Escolar Documentos
Profissional Documentos
Cultura Documentos
T he study presented in this paper examines how beliefs about coworkers affect group
discussion and performance. Two beliefs are considered: (1) Perceptions of coworker
task competence, and (2) achievement motivation. This study investigates whether these
perceptions can actually hinder group discussion and performance, and considers the con-
texts in which these detrimental effects are more or less likely to arise. Results indicate that
although perceptions of higher achievement motivation in coworkers lowered performance
when task information was partially shared, they raised performance when task informa-
tion was fully shared. A content analysis of group discussions reveals that the discussion
behaviors examined by this study, rather than the more frequently examined bias toward
discussing common information, mediated these results.
(Group; Information Sharing; Perception; Competence; Motivation)
on how his or her role relates to those of others in partners were already aware of their privately held
their group (Wegner 1987, Liang et al. 1995, Moreland information and, thus, lead them to exert less effort
1999). This research contends that, as a result of to communicate it than members of inexperienced
this change, groups are able to reduce their informa- groups.
tion loads and ultimately improve their performance. A more thorough explanation for Kim’s (1997) find-
Similarly, research on shared mental models indicates ings, however, is that experience may shape a range
that, over time, groups can develop collective under- of beliefs about coworkers (e.g., due to increased
standings or shared mental models of their situation familiarity with the task and/or team, members’
and that these organized bodies of knowledge can schemas about how relationships and/or capabilities
facilitate communication, improve coordination, and should develop over time, and so on) and that each of
improve group performance (e.g., Orasanu and Salas these beliefs can affect the sharing of information in
1993). groups. This notion is based on research in which per-
Recent evidence suggests, however, that a group’s ceptions of high-coworker competence and motiva-
ability to manage knowledge does not always tion were found to lower member efforts and collec-
improve over time. This evidence is provided by a tive performance (Williams and Karau 1991) as well
small number of studies that have investigated the as on the potential implications of such perceptions
implications of experience for group discussion and for group coordination and discussion (Wittenbaum
performance. Although the exchange of information et al. 1998, 1999). In this study, I consider these two
among members can be essential for performance, perceptions and their likely effects on the discus-
groups often focus on information they share in com- sion bias (including the mentioning, repetition, and
mon rather than the potentially valuable information proportion of critical versus noncritical information),
any individual member may hold (e.g., Stasser and teamwork (including performance monitoring, feed-
Stewart 1992, Winquist and Larson 1998). While one back, closed-loop communications, and backing-up
might expect that this discussion bias would diminish behaviors), and task performance.
over time, initial findings suggest that it may some-
times worsen as groups gain experience with the task Task Competence
and/or team (Kim 1997, Wittenbaum 1998, see also Although many kinds of ability can be identified, this
Gruenfeld et al. 1996 for results that depend on the study focuses on a type of ability that would be most
distribution of information). relevant for the work group conditions under investi-
Kim (1997), for example, found that groups in gation (i.e., task competence). Competence is defined
which members had previously worked together on as the degree to which one possesses the technical
a similar task displayed a larger discussion bias and and interpersonal skills required for one’s job (e.g.,
achieved lower task performance than groups with Butler 1991). Thus, we might expect that perceptions
no prior experience with coworkers or the task and of competence in coworkers should affect members’
groups whose prior experience was limited to either beliefs that their partners will be able to successfully
coworkers or the task. One explanation for this effect complete the task.
is the “curse of knowledge,” which refers to peo- Perceptions of coworker competence may affect
ple’s tendency to overestimate the ability of others group information sharing and performance in two
to accurately solve a problem (e.g., Camerer et al. ways. First, perceptions of coworker competence may
1989, Fussell and Krauss 1992, Nickerson et al. 1987). affect member orientations toward their task. Whereas
Kim (1997) suggested that members of experienced perceptions of high-coworker competence should fos-
groups may have exhibited a larger discussion bias ter the belief that one’s own contributions are less
and achieved lower performance because their greater necessary and entice members to exert less effort on
familiarity with both the task and team may have their task (i.e., engage in social loafing), perceptions of
made them more susceptible to the “curse of knowl- low-coworker competence should compel members to
edge,” which would lead them to believe that their increase their efforts to compensate for the inadequate
of low-coworker motivation when task information is the discussion bias and teamwork are expected to
fully shared. influence task performance, we might expect that
When information is only partially shared, how- (3) the predicted effects of perceived coworker com-
ever, members are not only unable to develop fully petence and motivation on task performance would
informed prediscussion opinions (given that none of operate through their influence on the discussion bias
them would possess all the necessary facts), but will and teamwork. In other words, the discussion bias
also find it much more difficult to evaluate their part- and teamwork are each predicted to mediate the
ners’ justifications for these opinions. Such conditions effects of perceived coworker competence and moti-
should lead members who perceive high motivation vation on task performance.
