info 5 970 - Tony Polito

advertisement
Information Exchange and Use in GSS and Verbal Group Decision Making:
Effects of Minority Influence
Alan R. Dennis, Kelly M. Hilmer, Nolan J. Taylor, Anthony Polito
Dept. of Management, Terry College of Business, University of Georgia, Athens, GA 30602
adennis@uga.cc.uga.edu
Abstract
The purpose of this study was to investigate the effects
of GSS use on the exchange and use of information in
group decision making under two conditions: when there
was and was not a majority/minority split of opinion in the
group. When there was a distinct majority/minority,
groups using a GSS exchanged more information, made
better decisions, and took no more time than when they
did not use a GSS. In the uniform treatment where there
was no majority/minority, groups using a GSS exchanged
more information, but made worse decisions, and took
more time than when they did not use a GSS.
1. Introduction
Individual managers seldom have access to all relevant
information, so when truly important decisions have to be
made, a group is typically formed to make the decision or
to advise the individual who must make it [9]. Group
discussion enables members to share information so that
the group as a whole can access a larger pool of
information than any one member acting alone [23].
However, information exchange in group decision
making is often done poorly; much unique information
known to some group members is never shared with the
group [24]. Individuals, particularly low status
participants, may withhold information due to
apprehension about the group’s reaction to it or may feel
pressured to conform to the views of the group majority [7,
9, 13]. This unique information can be important, leading
to poor decisions when it is not considered [8, 25].
Information technologies such as Group Support
Systems (GSS) have the potential to fundamentally change
the nature of information exchange in group discussion [6,
20]. The purpose of this study was to investigate the
effects of GSS use on the exchange and use of information
in group decision making under two conditions: when
there was and was not a majority-minority split of opinion
in the group.
2. Information and group decision making
2.1. Information distribution
In decision making situations, group members typically
know a host of information about possible alternatives.
This information shapes members' pre-discussion
preferences and, as members communicate information
during discussion, shapes the group's decision. The
information that members have can be distributed in a
variety of ways. It can be known to all members before
group discussion (“common”), known to only one before
group discussion (“unique”) or known to some but not all
before discussion (“partially shared”).
Unique information can be distributed so that it is
unbiased; that is, each group member has a balanced set of
information that indicates the optimal alternative. In this
case, the group should reach the same decision with or
without the unique information. However, information can
be distributed so that the optimal alternative is hidden; that
is, the common information indicates a non-optimal
alternative, but if the group were to pool the unique
information, they would be able to discover the optimal
alternative. This situation is called a "hidden profile task"
by Stasser [24]. In this case, if discussion does not raise all
available but unique information, the group likely will
make a poor decision. Here, the value of unique
information is high, but empirical studies have found that
much of it is not exchanged [26, 27].
In this paper, we focus on two conditions of information
distribution in hidden profile tasks. One well-studied
condition is when a majority/minority exists in the group
[19]. In this case, a majority of group members favor one
alternative (a sub-optimal one) and a minority favor a
second alternative (the optimal one). Another, lessstudied, condition is when no majority/minority exists in
the group; members are evenly divided among the
alternatives in their opinions. We term this a “uniform”
condition. We begin with the uniform condition and then
consider how this may change when a majority/minority
exists.
2.2. Information exchange and use
In order to reach a group decision, members engage in
three activities simultaneously: information recall (either
from memory or notes), information exchange (either
giving or receiving information), and information
processing (actually using the information: assessing the
cognitive and social implications of the information and
storing it in memory) [2]. Humans have a limited amount
of cognitive resources to spread across these three
activities [1]. Most people can engage in only one activity
at one time, so that engaging in any one activity limits the
ability to engage in the other two. For example, actively
listening to discussion blocks the processing of newly
received information and the recall of related information
[13]. We consider each activity in turn.
2.2.1. Information recall. Information recall of the
group as a whole is biased toward common information,
due to the laws of probability. Common information
known by all members is more likely to be remembered by
at least one individual than is unique information known
only to one member [27]. Thus, groups are more likely to
focus discussions on common rather than unique
information, because common information is more likely
to be recalled. Yet in a hidden profile task, it is the unique
information that is critical to success; without adequate
consideration of the unique information, group members
will miss the optimal decision.
2.2.2. Information exchange.
The exchange of
information in verbal discussion is not perfect. Members
must first decide to contribute the information, and then
have the opportunity to contribute it. Individuals are often
motivated to defend or support their initial preference, so
the information they choose to contribute often favors their
preferences or challenges an alternative [26]. Information
contradicting the preferences of the majority is also less
likely to be exchanged [9], because members may fear a
negative evaluation of their comments [7, 13].
In verbal discussion, only one group member can speak
at one time, so all other members are blocked from
contributing. This "production blocking" significantly
reduces the exchange of information because members
who are prevented from contributing information as they
choose, often forget or suppress it because it seems less
relevant or less important later in the discussion [7, 13].
