The International Professional Publishers
Newbury Park London New Delhi
i f ~ ·
CHARLES W. STEIN FIELD
6. A Social Influence Model of
The explosion of new communication technologies has generated widespread
controversy over their potential effects on the workplace. Accurate claims of
effects must be rooted in valid assumptions about just how the technologies
are used. Consequently, media-use behavior has resurfaced as a vibrant area
of inquiry. In this chapter, we take a close look at current media-use theories.
We argue that they rest on some very unrealistic assumptions about human
behavior in organizations. In particular, we take issue with the rationalist
bias and technological determinism that pervades these approaches. Cur-
rent media-use theories fail to recognize a central premise of current organi-
zation theory: Behavior occurs in a very social world which is far from
neutral in its effects.
Once we assume an active and influential social context of media use, we
then face the problem of specifying the social processes more precisely. An
appropriate starting point is to examine existing theories of social relations
in organizations with an eye to how their premises apply to media-use
behavior. This chapter presents a model of media use that is derived from
such an examination.
We begin by summarizing traditional models of media use. Then, we show
how these theories are incapable of adequately explaining a whole range of
findings on media use in real-life organizations. As a promising alternative,
we develop a model of social influences on media-related attitudes and
behavior in organizations. Then, we review research results that shed light
on the validity of the new model compared to traditional theory. Finally, we
Author's Note: We would like to thank George Huber, Lynne Markus, and
Ron Rice for their helpful comments. We are indebted to Peter Monge and
Ev Rogers for their insightful discussions over the course of this work. Gerry
Power provided important contributions during the initial stages of concep-
outline the implications of the social influence perspective for management
practice, and provide suggestions to guide research on communication tech-
nologies in organizations.
Rational Choice Models
Media use is known to be a function of a number of facilitating factors,
such as media accessibility, availability of communication partners, experi-
ence with the medium, personal style in using media, time and cost advan-
tages, and communication task requirements (for reviews see Culnan &
Markus, 1987; Rice, 1980, 1984; Steinfield, 1986b; Svenning & Ruchinskas,
1984). Media use theory proposes that individuals choose media through a
matching process. This matching involves assessing the requirements of the
particular communication task at hand and selecting a medium with com-
munication capabilities that match these requirements. Efficient communi-
cation takes place when the match is perfect: the medium has neither more
nor less communication capability than the task requires. This rational
choice approach is described using two illustrative theories. Note, however,
that this perspective is not the hallmark of any single theoretical model, but
rather a general tradition characterizing the study of media choice processes
Social Presence Theory. Short, Williams, and Christie (1976) conceptual-
ized communication media as falling along a single continuum of "social
presence." Social presence is the degree to which the medium facilitates
awareness of the other person and interpersonal relationships during the
interaction. Face-to-face communication has the greatest social presence,
followed by audio plus video (e.g., videoconferencing), audio-only (e.g., tele-
phone), and print.
Communication is efficient when the medium selected has a social pres-
ence level that matches the level of interpersonal involvement required for
the task. Highly involving tasks (e.g., conflict and negotiation) are best
completed using high social presence media (e.g., face-to-face). Similarly,
media with less social presence (e.g., written letters) are optimally efficient
for simple information exchange tasks.
Tests of the social presence model primarily have been conducted in the
laboratory using cooperative versus competitive tasks. The model received
moderate support in these contexts, but explained only a small proportion of
the variance in media-related behavior. However, results have limited gen-
eralizability beyond the laboratory setting (for reviews see Albertson, 1980;
Fowler & Wackerbarth, 1980; Short et aI., 1976; Williams, 1977).
Social Influence Model
Information Richness Theory. An alternative view is provided by Infor-
mation Richness Theory (Daft & Macintosh, 1981; Daft & Lengel, 1984,
1986; Daft, Lengel, & Trevino, 1987). From this perspective, communication
media are arrayed along a continuum of "information richness" based on four
criteria: speed of feedback, types of channels employed, personalness of
source, and richness of language carried. Face-to-face is the richest medium
followed in decreasing order of richness by telephone, electronic mail, writ-
ten personal, written formal, and then numeric formal media.
The key criterion for media choice is message ambiguity. Ambiguous tasks
should be completed using rich media, while unambiguous tasks require lean
media. Efficient managers select media whose richness matches the ambigu-
ity of the communication task. Recent statements of the model (Trevino,
Lengel, & Daft, 1987) include two additional factors. First, contextual deter-
minants such as geographical separation between communication partners
may constrain which media can employed. Second, the symbolic character of
formal written media may serve to increase its appropriateness in certain
situations in which the user wants to convey an impression of authority.
Direct tests of this newer model are as yet scarce. Trevino et al. (1987) and
Daft, Lengel, and Trevino (1987) found support for the basic information
richness proposition, although Steinfield and Fulk (1986) and Markus (1988)
found only weak support. Trevino, Daft, and Lengel (Chapter 4 of this
volume) present evidence that links rational media choice to performance
Social Presence Theory and Information Richness Theory are the best
known rational choice models of media use. They share with most other
media use perspectives a number of key assumptions about media and about
Assumptions of Rational Choice Models
Media. Each communication medium has fixed, inherent properties. A
medium's social presence or information richness is invariant regardless of
who is using it or what the context is. Task requirements, as well, are largely
based on objective features such as ambiguity. As Albertson (1980, p. 389)
noted about social presence theory, "people are regarded as 'black boxes'."
Differences among media in their objective features are salient to users.
That is, people are aware of inherent differences among media and they are
"differences that make a difference." Similarly, users also perceive variations
across tasks in such features as social presence requirements.