in their coworkers to believe that attempts to resolve
such differences would foster a fruitless debate over Hypothesis 3. The discussion bias will mediate the
whose facts are more correct, given that they were effects of perceived coworker competence and perceived
each highly motivated to succeed, yet could not even coworker motivation on task performance.
agree on the basic facts of the case. This belief should,
Hypothesis 4. The amount of teamwork will mediate
in turn, lead members to lower their assessments
of the potential value of group discussion and, con- the effects of perceived coworker competence and perceived
sequently, engage in social loafing (e.g., by simply coworker motivation on task performance.
adopting a majority rule voting scheme). Thus, per-
ceptions of high-coworker motivation should foster a Overview
larger discussion bias, less teamwork, and lower task In summary, this study examines two percep-
performance than perceptions of low-coworker moti- tions that may help explain why experience may
vation when task information is partially shared. We adversely influence group information sharing and
might, therefore, predict the following interactions. performance. Moreover, it explores the contexts in
which these detrimental effects are more or less
Hypothesis 2a. Perceptions of high-coworker motiva-
likely to arise. Whereas perceptions of high-coworker
tion will foster a larger discussion bias than perceptions of
task competence and achievement motivation should
low-coworker motivation when task information is partially
shared, but perceptions of high-coworker motivation will increase the discussion bias, reduce teamwork, and
foster a smaller discussion bias than perceptions of low- lower performance when information is partially
coworker motivation when task information is fully shared. shared, these effects should not occur and even
reverse (for perceived achievement motivation) when
Hypothesis 2b. Perceptions of high-coworker motiva- information is fully shared. The study also inves-
tion will foster less teamwork than perceptions of low- tigates the mechanisms underlying these relation-
coworker motivation when task information is partially ships by assessing several teamwork dimensions (i.e.,
shared, but perceptions of high-coworker motivation will performance monitoring, feedback, closed-loop com-
foster greater teamwork than perceptions of low-coworker munications, and backing-up behaviors) through a
motivation when task information is fully shared. content analysis of group discussions. These objec-
Hypothesis 2c. Perceptions of high-coworker motiva- tives are pursued through the laboratory experiment
tion will foster lower task performance than perceptions of described below.
low-coworker motivation when task information is partially
shared, but perceptions of high-coworker motivation will
foster greater task performance than perceptions of low-
Method
coworker motivation when task information is fully shared.
This study employed a 2(high vs. low competence) ×
Finally, given that (1) perceived coworker com- 2(high vs. low motivation) × 2(full vs. partial informa-
petence and motivation are expected to influence tion) between-subjects design. Three-person groups
the discussion bias and teamwork and (2) both were used as the unit of analysis.
The mentioning of critical information counted the Members exhibited performance monitoring by
number of critical clues that were stated at least once. asking for an opinion (e.g., “Who do you think commit-
For example, if a group stated 3 different critical clues ted the crime?”) or asking for a justification (“Why do
during its discussion, “X,” “Y,” and “Z,” the group is you think it was Billy Prentice?”). Members provided
considered to have mentioned 3 critical clues, regard- feedback by dissenting with a partner’s statement (e.g.,
less of how often each of these 3 critical clues was “I don’t think he did it.”) or confirming a partner’s
stated. The repetition of critical information counted statement (e.g., “That’s right.”). Members displayed
the number of times critical clues were stated after closed-loop communications by confirming a partner’s
they were first mentioned, and then divided these statement (see previous example) or asking a fact-
totals by the number of critical clues that were men- based question (e.g., “Eddie had a hearing aid?”). Mem-
tioned at least once. This measure yielded a per- bers exhibited backing-up behaviors by contributing
mentioned-item repetition rate for critical information an opinion (e.g., “I thought it was Mickey Malone.”) or
(see Winquist and Larson 1998). In the above example, contributing a justification (e.g., “Mickey did it because
if the group stated clue “X” one time after it was first Mr. Guion was going to ruin his business.”). Members
mentioned, clue “Y” two times after it was first men- also indicated the need for backing-up behaviors by
tioned, and clue “Z” three times after it was first men- stating that they had no opinion (e.g., “I don’t know.”).