In a situation where a group majority favors one
alternative, the tendency to support one’s own choice will
result in a majority of members contributing information
that supports the majority choice. Likewise, the tendency
not to contribute information contradicting the group
majority (e.g., information challenging it, or favoring a
competing alternative) may also result in incomplete
information exchange. The result may be a very biased
exchange of information that favors the majority
alternative, further reinforcing its apparent superiority.
2.2.3. Information processing and use.
Once
information is exchanged, members must actually process
it before it can affect decisions. Petty and Cacioppo [21]
argue that there are two “routes” by which information is
processed. The first is the “central route,” in which
members actively assess the information and integrate it
into their overall understanding of the situation and their
preferences (also called persuasive arguments or
information influence). This theory argues that unique
information not previously known should be more
persuasive than information already known and considered
[18, 32].
However, empirical evidence has found the opposite:
unique information is more likely to be ignored, while
common information known to all members prior to
discussion has a greater influence on the groups decision
[28]. There are at least two plausible reasons. First,
during verbal discussions, members must constantly pay
attention to other members or miss their contributions, and
therefore the unique information first received during
discussion is processed only superficially and not
considered as fully as information studied before
discussion. Second, information that supports prediscussion choices is more thoroughly processed and
integrated into an individual’s cognitive schema [21],
while information challenging initial decisions is
discounted [14, 34].
The processing of information is also affected by the
presence of a majority opinion within the group. When
faced by a group majority whose preferences are different
from theirs, members assume the majority to be correct,
and focus on comparing their preferences to the majority's.
"The world of alternatives is reduced to two" [19, p. 25].
This stimulates a comparison process of convergent
thought, of viewing the issues from the majority's
preference [17]. In contrast, when faced by a minority
with a different preference, members are motivated to
assume that minority members are incorrect, and initially
dismiss them [15, 19]. However, if the minority is
consistent in presenting information and opinions, and
does not yield to majority pressure, members realize that
the minority is convinced of its position. They make
attempts to understand it -- a validation process of
divergent thought [17]. Information is re-examined from
many perspectives, not just that of the minority, to
understand why it has led to different conclusions [15, 19].
Information from minority members may be more
thoroughly processed than information from the majority
because minority members are perceived to be more
independent or more dissimilar from one another than are
members of the majority [10].
The second route by which information is processed is
the “peripheral route,” in which members’ preferences are
shaped more by peripheral cues such the number of people
arguing for a position or their attractiveness, rather than
the quality of the information itself (also called social
comparison or normative influence). This theory argues
that members are motivated to conform to the opinions of
others to preserve a favorable self-perception and self-presentation [18]. After obtaining information about
others' preferences, members may change their preference
to more closely match others, either through choice or
coercion [9]. Also, publicly stating a preference may make
that preference stronger; public commitment may make
changing preferences more difficult without losing face
[22] and may cause stronger internalization to reduce
cognitive dissonance [18].
In a majority/minority situation, normative influence is
usually quite strong [15]. Members may not actually
change their preference, but just state a more socially
acceptable preference than their true preference. They
may simply comply with the majority preference to avoid
further stress, without examining any information [15].
2.3. GSS effects on information exchange and use
GSS provide new opportunities for information
exchange that may differ in important ways from non-GSS
environments. One of these is electronic communication
that can augment or replace verbal communication. This
electronic communication provides a package of many
different components, each of which may introduce new
dynamics into the information exchange process. Three
components in particular should affect information
exchange and use: parallelism, group memory, and
anonymity [20].
The first, parallelism, is the ability of all group
members to enter information at the same time. All group
members have computer workstations, which enable them
to contribute information and opinions by typing ideas,
which are immediately shared with all other members.
Since all members can type at the same time, no
participant need wait for others to finish before
contributing information. This parallelism mitigates the
blocking that inhibits the exchange of information in
verbally interacting groups [31].
The second, group memory, means that all remarks
typed into the computer are stored so that members can
refer to them later in the discussion. One of the key
problems in verbal discussion is that listening to group
discussion blocks processing of new information, and
information processing blocks the receipt of new
information from discussion. A group memory enables
members to enter or read information at their discretion;
they can more easily pause to process information without
risk of missing information.
The third component is anonymity; group members
make contributions without attaching their names, which
may motivate them to participate differently. Anonymity
may reduce the reluctance to contribute information that
contradicts the dominant group preference [20]. Members
are also often motivated to defend or support their initial
preference, so the information they choose to contribute
favors their preferences or attacks an alternative [26, 27].
To the extent that this motivation stems from the need to
save face about a publicly advocated decision preference,
anonymity, which can obscure both the initial public
statement of preference as well as any subsequent
arguments for or against it, may reduce the tendency to
contribute information supporting one's initial preferences.
