. “i {1'

c

.‘.
:2 '

.51 e .
1 Jo .

I 1..) 1
11.19.1133.
. rear“:

.0. V5. .
(cart...
.153“. . y .
x .
. 3%»!
.111

:Ali .

.wufimvm.
.. a

i.
1(2):.
335’»:

 

 

:3 2.12.1 . .

EECLM.‘ . .

.7...

 

 

1v“

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the

thesis entitled

CONCERNS ABOUT IMPLIED NEGATIVE EVALUATION
OF FRIENDS VERSUS SOCIAL COMPARISON:
POSSIBLE EXPLANATIONS FOR MATCHING EFFECTS
IN COHESIVE DYADS

presented by

William Allen Jellison

has been accepted towards fulfillment
of the requirements for

M. A . degree in Psychology

 

 

474' [Le/WK

Major professor

Date 02/08/2002

 

0-7 639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

 

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE ' DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c2/ClFiC/DateDue.965-p.15

CONCERNS ABOUT IMPLIED NEGATIVE EVALUATION
OF FRIENDS VERSUS SOCIAL COMPARISON:
POSSIBLE EXPLANATIONS FOR MATCHING EFFECTS
IN COHESIVE DYADS
By

William Allen Jellison

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF ARTS
Department of Psychology

2002

ABSTRACT
CONCERNS ABOUT IMPLIED NEGATIVE EVALUATION
OF FRIENDS VERSUS SOCIAL COMPARISON:
POSSIBLE EXPLANATIONS FOR MATCHING EFFECTS
IN COHESIVE DYADS
By

William Allen Jellison

Previous research has suggested that people may fail to compensate (i.e., increase one’s
effort) when working on an important task with a friend who is either unwilling or unable
to perform well (Karau & Williams, 1997). The current research was designed to explore
whether these previous results could be explained better by concerns over negative
evaluation or by social comparison. Even though results failed to demonstrate strong
evidence for the social compensation effect, indirect evidence in support of the negative
evaluation hypothesis was found. Furthermore, gender differences were found for both
perceptions of one’s own performance and that of one’s co-worker. These gender

differences are discussed in the context of previous research on motivation gains.

To Lynn, for being there when I needed her most.

iii

ACKNOWLEDGEMENTS

Foremost, I must express my gratitude to my committee, Norb Kerr, Debby
Kashy, and Larry Messé for their guidance and support throughout this process. I must
also acknowledge the assistance of Shira Gabriel who served as a committee member
during the proposal stage of this thesis. Furthermore, I must thank Ernest Park who
assisted in the development and pre-testing of the persistence task used in this
experiment, as well as offering beneficial insights and experience. Finally, I must
acknowledge the contribution of three outstanding undergraduate research assistants,
Sandy Robinson, Timothy McCarthy, and Dan Murphy, who assisted in data collection.

I must also thank my partner, James Botelho, for his unwavering love and support
and most of all for continuing to believe in me when it seemed I was incapable of

believing in myself.

iv

TABLE OF CONTENTS

LIST OF TABLES ...................................................... vi
INTRODUCTION ....................................................... 1
Previous research on motivation gains and losses ............................ 2
The effects of cohesion on motivation gains and losses ....................... 4
The nature of cohesion ................................................. 5
Avoiding negative evaluations ........................................... 6
Social comparison ................................................... 10
METHOD ............................................................ 12
Overview .......................................................... 12
Participants ........................................................ 13
Materials .......................................................... 14
Procedure ......................................................... 16
RESULTS ............................................................ 23
Overview of analyses ................................................ 23
Manipulation checks ................................................. 23
Persistence trial data ................................................. 25
Proposed hypotheses ................................................. 27
Perceptions of own performance ........................................ 3O
Perceptions of task success and co-worker’s performance .................... 30
Gender effects ...................................................... 33
DISCUSSION ......................................................... 36
APPENDICES ......................................................... 43
REFERENCES ........................................................ 69

LIST OF TABLES

Table 1. Experimental design and corresponding cell reference letters ............. 44

Table 2a. Means for similarity/liking manipulation check and ancillary post-
experimental questionnaire items .......................................... 53

Table 2b. Means for willingness to work with co-worker in the future for work condition
x similarity interaction ................................................... 54

