Group diversity and decision quality: Amplification and attenuation of the framing effect

Article (PDF Available)inInternational Journal of Forecasting 27(1):41-49 · March 2011with400 Reads
DOI: 10.1016/j.ijforecast.2010.05.009
  • 28.95 · Hebrew University of Jerusalem
Abstract
Do groups make better judgments and decisions than individuals? We tested the hypothesis that the advantage of groups over individuals in decision-making depends on the group composition. Our study used susceptibility to the framing effect as a measure of decision quality. Individuals were assigned to one of two perspectives on a choice problem. The individuals were asked to indicate their individual preference between a risky option and a risk-free option. Next, they were asked to consider the same (or a related) choice problem as a group. Homogeneous groups were composed of similarly framed individuals, while the heterogeneous ones were composed of differently framed individuals. In comparison to individual preferences, the homogeneous groups’ preferences were polarized, and thus the framing effect was amplified; in contrast, the heterogeneous groups’ preferences converged, and thus the framing effect was reduced to zero. The findings are discussed with regard to group polarization, the effects of heterogeneity on group performance, and the Delphi forecasting method.

Figures

International Journal of Forecasting 27 (2011) 41–49
www.elsevier.com/locate/ijforecast
Group diversity and decision quality: Amplification and
attenuation of the framing effect
Ilan Yaniv
Department of Psychology & Center for the Study of Rationality, Hebrew University of Jerusalem, 91905, Israel
Abstract
Do groups make better judgments and decisions than individuals? We tested the hypothesis that the advantage of groups
over individuals in decision-making depends on the group composition. Our study used susceptibility to the framing effect as
a measure of decision quality. Individuals were assigned to one of two perspectives on a choice problem. The individuals were
asked to indicate their individual preference between a risky option and a risk-free option. Next, they were asked to consider
the same (or a related) choice problem as a group. Homogeneous groups were composed of similarly framed individuals,
while the heterogeneous ones were composed of differently framed individuals. In comparison to individual preferences, the
homogeneous groups’ preferences were polarized, and thus the framing effect was amplified; in contrast, the heterogeneous
groups’ preferences converged, and thus the framing effect was reduced to zero. The findings are discussed with regard to group
polarization, the effects of heterogeneity on group performance, and the Delphi forecasting method.
c
2010 International Institute of Forecasters. Published by Elsevier B.V. All rights reserved.
Keywords: Homogeneous vs heterogeneous groups; Risk attitudes; Group polarization; Delphi method
1. Introduction
The study of group decision processes is of great
interest for both theoretical and practical reasons.
First, it provides new vantage points for looking into
the mind of the individual and for improving our
understanding of the origins of judgment and choice
biases (Bornstein & Yaniv, 1998; Kerr & Tindale,
2004). Second, it sheds light on the mode of operation
of expert governmental panels, university committees,
E-mail address: ilan.yaniv@huji.ac.il.
managerial business teams, and other groups which
are charged with the mission of forecasting social,
economic and environmental changes, and shaping the
policies of their organizations in response.
1.1. Group composition and diversity
The goal of the present research was to investigate
the effects of group composition on the quality of
group decisions. The effects of group diversity have
been of great interest in organizational research on
the performance of work teams, as well as in the
experimental study of group processes and outcomes
0169-2070/$ - see front matter
c
2010 International Institute of Forecasters. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.ijforecast.2010.05.009
42 I. Yaniv / International Journal of Forecasting 27 (2011) 41–49
in lab settings. Mannix and Neale (2005) reviewed
the literature on the advantages and disadvantages of
diversity in teams. Their review offers both favorable
and unfavorable views of the role of diversity. On
the negative side, diversity highlights members’ social
identities and social divisions, thereby increasing the
stereotyping of out-group members. Overall, such
processes tend to hurt the group performance. It
is important to note, though, that the evidence for
the unfavorable effects of diversity comes primarily
from studies involving work teams in organizational
settings, where members differ in both demographic
factors (chiefly gender, race, and age) and social
factors (such as seniority, tenure, and educational
background).
Favorable views of diversity tend to come
from information-processing approaches to the study
of group performance (Mannix & Neale, 2005).