2.4. Summary and hypotheses
2.4.1. Information exchange. There is often an
incomplete exchange of information in verbal discussions,
which may lead to poor decisions. GSS use may increase
the amount of information exchanged because it provides
parallelism (which reduces blocking) and anonymity
(which may reduce the reluctance to contribute information
challenging one's own preference or that of the majority).
These factors are likely to affect common and unique
information equally. Therefore:
H1: Members of GSS groups will exchange more
common information and more unique information than
members of non-GSS groups.
In a majority/minority situation, there is more pressure
on minority members to accede to the majority,
encouraging less discussion of the available information;
they may simply agree with the majority without much
discussion because swaying the majority will be more
difficult than swaying the few members required in the
uniform situation. The shield provided by anonymity will
be more strongly felt when there is a majority/minority
within the group. Minority members who oppose the
preference favored by the group majority will be more
likely to oppose the majority viewpoint when they use a
GSS. Because they will present their opinions more
strongly and are more likely to persist in their opinions,
they are likely to exchange more information. Therefore:
H2: The difference in information exchange between
GSS and non-GSS groups will be the greatest in
majority/minority groups.
2.4.2. Information processing and use. As argued
above, GSS should better enable group members to
process information because it provides a group memory
that enables members to review information at will.
Previous GSS research has found GSS use to increase the
use of factual information rather than members’
preferences; that is, to increase information influence
rather than normative influence [5, 12].
H3: Members of GSS groups are more likely to
consider the information exchanged in making their
decisions than are members of non-GSS groups.
There will be different effects on information
processing between majority/minority groups and uniform
groups. Minority members are more likely to accede to the
majority without the equal time provided by parallelism
and the shield of anonymity. In GSS groups, minority
members are likely to present their opinions more strongly
and persist longer, and therefore are more likely to trigger
the intensive validation process of Moscovici [17].
Therefore:
H4: The difference in information use between GSS
and non-GSS groups will be the greatest in
majority/minority groups.
2.4.3. Outcomes. The discussion and use of more
information should increase the time required to reach a
decision compared to non-GSS groups, unless the GSS
enables groups to contribute and process information
faster. There is evidence that GSS use takes longer, in part
because GSS require participants to type, which is slower
than talking. [3]. Therefore:
H5: GSS groups will take longer to reach a decision
than non-GSS groups.
In a majority/minority situation, there is more pressure
on the minority members to accede to the majority,
encouraging less discussion and less consideration of
information. The equal time provided by parallelism and
the shield of anonymity will encourage minority members
to present their opinions more strongly and to persist in
their opinions, requiring more time. Therefore:
H6: The difference in time taken between GSS and nonGSS groups will be the greatest in majority/minority
groups.
In certain situations, such as hidden profile tasks,
greater use of unique information will enable groups to
better identify the optimal alternative. Anonymity may
also reduce the pressure to conform to others' preferences,
and reduce the loss of face in backing down from previous
positions, making it easier for members to change their
preferences. Therefore:
H7: When information distribution is biased, members
of GSS groups will make better decisions than members of
non-GSS groups.
In a majority/minority situation, the minority members
are more likely to accede to the group majority, unless the
parallelism and anonymity encourages more thorough
exchange and use of the available information. Therefore:
H8: The difference in decision quality between GSS
and non-GSS groups will be the greatest in
majority/minority groups.
3. Method
3.1. Subjects
One hundred and fifty sophomore, junior, and senior
business students drawn from a core business course
participated as members of 15 ten-person groups (5
majority/minority task groups; 10 uniform task groups).
The average age was 21 years; 56 % were male.
3.2. Independent variables
This study used a 2 x 2 repeated measures design,
crossing GSS use with the decision situation
(majority/minority or uniform). Each group was assigned
into either the majority/ minority or uniform condition, and
performed two tasks, one using a GSS and one without.
The order of the two tasks and the GSS and verbal
treatments were reversed for half the groups.
In the non-GSS treatment, subjects interacted verbally.
In the GSS treatment, subjects used TCBWorks, a GSS
developed at the University of Georgia [30] that provided
electronic communication with parallelism, anonymity, and
group memory. Subjects were instructed to use the
software to exchange information until their first vote.
After the first vote they were instructed to use both the
software and to discuss the issues verbally.
The majority/minority task was designed such that
seven subjects received a sub-set of information favoring
the same sub-optimal alternative, while three received a
sub-set of information favoring the optimal alternative. In
the uniform task three subjects received a sub-set of
information favoring the optimal alternative, four received
a sub-set of information favoring a sub-optimal the
alternative and three received information favoring a
different sub-optimal alternative. Any session in which the
tasks did not produce the intended majority/minority
(defined as at least six members favoring the sub-optimal
alternative) or an uniform split (defined as no more than
five members favoring the sub-optimal choice and at least
three favoring the optimal choice) for both tasks was
discarded. We studied 21 groups, but only 15 met these
criteria and were used in the study; i.e., the intended task
split failed in one of the two tasks for six groups (a failure
rate of six in 42 tasks or 14%).