Table 2c. Means for feedback manipulation check and ancillary post-experimental
questionnaire items ..................................................... 55

Table 3a. Mean persistence duration in seconds on practice and performance trials for
dyad participants ....................................................... 56

Table 3b. Means for persistence trial duration in seconds for work conditions and by
gender ................................................................ 57

Table 4a. Mean difference in seconds between practice and performance persistence
trials for all experimental conditions ........................................ 58

Table 4b. Mean difference in seconds between practice and performance persistence
trials for all experimental conditions by gender ............................... 59

Table 5. Means for perceptions of own performance on post-experimental questionnaire
items for similarity, feedback and work conditions ............................. 60

Table 6a. Means for perceptions of responsibility for success and failure on the
performance trials for work condition and Similarity main effects ................. 61

Table 6b. Means for perceptions of responsibility for success and failure on the
performance trials for similarity x gender and similarity x gender x work condition

interactions ............................................................ 62

Table 60. Means for perceptions of co-worker’s performance for feedback main effects
and feedback x gender interactions ......................................... 63

Table 6d. Means for perceptions of co-worker for similarity x gender x work interactions
...................................................................... 64

Table 7a. Means for liking and concern over evaluation post-experimental questionnaire
items for gender and for similarity x gender interactions ........................ 65

vi

Table 7b. Means for co-worker concern over evaluation post—experimental questionnaire
items for feedback x gender interaction ...................................... 66

Table 7c. Means for perceptions of how hard participant worked on performance trials
for gender and work x gender interaction .................................... 67

Table 7d. Means for perceptions of how hard participant worked on performance trials
for work x gender x feedback interaction .................................... 68

vii

INTRODUCTION

Imagine you are working on a task with a friend. You like this person and value
his or her friendship. However, while working you realize that your friend is unlikely to
do well on the task. How would this influence the amount of effort you put in on the
task? Would you work harder to make up for your friend’s lower ability, or would you
end up working less hard? Contrary to what one might expect, social psychological
research suggests that people in situations like this may actually work less hard (Karau &
Williams, 1997). The present research explored two possible underlying mechanisms
that could account for why group members may tend to lower their effort when working
with a friend who is low in ability on a task.

One possible yet empirically untested explanation is that concerns over the
welfare of a friend may influence a person’s level of motivatiOn on the task.
Outperforming a friend may lead to uncomfortable feelings between the two individuals,
because if the dyad fails to succeed at the task the weaker member can be blamed, either
by the stronger member or by him or herself. This concern may lead to lower effort in
order to spare the friend’s feelings. Specifically, one question this research explored was,
when working in a cohesive group, will concerns over the negative evaluation of the dyad
partner for making them look bad influence the level of motivation on a group task?
Another possible explanation for this phenomenon is that fi'iends may be used as a
comparison standard in order to assess the level of effort needed to complete a task.
When we know that a friend will not perform well, we use this as an indicator of how

well we should perform. So conversely, do people lower their level of effort when

working with a low-ability partner who is a friend because they use the performance of
the friend as an indicator of their own expected performance? The current research was
designed to explore these two possible explanations for why group members would lower
their level of motivation when working in a cohesive group.
Previous research on motivation gains and losses

Research on motivation loss has demonstrated that under certain conditions
individuals working on a collective task, compared to when working alone or co-actively,
tend to reduce their level of motivation, a phenomenon Latane, Williams, and Harkins
(1979) describe as “social loafing.” Social loafing may occur for several reasons.
Individuals may reduce their effort when working in a group when their contribution is
not identifiable, or when the value of their effort is unclear (e.g., when working in a large
group on a collective task; Latane et al., 1979). Also, it seems to make sense to reduce
one’s effort when the additional effort is not needed. The free-rider effect (Kerr, 1983)
suggests that individual group members will reduce their effort when they believe their
contribution to the group project is unnecessary or dispensable. In addition, an individual
may reduce his or her effort in order to restore equity among the contributions of group
members (Jackson & Harkins, 1985). The sucker effect (Kerr, 1983) suggests that a
group member may reduce individual effort when individual members believe that their
partner is free-riding off of them. The idea that a free-riding partner is not contributing
an equitable share is unpleasant to the more capable and/or task-motivated member of the
group. One way of relieving this negative feeling is to reduce one’s own effort. In fact,
the idea of playing the “sucker” can be so discomforting that group members would

rather fail at a task than carry a “free-rider” (Kerr, 1983).

