Information Technology Reference
In-Depth Information
ber had to respond to each topic before seeing the
responses of others or joining the discussion on the
topic. All interactions took place asynchronously
through EIES 2.
fect with the communication mode on objective
decision quality.
Although it made no significant difference in
objective decision quality, the communication
mode had a significant impact on perceived deci-
sion quality. Parallel communication groups felt
that they made significantly better decisions than
sequential communication mode groups (F=5.26
and p=0.0268). This finding can be explained
by the preference for procedural order construct
(Putnam, 1979), which states that individuals enter
groups with a predisposition for particular work
habits, ranging from tightly organized procedures
to loosely structured ones. In making a decision in
a group, all group members have implicit cognitive
maps for structuring group activity that interacts
with the group's contingency factors that influence
group performance. These cognitive maps are
relatively inflexible, regardless of their previous
success or failure. Satisfaction with the decision
process may have resulted from the fact that in
the parallel communication mode, group members
were able to rearrange the sequence of topics in
a way that was compatible with their cognitive
maps. Hence, they gave a high rating to perceived
decision quality, which is correlated highly with
satisfaction with the decision process. Indeed,
perceived decision quality was found to correlate
highly with satisfaction with the decision process
(ρ=0.7718). On the other hand, group members
with the sequential communication mode might
have had a conflict with a GSS-enforced decision
procedure, if it was not their preference. This pref-
erence for procedural order also was confirmed in
the study of how group members appropriate and
react to GSS technology and structured heuristics
(Wheeler & Mennecke, 1992).
Another finding on decision quality is the
relationship between objective and perceived
decision quality. Although objective and per-
ceived decision qualities were not expected to be
different (Gopal, Bostrom, & Chin, 1992-1993),
Pearson's Correlation Coefficient Rho between
Findings and disCussion
statistical measures
Except for the objective decision quality, all other
dependent variables were measured by compos-
ite variables of multiple questionnaire items. A
composite variable was used to test a hypothesis
only when Cronbach's Coefficient Alpha was
higher than 0.8. Because of the unequal number
of subjects and groups for each experimental con-
dition, the General Linear Model procedure was
used instead of ANOVA for hypothesis testing.
Whenever an interaction effect was significant,
Fisher's Least Significant Difference Test (LSD)
was used for pair-wise comparison of means
among all experimental conditions. The results
of the statistical analysis are summarized in Table
3, and a summary of the hypothesis testing is in
Table 4.
discussion
Objective decision quality was measured twice
by comparing the dollar values of portfolios in
six months and one year after the experiment,
and perceived decision quality at the end of the
experiment with the questionnaire. Although
it did not play any significant role in perceived
decision quality, the presence of a group leader
significantly improves the objective decision
quality after six months and after one year. The
leadership variable was expected to have an impact
on group performance in interaction with another
variable (George et al., 1990; Hiltz et al., 1991;
Ho & Raman, 1991)—communication mode in
this study. However, there was no interaction ef-
Search WWH ::
Custom Search