Choice-Making Processes. Individuals make independent choices. The in-
terpersonal setting in which choice occurs does not intrude directly upon the
decision process. The only exception occurs if the sender wants to symboli-
cally convey impressions of formality or authority, as specified in information
Choice-making is a cognitive process. Attitudes and behaviors follow from
cognitive evaluations of media attributes and message requirements. Choice
is also prospective in that it depends upon assessments of current needs and
future goals. Purpose is posited to occur before action.
Choice-making is objectively rational. The user objectively evaluates the
characteristics of tasks and media, and then chooses that combination that
most closely matches task requirements with media capabilities. Rational
choices are based upon inherent features of media and tasks; little place is
accorded for individual variation in deciding what is an optimal match.
Behavior is efficiency-motivated. Use of overly rich media for unambigu-
ous tasks is inefficient. The efficient user treats media capacity as a resource
not to be squandered. At the same time, insufficient media capacity is
ineffective. Exact matches are optimally efficient (Williams, 1977).
Evidence for Social Context Effects
Recent investigations of new information technologies cast doubt on the
overarching importance of inherent media characteristics and objectively
rational choice processes in predicting media use:
• Highly interpersonally involving interactions such as conflict resolution
and negotiation consistently occur over electronic mail, a low social
presence medium (Hiltz & 'lliroff, 1978; Kiesler, Siegel, & McGuire,
1984; Rice & Love, 1987; A. Phillips, 1983; S. Phillips, 1988; Steinfield,
1985; Markus, 1988).
• Use of personal computers and computer-mediated communication sys-
tems is facilitated by a network of supportive relationships (Kling &
Gerson, 1977), such that "being a member of one group (or subculture)
rather than another seems to shape the experiences of the members and
the quality of their (electronic) life" (Hiltz, 1984, p. 90).
• Recent investigations in a large office products firm (Steinfield, 1986a)
and a city government (Schmitz, 1987) discovered that a potent predic-
tor of electronic mail use was the extent of use by relevant coworkers,
including supervisors. A similar (but less strong) relationship was
reported for voice mail use in a large insurance firm (Shook, 1988).
• Research in a major petrochemical firm found that a significant predic-
tor of attitudes toward videoconferencing was an individual's percep-
Social Influence Model
tions of the OpInIOnS of coworkers and supervisors about videocon-
ferencing (Svenning, 1982).
Rational choice models cannot account for these findings.
Where do we search for explanations of the apparent influence of the
social context on media use behavior? A logical source is theory that focuses
on the dynamics of social influence. In the next section we develop a model
and predictions of media use that has roots in various approaches known as
symbolic interactionism (Mead, 1934; Cooley, 1902; Dewey, 1922), attribution
theory (Bem, 1972), cognitive dissonance theory (Festinger, 1954), social
learning theory (Bandura, 1977, 1986), and social information processing
(Salancik & Pfeffer, 1978). The latter work is particularly helpful because it
focuses specifically on the work environment.
A Social Influence Model of Media Use
The social influence model starts with the same basic assumption that
individuals cognitively process stimuli. However, it departs substantially
from rational choice models in positing how cognitions develop and change.
Media perceptions are, in part, subjective and socially constructed.
Clearly, they are determined to some degree by objective features such as
ability to provide a permanent record, asynchroneity and the like (see
Culnan & Markus, 1987 for a review of these features). However, they are
also determined to a substantial degree by the attitudes, statements, and
behaviors of coworkers.
How do coworkers exert social influence? The most direct way is by overt
statements about characteristics of media or tasks that individuals assimi-
late into their own evaluations (Salancik & Pfeffer, 1978). These direct
statements also have an indirect effect. By discussing particular features,
coworkers increase the saliency of those features. Coworkers also voice
judgments and interpretations of events that may be accepted by the indi-
vidual (Bandura, 1986; Salancik & Pfeffer, 1978). The net effect is that media
perceptions are not fixed and objective; instead they will vary across individ-
uals and situations.
One example of this cuing occurred in the production research division of
a large petrochemical company. An R&D manager reported to a division
manager that rapid information exchange with field sites via PROFS elec-
tronic mail helped to speed up the product development cycle. This social cue
had at least three points of potential influence on media evaluations. First,
it highlighted speed of feedback as an important criterion for media assess-
ment. Second, it described one feature of electronic mail (ability to rapidly
convey messages to a remote site) as particularly salient. Third, it provided
an interpretation of the shortened product development cycle as resulting
from electronic mail rather than other potential causes. Thus, it implied a
very favorable evaluation of the medium's effect on a key organizational goal.
Statements by coworkers also directly affect choice making. Remarks may
cue an individual about the requirements of a communication task. They
may also help define the individual's media options in relation to the task
requirements. More explicit statements may make specific recommenda-
tions about media choice.
An example of this overt cuing occurred at an electrical equipment man-
ufacturer. An engineer needed to inform the manufacturing manager about
a last-minute product design modification. Another engineer advised against
using a written memo for what she saw as a politically sensitive issue. She
recommended that the engineer personally visit the manufacturing man-
ager's shop as soon as possible. She believed the personal visit would get the
changes made quickly and not damage relations between the two units.
Of course, sources of influence may conflict in their statements. In the
previous example, a third engineer recommended using a written memo. He
believed that the use of proper channels and established procedures would
show the most respect for the manufacturing group and would also protect
the design modifications. A fourth person even got involved by telling a story
of what happened with manufacturing several years ago when an engineer
sent last-minute product design changes in a written note. The last element
ofthe story is particularly instructive because it illustrates that choice occurs
in a rich historical context as well.
Social influence also may take the form of vicarious learning from observ-
ing the experiences of others (Bandura, 1986). When the choices have lead to
positive consequences, behavior modeling may occur. Thus, effective behav-
ior by one person may well be repeated by others through a process of
observational learning. Similarly, choices that lead to undesirable conse-
quences may be avoided by others.