Given that individual group members may tend to loaf or free-ride off the efforts
of others, and that other group members may reduce their efforts because of the
perception that a person may be free-riding, it appears as if group projects are doomed
from the start. However, other research has demonstrated situations for which
individuals may be motivated to increase their efforts when working on a collective task.
For example, the KOhler effect (1926, cited in Stroebe, Diehl, & Abakoumkin, 1996)
suggests that if there is a moderate level of discrepancy between the abilities of two
members of a dyad, and the task is conjunctive (i.e., all group members must succeed for
the group to succeed; Steiner, 1972), the weaker member may actually work harder in
order to insure that the dyad will succeed (Messé, Hertel, Kerr, Lount, & Park, in press).

Motivation gains have also been demonstrated by the increased efforts of the
stronger member of a dyad on additive tasks. Williams and Karau (1991) suggested that
group members might actually increase their performance to make up for the insufficient
performance of another group member. They proposed that individuals increase their
efforts on additive, collective tasks (i.e., those that involve the summing together of
individual group outputs; Steiner, 1972) to compensate for the anticipated poor
performance of other group members. For example, a dyad member might work harder
for a partner who is either unable and/or unwilling to perform. However, Williams and
Karau (1991) suggested that certain conditions must be met for the social compensation
effect to occur. First, the task must be additive such that the contribution of each group
member is added to the contributions put forth by other group members. Second, group
members must be working on a collective task in which individual effort cannot be

monitored nor rewarded and success is determined by the combined effort of both

members of the dyad. Third, the dyad partner is either unwilling and/or unable to
perform well. Finally, the task is seen as sufficiently meaningful and/or important to the
stronger partner. Furthermore, Karau and Williams (1993) also suggest that the stronger
member must believe that the additional effort that he or she puts forward will lead to
success at the task (i.e., the stronger member believes he or she will be able to
compensate for the weaker member). Under these conditions, the stronger partner will
tend to work harder to make up for the insufficient output of his or her partner so that the
dyad may succeed.

The meaningfulness and importance of the task, as well as the value of an
individual’s input to the group product, are crucial for social compensation to occur. The
task can be meaningful or important because of any of several factors (Karau and
Williams, 1997). For example, a task is meaningful if the issue is personally relevant
(e.g., raising money for a charity), the outcome or reward is personally relevant (e.g.,
cash incentive), or because of concerns over the evaluation of others (e.g., the
experimenter). In short, personal goals or concerns can motivate the stronger group
member to compensate for their partner who is not adequately contributing.

The effects of cohesion on motivation gaiwd losses

Karau and Williams (1997) further elaborated on the social compensation and
social loafing phenomena by demonstrating that the nature of the relationship with one’s
work partner can affect motivation. In one experiment, secretarial students working on a
typing task tended to work harder when working with friends than when working with
strangers, regardless of whether they were working co-actively or collectively. In

addition, in a second experiment, when working on an idea generation task with a

stranger, participants tended to display behavior consistent with the established social
loafing and social compensation effects (i.e., worked less hard with a capable work
partner on a collective task than a co-active task, and worked harder to make up for the
loss of an incapable partner on a collective versus co-active task). However, when
working with a close friend of the other sex on the same task, these standard effects did
not hold. Results demonstrated that when dyad partners were highly cohesive they
tended to match the performance of their partner (i.e., work harder when dyad partner
was capable and less hard when partner was incapable).

Similar matching effects were found in a social loafing study conducted by Karau
and Hart (1998). In this study, cohesion was manipulated by convincing dyad members
that they either held similar views or dissimilar views on various social issues. Results
were comparable to Karau and Williams (1997) such that social loafing was found when
working in dyads with low cohesion (i.e., partner was seen as dissimilar). However, in
the high cohesiveness condition participants tended to perform moderately well in both
the co-active and collective conditions. Thus, when partners were seen as similar,
participants put forth equal effort in both the collective and co-active conditions. The
authors (i.e., Karau & Hart, 1998 and Karau & Williams, 1997) suggest in their
conclusions that it is not clear what aspect of cohesion may be influencing these results.
The mture of cohesion