3.3. Tasks
There were two tasks. Subjects performed both tasks,
one in the GSS condition and the other in the non-GSS
condition. Subjects were informed at the start of each task
that they each had only a subset of all available
information and would have to share their information to
make a good decision. Both tasks underwent extensive
design and pilot testing for both the complete information
version and the incomplete information versions of the task
given to each individual subject (i.e., at least six rounds of
redesign and pilot testing and as many as ten for one set of
incomplete information versions that proved difficult to
refine).
One task asked subjects to select one student from a set
of four students for admission to the university. The
complete information set included positive, negative, or
neutral information on 18 criteria for each alternative for a
total of 72 pieces of information. Five pieces of
information for all four alternatives (verbal SAT, math
SAT, GPA, sex, major) were "common" -- i.e., given to all
subjects. The remaining information was "unique."
This task was designed with the assistance of the
university Admissions Office. The complete information
version of the task was independently validated by three
admissions officers (the director of admissions and two
associate directors) to ensure the alternative designed to be
optimal, was optimal. The complete information version
was also validated by 28 subjects drawn from the same
subject pool but who did not participate in the study.
Twenty-two (79%) chose the optimal choice; five (19%)
chose next alternative; one (2%) chose the third
alternative.
The second task asked subjects to select a computer to
be the “standard” computer required of all College of
Business students from a set of four computers. The
complete information set included positive, negative, or
neutral information on 18 criteria for each alternative for a
total of 56 pieces of information. Four pieces of
information for all four alternatives (price, CPU, RAM,
hard drive size) were "common."
The remaining
information was "unique."
This task was designed in conjunction with the
College’s computer lab staff. The chair of the College’s
computer users committee (who was also a lab manager)
and two instructors independently validated the task to
ensure the alternative designed to be optimal, was optimal.
The complete information version of the task was validated
by 21 subjects drawn from the same subject pool but who
did not participate in the study. Sixteen (76%) chose the
optimal choice; four (19%) chose the second best
alternative; one (5%) chose the third alternative.
3.4. Procedures
The procedures followed those of Stasser and Titus
[26]. Subjects first read the task and made an individual
decision. Then they worked together as a group (using
either a GSS or no GSS) to arrive at a unanimous group
decision or until 25 minutes had elapsed. (Some subjects
using the GSS may have been less experienced with
computers than others, but this was not seen as a hindrance
since groups using GSS were trained on the GSS before
starting the group task. These subjects were shown how to
add and view comments using the GSS.) Subjects then
individually completed the post-session questionnaire that
asked them to make another individual decision (the same
as or different from the group decision) and to report
several perceptions. After a five-minute break, these steps
were repeated for the second treatment.
3.5. Measures
3.5.1. Information exchange.
The amount of
information exchanged was measured by three raters, who
counted the pieces of correct information in the group
discussions (video-tapes for non-GSS treatments; printed
computer records and video-tapes for GSS treatments).
The video camera failed during the last three sessions (it
appeared to be recording but did not), so only the data for
the 12 sessions for which the video tapes were available
were used in these analysis). A set of six randomly
selected groups (three uniform, three majority/minority)
was selected to determine inter-rater reliability. The raters
agreed on .94 of the ratings (measured as 1 - (number of
disagreements/information
exchanged)),
indicating
adequate agreement.
3.5.2. Information processing and use. Determining
information processing and use was done in three ways.
First, a perceptual measure was used. The post-session
questionnaire included two scales designed to measure
perceived information usage: “I-thought” (i.e., the extent to
which the respondent thought about the information
contributed by others, 3 items, Cronbach alpha=.84,
1=low, 7=high), and “others-thought” (i.e., the extent to
which the respondent believed other group members
thought about the information contributed by respondent, 3
items, alpha=.85, 1=low, 7=high). While perceptions of
information usage can be unreliable, they are one way of
triangulating on this concept [21].
Second, information use was considered by examining
the information recalled by each group member at the end
of the group discussions [4]. Information recalled at the
end of the experiment must have been received and
processed during discussions because information must be
processed to be stored in memory [27]. Subjects were
asked to list all information they remembered in a free
recall question on the post-session questionnaire. Two
raters coded the questionnaires and counted the number of
correct information provided. Inter-rater agreement was
again adequate (.93).
Third, information use was considered by examining
the unique information learned by each group member at
the end of the group discussions. Information learned is the
information subjects recalled at the end of the experiment
that they did not have in their initial task descriptions [4].
This was done using the information coded above, but
counting only the information that the subject did not have
at the beginning of group discussion.