tioned, the per-mentioned-item repetition rate for crit- Finally, each of these behaviors was aggregated to
ical information would be two (i.e., 1 + 2 + 3/3 = 2). determine the frequency with which members con-
Analogous procedures were used to calculate the men-
tributed a task-relevant phrase, which provided a gen-
tioning and repetition of noncritical clues.
eral indication of the time and effort groups devoted
The content analysis assessed the discussion focus
to their task.
by examining the degree to which groups discussed
Two independent raters were trained to perform
critical versus noncritical information. Thus, two vari-
these content analyses. The raters were given cod-
ables were coded: (1) The total number of times group
ing sheets that identified all the discussion bias and
members stated critical clues (i.e., by counting each
discussion behavior variables and were instructed to
time any critical clue was mentioned or repeated), and
record, for each line of the transcripts, every occa-
(2) the total number of times group members stated
sion in which any of these variables were found. The
noncritical clues (i.e., by counting each time any non-
desired measures were then obtained through fre-
critical clue was mentioned or repeated). The discus-
sion focus was obtained by dividing the total number quency counts. Some utterances qualified as both a
of times group members stated critical information discussion bias and discussion behavior variable and
by the total number of times they stated noncritical were coded for both. The utterance, “Eddie had a
information. hearing aid?,” for example, represents the mention-
Finally, the content analysis coded 9 additional ing of critical information as well as the asking of a
discussion behaviors. These behaviors were found, fact-based question and was, thus, counted for each
through a perusal of transcripts, not only to cap- category. These kinds of dual-coded utterances helped
ture the essence of these group discussions but also foster significant correlations between the discussion
to illustrate the teamwork dimensions identified by bias and discussion behavior measures. Despite these
Morgan et al. (1986)—performance monitoring, feed- correlations, the measures still differed in the roles
back, closed-loop communications, and backing-up they played in this study (see Results).
behaviors. These behaviors included the frequency The raters were blind to the hypotheses and treat-
with which group members contributed an opinion, ment conditions. Nevertheless, the content analy-
asked for an opinion, contributed a justification, asked ses from the two independent raters demonstrated
for a justification, confirmed a partner’s statement, high interrater reliability, as indicated by a ran-
dissented with a partner’s statement, asked a fact- dom sample of 26 transcripts (noncritical informa-
based question, stated that they had no opinion, and tion (Kappa = 094), critical information (Kappa =
contributed a task-relevant phrase. 085), asked for opinion (Kappa = 091), asked for
justification (Kappa = 089), dissented with partner would foster a larger discussion bias relative to
comment (Kappa = 087), confirmed partner statement perceptions of low-coworker competence and motiva-
(Kappa = 086), asked fact-based question (Kappa = tion, respectively, when task information is partially
090), contributed opinion (Kappa = 086), contributed shared, these effects should not arise and even reverse
justification (Kappa = 085), and stated they had no (for perceived motivation) when information is fully
opinion (Kappa = 074)). A third, blind rater resolved shared. These predictions were tested by examining
any differences between these two sets of ratings. the mentioning and repetition of critical and noncrit-
ical information and the proportion of critical versus
Results noncritical information discussed.
