interested and resourceful people willing to contribute even if others do not--critical mass--can get contributions started and going. Successive contributions increase the possibility of more participants getting benefits and more participants contributing and getting to use the system. Benefits of participating increase with the number of users thus getting higher contribution rates and higher rates of new participants in larger groups is likely (see Fig. 1).

CRITICAL MASS THEORY

REGULAR CONTRIBUTORS

(Critical mass)

COSTS

ADOPTION RATE

SIZE

> BENEFITS

CONTRIBUTION RATE

HETEROGENEITY

CONTENT > DIVERSITY

HETEROGENEITY OF BENEFITS

Markus (1990) pointing out that the unit of analysis is the community, described the difficulties of defining the boundaries of the community of interest (many communities consist of nested subunits) but the context for their problem is the organization, a closed system. For open systems as public collaborative mass media, the problem is that they have self-defining communities of interest. Rafaeli and LaRose (1993) define the community of interest for electronic bulletin boards as "people with personal computers and modems who access bulletin board systems and who are interested in the topics the board specializes in" (Rafaeli and LaRose, 1993; p.294). I followed a similar solution for community of interests in SEFs.

Discretionary database theory

Connolly and Thorn (1990) conceptualized a communication structure called discretionary database that can be used to represent collaborative mass media such as electronic bulletin boards and on- line forums. This communication structure has the following features: 1) a shared pool of information to which members are free to contribute or not; 2) contributing to the pool involves costs and being able to access the available information involves benefits; 3) information is freely available to all members (Connolly and Thorn, 1990). This configuration results in a lack of incentives to contribute because the cost of contributing the information falls on one person while the benefits spread to everyone except the provider whose "benefits arise only from the contribution of others, with no direct link to his own decisions" (Connolly and Thorn, 1990, p.225). Thus, Connolly and Thorn (1990) predict that discretionary information in this kind of media will be generally undersupplied. They posit that the higher the costs the lower is the incentive to contribute. Next, they posit that asymmetry among participants, in terms of cost, benefits and resources (members' heterogeneity), diminishes the rate of contribution because the hope for reciprocity in the exchange is decreased (e.g., exchange between someone possessing "good" information and someone possessing "bad" information). The hope for reciprocity is also decreased in large systems because the possible link between contributor and beneficiary is weakened thus getting lower contribution rates in larger groups is likely (see Fig. 2).

 
 
Fig. 2 
DISCRETIONARY DATABASE THEORY 
 
   COSTS 
     
    HETEROGENEITY 
 
   CONTRIBUTION 
 
      RATE 
 
  HOPE FOR 
 
RECIPROCITY 
 
   ASSIMMETRY 
 
    SIZE 
 
HETEROGENEITY 
OF BENEFITS 

2.3.1.3 Applying public goods-based theories

Thorn and Conolly (1990) tested their hypotheses in the laboratory with a number of small groups (4-8 students in each group). The participants' task consisted in playing the role of production manager in charge of the agricultural output of an imaginary country. They could contribute information of their estimation of the demand in their own country to a database available to other players, incurring a charge for doing so. Then they had access to the content of the database containing information about the demand of each nation that had contributed market estimates. Their predictions were confirmed: the higher the group's contribution costs the lower the rate of contribution; asymmetry in resources and benefits reduced the contribution rate. However, the group size prediction was not confirmed. The researchers explain this absence as a result of the test being carried out in very small groups.

Field study

Rafaeli and LaRose's (1991, 1993) study on electronic bulletin boards was the first to apply the public goods-based theories in the field. To integrate the two theories they utilized the concept of media success, connoting with it adoption, contribution, usage and longevity. These researchers derived four hypotheses from public goods theories:

Hy 1: "Bulletin boards success will be negatively related to access restrictions placed on users" (Rafaeli and LaRose, 1993; p. 382).

Hy 2: "Bulletin boards success will be positively related to the diversity of content available" (Rafaeli and LaRose, 1993; p. 382).

Hy 3: "Group size will be negatively related to contribution levels but positively related to other measures of bulletin board success" (Rafaeli and LaRose, 1993, p.383).

Hy 4: "Symmetry in contribution levels will be positively related to bulletin board success" (Rafaeli and LaRose, 1993; p. 383).