3.5.3. Outcome measures.
Decision time was
measured by the number of minutes required for the group
to reach a decision. Decision quality was measured by the
number of groups making the optimal decision. Three
additional perceptual measures were also included on the
post-session questionnaire, all with scales of 1=low,
7=high: satisfaction (4 items, alpha=.83), apprehension (4
items, alpha=.78), and information credibility (i.e., the
degree of accuracy that subjects attributed to information
contributed by others, 4 items, alpha=.82).
4. Results
4.1. Information exchange
H1 was partially supported. Groups exchanged more
unique information when using the GSS than when not but
there was no difference in the amount of common
information exchanged (see Tables 1 and 2 for the means
and all statistical results). H2 was not supported. There
was no interaction between the majority/minority status of
the groups and GSS use.
We used three measures of information use. The first
were perceptual data. Participants perceived themselves
and others to have thought about information to a lesser
extent when using the GSS than when not. There were
interaction effects indicating that members of the
majority/minority groups perceived themselves and others
to have thought less about the information than members
of the uniform groups.
The second measure was the recall of information.
There were significant interaction effects on the recall of
both common and unique information between the use of
GSS and the majority/minority and uniform treatments.
Participants in the uniform treatment recalled less common
and less unique information when using the GSS. Members
of majority/minority groups recalled more common
information when using the GSS, but there were no
differences in the recall of unique information.
The third measure was the recall of unique information
unknown at the start of the discussions (i.e., the
information learned from the discussion). Participants
learned less initially unknown information when working
with the GSS, suggesting that they were less likely to
process previously unknown unique information received
during the discussion.
4.2. Information processing and use
Table 1: Means and Standard Deviations
Uniform
GSS
Verbal
Group Measures
mean
std
mean
std
Common information exchanged
9.40 4.84
5.88
3.18
Unique information exchanged
12.30 2.63
5.00
3.55
Common information recalled
3.91 3.67
4.58
3.61
Unique information recalled
2.12 2.10
3.78
2.86
Information learned
1.19 1.44
2.32
2.25
Time
1400
205
422
227
Decision quality
0.10 0.32
0.80
0.42
Subject Perceptions (1= low, 7=high)
I thought
5.25 1.52
5.63
1.19
Others thought
4.83 1.30
5.12
1.31
Satisfaction
5.44 1.33
5.68
1.18
Apprehension
1.97 1.12
2.52
1.28
Information credibility
4.21 1.65
4.80
1.42
Majority/minority
GSS
Verbal
mean
std
mean
std
10.20
6.05
8.25 1.71
13.40
7.83
12.25 4.35
5.28
4.04
4.20 2.93
3.54
4.63
3.76 4.53
2.12
3.11
2.72 2.84
943
331
848
603
0.60
0.55
0.00 0.00
4.92
4.91
5.85
1.68
4.17
1.75
1.38
1.10
0.78
1.80
6.01
5.75
6.11
1.96
5.17
Table 2: Statistical Results
Group Measures
Common information exchanged
Unique information exchanged
Common information recalled
Unique information recalled
Information learned
df
8
8
146
145
145
GSS
F
p
1.91
ns
7.54 .025
0.34
ns
10.12 .002
15.28 .001
Minority
F
p
1.17
ns
1.06
ns
0.01
ns
1.68
ns
0.28
ns
GSS x Minority
F
p
0.03
ns
3.28
ns
6.36 .013
9.80 .001
2.46
ns
0.91
0.98
0.75
1.03
1.42
Time
Decision quality
Subject Perceptions
I thought
Others thought
Satisfaction
Apprehension
Information credibility
11
11
16.11
0.17
.002
ns
0.22
1.41
ns
ns
7.69
21.48
.018
.001
146
146
146
146
146
22.42
18.19
4.95
14.66
21.63
.001
.001
.028
.001
.001
0.19
2.48
5.00
3.72
0.30
ns
ns
.027
ns
ns
4.99
4.83
0.08
1.15
1.05
.027
.030
ns
ns
ns
These three measures indicate that groups in the
uniform treatment were less likely to process and use the
information present in their discussions when using the
GSS. The perceptual and learning measures suggest that
participants in majority/minority groups were less likely to
process and use the information when using the GSS, but
the recall measures suggest otherwise.
H3 is not
supported. H4 is partially supported.
4.3. Outcomes
There was a main effect on time due to GSS use, but
also an interaction effect. Groups took longer to reach a
decision when using a GSS in the uniform condition, but
there were few differences in the majority/minority
condition. H5 was partially supported, but H6 was not.
The effects on decision quality were rather unusual,
producing an interaction effect but no main effects. In the
majority/minority treatment, groups made significantly
better decisions when using the GSS. The reverse was true
for the uniform condition: groups made significantly better
decisions when not using the GSS. H7 was not supported,
but H8 was supported.
Participants in the uniform treatment reported less
satisfaction, less apprehension, and less credibility in the
information discussed when using the GSS than when
interacting verbally. Participants in the majority/minority
treatments reported more satisfaction but no differences in
apprehension or information credibility.