All 276 subjects (92 groups) completed the study. A multiple analysis of variance (MANOVA) was
However, audiotapes for three groups could not be conducted to determine whether the experimental
transcribed or coded due to the poor quality of the manipulations affected the mentioning and repetition
recordings. As a result, 89 groups were left for anal- of critical and noncritical information and the pro-
ysis. Analyses were first conducted to examine the portion of critical versus noncritical information dis-
independent effects of the study variables. Tests of cussed. The distribution of information was found
mediated relationships among these variables were to exert an overall effect on these discussion bias
then explored through a series of hierarchical regres- measures (F 5 77 = 490 p < 0001. Follow-up anal-
sions (Baron and Kenny 1986, James and Brett 1984). yses of variance (ANOVAs) revealed that groups
A summary of means, standard deviations, and corre- with full information mentioned more critical infor-
lations is presented in Table 1. Means, standard devi- mation (M = 222 sd = 020) than groups with par-
ations, and sample sizes per condition for the study tial information (M = 161 sd = 020) ((F 1 88 =
variables are presented in Table 2. 481 p < 005); (Effect Size ES = 305). No other main
effects or higher order interactions were found ((F s <
Effects on the Discussion Bias 322 ns ES < 048)). These results do not support
Hypotheses 1a and 2a predicted that whereas percep- predictions that perceived coworker competence or
tions of high-coworker competence and motivation motivation would affect the discussion bias.
High comp. Low comp. High comp. Low comp. High comp. Low comp. High comp. Low comp.
Variable N = 12 N = 11 N = 10 N = 11 N = 11 N = 12 N = 12 N = 10
Task performance 017 018 054 045 067 042 042 018
014 014 014 014 014 014 014 014
Critical clues mentioned 167 164 160 154 218 225 217 230
150 103 135 144 108 136 134 134
Noncritical clues mentioned 625 573 610 600 527 550 558 490
191 210 179 184 200 211 188 264
Critical clues repeated 139 148 045 114 082 067 106 092
166 217 055 134 116 083 107 052
Noncritical clues repeated 125 106 080 128 094 113 141 101
102 065 081 096 081 110 103 106
Critical/noncritical 029 037 027 024 037 034 028 077
022 032 028 023 019 030 016 094
Task-relevant phrase 6117 4936 6440 7409 5727 6183 6692 4730
3831 3459 4092 5601 3523 2790 3129 2529
Contribute opinion 1392 991 1730 1836 1309 1408 1742 1060
634 579 1268 1328 699 568 804 432
Ask for opinion 392 309 350 282 554 433 442 270
300 181 264 209 491 192 162 170
Contribute justification 858 691 1050 1354 1054 1092 900 730
650 446 952 705 873 396 544 337
Ask for justification 292 345 150 254 118 108 183 170
211 344 108 284 178 144 175 134
Confirm 600 382 930 882 445 525 492 320
445 322 706 935 294 441 368 132
Dissent 150 091 500 300 191 217 408 240
157 114 518 322 378 310 487 237
Fact-based question 442 300 340 482 445 483 483 220
614 452 232 623 301 551 426 148
No opinion 058 127 040 036 036 142 142 040
116 205 052 050 067 116 178 052
Effects on Discussion Behaviors shared. The following analyses tested these pre-
Hypotheses 1b and 2b predicted that whereas per- dictions as well as the relationship between the
ceptions of high-coworker competence and motiva- discussion behaviors and the discussion bias.
tion would foster less teamwork (assessed by the A MANOVA was conducted to examine the overall
discussion behaviors) relative to perceptions of low- effects of the experimental manipulations on the
coworker competence and motivation, respectively, 9 discussion behaviors. Both perceived motivation
when task information is partially shared, these (F 9 73 = 285 p < 001) and the distribution of infor-
effects would not arise and even reverse (for per- mation (F 9 73 = 332 p < 0005) were found to
ceived motivation) when task information is fully affect these discussion behaviors. These main effects
were qualified, however, by an interaction between noncritical information F 9 79 = 2408 p < 00001
perceived motivation and the distribution of informa- and the repetition of critical F 9 79 = 776 p <
tion F 9 73 = 297 p < 0005. 00001 and noncritical information F 9 79 =
ANOVAs were then conducted to isolate the 4367 p < 00001 were all significantly related to
specific variables driving these multivariate effects. these discussion behaviors. Univariate correlations
Groups with high-perceived motivation were found between the discussion bias measures and each of
to make fewer dissenting comments M = 162 sd = the discussion behaviors were then examined to iso-
050 than groups with low-perceived motivation late the particular variables driving these multivari-
M = 362 sd = 052 F 1 88 = 758 p < 001 ES = ate effects (see Table 1). These univariate correlations
392. Groups with partial information were also revealed that the mentioning and repetition of critical
found to ask for justifications and confirm partner and noncritical information were each positively cor-
statements more frequently M = 260 sd = 032 and related with the majority of the discussion behaviors,
M = 698 sd = 077, respectively) than groups with indicating that the level of teamwork was positively
full information M = 145 sd = 032 F 1 88 = associated with the degree to which both critical and
658 p < 005 ES = 359 and M = 446 sd = 076 noncritical task-relevant information was discussed.