Cohesion as a psychological construct has been widely explored as a group
phenomenon, yet remains difficult to define and tends to be poorly operationalized
(Mudrack, 1989). Cohesion is generally viewed as a multi-dimensional construct

consisting of numerous social forces that include interpersonal attraction, an individual’s

attraction to the group as a whole, an individual’s level of commitment to the group, and
feelings of similarity toward other group members (Brawley, 1990; Carron, 1988;
F orsyth, 1999). Cohesion has generally been operationalized in the social compensation
literature as interpersonal attraction. For example, Karau and Williams (1997) defined
cohesion as “the degree to which membership in the group was valuable or important to
its members” (p. 157) and was operationalized by individuals who were friends versus
strangers. In Experiment 1, friends consisted of “classmates with whom they would most
like to wor ” (p. 160). In addition, in Experiment 2, the cohesiveness index that was
used assessed the nature of the relationship with one’s co-workers (e.g., how well they
liked their co-worker, desire for future interactions).
Avoiding negative evaluations

In these studies conducted by Karau and colleagues (Karau & Hart, 1998; Karau
& Williams, 1997), the individual’s level of desire to maintain cohesiveness was assessed
(i.e., desire for future interactions, liking of group members) and manipulated across the
cohesiveness conditions. This level of cohesiveness influenced changes in one’s
motivation to perform the group task well. When working with strangers, participant’s
motivation was influenced by the effort necessary to succeed at the task (i.e., work harder
when paired collectively with a stranger who wasn’t expected to do well, work less hard
when working collectively with a stranger who was expected to do well or when working
individually beside a stranger who was not expected to well). In these situations, the
importance of succeeding at the task (or at least performing about as well in a co-active
setting) seemed to determine the level of effort necessary for the task. However, when

working with good friends, fluctuations in performance may have been due to the

importance of one’s friendship. Karau and colleagues (Karau & Hart, 1998; Karau &
Williams, 1993) have argued that there are times when working in a cohesive group on a
collective task that individuals will be more concerned with the welfare of the group than
with their own individual gains or losses. Friends may reduce their efforts on a task in
order to “refrain from attaching blame or stigma to individual members” (Karau &
Williams, 1997, p. 166).

However, can this explanation account for all of the results found in these studies?
First, it is necessary to explain how the matching of a cohesive partner’s performance
serves to maintain a person’s friendship. As previously stated, the meaningfulness of
one’s performance and task importance are necessary for a person to work harder on a
task and one of the factors that make an outcome meaningful or important is the
evaluation of others (e.g., the experimenter). However, when working with a friend,
people would also be concerned about the evaluation of their partner and how the partner
would feel about their performance, especially if they intend to interact with their partner
in the future (i.e., as in the case of friends). So, they may be more concerned with not
wanting to outperform their friends when they know that their friends will do poorly
(e.g., Experiment 2; Karau & Williams, 1997) and thus work less hard. However, when
individuals expect their friend to do well (e.g., when both are part of a typing class;
Experiment 1; Karau & Williams, 1997), and thus will not have to worry about
outperforming them, they may work harder in cohesive groups. In both Situations,
participants would be concerned with how a partner would evaluate them based on their

performance. The partner may evaluate them negatively if they out perform him or her

(i.e., in low ability Situations), or evaluate them negatively if they do not put forth the
necessary effort to succeed (i.e., in high ability situations).

Second, concern over not outperforming a friend (and thus not making him or her
feel bad) could also explain why differences were not found between the individual (i.e.,
co-active) and collective conditions for the cohesive groups. In these previous studies,
across work conditions the poorer performing partner would know if their partner/friend
outperformed them. In the co-active trial, individual scores could be assessed and
compared. Similarly in the collective condition, even though output is combined, poorer
performing partners would have a rough idea how their partner did (e.g., how many ideas
they generated) based on their own performance. In fact, in Karau and Williams (1997,
Experiment 2) participants were explicitly told that they would be told how they did,
either individually (in the co-active condition) or as a team (in the collective condition).
So, if the stronger partner works hard and his or her dyad succeeds at the task, poorer
performing partners will know that they did not contribute their fair share. Thus, in this
case, the situation would work out well, given that the dyad still succeeded. or at least did
well.