5. Discussion
The objective of this study was to examine the
information exchange and use processes in group decision
making. The results were rather surprising.
5.1. Majority/minority groups
In groups that had a pre-discussion majority favoring a
sub-optimal alternative, verbal interaction without a GSS
led to poor decisions: groups almost uniformly choose the
sub-optimal alternative favored by the group majority. The
exchange of unique information key to the optimal
decision was done poorly; very little of the information
needed to identify the optimal alternative was contributed.
The final group decision depended more on the biased
information members received before the meeting than the
unique information first exchanged during the meeting.
For these groups, the use of GSS added considerable
value. When using the GSS, these same groups exchanged
more unique information and almost uniformly made the
optimal decision in no more time than when not using the
GSS.
Without the GSS, normative influence dominated, and
the group majority convinced the minority to accede to its
preference.
With the GSS, information influence
dominated, and the minority was able to induce the
majority to listen to the information presented and to
convince them to change their minds to the optimal
alternative. We attribute the value of the GSS to its ability
to provide parallelism, anonymity, and group memory.
5.2. Uniform groups
The situation was quite different in the uniform
treatment. When not using the GSS, members exchanged
only a small portion of the available information, appeared
to think about it more, and yet generally quickly found the
optimal decision. When using the GSS, members
exchanged more information, but appeared to think about
it less, took a longer time, and made a sub-optimal
decision or could not agree on a unanimous decision.
In short, in the uniform treatment, GSS use impaired the
groups’ decision making process. We believe that when
there was no majority in the group, the groups tended to
focus more on the information than on peripheral
normative influence cues. Without the GSS, the groups
quickly shared the critical unique information, discussed it,
and identified the optimal decision (one group doing so in
less than two minutes). Three pieces of evidence suggest
that the problem with GSS use in the uniform condition
was that participants were less likely to have processed and
used the information they received during group discussion
when using the GSS: participants perceived themselves
and others to have used less information, recalled less
common and less unique information, and learned less
information when using the GSS than when not using it.
The question, is why did this occur?
5.3. Lack of information processing
The reasons for the inhibited information processing in
uniform groups when using the GSS is unclear. We offer
three plausible interpretations for future research:
information integration, information credibility, and lack
of focus.
5.3.1. Information integration. One possibility may lie
in group members' abilities to integrate newly received,
unique information into their existing base of information.
During discussion, group members must engage in three
simultaneous activities (information recall, information
exchange, and information use), and because humans have
a limited amount of cognitive resources, engaging in any
one activity limits the ability to engage in the other two.
Factual information requires more cognitive processing
than information about others' preferences [21].
Streufert and Swezey [29] argue that if factual
information arrives faster than one item every three
minutes, individuals have difficulty integrating it with
existing information. While we do not believe the volume
of information in these studies was very high [27], it
clearly required the processing of more than one piece of
information every three minutes. In the uniform treatment,
there were three competing alternatives (versus two in the
majority/minority treatment), and no minority advocating
its opinions to focus the analysis on just two alternatives.
There was simply more to consider.
When using the GSS, members generated about twice
as much information as when not using it. Therefore, one
plausible explanation for the lack of information
processing may lie in the cognitive limitations of
individuals' ability to integrate information. Participants
may have suffered from information overload and an
inability to focus their analysis on the dominant
alternatives to a greater extent when using the GSS and
therefore failed to process the information it contained as
thoroughly as information received in the verbal
discussions. By limiting the amount of information
available, the production blocking in the non-GSS groups
may actually have helped by filtering the available
information [11].
Under this interpretation, there are two implications for
future research. First, we need research varying the
information load to determine if GSS use helps or hinders
in situations of high and low information load. Second, we
need to better understand how to structure the information
in the GSS to minimize overload and enable better
integration of information.
5.3.2. Information credibility. Another possibility for
the reduced processing of unique information in GSS
groups may lie in the credibility of the information. Morley
and Walker [16] suggest that to cause a change in a
previous decision (in this case, the individual decision
made prior to group discussion), information must be
important, novel (i.e., not previously processed), and
credible. In this case, the unique information received
during discussion was both important and novel.
However, participants were less likely to find it credible
when using the GSS, and unlike in the majority/minority
treatment, there was no subgroup lobbying hard for a
specific alternative.
Thus participants may have
intentionally disregarded the information in the GSS.
There are at least two plausible explanations for this
reduced credibility.
The first is a negative anonymity effect on credibility.
The credibility of the source of information can be
important to the acceptance and processing of a message,
particularly when the information is ambiguous or difficult
to process [21]. It may be that the anonymity of the GSS
used in this study made the information more suspect,
because it was more difficult to verify the source's
credibility. Anonymity may also conceal the number of
information sources. Information perceived to be from
several independent sources is seen as more credible than
information from one source [10, 31]. Thus information
from several distinct individual sources in the non-GSS
groups may have been seen as more credible than the pool
of information from an unknown number of sources in the
GSS.