Such studies tend to involve ad-hoc groups in
experimental settings (making their findings more
relevant to the present study), and find that diverse
teams are especially appropriate for tasks involving
innovation and the exploration of choices and new
opportunities (e.g. Sommers, 2006). Such tasks benefit
from the multiplicity of sources of information,
heterogeneous skills, and divergent perspectives.
Diverse perspectives sometimes create disagreement
among group members and therefore reduce members’
confidence, yet disagreement is often associated with
improved judgmental accuracy (e.g. Sniezek, 1992).
1.2. The framing paradigm
In our study we used the susceptibility to framing
effects as an index of the quality of group decisions.
The framing effect refers here to the phenomenon
that seemingly superficial changes in the description
of the choices can have a large effect on behavior.
In the classic demonstration of the effect on risky
choice (Kahneman & Tversky, 1984), a decision
problem is presented along with two choice options
(one risk-free and one risky), and the participants
are asked to select the more attractive of the two. A
standard finding is that individuals prefer the risk-free
option (over the risky one) significantly more often
when the options are framed in positive (gain) terms
than when the options are framed in negative (loss)
terms. According to Kahneman and Tversky, this
description-dependent preference for the risky option
(framing effect) violates the normative principle of
description invariance, whereby the rational choice
ought to remain invariant under superficial changes to
the descriptions of the options.
Earlier studies on the susceptibility of groups to
framing effects have yielded varied results. Neale,
Bazerman, Northcraft, and Alperson (1986) examined
risky-choice framing effects among individuals and
groups. They obtained the typical framing effect for
individual decisions, but a reduced framing effect
in group decisions. Paese, Bieser, and Tubbs (1993)
also examined risky-choice framing effects among
individuals and groups. They also found a framing
effect at the individual level, but found an increased
framing effect at the (homogeneous) group level, on
only two (of four) decision problems. Finally, a recent
study by Milch, Weber, Appelt, Handgraaf, and Krantz
(2009) found a framing effect among individuals,
but did not find evidence for either a reduced or an
increased framing effect in group choice, compared to
individual choice.
We used the framing paradigm to investigate the
effects of group composition on performance. This
research paradigm is akin, but not identical, to ones
used in the past (Milch et al., 2009; Neale et al.,
1986; Paese et al., 1993; Tindale, Sheffey, & Scott,
1993). With this paradigm, individuals are first asked
to indicate their choices between risk-free and risky
options, then they are convened and asked to reach
a decision as a group. Two kinds of groups are
created. Homogeneous groups are composed entirely
of members who had been assigned to the same
framing condition of the decision problem, while
heterogeneous groups are composed of members who
had been assigned different framings of the problem
at the individual stage. At the group level decision,
each group is given only one frame for consideration
(details below).
1.3. The present hypotheses
Since our study engaged ad hoc groups in an
experimental task, we hypothesized, in line with the
review by Mannix and Neale (2005), that higher
levels of diversity will have beneficial effects on
the performance. In what follows, we focus on
the psychological mechanisms that underlie our
I. Yaniv / International Journal of Forecasting 27 (2011) 41–49 43
expectations about the effects of group homogeneity
and heterogeneity on group performance. Using this
paradigm, we investigated the way in which the
composition of the group affects its susceptibility
to framing effects. In particular, we suggest below
that group deliberation amplifies the framing bias in
homogeneous groups, while it attenuates the bias in
heterogeneous groups.
1.3.1. Homogeneous groups
Previous research has shown that interactions in
homogeneous groups are conducive to the polarization
of opinions. Specifically, groups composed of like-
minded individuals tend to make judgments that are
more extreme than the average of the individual
members’ judgments (Isenberg, 1986). In one study,
for example, student participants were categorized
as conservatives or liberals based on a measure of
political attitude. Then groups of like-minded students
(either conservatives or liberals) were created. The
students in each group were to discuss and give
their opinions on various social and political issues.
The group discussion amplified the group members’
prior inclinations, increasing the gap between the
conservatives and liberals, a phenomenon called group
polarization (Myers & Lamm, 1976; for a recent
review see Sunstein & Hastie, 2008).