F 1 88 = 544 p < 005 ES = 332, respectively). These results support predictions that perceptions
Moreover, the interaction between perceived motiva- of coworker motivation affect teamwork (assessed by
tion and the distribution of information was found the discussion behaviors) and that these relationships
to affect three discussion behaviors: the frequency depend on the distribution of task information (i.e.,
with which group members contributed justifications Hypothesis 2B was supported). Follow-up univari-
F 1 88 = 631 p < 005 ES = 361, asked for jus- ate analyses, furthermore, suggest that the frequency
tifications F 1 88 = 399 p < 005 ES = 202, and with which group members contributed justifica-
confirmed partner comments F 1 88 = 520 p < tions, asked for justifications, and confirmed partner
005 ES = 229. Groups with high-perceived moti- comments were driving this multivariate interaction
vation were found to contribute justifications less effect.
frequently M = 775 sd = 134, ask for justifica-
tions more frequently M = 318 sd = 044, and Effects on Performance
confirm partner statements less frequently M = Hypotheses 1c and 2c predicted that whereas per-
491 sd = 106 than groups with low-perceived ceptions of high-coworker competence and motiva-
motivation when information was partially shared tion would foster lower task performance relative to
M = 1202 sd = 140 M = 202 sd = 046 M = perceptions of low-coworker competence and moti-
906 sd = 111, respectively), but groups with high- vation, respectively, when information is partially
perceived motivation were found to contribute justi- shared, these effects should not arise and even reverse
fications more frequently M = 1073 sd = 134, ask (for perceived motivation) when information is fully
for justifications less frequently M = 113 sd = 044, shared. The following analyses tested these predic-
and confirm partner statements more frequently M = tions as well as the effects of the discussion bias and
485 sd = 106 than groups with low-perceived moti- discussion behaviors on performance.
vation when information was fully shared M = An ANOVA was conducted to assess whether
815 sd = 137 M = 177 sd = 045 M = 406 sd = the manipulations affected task performance. The
109, respectively). results revealed a two-way interaction between per-
Canonical correlations were also conducted to ceived motivation and the distribution of infor-
examine the relationship between the discussion bias mation. Groups with high-perceived motivation
and the 9 discussion behaviors. As might be expected, achieved lower task performance M = 017 sd =
given that some utterances were coded as both discus- 010 than groups with low-perceived motivation
sion bias and discussion behavior variables, the men- M = 050 sd = 010 when task information was
tioning of critical F 9 79 = 1485 p < 00001 and partially shared, but groups with high-perceived
motivation achieved higher task performance M = A multivariate regression was conducted to deter-
054 sd = 010 than groups with low-perceived moti- mine whether the perceived motivation X infor-
vation M = 030 sd = 010 when task information mation distribution interaction affected the group
was fully shared F 1 91 = 824 p < 001 ES = discussion variables. This analysis revealed a sig-
285. These results support predictions that percep- nificant relationship between the perceived motiva-
tions of coworker motivation affect performance and tion X information distribution interaction and the
that these relationships depend on the distribution discussion variables, in general F 9 76 = 304 p <
of information (i.e., Hypothesis 2C was supported). 0005, and the frequency with which groups con-
Whereas perceptions of high-coworker motivation tributed justifications t1 88 = 252 p < 005, asked
lowered performance when information was par- for justifications t1 88 = −203 p < 005, and con-
tially shared, this relationship was reversed when firmed partner statements t1 88 = 230 p < 005,
information was fully shared. in particular. Thus, Condition 1 was satisfied for the
discussion behaviors as a whole and, specifically, for
Mediated Regression Results the frequency with which groups contributed justifi-
Hypotheses 3 and 4 predicted that the effects of cations, asked for justifications, and confirmed part-
perceived coworker competence and motivation on ner statements (i.e., the independent variable affected
task performance would be mediated by the dis- the potential mediators).