Rafaeli and LaRose (1993) operationalized dependent variables in this way: contribution level as the ratio of files contributed or uploaded in a week to the total number of weekly file transactions; adoption rate as the ratio of regular callers (who call at least once a week) to the total number of users; longevity as the time that the board was in existence; usage as the average number of calls on an average day. Independent variables were operationalized like this: ratio restrictions as a dichotomous variable, scored 1 if the board operator had upload-download restrictions, 0 if not; content diversity as the percentage of the board content that is not computer related; group size as the total number of users; symmetry as the percentage of users characterized by the board operator as being "givers" or "exchangers" (sharing resources or participating in a fair exchange) as opposed to "takers" (those looking for something for nothing). It is important to notice that all of these measures were not direct measures but the board operators' estimates about the issues. Data were collected through a questionnaire uploaded to the system operator in the sample.

Results partially supported two of the hypotheses. In regard to hypothesis 2, Rafaeli and LaRose found that contribution levels were significantly related to diversity of content; the relationship between diversity of content and longevity was in the predicted positive direction but was non significant. A negative (non significant) relationship was found between diversity of content and adoption and usage. In regard to hypothesis 4, they found that symmetry in the exchange was positively related to the four measures of bulletin board success but the relationship was statistically significant only for contribution level and adoption rate. Rafaeli and LaRose (1993) conclude that public goods theory better predicts contribution levels than the other measures of bulletin board success and suggest that the operationalization of the adoption rate could have to do with that fact.

2.3.1.4 Summary

Both Critical Mass theory (Markus, 1990) and Discretionary Database theory (Connolly and Thorn, 1990) are based on the assumption that participation in collaborative mass media entails costs and benefits and the conception that the information in the system is a public good--all participants have free access to any information contributed by others. Both theories predict that cost of contributing is negatively related to contribution. However, they differ in their perspectives on the influence of heterogeneity of participants and forum size on contribution. For the Critical Mass theory (Markus, 1990), heterogeneity (variability in benefits obtained and resources to contribute) increases the possibility of adoption and contribution. Because benefits of participating increase with the number of users, higher contribution rates in larger groups are likely. For the Discretionary Data Base theory (Connolly and Thorn, 1990), heterogeneity (variability in benefits obtained and asymmetry in the value of information to contribute) diminishes the rate of contribution because the hope for reciprocity in the exchange is decreased. The hope for reciprocity is also decreased in large systems because the possible link between contributor and beneficiary is weakened; thus, lower contribution rates in larger groups are likely.

There is a need to elaborate in detail some equivalencies and differences in the overlapping of the two public goods-based theories. For example, a similar concept in both theories but more elaborated in Discretionary database theory than in Critical mass theory, is the heterogeneity of resources concept. In the Discretionary database theory the variability of resources refers to asymmetries in the information quality. In the Critical mass theory there is no mention of any specification of the dimension that varies in the resources to be supplied. For this study I devised two variables articulating these concepts: 1) heterogeneity of resources (content diversity), operationalized as the number of users' research (focus of) interests in a forum and 2) asymmetry of resources (asymmetry in the quality of information possessed) operationalized as users' perceptions of the quality of information supplied in the forum in terms of the comparison between their own knowledge and others' knowledge (other members have higher, the same or less knowledge than me with respect to the topics discussed in the forum).

Connolly and Thorn's (1990) laboratory study found positive evidence for all their predictions but the group size. Rafaeli and LaRose's (1993) field study on electronic bulletin boards found positive evidence for the relationship between content diversity (heterogeneity of resources) and contribution and for the relationship between symmetry (critical mass) and contribution.

The users' perspective

Dervin (1989) developed an approach that allows research to move from questions arising from observers' perspectives that leaves out the experiential world of users to questions arising from actors' perspectives. Dervin started with the assumption that the uses a person makes of an information or communication system arise from the intention of making meaning, bridging gaps. Then, she developed categories that allow one to investigate "universals" of human experience in relation to communication. These alternative categories involve entering the world of users from the inside, from the actors' perspective. The categories are valid for all users in all situations. The categories are:

a) the actor's situation--category designed to understand what in a situation induces a person to use a communication medium;

b) gaps in sense making--category designed to uncover gaps that the communicator is attempting to bridge;

c) actor-defined purposes--category that deals with the actor's expected outcomes from using the system;

d) information-using strategy--category referring to strategies for seeking and using information;

e) information values--category designed to describe the user's criteria for evaluating information;

f) information traits--category referring to the information characteristics that would match the user's needs.

These open categories are a tool allowing an exploration of people's day-to-day usage from their own perspective: what they use the medium for, how they interact with the system. They seem particularly appropriate for investigating the adoption process.

2.3.3 Summary

| Ph. D. Thesis Index |