However, what if the Stronger partner works hard and the group Still does not do
well? In this case, blame can be placed on the poorer performing member. This Situation
would have little consequence when working with a stranger whom the person will never
interact with again. However, participants in cohesive dyads are working with friends
with whom they would most likely wish to maintain a positive relationship. This concern
over the welfare the dyad partner would also be compounded, especially in mixed gender

dyads, given the level of interpersonal attraction between the dyad members (Clark &

Mills, 1979). By reducing their efforts to match the poorer performance of the friend,
people can avoid the possibility of an invidious comparison with their friend and thus
avoid any negative feelings between themselves and their friend, because neither partner
alone can be blamed for the poor performance.

It is important to note that this comparison can be made in both the collective and
co-active conditions. So, even though in the co-active condition, a poorer performing
friend cannot be blamed for a partner not succeeding, the poorer performing friend can
still be viewed as less capable than the outperforming partner. In these situations, a
person’s apprehension that they may be evaluated negatively by a friend whom they have
outperformed on a task may be more salient than succeeding at the task. When
outperformed by a friend, he or she may choose to distance himself or herself from a
friend in an effort to maintain their self-evaluation (Pleban & Tesser, 1981; Tesser,
1999). To avoid this from happening, the better performer may attempt to match the
performance of the close friend, especially on non-self-relevant tasks (Tesser, 1999).

In summary, as previously stated, the evaluation of others can influence a
person’s performance on a task. When they are concerned about being evaluated
negatively for a poor performance and working collectively with a partner who will not
perform well, people will tend to work harder. However, research on motivation losses
has demonstrated situations when people will lower their performance on a task because
of the anticipated poor performance of another, especially when they believe that their
partner is able (e.g., maintain equity, not look like a sucker). In these situations, their
own self-focused concerns over equity or being made to look like a “sucker” motivates

them to lower their efforts. In such contexts, people would rather fail at the task then be

taken advantage of, thus demonstrating situations when the behavior of their partner
diminishes the significance of succeeding at a task: However, under what conditions
would the evaluation from others lead to a decrease in performance? The current
research suggests that people would decrease their performance when the evaluation is
from a friend who is unable to do well. This would be because of the possible risk of
negative feelings that could arise if the dyad does not succeed and the poorer performing
partner can be blamed, or Similarly, the poorer performing partner would look to be
inferior.
Social comparison

The performance matching found in previous social compensation studies could
be due to the effects of comparing oneself with Similar others. It is plausible that the
manipulation of cohesion utilized in previous social compensation research could have
led participants to compare themselves more readily with their partners to determine the
appropriate level of effort that was expected by their co-workers (i.e., they were working
with friends or similar others). Festinger’s (1954) theory of social comparison processes
suggests that we compare ourselves to similar others in cohesive groups in an attempt to
assess our own ability. In addition, through social comparison a person can assess how
hard he or she has to work on the task, especially if the task requirements are ambiguous.
When working with friends on a task, individuals would know how hard the other friend
may work and what their partners would most likely expect of them. For example, in
Karau and Williams (1997; Experiment 1), participants were all members of a typing
course and partners in the cohesive condition were chosen by the participant. Thus,

participants were very likely aware of how well their partner would perform. In addition,

10

in Experiment 2, participants were informed that their partners expected to generate a
small or large number of ideas on the task (i.e., low or high in ability). It is possible that
partners used this information as a relevant standard of comparison and a goal for their
own performance. Research on group norms has demonstrated that when group members
are aware of the performance norms in a group, they tend to match the efforts of others
(see Forsyth, 1999; Levine & Moreland, 1998). Members of work groups learn what is
expected by others and tend to perform accordingly (Forsyth, 1999).
In summary, the following research will address these two competing questions:
1) Can concern over the negative evaluation from someone we like (i.e., negative
affect between friends because one performed poorer than the other) account
for the matching of effort effects found with cohesive groups in previous
social compensation studies?
2) Can social comparison account for the matching of effort effects found with

cohesive groups in previous social compensation studies?

11

METHOD

W

As displayed in Table 1 (Appendix A), the current study consisted of six
conditions that varied on three separate factors (Similarity, Work condition, Feedback).
Participants completed two phases of an experiment ostensibly investigating
interpersonal dynamics in the workplace. All were led to believe that they worked on the
tasks as part of a mixed sex dyad. The first phase consisted of a demographic and
attitude questionnaire that served as the basis for a similarity manipulation, and the
second phase consisted of two sets of trials of a persistence task that involved the work
condition and feedback manipulations. The similarity manipulation was used to increase
participants’ liking for the other member of the dyad. Participants in the similarity
condition (cells A, B, C, & D in Table 1) received false feedback that the other dyad
member held similar attitudes on a variety of topics. Participants in the no similarity
manipulation condition (cells E & F in Table 1) completed the same attitude
questionnaires, but did not share this information.