A second reason for reduced credibility may be a
reduced ability to challenge the information’s contributor.
In most non-GSS groups, we observed at least one (usually
several) occasions when unique information presented to
the group was immediately challenged (e.g., "Really? I
didn't get that."). The contributor would then confirm the
validity of the information and discussion would move on.
Similar challenges were seen in GSS groups, but the
response would come much later or not at all. It may be
that the reduced ability to provide immediate confirming
feedback reduces the credibility of the information.
Under this interpretation, we need additional research
to determine if anonymity is useful in this situation.
Anonymity has been found to be important for groups
whose members have different power and status, but to
have fewer benefits for peers [20].
5.3.3. Information focus. A third possibility may be
the ease of ignoring the information because its
presentation is unfocused. When a person speaks verbally
and contributes information it may focus the group's
attention on that information [10] -- it can be difficult to
ignore a person speaking right in front of you. In contrast,
when using a GSS, information is contained in lines of text
in a pool of information. Without a GSS, information is
presented to you; with a GSS, you must actively peruse the
group memory. When using the GSS, participants may not
have actively searched out and reviewed all available
information.
Under this interpretation, we need research to discover
how GSS can better engage participants and encourage
them to actively seek and use the information it contains
(i.e., a GSS like a video game that draws participants in).
Systems," presented at INFORMS, Singapore, 1995.
5.4. Implications for practice
[4] Cacioppo, J.T., Petty, R.E. and Morris, K. “Effects of need
for cognition on message evaluation, recall, and persuasion,”
Journal of Personality and Social Psychology, (45) 1983, 805818.
This study has several implications for managers
striving to understand and improve the exchange and use
of information in organizational problem solving. For
groups that had a majority/minority, GSS use improved the
exchange of information and the quality of the decision.
However, for groups without a majority/minority, GSS use
impaired the quality of decision making. One implication
is to use a GSS only when there are strong differences of
opinion within a group, where the GSS’s ability to focus
the group on the information will be valuable.
We believe that the key problem in using this
information for both GSS and non-GSS groups was the
difficulty in assessing and integrating the new information
received during group discussion into their existing
information base, which encouraged members to overlook
this information. A second implication is therefore to
structure group meetings to provide sufficient time for
members to study, assess, and integrate this new
information before they must make decisions. Separating
information exchange from the decision making that uses it
is one option. Members could hold one meeting to discuss
a problem or opportunity, break for several hours or days
to evaluate and integrate the information, and hold a
second meeting to make the actual decision. This structure
is similar to the "incubation" period recommended for idea
generation.
A third implication deals with knowledge of having all
of the information. We informed subjects that they had
incomplete information and would need to share
information. In a meeting, participants do not know if they
have complete information. Do they assume that they do
have all of the information needed? The answer to this may
change how GSS is used or not used.
Clearly another alternative is to develop new group
processes and GSS tools to help members better integrate
information, but this will have to wait for more research.
6. References
[1] Ball, S.A. and Zuckerman, M. “Sensation Seeking and
Selective Attention: Focused and Divided Attention on a
Dichotic Listening Task,” Journal of Personality and Social
Psychology, (63) 1992, 825-831.
[2] Briggs, R.O. "Focus Theory: An Explanation of Group
Productivity for Developers and Users of Electronic Meeting
[3] Benbasat, I. and Lim, L.H. “The Effects of Group, Task,
Context, and Technology Variables on the Usefulness of Group
Support Systems: A Meta-Analysis of Experimental Studies,”
Small Group Research, (24), 1993, 430-462.
[5] Clapper, D.L., McLean, E.R. and Watson, R.T. “An
Experimental Investigation of the Effect of a Group Decision
Support System on Normative Influence in Small Groups,”
Proceedings of the 12th International Conference on Information
Systems, New York, 1991, 77-87.
[6] DeSanctis, G.L. and Gallupe, R.B. “A Foundation for the
Study of Group Decision Support Systems,” Management
Science, (33) 1987, 589-609.
[7] Diehl, M. and Stroebe W. “Productivity Loss in
Brainstorming Groups: Toward the Solution of a Riddle,”
Journal of Personality and Social Psychology, (53), 1987, 497509.
[8] Gigone, D. and Hastie, R. “The Common Knowledge
Effect: Information Sharing and Group Judgment,” Journal of
Personality and Social Psychology,” (65), 1993, 959-974.
[9] Hackman, J.R and Kaplan, R.E. “Interventions into Group
Process: An Approach to Improving the Effectiveness of
Groups,” Decision Sciences, (5), 1974, 459-480.
[10] Harkins, S.G. and Petty, R.E. “Information Utility and the
Multiple Source Effect,” Journal of Personality and Social
Psychology, (52), 1987, 260-268.