An example of vicarious learning occurred in an advertising agency. An
account representative lost a major client because the client misinterpreted
a brief electronic mail message. One consequence was that other account
reps used the telephone more frequently for similar transactions because
they wanted to avoid that same fate. The opposite could easily have occurred
as well: An account rep gaining a prized client presumably because of a
well-timed electronic mail message. Electronic mail use for courting clients
Social Influence Model
might then increase throughout the agency in the hope of achieving a similar
Thus far, we have said that media perceptions are subjective and socially
constructed. Two processes of social influence were described: (1) overt
statements by coworkers about media characteristics, task features, and
choice making; and (2) vicarious learning.
Assumptions about rationality are quite different when viewed from a
social influence perspective. Rationality is subjective, retrospective, and in-
fluenced by information provided by others. Cognitions are used for rational-
izing and giving meaning to behavior. However, cognitions may arise from
prior history of social interaction. That is, sense-making may well be created
after the occurrence of the behavior. In this case, it is used to interpret the
behavior retrospectively rather than to direct the choice prospectively. Indi-
viduals observe their own present and past behavior and develop explana-
tions for it. In the process, goals and intentions may be created after the fact
to make sense of what has occurred (Weick, 1979). The inferred goals are
sensible and rational in the current interpretation, whether or not they were
at the time the behavior occurred. One outcome of retrospective sense-mak-
ing is the reconciliation of goals with current and past behavior. A simple
form of this process is the classic response to failure: ''Well, I wasn't really
Interpretation also involves a process of attribution in which past behav-
ior serves as a source of current attitudes (Bem, 1972). Thus, people make
assumptions about what their attitudes are about an issue, event, or person
by recalling their own behavior surrounding the issue, event, or person. "I
did it so I must like it" is an only slightly exaggerated view of the attribution
process. Like goals, attitudes are used to interpret and make sense of
behavior that has already occurred. Salancik and Conway (1975) note that
inferences about attitudes are based not on one's actual behavior, but rather
upon what someone actually knows or remembers about his or her behavior.
As Pfeffer (1982) indicates, this means that different attitudes can be created
by altering the information provided about past behavior.
A key point is both sense-making and behavior are subject to social
influence. One form of direct influence is group norms. Social psychologists
(e.g., Asch, 1952) have shown that group norms have powerful effects on
individual cognitions and behaviors. Social cues regarding appropriate
media use may well be embedded within the norms of a particular group.
For example, in one work unit of an employment agency, members exerted
pressure against using written memos within the group-memos were
viewed as too impersonal for almost any kind of group interaction. In this
group, a simple information exchange frequently required face-to-face inter-
action. The choices that individuals made to adhere to social pressures were
rational from their own subjective view, even though they did not meet
efficiency criteria (March, 1974). Thus, although choice-making may be
supported by norms favoring prospective and objectively rational choice, it
need not be. It can equally well be subject to normative pressures to behave
in a nonefficient manner. The key implication is that behavior is subjectively
The social environment also influences retrospective sense-making.
Social contexts embody requirements for "meaningful and justifiable behav-
ior." They also provide "norms and expectations that constrain their rational-
ization or justification of those activities" (Salancik & Pfeffer, 1978, pp. 232,
233). In essence, the social environment creates requirements for sense-mak-
ing but also constrains the types of sense-making that are acceptable. What
is sensible in one social environment may not be in another.
For example, a rationale of "don't fix what ain't broken" may support
resistance to experimenting with new communication technologies in one
unit. At the same time, another unit may support this experimentation with
the axiom "you don't get anywhere by standing still." In each case, the
justification is framed to reflect what appears to be quite sensible behavior.
One key constraint in social environments is that people are typically
expected to be consistent over time in their statements and behaviors.
Deviations require justifications that are sensible within social definitions of
rational behavior. This may lead to commitment to courses of action beyond
the point when those actions are sensible from other points of view (Becker,
For example, an executive in the auto industry initially opposed the
installation of terminals for a computer-based integrated office system for his
level of management. He continued to maintain resistance to the system,
even after it had become considerably more user-friendly. His rationale for
opposing the innovation was consistent with social definitions of rational
behavior in that context-"a manager is paid a lot of money to make tough
decisions, not to sit at a keyboard and type." To overcome this opposition, the
manager must perceive a reason to change, but he also needs a socially
acceptable rationale that defines both past resistance and current accep-
tance as appropriate.
Social Influence Model
Comparison of Assumptions
RATIONAL CHOICE MODELS SOCIAL INFLUENCE MODEL
Media and Task Features
Subjective; socially constructed
Subject to social influence
Can be retrospectively rational
Can be efficiency-motivated but need not be
Summary. Media-use behavior is, at its core, subjectively rational. Be-
havior need not meet efficiency criteria in order to be considered rational
within the particular social context. Behavior is subject to social influence in
the form of widespread norms and pressure for sense-making. Although
sense-making may be prospectively rational, it is often better described as
retrospectively rational. Thus, the social influence perspective starts from an
assumption of cognitive processing of stimuli, then departs radically from
rational choice theories. A comparison summarizing the contrasting assump-
tions of the two approaches is provided in Table 6.l.
Under what conditions are social influence processes strongest? One
consistent finding is that individuals rely more strongly on social comparison
processes in ambiguous situations (Festinger, 1954; Salancik & Pfeffer,
1978). In fact, Thomas and Griffin (1983) found that greater experience with
a task reduces the influence of social information about that task. In a
parallel fashion, we may expect that social information regarding a particu-
lar medium will be more influential for individuals who have less experience
and knowledge of that medium. Experienced users will have longer histories
of their own behavior upon which to base attributions. Also, greater mastery
of the skills necessary to employ the newer communication technologies such
as electronic mail or computer conferencing is likely to directly facilitate use
of those media. Evidence from four studies shows electronic mail use to be
directly linked to experience with the medium and knowledge mastery (Kerr
& Hiltz, 1982; Johansen, 1988; Schmitz, 1988; Fulk, Schmitz, & Steinfield,
A host of factors beyond those described above come into play in any
organizational context. A detailed treatment of these factors necessarily
would complicate the social influence model to the point where it becomes
the proverbial "Indian war," with causal arrows flying in every direction.
