The Impact of Group Size and Social Presence on Small-group communication:

Does Computer-Mediated Communication make a Difference?

Project groups are becoming a mainstay in todays work environment. This trend,
coupled with globalization, has virtualized many teams, and has made communication
among group members vital to project success. This study evaluates the impact of
varying group size and social presence on small-group communication. It compares
key communication factors for three different treatments: face-to-face (FtF) without
computer-mediated communication (CMC) support, FtF with CMC support, and
virtual with CMC support. This study also evaluates these impacts on two different
small group sizes (3 and 6). The results indicate that smaller groups establish and
maintain higher levels of communication quality. In addition, the results indicate that
FtF with CMC support groups had higher levels of communication quality than
virtual with CMC support groups; however, we did not find a significant difference
between traditional FtF groups and virtual groups with CMC support.

METHOD

The participants in this experiment were students in a sophomore-level

information systems class at a large Midwestern university. Approximately 500
students were enrolled in the course (in two sections over two semesters).
The researchers received human-subjects approval for student participation
as part of the course. All had the option to participate in this HE experiment,
or an alternative assignment, for course credit for an inconsequential amount
of their course grade. Those who chose to participate gave informed consent.
Those who chose not to participate were given an alternative assignment
designed to be similar in scope and effort to participation in this study. A total
of 439 students participated in the study.

Method of Analysis

WABA was developed to assess both variation and covariation in variables

within and between levels of analysis (Cogliser & Schriesheim, 2000;
Dansereau et al., 1984; Yammarino, 1998). In our case, this analysis compared
two levels: individuals and groups. The E test and F test under WABA I
indicate whether variance is between or within groups or both or neither.
Similarly, WABA II assesses covariance. Finally, the WABA equation combines
WABA I and WABA II to examine correlation components. For this study, WABA I
results indicated a slight parts or individual result for communication richness
while the remaining results were found to be equivocal. WABA II results found
slight indicators for wholes or groups for the covariance between group
discussion and group work. The final WABA equation produced equivocal
results.

Limitations

The primary limitation of this study is its restricted generalizability. Using a

controlled laboratory environment with tightly scripted conditions under a
limited time frame causes limited generalizability of the results of this
research. We believe that different results may occur in groups larger than
six, in groups that work together over extended periods of time, in groups
that include usability experts, or in groups that perform different tasks.

Conclusion

This study evaluated the impacts of group size and social presence on group
communication. We compared key communication factors (appropriateness,
openness, richness, discussion quality, and accuracy) for three different
treatments (proximate without CMC support, proximate with CMC support,
and virtual with CMC support). In addition, our study evaluated these impacts
with two different group sizes (i.e., three member groups and six member
groups).

We found that three-person groups maintained higher levels of

communication quality than did six-person groups. Discussion quality was not
significantly improved by simply being in a three-person group instead of a
six-person group; however the other four communication variables of
appropriateness, openness, richness, and accuracy were greater within the
three-person groups that in the six-person groups. This finding is of particular
interest to practitioners because it suggests more complex projects may
benefit from using much smaller groups.

The Effects of Interpersonal Trust on

Work Group Performance

This study is intended to explore two questions: Does the level of trust within
a group effect group performance? If so, how does this relationship operate?
To answer these questions I use an experimental method to examine two roles
through which interpersonal trust could effect group performance: a main
effect and a moderating effect. The data do not support the main effect
that has dominated the literature on interpersonal trust. The data do support
the moderating role: trust seems to influence how motivation is converted
into work group processes and performance. On the basis of these findings, I
suggest that trust may be best understood as a construct that influences
group performance indirectly by channeling group members energy toward
reaching alternative goals.

Method
The hypotheses were examined using an experimental method with a between-subjects
design, and the unit of analysis being the work group. Trust, the focal variable, was the
single factor examined at two treatment levels (high - low). In accordance with prior
literature (e.g., Berkowitz & Donnerstein, 1982; Dobbins, Lane, & Steiner, 1988; Ilgen,
1986), I chose a laboratory setting as it improved the ability to (a) draw conclusions of
causality and specific mediating variables, (b) use reliable measures from different sources
and methods, (c) use hard performance measures, and (d) isolate trust.
Subjects
Forty-two three-person work groups consisting of undergraduate students participated in the
experiment. The students received extra-credit points in their class for participating. In
addition, they
could earn chances at winning one of three $40 prizes. Groups were randomly assigned to a
treatment.

Results
The effectiveness of the manipulation was examined by having participants
complete the survey prior to performing the group task. The results from a oneway ANOVA indicate that the manipulation was highly effective as the difference
in scores between the high and low trust conditions is substantial and significant
(p < .000). The mean score reported in the low trust groups was 2.5 and in the
high trust groups the mean score reported was 5.7. The data gathered through
open-ended questions at the end of the experiment supported the survey results.
For example, in groups in the high-trust treatment, subjects made remarks such
as I trusted the other group members and Going into it, [I trusted my
coworkers]. I felt from reading their descriptions that they were team players
and that I could depend on them to work for the team as a whole.