cussion bias (Hypothesis 3) and teamwork (assessed Multiple regressions were conducted to examine
by the discussion behaviors) (Hypothesis 4). Results the effects of the perceived motivation X information
from the prior analyses indicate that the experimen- distribution interaction and group discussion behav-
tal manipulations affected both the discussion bias iors, respectively, on task performance. These analyses
and the discussion behaviors. However, only the dis- revealed a significant relationship between the per-
cussion behaviors were significantly related to both ceived motivation X information distribution interac-
the perceived motivation X information distribution tion and task performance t1 91 = 291 p < 0005.
interaction F 9 73 = 297 p < 0005 and task per- These analyses also revealed significant relationships
formance F 9 79 = 507 p < 00001 (see Table 1). between the group discussion behaviors as a whole
These behaviors, therefore, represent the only factor and task performance F 9 88 = 507 p < 00001, as
that might plausibly mediate the reported relationship well as between the frequency with which groups
between coworker perceptions and performance. This contributed justifications t1 88 = 450 p < 00001
possibility was tested through a series of hierarchical and confirmed partner statements t1 88 = 440 p <
regressions (Baron and Kenny 1986, James and Brett 00005, respectively, and task performance. Thus,
1984). Condition 2 was satisfied for the perceived motiva-
Tests of mediation require that four conditions be tion X information distribution interaction, the group
met (James and Brett 1984). To demonstrate the medi- discussion behaviors as a whole and, specifically, for
ating effect of variable Y in the relation X → Z: the frequency with which groups contributed justi-
(1) X must be significantly related to Y , (2) X and Y fications and confirmed partner statements (i.e., the
must be significantly related to Z, (3) the variance in Z independent variable and potential mediators affected
predicted by X must be nonsignificant after mediator the dependent variable).
Y is controlled (the satisfaction of all but this condi- A two-step procedure was performed to determine
tion indicates partial mediation), and (4) Y should be whether the variance in task performance predicted
significantly related to Z after X is controlled. Each of by the perceived motivation X information distribu-
these conditions was examined to determine whether tion interaction was nonsignificant after controlling
the discussion behaviors Y mediated the relation- for the group discussion behaviors. Task performance
ship between the perceived motivation X information was, first, regressed on the group discussion behav-
distribution interaction X and task performance Z. iors. The residuals from this first step were then
regressed on the interaction between perceived moti- frequency with which groups contributed justifica-
vation and the distribution of information (i.e., after tions and confirmed partner statements, in partic-
the effects of the group discussion behaviors were ular, completely mediated the relationship between
controlled). This procedure revealed a nonsignificant the perceived motivation X information distribution
relationship between the perceived motivation X interaction and task performance (i.e., Hypothesis 4
information distribution interaction and performance was supported). These results, therefore, provide a
t1 88 = 147 ns. Analogous procedures were clear explanation for the performance differences
also conducted on the specific group discussion observed in this study. The evidence indicates that the
behaviors. These procedures revealed nonsignificant discussion behaviors assessed in this study can play
relationships between the perceived motivation X a critical role in the information-sharing process by
information distribution interaction and task perfor- mediating the effects of coworker perceptions on per-
mance when controlling for the frequency with which formance. The results also help substantiate, to some
groups contributed justifications t1 88 = 191 ns degree, the univariate analyses conducted in the pre-
and confirmed partner statements t1 88 = 197 ns. ceding section. Whereas the effect of the perceived
Thus, Condition 3 was satisfied for the group discus- motivation X information distribution interaction on
sion behaviors as a whole and, specifically, for the fre- any of the 9 discussion behaviors may have occurred
quency with which groups contributed justifications by chance, the finding that the frequency with which
and confirmed partner statements (i.e., the indepen- groups contributed justifications and confirmed part-
dent variable did not affect the dependent variable ner statements also met the three other conditions
after controlling for the potential mediators). required for mediation is less likely. More substantive
Finally, a procedure similar to that performed in verification of the role(s) played by these discussion
Condition 3 was used to determine if the group dis- behaviors, however, will require further investigation.