Then, the model would lose its unique value as a guide for highlighting the
social influence process as applied to media use. For purposes of simplicity,
we simply label these other contextual features situational factors. This
section briefly describes several categories of more obvious situational influ-
ences on media use.
One set offactors is individual differences. Perceptions of computing have
been found to relate to an individual's cognitive style (Aydin, 1987) and to
individual media style (Rice & Case, 1983). Although this latter moderator
is defined behaviorally, it is intended to reflect a psychological factor that
predisposes individuals to select certain media over others.
A second group are the facilitating factors mentioned earlier. This would
include such variables as accessibility of the medium, training support for
new media, critical mass of users, protection of documents, nonpunitive
budget and pricing policies, reliability and flexibility of the technology,
compatibility of the technology with current values, and organizational
support for the medium, as well as specific hardware and software features.
A final category is direct constraints on media use. This includes such
considerations as barriers of geography and time that prevent utilization of
certain media (Steinfield & Fulk, 1986; Trevino et aI., 1987).
The social influence perspective does not imply that socially constructed
reality is idiosyncratic or random. Social construction of reality entails some
degree of agreement on the nature of past events and on appropriate future
behavior (Rose, 1962). The process of developing understandings is ongoing
and somewhat self-correcting. Over time, people come to share similar
interpretations and parallel realities which vary as a function of their group
membership and of their personal interaction history. The net effect is this:
Although it is unlikely that perceptions of media and tasks will be invariant
across individuals, neither will they be random. There will be patterns of
systematic variation which are reliably linked to the social context.
The perspective articulated here in no way excludes objectively rational
media choices as one outcome. Rather, it subsumes rational choice as simply
; I \ ;
I ' ; ~
Social Influence Model
one of many options that may emerge from the social influence process in
organizations. The limitation of traditional media use theories is their
over-reliance on rational processes to explain the entire range of media-
choice situations. A realistic understanding of behavior requires knowledge
not simply of objective features of the environment, but also the social milieu
that alters and adjusts perceptions of that environment. The advantage of
the social influence model is its potential to explain a much wider range of
media-use behavior across a greater variety of situations.
Figure 6.1 provides a schematic that indicates the pivotal role of social
influence in media evaluations and behavior. Formally s t a t e d ~ the proposi-
tions supporting Figure 6.1 are:
Proposition 1: Media evaluations (perceptions and attitudes) are a function
of: (a) objective media features; (b) media experience and
skills; (c) social influence, in the form of direct statements by
coworkers, vicarious learning, group behavioral norms, and
social definitions of rationality; and (d) prior media-use
Proposition 2: Task evaluations are a function of: (a) objective task features;
(b) task experience and skills; and (c) social influence, in the
form of direct statements by coworkers, vicarious learning,
group behavioral norms, and social definitions of rationality.
Proposition 3: For any application, an individual's media use is a function of:
(a) media evaluations (perceptions and attitudes); (b) media
experience and skills; (c) social influence in the form of direct
statements by coworkers regarding the application, vicarious
learning, group behavioral norms, and social definitions of
rationality; (d) task evaluations; and (e) situational factors
such as individual differences, facilitating factors, and con-
Assessing the Social Information Model
One way to assess how accurate a picture the social influence model
paints is to identify specific predictions of the model for real-life organiza-
tions. Then we can examine findings from organizational studies to see
whether these predictions hold up.
Predictions for Media Evaluations
The social influence model predicts that people will vary in how "rich"
they perceive a particular medium to be. It also predicts that this variation
will not be random or idiosyncratic. Rather, variation will be systematically
~ . . . . <
~ ; . : :
r.:I tJ .: ~
Z " .. ".-
r.l E ~ E·':::
~ E . ~
~ . : : . ~
g • • • •
~ l : 3 ]
U ; t ~ ~
<.- .... '"
~ '" bJ) "
<';:5 ~ b ~
z._._ bJ) '"
8 .:: =s w c ~
< . . : : ~
: ~ . ~ ~ .
~ ._ bJ)
U c CI,)
~ .. .
; . : : ~
Social Influence Model
linked to variation in the social context and media experience/skills. Alter-
natively, rational choice models predict no variation in media perceptions,
except for random variation or errors in measuring media perceptions. In
effect, if nonrandom variation exists that is not attributable to measurement
error, social influence theory provides explanatory power, while rational
media choice models do not.
What evidence exists regarding variation in media perceptions? Several
studies have asked respondents to rate different media on perceived infor-
mation richness. A review of this combined evidence (Steinfield & Fulk,
1989) showed that the rank order of media ratings was consistent across the
studies, with one exception. Electronic mail was sometimes ranked above
written media, and sometimes below. The rank order of other media was
predicted by rational choice models. A rank order, however, may mask
considerable variation in people's perceptions of individual media. In some
studies, the mean ratings for certain mid-range media were quite close to
each other, while in others these same media were much farther apart on the
richness scales used. Yet, both patterns produced the same rank order. These
results are consistent with earlier findings that individuals varied in their
perceptions of the social presence of a medium, but that the rank order of
mean ratings stayed the same (Short et al., 1976).
A recent pair of studies provides more direct evidence (Fulk, Schmitz, &
Steinfield, 1988; Fulk & Ryu, 1990). Each study used the exact same mea-
sure of information richness, one taken directly from written statements of
this theory. Both samples were R&D units involved in applied petrochemical
research. The researchers found the same pattern as the earlier studies:
rankings were similar across studies, although electronic mail was lower in
richness than was predicted by information richness theory. More impor-
tantly, the mean ratings in each unit were different for the same media.