In our decision framing paradigm, group delibera-
tions among similarly framed group members were ex-
pected to polarize their predominant risk attitudes. The
framing of choices as gains and losses evokes com-
mon risk attitudes. Hence, under gain framing, groups
will endorse the safe option more strongly than the
average member’s rating; under loss framing, groups
will endorse the risky option more strongly than the
average member’s rating. Our expectations here are
in agreement with an earlier theoretical analysis by
Whyte (1989), which suggested that decision-making
groups faced with losses tend to be risk seeking, par-
ticularly if they share common values, goals, and per-
spectives. In summary, we expected the amplification
of the framing effect in groups composed of similarly
framed members (homogeneous).
1.3.2. Heterogeneous groups
Diversity could be defined in terms of numer-
ous different factors (demographic, background, func-
tional, and so on). In our study, the heterogeneous
groups were composed of differently pre-framed indi-
viduals. Each group included members who were ex-
pected to have diverging risk attitudes (on average),
based on the framing presented to them at the indi-
vidual level. The group level framing effect was ex-
pected to reduce in size, since members holding dif-
ferent (initial) risk attitudes (due to their different ini-
tial perspectives) were expected to mitigate each oth-
ers’ initial preferences through interaction. Since the
same mitigating influence was expected in either fram-
ing condition, the framing effect should be reduced or
eliminated.
1.4. Summary
Our study uses a simple manipulation of diversity
to test our hypotheses. Building on earlier research,
we will consider the following hypotheses. First,
we should observe the framing effect in individual
choices. Second, homogenous groups should make
more extreme (polarized) decisions than individuals,
thereby enhancing the framing effect. Third, the
group composition should have diverging influences
on the framing effect. Thus, while homogenous groups
should amplify the framing effect, heterogeneous
groups should attenuate it.
2. Method
2.1. Materials
We created a hypothetical decision scenario
involving a plan by the university administration
to raise tuition fees. The scenario describes two
alternative courses of action to confront the planned
tuition fee hike: a sure option (A) and a risky option
(B). The two options were modeled on Kahneman
and Tverskys (1984) gain/loss framing of the Asian
disease problem. In particular, they were framed
in terms of either paying or saving tuition money.
Consider first the pay (loss) frame (note that 1 Israeli
Shekel was about $0.30 at the time):
Imagine that you have been nominated to repre-
sent the Student Association at the university. The
administration has announced that due to govern-
mental budget cuts to higher education, the annual
tuition will be raised by 6000 IS. In view of the up-
coming negotiations with the association, a friend
44 I. Yaniv / International Journal of Forecasting 27 (2011) 41–49
of yours from the economics department has pre-
pared two alternative plans to cope with the ex-
pected tuition hike. The outcomes of these plans are
as follows:
If plan A is adopted, the students will pay an extra
4000 IS.
If plan B is adopted, then there is a probability of
1/3 that students will not have to pay any additional
tuition, and a probability of 2/3 that students will
have to pay the additional 6000 IS.
Note that this problem makes sense to Israeli
students, since tuition in major universities is subject
to multiparty negotiations. While the alternatives
above indicated how much students would be
expected to pay (lose) under each course of action,
the alternatives below indicate how much students
stand to save (gain) through negotiations with the
administration:
If plan A is adopted, students will save 2000 IS of
the planned tuition hike.
If plan B is adopted, then there is a probability of
1/3 that students will save the entire 6000 IS tuition
hike, and a probability of 2/3 that students will not
save anything.
The question was presented along with an 11-point
Likert scale anchored at 5 (preference for plan B)
and +5 (preference for plan A).
2.2. Procedure
Participants were recruited for a lab study in groups
of four. They were seated separately, and were each
given one of the two versions of the tuition negotiation
problem, with instructions to indicate their individual
preference between the options on the 11-point scale.
They were also asked to indicate their reasons for
their preferences. The individual questionnaires were
then collected, and the participants convened in groups
and were handed the instructions and questionnaire
materials for the group decision. They were told that
they would need to discuss the problem in order to
make a collective decision on the two options (using
the 11-point scale) within 10 minutes. No instructions
were given as to how they should come to the decision.
All of the groups ended their deliberations and made
their decisions within the allotted time period. They
also listed the reasons for their ratings. At the end they
were debriefed and thanked.