cussion behaviors were still significantly related to
task performance after the effects of the perceived
motivation X information distribution interaction was Discussion
controlled. Task performance was first regressed on The purpose of this study was to investigate how
the perceived motivation X information distribution beliefs about coworkers may affect group informa-
interaction. The residuals from this analysis were tion sharing and performance. In addition, it sought
then regressed on the group discussion variables to identify contexts in which these adverse effects are
(i.e., after controlling for the perceived motivation more or less likely to arise. Finally, this study inves-
X information distribution interaction). This proce- tigated the underlying mechanisms for these find-
dure revealed that the group discussion behaviors, ings through an exploratory content analysis of group
in general F 9 88 = 409 p < 00005, and the fre- discussions.
quency with which groups contributed justifications The evidence from this study suggests that beliefs
t1 88 = 397 p < 00005 and confirmed partner about coworkers can exert important effects on group
statements t1 88 = 434 p < 00001, in particular, discussion and performance. The finding that percep-
still affected task performance. Thus, Condition 4 tions of higher coworker motivation lowered perfor-
was satisfied for the group discussion behaviors as a mance when task information was partially shared is
whole and, specifically, for the frequency with which consistent with Kim’s (1997) observation that mem-
groups contributed justifications and confirmed part- bers’ beliefs about their coworkers can sometimes hin-
ner statements (i.e., the potential mediators affected der group interactions and outcomes. This study also
the dependent variable even after controlling for the extends this insight, however, by suggesting that we
independent variable). broaden our inquiry beyond mechanisms such as the
The satisfaction of all four conditions at both the “curse of knowledge” to more fully understand these
multivariate and univariate levels supports the notion effects. It may be useful, therefore, to consider how a
that the discussion behaviors, in general, and the range of coworker perceptions may influence group
discussion and performance and consider both their imply that we ignore this measure. The fact that the
additive and interactive effects. frequency with which group members mentioned crit-
The influence of different coworker perceptions ical information was found to affect task performance,
may also help explain an apparent discrepancy for example, does provide some indirect justification
between studies that have examined the effects for investigating this discussion measure. However,
of experience on group information sharing and the fact that the discussion bias failed to mediate the
performance. Whereas Kim (1997) found that ini- effects of coworker perceptions on task performance
tial experience with both the task and team could suggests that we should broaden our inquiry to exam-
impair group information sharing and performance, ine how a variety of additional factors may affect
Gruenfeld et al. (1996) found that group member performance in this context.
familiarity with their partners could either improve These considerations suggest that it may be use-
or worsen these outcomes depending on the dis- ful to investigate how various discussion behaviors
tribution of task information. Familiars were more may provide additional avenues for assessing infor-
likely than strangers to effectively exchange informa- mation sharing in groups. The content analysis per-
tion and identify the correct solution when informa- formed by this study provides an example of such
tion was partially shared, but strangers outperformed an effort. The exploratory nature of this analysis per-
familiars on these measures when information was mits only qualified conclusions. However, its dis-
fully shared. This study offers a potential explanation covery that the discussion behaviors in general, and
for these findings by suggesting that Kim (1997) and two behaviors in particular (i.e., the frequency with
Gruenfeld et al.’s (1996) operationalizations of expe- which group members contributed justifications and
rience may have fostered different kinds of coworker confirmed partner statements), completely mediated
perceptions. Whereas the initial levels of task and the relationship between the perceived motivation X
team experience examined by Kim (1997) may information distribution interaction and performance
have heightened member expectations regarding their offers a provocative set of findings. These behaviors
likely performance on the subsequent task (i.e., per- may have enhanced teamwork and, thus, fostered
ceive high-coworker competence), the type of famil- interactions that allowed group members to work
iarity examined by Gruenfeld et al. (1996) (i.e., friends more effectively. This interpretation is not only con-
versus strangers) may have created the belief that sistent with the notion that coworker perceptions can
coworkers would value group cohesion over perfor- influence social loafing versus social compensation
mance (i.e., perceive low-coworker motivation). The in groups (Williams and Karau 1991), but also sup-
operation of these different perceptions would not ports the notion that the extent to which members
only explain the patterns of results observed by those repeat and validate each other’s ideas can affect the
studies, but is also consistent with the interpretations nature of group decisions (Brauer et al. 1995). These
they offered to explain these effects.3 possibilities will, of course, require verification. How-
It is important to note, however, that although per- ever, this approach to examining group processes may
ceived coworker motivation affected performance in prove beneficial by allowing us to move beyond sim-
this study, they did not do so through changes in ple measures of the extent to which information is
the discussion bias. These results, therefore, support shared to provide insight into the manner in which it
Gigone and Hastie’s (1993, 1997) suggestion that the is used.