Was this variation in richness ratings systematically linked to variation
in social context and media experience/knowledge? To address this question,
the researchers asked each respondent to name the five persons he had most
communication with-the respondent's important potential sources of social
influence within the organization. For each source of influence plus the
supervisor, respondents rated the person's attitudes about electronic mail.
The researchers also gathered information on the respondent's experience
with electronic mail, computing experience, and training in system use. The
findings from both studies were that: (1) People's perceptions of electronic
mail were significantly related to the attitudes of both the supervisor and the
five frequent communication contacts; and (2) persons with more electronic
mail experience and training rated electronic mail as richer than those
without such experience and training. These findings are consistent with an
earlier study in a large petrochemical company, which found that attitudes
toward videoconferencing were positively related to perceptions of the atti-
tudes held by coworkers toward the same system (Svenning, 1982).
In combination, these findings suggest that: (1) richness perceptions vary
in a nonrandom fashion, and (2) they are directly linked to social context and
media experience factors, as predicted by the social influence model.
Predictions for Media Use
Less explicit evidence exists for media use. An indirect test is the ability
of the models to explain existing findings. Below we describe briefly three
sets of research findings that demonstrate the enhanced explanatory ability
of the social influence model.
Pattern Matching Within Groups. Rational choice models predict that we
will find similar patterns of media use across individuals in situations where
their tasks are similar. By contrast, the social influence model predicts some
similarity of media attitudes and use behavior within groups, even across
tasks with different communication requirements.
What evidence exists to date? A study in a large insurance firm found
similar patterns of usage of voice mail among coworkers occupying the same
structural network position (Shook, 1988). Rice, Grant, Schmitz, and Torobin
(1988) found similar patterns of electronic mail adoption among closely
connected coworkers. Fulk, Schmitz, Ryu, and Steinfield (1989) found that
electronic mail use (number of messages sent and percent of time using the
medium) was predicted by the perceptions of the medium's usefulness held
by communication network partners. Research in a large office products firm
(Steinfield & Fulk, 1989) found that 25% of the variation in electronic mail
use could be explained by the proportion of coworkers who used electronic
mail, while no variation in use was explained by a global measure of task
ambiguity. To make sense of these findings using rational choice models, we
would have to assume that the tasks were similar across individuals in the
same network. This is a considerably more restrictive assumption than social
influence models would require.
Information richness theory also predicts that leaner media will be used
relatively less frequently at each higher management level (Daft & Lengel,
1984), because higher level jobs confront greater ambiguity. The social
influence model predicts less discrepancy between adjacent levels if there is
intense social interaction, such as between supervisor and subordinate. It
I ~ j
~ , r
~ ! :
Social Influence Model
also proposes other factors that influence media selection, so that lean media
may be used in spite of a lack of "fit" with task ambiguity.
Some evidence is available. Both survey and computer-captured usage
data from a city government showed that 20% of the variation in electronic
mail use was explained by organizational level (Schmitz, 1987). However, the
direction of the relationship was contrary to predictions from information
richness theory: the use of this relatively lean medium increased with
organizational level. And, an additional 10% of variation in electronic mail
usage was explained by the supervisor's usage patterns. Rice, Hart, Torobin,
Shook, and Tyler (1989) found that electronic mail was used more heavily by
managers than by technicians or clerical staff. These findings are generally
consistent with the more robust social influence explanation.
Pattern Mismatching Across Groups. Several studies have found differ-
ences in attitudes or patterns of use of the same communication technology
across groups with relatively equal access to the technology (e.g., Hiltz &
Johnson, 1987; Kerr & Hiltz, 1982; Schmitz, 1987). Without descriptions of
task features, the rational-choice and social-influence explanations for these
differences cannot be compared conclusively. Theoretically, the social influ-
ence model predicts the observed findings from its basic premises. Rational
choice theories could explain the findings only by resorting to assumptions
of similar levels of task ambiguity or unusual contextual determinants.
Two other studies of similar R&D units did present more information on
tasks (Fulk et al., 1988; Fulk & Ryu, 1990). Both samples were engaged in
production research for the same corporation, used the same PROFS elec-
tronic mail system, and were subject to similar organizational budgets and
corporate policies. Levels of experience were relatively similar, as both units
had been using PROFS for several years. In this situation, rational choice
models predict similar patterns of use. The social influence model predicts
that usage will vary between units because their social norms and interac-
tion histories are different. In fact, the patterns of use were different. In Unit
A, more time was devoted to using the system, and almost three times the
number of electronic mail notes were sent. Also, Unit A members considered
this system more effective across the board for all types of communication
Inefficient Choice-Making. Several investigators have reported that low
social presence media are used for high social presence tasks. Studies of
electronic mail and computer conferencing have found socioemotional uses
such as getting to know someone, maintaining relationships, resolving con-
flict and disagreements, negotiation and bargaining, and expressing anger
or gripes (Hiltz & Throff, 1978; Kiesler, 1986; Kiesler et aI., 1984; Phillips,
1983; Rice & Love, 1987; Steinfield, 1986a). The other type of mismatch is
illustrated in the earlier example of the employment agency where people
met face-to-face for nearly all communication tasks. Rational choice models
cannot readily explain these findings. The social influence model directs us
to search for potential explanations in social norms which can develop to
encourage experimentation with a wide variety of applications for new
A common observation is that old media-use patterns persist even after
more efficient communication options become available (Panko, 1984). Ratio-
nal choice models imply that when a new medium is introduced into the work
situation, media-use patterns should change; for example, electronic mail
should replace some telephone calls or some written letters for low social
presence tasks. Evidence exists to support this media substitution prediction
(Picot, Klingenberg, & Kranzle, 1982; Rice & Bair, 1984), but that evidence
also suggests that substitution effects are somewhat limited. Retrospective
sense-making and the influence of social norms would explain those addi-
tional situations in which new and more efficient media continue to be
The evidence to date indicates that rational choice models effectively
predict some observed patterns of media use in organizations. Nevertheless,
these models fall short in their ability to account for behavior patterns that
appear nonrational on the surface. From the rational choice perspective,
such findings are anomalies that leave the researcher puzzled. The number
of such anomalies in the research literature is not small and is rapidly
growing. There is clearly a need for a theory which explains more of these
anomalies-a theory that solves more of the puzzle. While no theory can
solve the whole puzzle, the social influence model takes us further in explain-
ing patterns found in today's organizations. The model also offers one addi-
tional advantage. It takes into account the social context of behavior-a
factor known by organizational theorists to be an integral part of organiza-
Implications for Management Practice
Given that social influence is a key factor in media use, what does this
knowledge offer to management? To illustrate the benefits of a management
approach informed by a social influence perspective, we sketch two contrast-
ing situations involving the introduction of a new medium, videoconfer-
Social Influence Model
encing, into two organizations. Each example is an actual occurrence from a
Company One: The Costs of Neglecting the Social Network
An aerospace firm decided to try videoconferencing. Engineers in Califor-
nia and Arizona needed to consult frequently about aircraft design. Written
reports and letters were slow. Fax was quicker, but the engineers couldn't
jointly view and modify documents. The telephone provided a synchronous
medium, but not shared text. The company was experiencing delays in the
product delivery schedule because of communication lags. The solution was
to install a two-way, full-motion videoconferencing system that linked both
sites. The system was "almost as good as face-to-face."