2.3. Design
Two factors were manipulated, group diversity
and framing. Half of the groups were homogeneous
and half were heterogeneous. Specifically, in the
homogeneous condition, similarly framed individuals
convened as a group to discuss and decide on the
tuition problem in the same frame they had each
considered alone. Half of the homogeneous groups
were given (individually and as groups) the pay (loss)
frame; the other half of the groups were given the save
(gain) frame.
In the heterogeneous condition, the groups were
composed of differently framed members. Each group
included two members who had previously considered
the problem framed in negative terms (tuition money
paid) and two who had previously considered the
problem framed in positive terms (tuition money
saved). For half of the heterogeneous groups the
tuition problem to be considered at the group level was
framed in negative terms (money paid), while for the
other half the problem was framed in positive terms
(money saved).
2.4. Participants
One hundred and sixty Hebrew University students
were recruited (as part of course requirements) in
groups of four to participate in a lab experiment on
group decision-making. Thus, 40 participants were
allocated to each of the four conditions, creating 10
four-member groups in each of the four conditions.
3. Results
3.1. Framing effects
The dependent measure was based on the
individual and group ratings on the 11-point scale
(from 5 to +5). Higher ratings indicated a stronger
preference for the sure option (A). Participants found
the sure option more attractive in the save (gain)
than in the pay (loss) frame (the means were 2.22
and 1.02, respectively), t (158) = 2.50, p < 0.05.
Thus, the framing effect was evident in the individual
responses. The framing effect was also evident
when the individual data were divided into two
subsets, based on their subsequent assignment to
I. Yaniv / International Journal of Forecasting 27 (2011) 41–49 45
Table 1
Homogeneous group polarization of the framing effect (n = 80).
Decision maker Framing of tuition problem
Save Pay Magnitude of effect
Individuals 2.73 1.28 1.45
Groups 3.50 0.50 3.00
the homogeneous and heterogeneous groups. The
ratings of participants who were later assigned to the
homogeneous condition averaged 2.73 and 1.28 under
the save and pay framings, respectively, t (78) = 2.27,
p < 0.05. The ratings of participants assigned to
the heterogeneous condition averaged 1.73 and 0.70
under the save and pay framings, showing a marginally
significant trend in the expected direction, t (78) =
1.45, p < 0.08 (one tail).
3.2. Polarization: comparing individuals and groups
For the next analysis we considered the data from
the homogeneous condition alone (n = 80). The
magnitude of the framing effect is assessed as the
difference between the mean ratings under the two
frames, as shown in Table 1. The ratings of the sure
option averaged 2.73 under the save framing and
1.28 under the pay framing. The mean group ratings
averaged 3.50 under the save framing and 0.50 under
the pay framing. Thus, compared with the individual
ratings, the respective group ratings were closer to the
A and B poles of the preference scale. In other words,
the group judgments were more polarized than the
individual judgments. The significance of these two
polarization effects was assessed by t -tests. In the save
(gain) frame, the group ratings were higher than the
individual ratings, t (39) = 2.09, p < 0.05, whereas
in the pay (loss) frame, the group ratings were lower
than the individual ratings, t(39) = 1.31, p < 0.1
(one tail), as predicted by the polarization hypothesis.
3.3. Effects of group diversity
Next we compared the magnitude of the framing
effect as a function of the group diversity. The
mean ratings and the sizes of the framing effect
(see Table 2) suggest that homogeneous groups
(similarly framed members) enhanced the framing
effect (magnitude = 3.00), whereas heterogeneous
groups (differently framed members) lowered the
Table 2
Framing effects in heterogeneous and homogeneous groups (n =
160).
Group composition Framing of tuition problem
Save Pay Magnitude of effect
Homogeneous 3.50 0.50 3.00
Heterogeneous 1.20 1.10 0.10
framing effect to almost zero (magnitude = 0.10).
A two-way analysis of the variance was conducted,
with the problem framing (save/pay) and group
composition (homogeneous/heterogeneous) as factors.
We found a framing effect, F(1, 37) = 3.28, p < 0.1,
and an interaction between group composition and
framing, F(1, 37) = 2.87, p < 0.1. In particular, the
interaction effect suggests that the framing effect was
less in the heterogeneous than in the homogeneous
condition.
In summary, the trends in the individual and group
data are orderly and coherent with our theoretical
framework, although they tend to be weaker (sig-
nificant at the 10% level). The study involved 160
participants, but there were only 40 group data points
for testing the group level hypotheses, since each
group produced only one data point. Future research
should involve larger samples or engage methods that
allow for the collection of multiple observations per
group.