[11] Hiltz, S.R. and Turoff, M. “Structuring ComputerMediated Communication Systems to Avoid Information
Overload,” Communications of the ACM, (28:7), 1985, 680.
[12] Huang, W., Raman, K.S., and Wei, K.K. “A process
Study of the Effects of GSS and Task Type on Informational
and Normative Influence in Small Groups,” Proceedings of the
14th International Conference on Information Systems,
Orlando, 1993, 77-87.
[13] Lamm, H. and Trommsdorff, G. “Group versus Individual
Performance on Tasks Requiring Ideational Proficiency
(Brainstorming): A Review,” European Journal of Social
Psychology, 1973, 361-387.
[14] Lord, C.G., Ross, L., and Lepper, M.R. “Biased
Assimilation and Attitude Polarization: The Effects of Prior
Theories on Subsequently Considered Evidence,” Journal of
Personality and Social Psychology, (37), 1979, 2098-2109.
[15] Maass, A. and Clark, III, R.D. “Hidden Impact of
Minorities: Fifteen Years of Minority Influence Research,”
Psychological Bulletin, (95), 1984. 428-450.
Organizations, 1986, Academic Press, Orlando.
[30] TCBWorks at http://tcbworks.mgmt.uga.edu:8080.
[16] Morley, D.D. and Walker, K.B. “The Role of Importance,
Novelty, and Plausibility in Producing Belief Change,”
Communication Monographs, (54), 1987, 436-442.
[17] Moscovici, S. “Toward a Theory of Conversion
Behavior,” in L. Berkowitz (ed.), Advances in Experimental
Social Psychology, Volume 13, New York: Academic Press,
1980, 209-239.
[18] Myers, D.G. and Lamm, H. “The Group Polarization
Phenomenon,” Psychological Bulletin, (83), 1976, 602-627.
[19] Nemeth, C., “Differential Contributions of Minority and
Majority Influence," Psychological Review, (93), 1986, 1-10.
[20] Nunamaker, Jr., J.F., Dennis, A.R., Valacich, J.S., Vogel,
D.R., and George, J.F. “Electronic Meeting Systems to Support
Group Work,” Communications of the ACM, (34:7), 1991, 4061.
[21] Petty, R.E. and Cacioppo, J.T. Communication and
Persuasion, Springer-Verlag, 1986. New York.
[22] Salancik, G.R. “Commitment and the Control of
Organizational Behavior and Belief,” in B.M. Staw and G.R.
Salancik (Eds.), New Directions in Organizational Behavior,
Chicago: St. Clair, 1977, 1-54.
[23] Shaw, M. Group Dynamics: The Psychology of Small
Group Behavior, (3rd ed.) 1981, New York: McGraw Hill.
[24] Stasser, G. “Information Salience and the Discovery of
Hidden Profiles by Decision-Making Groups: A ‘Thought
Experiment,’” Organizational Behavior and Human Decision
Processes, (52), 1992, 156-181.
[25] Stasser, G. and Stewart, D. “Discovery of Hidden Profiles
by Decision-making Groups: Solving a Problem versus making a
Judgment,” Journal of Personality and Social Psychology, (63),
1992, 426-434.
[26] Stasser, G. and Titus, W. “Pooling of Unshared
Information in Group Decision making: Biased Information
Sampling During Group Discussion” Journal of Personality
and Social Psychology, (48), 1985, 1467-1478.
[27] Stasser, G. and Titus, W. “Effects of Information Load
and Percentage of Shared Information on the Dissemination of
Unshared Information During Group Discussion," Journal of
Personality and Social Psychology, (53), 1987, 81-93.
[28] Stasser, G., Taylor, L.A., and Hanna, C. “Information
Sampling in Structured and Unstructured Discussions of Threeand Six-Person Groups” Journal of Personality and Social
Psychology, (57), 1989, 67-78.
[29] Streutfert, S. and Swezey, R.S. Complexity, Managers, and
[31] Valacich, J.S., Dennis, A.R., and Connolly, T. “Group
Versus Individual Brainstorming: A New Ending to an Old
Story," Organizational Behavior and Human Decision
Processes, (57), 1994, 448-467.
[32] Vinokur, A., Trope, Y., and Burnstein, E. “A Decision
Making Analysis of Persuasive Argumentation and the ChoiceShift Effect,” Journal of Experimental Social Psychology, (11),
1975, 127-148.
[33] Wilder, D.A. “Some Determinants of the Persuasive Power
of In-Groups and Out-Groups: Organization of Information and
Attribution of Independence,” Journal of Personality and Social
Psychology, (59), 1990, 1202-1213.
[34] Wood, W. “Retrieval of Attitude Relevant Information from
Memory: Effects on Susceptibility to Persuasion and on Intrinsic
Motivation,” Journal of Personality and Social Psychology,
(42), 1982, 798-810.
Download