Mter an initial test period, the system fell into disuse. Why would a
system specifically designed to meet well-articulated communication needs
sit idle? In this case, the answer lay not in the technology, but in the social
system. A key engineer, widely recognized as a knowledgeable computer buff,
refused to use the system. This engineer claimed that the system was not
secure and that competitors might intercept the satellite signal linking the
two sites. His opinion quickly diffused throughout the engineering division;
no engineer would use the system.
The example is particularly instructive because the conferencing system
was among the most secure in the nation; it used scrambled signals and
other sophisticated protection procedures. Technical experts on videocon-
ferencing were convinced that the system was secure. Yet, widespread per-
ceptions of security flaws were conveyed by social information communicated
among users-by persons who lacked the telecommunications expertise
necessary to evaluate the security measures.
Could this innovation have been salvaged? Possibly. Once the opinion
leader had communicated his security reservations to others, he made a
public commitment to that point of view. To change his position, he would
need a socially acceptable reason to act inconsistently with his expressed
position. The critical deficiency occurred when system designers and manag-
ers failed to anticipate the importance of the security issue for users. There-
fore, they failed to inform users, especially opinion leaders, of existing and
extensive safeguards. A preimplementation evaluation of user needs should
have surfaced the engineers' concern with security, a concern typical in the
aerospace industry. In addition, keeping in touch with informal discussions
in the social network during the early implementation stages might have
made early detection and correction possible.
Company Two: Using a Social Influence Strategy
Several years ago a petrochemical firm introduced a corporate-wide video-
conferencing system. Each of seven major locations within the continental
United States and Alaska was provided with a completely equipped video-
conferencing facility and staff support to operate it. The goal was to provide
better communications among corporate locations ranging from Texas to
Implementation began with an organization-wide survey of communica-
tion needs and attitudes toward videoconferencing. Survey results pin-
pointed employees who were most likely to have success experiences. This
group included persons who met two criteria. First, they had high cross-Ioca-
tional communication needs, and thus were likely to try the new technology.
Second, they held positive attitudes toward videoconferencing, and thus
were motivated to maintain these positive evaluations. These individuals
were enlisted as initial system users during the project's first stage. Oppor-
tunities were created for vicarious learning by using individuals who were
prominently located. In diffusion of innovations parlance, an attempt was
made to increase the "observability" of the innovation (Rogers, 1983). Finally,
to the extent that these individuals were opinion leaders, rapid diffusion of
positive attitudes toward the system resulted through their overt statements
in support of the new medium. The success of this strategy, described in
detail by Ruchinskas, Svenning, and Steinfield (1989), is particularly im-
pressive given the relatively poor track record of videoconferencing to date
(Olgren & Parker, 1983; Svenning & Ruchinskas, 1984).
These two contrasting situations highlight some of the ways that social
influence processes may be harnessed to introduce and/or enhance the use of
communication and information technologies on the one hand or lead to
expensive but underutilized systems on the other. One other information
technology strategy that relies on social influence processes is informal peer
training. Strassman (1985) contends that informal training in the peer-group
is likely to be the most effective and the least costly training method. In
contrast, formal training conveys an initial but limited competence that is
often rapidly forgotten. During interviews, members of an R&D unit using
the PROFS electronic mail system often described how they relied on nearby
coworkers for assistance in using unfamiliar features of the system. Nearly
all members had gone through a formal training process, but lack of regular
use of many system features soon eliminated the requisite knowledge from
members' repertoires. Sustained usage is more readily achieved if users are
members of a group that shares and reinforces the common use of a commu-
i l ~ :
I ~ :
~ I H I '
; . r . \ ~ . '
~ f s
' ~ ~ ..
Social Influence Model
nication medium. Quite often, secretarial staff who used more advanced
features on a regular basis became informal consultants to both R&D profes-
sionals and department heads.