3.4. Qualitative analysis of reasons
Qualitative analyses of the verbal explanations for
the preference ratings were conducted in order to
shed light on the decision processes. Based on a
preliminary analysis of the explanations, four types
of reasons were identified. Two coders coded each
written explanation for the occurrence of these types.
The agreement rate between the two coders was 84%.
3.4.1. Individual decisions
Table 3 shows the frequencies of the types of
reasons given by participants for their individual
decisions. Reasons involving risk aversion (e.g., “Prob-
ability of one third is too low”, “I would not want
to bet on other people’s money”) were mentioned by
76% of participants in the save frame, and 46% of the
participants in the pay frame. Reasons involving risk
seeking (“I would rather take the risk than pay”, “The
46 I. Yaniv / International Journal of Forecasting 27 (2011) 41–49
Table 3
Frequencies of the various types of reasons indicated by individuals
for their decisions.
Type of argument Framing of tuition problem
Save (n = 80) Pay (n = 80)
Risk avoidance 61 37
Risk seeking 15 25
Equal expectation 4 16
Other arguments 0 2
possibility of avoiding the tuition hike is tempting”)
were given by 19% and 31% of the participants in the
save and pay conditions, respectively. The frequen-
cies of these two types are in accord with the quan-
titative data suggesting a greater risk aversion under
the save framing. Reasons suggesting equal expecta-
tions (e.g., “Both alternatives are the same in expected
value”) were given by 5% and 20% of the participants
in the save and pay conditions, respectively. Finally,
other reasons (“Both options are bad for the students”)
occurred infrequently (0% and 3% in the save and pay
frames, respectively).
3.4.2. Group decisions
The distribution of the reasons given by the groups
is shown in Table 4. Since more than one type of
reason was mentioned in some cases, the frequency
counts of the various types could add up to more
than 100%. The homogeneous groups mentioned risk
aversion under the save framing more often than
under the pay framing (90% vs 50%); while they
mentioned risk seeking reasons far less under the save
framing than under the pay framing (0% vs 40%,
respectively). These large differences in frequencies
are consistent with the finding that the homogenous
groups polarized in their decisions. In contrast, the
heterogeneous groups mentioned risk aversion reasons
with almost the same frequency under the save and
pay framings (60% vs 70%, respectively); they also
mentioned risk seeking reasons about equally often
under the two framings (20% vs 30%). The fact that
the heterogeneous groups mentioned the two types of
reasons about equally often under the two framings is
consistent with our hypothesis that they gave weight
to both risk attitudes under either framing, thereby
attenuating (canceling) the framing effect.
Fig. 1. The magnitude of the framing effect in the different
conditions.
4. Discussion
Do groups make better judgments and decisions
than individuals? Our measure of decision quality
was the decision makers’ susceptibility to the framing
effect. We tested the idea that group decisions could be
predictably better or worse than individual decisions,
depending on the diversity of group members’
perspectives on the decision problem. Our findings
are summarized in Fig. 1 in terms of the magnitudes
of the framing effect in the individual and group
choices.
Consistent with our expectation, groups composed
of similarly framed individuals (homogeneous) re-
vealed a stronger framing effect than individuals.
These findings follow, as we suggested, from the group
polarization effect (Myers & Lamm, 1976); that is,
the general tendency for group discussion to amplify
the prevailing pre-discussion inclinations. Indeed, the
preferences of homogeneous groups were more ex-
treme than the average of the group members’ individ-
ual ratings, such that groups endorsed more strongly
(than individuals) the safe option, under gain framing,
and the risky option, under loss framing.
On the other hand, groups composed of differently
framed members (heterogeneous), did not differ in
their preferences under the save (gain) and pay (loss)
framings of the tuition problem; thus, they showed
almost no framing effect (Fig. 1). Our explanation
in terms of group processes suggests that, although
in the group decision phase members were supposed
to consider a single framing of the tuition problem,
members’ initial risk attitudes actually varied—
I. Yaniv / International Journal of Forecasting 27 (2011) 41–49 47
Table 4
Frequencies of the various types of reasons indicated by groups for their decisions.