discussion bias may provide an insufficient account Taken together, the results of this study highlight
of the information-sharing processes that affect group the importance of considering more thoroughly the
decisions. Of course, these results do not necessarily social-psychological processes operating in groups
and the reasons why they arise. This importance is
3
A potential reason why perceived coworker competence was not
heightened by the fact that research on the group
found to affect information sharing and performance in this study information-sharing process has primarily focused
is considered in the last section—Limitations and Future Directions. on the structural factors that operate to determine
the nature of group decisions (group size, informa- attempt to meet the perceived demands of their col-
tion load, and so on). Although these factors play lective task (Wittenbaum et al. 1996). Thus, it is possi-
an important role in determining the interactions of ble that effects would have been found for perceived
groups, the neglect of social-psychological processes competence if they were introduced at this earlier
may impede our attempts to understand their behav- point in time.
ior and performance. Indeed, it is only by considering Fourth, although this study focused on the condi-
such processes that we could predict and explain the tions in which group members were not told how
patterns of data observed in this study. This evidence their partners perceived them, there are many occa-
emphasizes the fundamental role played by social- sions in which such perceptions become known. This
psychological processes in groups and the need to raises a natural question about how members might
examine their effects more thoroughly through further react to their partners’ perceptions of them, given that
study. they may sometimes agree and sometimes disagree.
Is it a good idea, for example, to provide such feed-
Limitations and Future Directions back to one’s partners, and does this depend on the
The need for additional research is also underscored kind of feedback we provide? The possibility that
by the limitations faced in this study. One concern such feedback might affect members’ subsequent ori-
arises from the large number of analyses conducted entations, behaviors, and performance highlights the
in this study. Although the analyses attempted to
potential value of research that explores these issues
explore the data and provide a more comprehensive
for improving group performance (see Kim et al.
picture of what occurred in these groups, they also
2003 for a study that examines this question in a
increase likelihood of Type 1 error. Thus, readers may
negotiation context).
choose to rely on a more stringent (0.01) significance
Finally, it would be useful to directly examine the
level and/or focus on the multivariate findings when
kinds of perceptions that arise from different kinds
interpreting the results. Such steps would not alter the
of experiences, and the ways in which these percep-
main conclusions.
tions might affect group behavior and performance
A second issue raised by this study is the need to
on more traditional work group tasks. This investi-
examine the effects of other beliefs about coworkers
gation would offer insight into the dynamic changes
on group discussion and performance. Given that this
that might occur in these perceptions and the impli-
study has examined only two perceptions, expanding
this inquiry would provide a more detailed account cations of such changes for group behavior and per-
of these social effects. Moreover, although this study formance. Moreover, although the murder mystery
gave each group member the same perceptions of task used in this study has offered many researchers
their coworkers, group members often differ on these the opportunity to study a wide variety of organi-
and other perceptions. Thus, it would be valuable to zationally relevant questions, external validity would
investigate the effects of differences in these and other be improved by examining the implications of these
perceptions in such groups. kinds of coworker perceptions in tasks that orga-
Third, it would be worthwhile to investigate nizational members are more likely to face. These
whether the effects of these manipulations would avenues for future research suggest that we expand
have differed if they were introduced at different our inquiry to consider the myriad ways in which
points in time. The perceptions under investigation social processes can shape group interactions, and to
were introduced only after participants had read the do so in a variety of work group contexts. This study
task materials and, thus, could only exert their influ- represents only the beginning of this endeavor, and it
ence during group discussion. Evidence suggests, is clear that more work must be done in this arena.
however, that beliefs members have about cowork- It is through such efforts, however, that we can begin
ers before reading their task materials can affect their to develop a more comprehensive understanding of
processing of information before discussion as they knowledge management in groups.
Accepted by Linda Argote, former department editor; received March 1, 2001. This paper was with the author 8 months for 3 revisions.