Implications for Research on Communication Technologies
The recognition that media use occurs within a web of social relationships
has important implications for the conduct of research. Research must
broaden methodological repertoires with designs and approaches that are
robust enough to capture social influence processes. We briefly highlight a
few of the considerations that are suggested by the above review and analy-
First, although experimental designs are useful starting points for re-
search on social cuing, they lack external validity. The effects of naturally
occurring communication among coworkers about media use must be studied
in field settings. Actual communication flows also can be modeled using
communication network analysis procedures. This procedure will permit us
to more directly track the flow of social information throughout the natural
Second, behaviors are responsive to social cuing during prolonged and
intense social interaction. This imposes the need for longitudinal research
d'esigns. Miller and Monge (1985) point out that little is known about the
relative influence of immediate versus accumulated social information. Also,
previous research provides little indication of the temporary versus perma-
nent nature of the effects of social influence. This is particularly important
for research on new technologies, because patterns identified in the initial
phases of an implementation may be substantially different from the equi-
librium patterns attained over time (Huber, this volume; Johansen, 1976).
Third, research designs should incorporate measures of group level atti-
tudes and behavior. This could include, for example, perceptions of coworkers
sentiments (Svenning, 1982), coworker media use (Fulk et aI., 1988), or
actual computer-monitored usage of significant others (Rice & Borgman,
1983; Schmitz, 1988). One promising research strategy is to explicitly model
relational ties among individuals with communication network models of
social influence. Group level measures will also enable comparisons across
social or organizational groupings, in addition to comparisons across types
of jobs or tasks. A last note on group level measures is that the study of the
dispersion of perceptions may be as fruitful as the study of central tendencies
(Zalesny & Farace, 1986). Danowski (1980), for example, found lower stan-
dard deviations in attitude scores for more highly connected groups, reflect-
ing a social influence that would have gone unnoticed if he had only exam_
ined mean group scores.
Fourth, we should treat rationalizations as valuable objects for study.
Investigation of the etiology and development of these rationalizations
should provide important insights into how individuals learn to make sense
of their media environments. Effective social influence research may depend
as much upon qualitative methodologies as upon quantitative strategies.
This work was motivated by our perception that although communication
scientists are aware of significant social effects on the attitudes and behav-
iors of individuals, much greater efforts are needed to explicitly specify these
factors in models of organizational media use. Accurate predictions of tech-
nological effects critically rely on valid assumptions about how individual
and organizations interact with the technology. If we are to rest our claims
on valid premises, social forces must be incorporated into our knowledge base
about new information technology.
Albertson, L. A. (1980). Trying to eat an elephant. Communication Research,
Asch, S. E. (1952). Social psychology. Englewood Cliffs, NJ: Prentice-Hall.
Aydin, C. E. (1987). The effects of social information and cognitive style on
medical information system attitudes and use. In W. W. Stead (Ed.),
Proceedings of the Eleventh Annual Symposium on Computer Applications
in Medical Care (pp. 601-606). New York: Institute of Electrical and
Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice-
Bandura, A. (1978). The self system in reciprocal determinism. American
Psychologist, 33, 344-358.
Bandura, A. (1986). Social foundations of thought and action. Englewood
Cliffs, NJ: Prentice-Hall.
Becker, H. (1960). Notes on the concept of commitment. American Journal of
Sociology, 66, 32-40.
Bem, D. J. (1972). Self perception theory. In L. Berkowitz (Ed.), Advances in
experimental psychology (Vol. 6, pp. 1-62). New York: Academic Press.
Cooley, C. H. (1902). Human nature and the social order. New York: Charles
: ' I
t ', ..... .
. j ' ~ ' t
) ~ ;:
i ~ ' l
Social Influence Model
Culnan, M. J., & Markus, M. L. (1987). Information technologies. In F. M.
Jablin, L. L. Putnam, K. H. Roberts, & L. W. Porter (Eds.), Handbook of
organizational communication (pp. 420-443). Newbury Park, CA: Sage.
Daft, R L., & Lengel, R H. (1984). Information richness: A new approach to
managerial behavior and organization design. In L. L. Cummings & B. M.
Staw (Eds.), Research in organizational behavior (Vol. 6, pp. 191-233).
Greenwich, CT: JAI Press.
Daft, R L., & Lengel, R H. (1986). Organizational information require-
ments, media richness and structural design. Management Science, 32,
Daft, R L., Lengel, R H., & Trevino, L. K. (1987). Message equivocality,
media selection, and manager performance: Implications for information
systems. MIS Quarterly, September, 355-366.
Daft, R L., & Macintosh, N. B. (1981). A tentative exploration into the
amount and equivocality of information processing in organizational work
units. Administrative Science Quarterly, 26,207-224.
Danowski, J. A. (1980). Group attitude uniformity and connectivity of organ-
izational communication networks for production, innovation and main-
tenance content. Human Communication Research, 6,299-308.
Dewey, J. (1922). Human nature and conduct. New York: Holt.
Festinger, L. (1954). A theory of social comparison processes. Human Rela-
tions, 7, 114-140.
Fowler, G., & Wackerbarth, M. (1980). Audio teleconferencing versus face-to-
face conferencing: A synthesis of the literature. Western Journal of Speech
Communication, 44, 236-252.
Fulk, J., & Ryu, D. (1990).Perceiving electronic mail systems: Apartial test of
the Social Information Processing Model of communication media use in
organizations. Paper presented to International Communication Associa-
tion, Dublin, Ireland.
Fulk, J., Schmitz, J., & Steinfield, C. w. (1988). Social information and
technology use in organizations. Paper presented at the Annual Academy
of Management Convention. Anaheim, California.
Fulk, J., Schmitz, J., Ryu, D., & Steinfield, C. w. (1989). Communication in
R&D via electronic mail III: Final report. University of Southern Califor-
nia, Annenberg School of Communications.
Hiltz, S. R (1984). Online communities: A case study of the office of the future.
Norwood, NJ.: Ablex.
Hiltz, S. R, & Johnson, K. (1987). Measuring acceptance of computer-medi-
ated communication systems. Paper presented at the International Com-
munication Association, Montreal, Canada.
Hiltz, S. R, & Turoff, M. (1978). The network nation: Human communication
via computer. Reading MA.: Addison-Wesley.