Type of argument Homogeneous Heterogeneous
Save (n = 10) Pay (n = 10) Save (n = 10) Pay (n = 10)
Risk avoidance 9 5 6 7
Risk seeking 0 4 2 3
Equal expectation 0 1 2 1
Other arguments 1 0 3 4
Note: The frequencies may add up to more than n = 10, since the groups indicated more than one kind of reason at times.
either risk-seeking or risk-avoidance—depending on
the frame presented to them during the individual
decision phase. Discussion among members espousing
opposite risk tendencies should naturally mitigate each
others’ preferences, producing a null framing effect.
Interestingly, our qualitative data (debriefing con-
ducted by the experimenters) suggest that only 3 out
of the 20 heterogeneous groups (15%) had noted that
equivalent decision frames were used. The decisions
of the groups in which the members had noticed the
equivalence were not different in any conspicuous way
from the decisions of those who did not notice the
equivalence. It appears that, whether groups notice the
equivalence or not, they still need to negotiate their
risk attitudes and the pros and cons of each option,
and hence, as far as we can tell, they engage in the
same deliberative process of reaching a decision.
The present findings join other findings in
the literature on the advantages of diversity in
organizations (Mannix & Neale, 2005; Sommers,
2006). Importantly, social actors believe that diversity
could be defined in terms of any factor that
makes a difference (demographic, personality traits,
background, professional education, organizational
position, and so on). In our study, heterogeneity was
created by assigning participants (who were otherwise
undifferentiated systematically) to different versions
or framings of the same problem. This minimal
manipulation was sufficiently potent to produce a
discernable impact on the decision quality, suggesting
that perspective-taking would potentially be a useful
manipulation in other settings as well.
4.1. Implications for group forecasting
The present study concerned the coherence rather
than the accuracy of group and individual choices.
However, we believe that the present results are rel-
evant for the study of judgmental forecasting in inter-
active (discussion) groups, as well as in Delphi groups
(Rowe & Wright, 1999). Our study suggests that inter-
active groups have an advantage over individual deci-
sion makers if their members’ perspectives are hetero-
geneous rather than homogeneous. That a similarity or
interdependence of opinion is detrimental to the per-
formance has previously been shown and discussed in
the forecasting literature. Indeed, recent work using an
advice-taking paradigm (e.g., Yaniv, 2004) has shown
that, when provided with advisory estimates that are
similar or correlated, decision makers’ confidence in-
creases, although their accuracy gains decrease (Yaniv,
Choshen-Hillel, & Milyavsky, 2009). Such dissocia-
tions of confidence and accuracy suggest that people
are not always aware of the extent to which “spurious
consensus” (defined as a set of agreeing interdepen-
dent opinions) might be detrimental for accuracy.
The problems of spurious consensus and po-
larization are relevant to the composition of ex-
pert panels, as was illustrated in a study by
Fraser, Pilpel, Kosecoff, and Brook (1994). These
researchers used Delphi groups to create medical
guidelines; specifically, indications for the appropri-
ateness of a cholecystectomy (surgical removal of
the gallbladder). Fraser et al. found that an ex-
pert panel composed of surgeons approved more
indications for surgery than a mixed expert panel com-
posed of family physicians, gastroenterologists, in-
ternists, and surgeons. The policy recommendation of
the surgeons’ panel would have led to more opera-
tions, even though the members of both panels were
provided with a scientific summary of the literature
on the efficacy of the procedures. This variation in the
recommendations of the two types of panels raises se-
rious issues with respect to the desired composition of
such panels.
48 I. Yaniv / International Journal of Forecasting 27 (2011) 41–49
Might panels of correlated experts inadvertently
polarize? Might more diverse groups of experts
reach better guidelines? Our present findings suggest
that concerns about the risk of polarization in
homogeneous panels are not unfounded, and the
prospect of improving the performance is not unlikely.
By recruiting experts from different backgrounds it is
possible to reduce or minimize individual judgment
biases, and thereby improve their forecasting accuracy
as well.
4.2. Practitioner-oriented applications: the perspec-
tive assignment method
The literature on group performance has already
noted the potential advantages of groups composed of
members with heterogeneous backgrounds (e.g. Rowe
& Wright, 2001). The present study goes one step
further, suggesting that practitioners of group methods
could actively create heterogeneity by assigning roles
to members. Thus, the administrator of a Delphi group
could assign specific, scripted roles to group members,
requesting them to take one perspective or another. We
take it as a truism that almost any problem involving
uncertainty could be framed in terms of either “glass
half full” or “glass half empty” (McKenzie & Nelson,
2003).
Then, based on the perspectives assigned to them
(e.g., conservative, pessimistic scenarios or, alterna-
tively, daring, optimistic scenarios), members might
generate either risk averse, timid forecasts or bold, risk
taking ones. Such a variety of forecasts, when commu-
nicated (with reasons) either in a structured, face-to-
face group discussion or anonymously, could enrich
and benefit the performance (Rowe, Wright, & Mc-
Coll, 2005). Our present work provides preliminary
support for these ideas. Further research is needed to
establish the effectiveness of perspective assignment
and the method of nominating members to scripted
roles. An interesting issue concerns the exact numer-
ical distribution of perspectives in a group. For ex-
ample, it may be interesting to investigate the condi-
tions under which a single devil’s advocate could be
effective in introducing a complementary perspective
to that of the group.
5. Conclusion
Research conducted on group decision processes
over several decades has produced a fascinating
spectrum of findings. An important conclusion has
been that the advantages of group decision processes
depend on situational and processing parameters, such
as the composition of the group, the nature of the
task, the distribution of information among group
members, the nature of the interactions in the group,
and the group decision rule (Kerr & Tindale, 2004).
The present research highlights a way of improving
interactive, group-based forecasting methods. By
assigning group members to different perspectives
(framings of a problem), the susceptibility of groups
to decision biases is reduced dramatically.
Acknowledgement
This research was supported by Grant No. 344/05
from the Israel Science Foundation.
References
Bornstein, G., & Yaniv, I. (1998). Individual and group behavior
in the ultimatum game: are groups more rational players?
Experimental Economics, 1, 101–108.
Fraser, G. M., Pilpel, D., Kosecoff, J., & Brook, R. H. (1994). Effect
of panel composition on appropriateness ratings. International
Journal for Quality Health Care, 6, 251–255.
Isenberg, D. J. (1986). Group polarization: a critical review and
meta-analysis. Journal of Personality and Social Psychology,
50, 1141–1151.
Kahneman, D., & Tversky, A. (1984). Choices, values and frames.
American Psychologist, 39, 341–350.
Kerr, N. L., & Tindale, R. S. (2004). Group performance and
decision making. Annual Review of Psychology, 55, 623–655.
Mannix, E., & Neale, M. A. (2005). What differences make
a difference? The promise and reality of diverse teams in
organizations. Psychological Science in the Public Interest, 6,
31–55.
McKenzie, C. R. M., & Nelson, J. D. (2003). What a speaker’s
choice of frame reveals: reference points, frame selection,
and framing effects. Psychonomic Bulletin and Review, 10,
596–602.
Milch, K. F., Weber, E. U., Appelt, K. C., Handgraaf, M. J. J., &
Krantz, D. H. (2009). From individual preference construction
to group decisions: framing effects and group processes.
Organizational Behavior and Human Decision Processes, 108,
242–255.
Myers, D. G., & Lamm, H. (1976). The group polarization
phenomenon. Psychological Bulletin, 83, 602–627.
I. Yaniv / International Journal of Forecasting 27 (2011) 41–49 49
Neale, M. A., Bazerman, M. H., Northcraft, G. B., & Alperson, C.
(1986). Choice shift effects in group decisions: a decision bias
perspective. International Journal of Small Group Research, 2,
33–42.
Paese, P. W., Bieser, M., & Tubbs, M. E. (1993). Framing effects
and choice shifts in group decision making. Organizational
Behavior and Human Decision Processes, 56, 149–165.
Rowe, G., & Wright, G. (1999). The Delphi technique as a
forecasting tool: issues and analysis. International Journal of
Forecasting, 15, 353–375.
Rowe, G., & Wright, G. (2001). Expert opinions in forecasting: the
role of the Delphi technique. In J. S. Armstrong (Ed.), Principles
of forecasting: a handbook for researchers and practitioners
(pp. 125–144). Norwell, MA: Kluwer Academic Publishers.