Johansen, R (1976, May). Pitfalls in the social evaluation of teleconferencing
media. Paper presented at the meeting of Second Annual International
Communications Conference, University of Wisconsin, Madison.
Johansen, R. (1988). Groupware: Computer support for business teams. New
York: Free Press.
Kerr, E. B., & Hiltz, S. R. (1982). Computer mediated communication sys-
tems. New York: Academic Press.
Kiesler, S. (1986). Thinking ahead: The hidden messages in computer net-
works. Harvard Business Review, 64 (January-February), 46-60.
Kiesler, S., Siegel, J., & McGuire, T. (1984). Social psychological aspects of
computer-mediated communication. American Psychologist, 39, 1123-
Kling, R., & Gerson, E. M. (1977). The social dynamics of technical innova-
tion in the computing world. Symbolic Interaction, 1 (11), 132-146.
March, J. G. (1974). Bounded rationality, ambiguity, and the engineering of
choice. Bell Journal of Economics, 9, 338-353.
Markus, M. L. (1988). Information richness theory, managers, and electronic
mail. Paper presented at the Annual Academy of Management Conven-
tion, Anaheim, California.
Mead, G. H. (1934). Mind, self and society. (C. Morris, Ed.). Chicago: Univer-
sity of Chicago Press.
Miller, K. I., & Monge, P. R. (1985). Social information and employee anxiety
about organizational change. Human Communication Research, 11, 365-
Olgren, C. & Parker, L. (1983). Teleconferencing technology and applications.
Dedham, MA: Artech House.
Panko, R. (1984). Electronic mail: The alternatives. Office Administration
and Automation, 45(6), 37-43.
Pfeffer, J. (1982). Organizations and organization theory. Boston: Pitman.
Phillips, A (1983). Computer conferences: Success or failure? In R. Bostrom
(Ed.), Communication yearbook 7 (pp. 837-856). Beverly Hills, CA: Sage.
Phillips, S. R. (1988). Electronic persuasion: The uses of electronic mail for
interpersonal influence in organizations. Unpublished doctoral disserta-
tion, University of Southern California, Los Angeles.
Picot, A, Klingenberg, H., & Kranzle, H-P. (1982). Organizational commu-
nication between technological development and socio-economic needs:
Report from field studies in Germany. In M. Burgoon (Ed.), Communica-
tion yearbook 6. Beverly Hills, CA: Sage.
Rice, R. E. (1980). The impacts of computer-mediated organizational and
interpersonal communication. In M. Williams (Ed.), The annual review of
information science and technology (Vol. 15, pp. 221-249). White Plains,
NY.: Knowledge Industry Publications.
Rice, R. E., and Associates. (1984). The new media: Communication, research,
and technology. Beverly Hills, CA: Sage.
Rice, R. E., & Bair, J. H. (1984). New organizational media and productivity.
In R. E. Rice (Ed.), The new media: Communication, research, and tech-
nology (pp. 185-215). Beverly Hills, CA: Sage.
Rice, R. E., & Borgman, C. L. (1983). The use of computer-monitored data in
information science and communication research. Journal of the Ameri-
can Society for Information Science, 34,247-256.
Social Influence Model
Rice, R. E., & Case, D. (1983). Electronic message systems in the university:
A description of use and utlity. Journal of Communication, 33,131-152.
Rice, R. E., Grant, A, Schmitz, J., & Torobin, J. (1988). A network approach
to predicting the adoption and outcomes of electronic messaging. Paper
presented at the Annual Academy of Management Convention, Anaheim,
Rice, R. E., Hart, P., Torobin, J., Shook, D., & Tyler, J. (1989). Task analyz-
ability, media use and performance: A multi-site test of information rich-
ness theory. Paper presented to International Communication
Association, San Francisco.
Rice, R. E., & Love, G. (1987). Electronic emotion. Communication Research,
Rogers, E. M. (1983). Diffusion of innovations (3rd ed.). New York: Free
Rose, A M. (1962). Human behavior and social processes: An interactionist
approach. Boston: Houghton Mifflin.
Ruchinskas, J., Svenning, L., & Steinfield, C. (1989). Video comes to organi-
zational communications: The case of ARCOVISION. In B. Sypher (Ed.),
Case studies in organizational communication. New York: Guilford.
Salancik, G. R., & Conway, M. (1975). Attitude inferences from salient and
relevant cognitive content about behavior. Journal of Personality and
Social Psychology, 32,829-840.
S alancik, G. R., & Pfeffer, J. (1978). A social information processing approach
to job attitudes and task design. Administrative Science Quarterly, 23,
Schmitz, J. (1987). Electronic messaging: System use in local governments.
Paper presented at the International Communication Association, Mon-
Schmitz, J. (1988). Electronic communication: A longitudinal view. Paper
presented at the Annual Academy of Management Convention, Anaheim,
Shook, D. E. (1988). A structural equivalence and contingency theory perspec-
tive on media usage and communication performance: The case of voice
messaging. Unpublished doctoral dissertation, University of Southern
California, Los Angeles.
Short, J., Williams, E., & Christie, B. (1976). The social psychology of
telecommunications. London: John Wiley.
Steinfield, C. W. (1985). Dimensions of electronic mail use in organizations.
In J. Pearce & R. Robinson (Eds.), Proceedings of the Annual Meeting of
the Academy of Management (pp. 239-243). Mississippi State University:
Academy of Management.
Steinfield, C. W. (1986a). Computer-mediated communication in an organi-
zational setting: Explaining task-related and socioemotional uses. In
M. McLaughlin (Ed.), Communication yearbook 9 (pp. 777-804). Beverly
Hills, CA: Sage.
Steinfield, C. W. (1986b). Computer-mediated communication systems. In
M. Williams (Ed.), The annual review of information science and technol-
o - o "