Rowe, G., Wright, G., & McColl, A. (2005). Judgment change
during Delphi-like procedures: the role of majority influence,
expertise, and confidence. Technological Forecasting and Social
Change, 72, 377–399.
Sniezek, J. A. (1992). Groups under uncertainty: an examination of
confidence in group decision making. Organizational Behavior
and Human Decision Processes, 52, 124–155.
Sommers, S. R. (2006). On racial diversity and group decision
making: identifying multiple effects of racial composition
on jury deliberations. Journal of Personality and Social
Psychology, 90, 597–612.
Sunstein, C. R., & Hastie, R. (2008). Four failures of deliberating
groups. John M. Olin Law and Economics working paper
no. 401. University of Chicago. http://ssrn.com/abstract id=
1121400.
Tindale, R. S., Sheffey, S., & Scott, L. A. (1993). Framing
and group decision-making: do cognitive changes parallel
preference changes? Organizational Behavior and Human
Decision Processes, 55, 470–485.
Whyte, G. (1989). Groupthink reconsidered. The Academy of
Management Review, 14, 40–56.
Yaniv, I. (2004). The benefit of additional opinions. Current
Directions in Psychological Science, 13, 75–78.
Yaniv, I., Choshen-Hillel, S., & Milyavsky, M. (2009). Spurious
consensus and opinion revision: why might people be
more confident in their less accurate judgments? Journal of
Experimental Psychology: Learning, Memory, and Cognition,
35, 558–563.
    • "Further evidence suggests that introduction of surface-level (demographic) differences such as gender allows groups to better capitalize on deep-level (psychological) differences such as risk-taking propensities in their decision-making (Harrison, Price, & Bell, 1998; Harrison, Price, Gavin, & Florey, 2002; Harvey, 2014). This processing of divergent information among group members would unearth different risk-taking preferences and foster more debate; eventually it would temper the automated and category-reinforcing information processing that moves groups toward a risky pole (Nieboer, 2013; Yaniv, 2011). This reduction of strategic risk-taking, through elaborated information processing in the TMT (first-stage effect) should result in an improvement of long-term fiscal outcomes for the firm (second-stage effect). "
    Article · May 2016
    • "For instance, a Delphi panel comprising a majority of like-minded experts sharing a certain desirability perspective will probably provide initial estimates being desirabilitybiased . The consequently biased feedback could make desirability contagious (Ecken et al., 2011) as other experts converge towards this biased feedback value (Kerr and Tindale, 2011; Yaniv, 2011), e.g. because the bandwagon effect makes them move away from their un-biased minority opinion (while the belief-perseverance bias keeps the majority participants at their erroneous position). "
    [Show abstract] [Hide abstract] ABSTRACT: Delphi is an established information gathering and forecasting approach that has proven to deliver valuable results in a wide variety of specialist fields. Yet, Delphi studies have also continuously been subject to critique and doubt, particularly concerning its judgmental and forecasting accuracy. To a large part this can be attributed to the substantial discretion researchers have in their design and implementation. Awkwardly designed Delphi studies may lead to severe cognitive biases that adversely affect the research results. This paper takes a cognitive perspective by investigating how different cognitive biases take effect within future-oriented Delphi studies and how their unfavorable impacts can be mitigated by thoroughly adapting specific Delphi design features. The analysis addresses cognitive biases affecting panelists' initial estimates — namely framing and anchoring as well as the desirability bias — as well as such cognitive biases taking effect during feedback and revision loops — namely the bandwagon effect and belief perseverance.
    Full-text · Article · Apr 2016
    • "A consequence for group decision environments is that decision support has to include mechanisms that pro-actively encourage knowledge exchange. One reason for increased knowledge exchange between group members is group diversity (in terms of dimensions such as demographic and educational background ), i.e., the higher the degree of diversity the higher the probability of higher quality decision outcomes (measured , e.g., in terms of the degree of susceptibility to the framing effect [23]). Schulz-Hardt et al. [17] discuss the role of dissent in group decision making: the higher the dissent in initial phases of a group decision process, the higher the probability that the group manages to share the decisionrelevant information (discover the hidden profile). "
    Full-text · Conference Paper · Sep 2015 · Technological Forecasting and Social Change
Show more

    Recommended publications

    Discover more