For the persistence task, the work and feedback manipulations were presented
between the first set of trials, or ‘practice’ trials, and the second set of ‘performance’
trials. Work condition was manipulated by informing participants that on the
performance trials, their performance scores would be combined with the performance
score of their partner in the collective conditions (cells A, C, & E in Table 1), or that their
performance scores would be calculated separately while working beside a co-worker in

the co-active conditions (cells B, D, & F in Table 1).

l2

Finally, a false feedback manipulation was used to increase concerns of being
evaluated by one’s co-worker.l In the full feedback condition (cells A & B in Table 1),
participants were told that both they and their co-worker were given feedback regarding
each dyad member’s performance on the practice, as well as the dyads’ upcoming
performance, trials. However, in the partial feedback condition (cells C & D in Table 1),
participants were informed that they were the only one to receive this information.
Participants in the no similarity manipulation condition (cells E & F in Table 1) were
given feedback similar to those in the full feedback condition. Participants in cells E and
F are similar to the conditions utilized in previous social compensation studies and,
therefore, should demonstrate the standard social compensation effect. In all dyads,
participants were given feedback that they performed better on the practice trials than
their co-worker.

Differences in persistence task duration between the practice and performance
trials served as the primary dependent measure. Thus, a 2 (Similarity work dyad:
Collective vs. Co-active) x 2 (Feedback: Full vs. Partial) + 2 (No similarity work dyad:
Collective vs. Co-active) between-subj ects design was utilized.

Participants

Data were collected from 147 participants that were recruited through psychology

classes at Michigan State University and received course credit for their participation. A

total of 15 participants were excluded from the final analyses for the following reasons.

 

|In the collective work condition, the other member of the dyad was referred to by the term “partner” or in
reference to the “team.” In the co-active work condition, the other member of the dyad was referred to as a
“co-worker.” However, for the ease of the reader, and consistent with previously published research on
motivation gains (e.g., Williams & Karau, 1991), in this article, when mentioning the other dyad member
across work conditions, this alleged person will be referred to as a co-worker.

13

In response to post-experimental suspicion items, 6 participants expressed suspicion or
obviously did not understand the procedure (e.g., “I know the feedback was fake,” or “I
don’t understand who the third person was”). In addition, in regards to the performance
data, two histograms demonstrated substantial breaks in the data at the end of the
distribution suggesting 3 outliers for the practice trials and 3 outliers for the performance
trials.2 Finally, because all participants were given feedback that they persisted, on
average, 28 seconds longer than the average participant, 3 participants who took very
little time (i.e. less than 30 seconds total) on the practice trials, and thus should have been
suspicious regarding the validity of the performance feedback were also omitted. The
final sample consisted of 132 participants (Men (n = 32), Women (n = 100) .with a mean
age of 19.08 (SD = 1.15).
Materials

Demographics. Questionnaire consisting of first name, sex, age, class level,
college major, marital status, and region of country where raised. In addition to gaining
participant information, these items were used to create a profile of the participant’s co-
worker.

Attitude Items. Questionnaire of 12 attitude items assessing opinions on various

topics relevant to college students using a 7-point Likert scale (e.g., In college, how much

 

2F or the combined practice trials, the break in the data was .7 standard deviations (102 seconds) beyond the
tail of the distribution. The z-scores for the omitted outliers for the practice trials were 2.77, 3.18, and 5.57.
For the combined performance trials, the break in the data was .67 standard deviations ( 105 seconds)
beyond the tail of the distribution. The z-scores for the omitted outliers for the performance trials were
2.73, 5.23, and 5.39. In addition, because these extreme scores were not consistent within subjects across
both practice and performance trials, these outliers also contributed to extreme scores on the calculated
motivation gain variable. Although one of the practice trial scores and one of the performance trial scores
fell within 3 standard deviations of their respective means, and thus fell within the range of a normal
distribution, given the potential for cheating on this task, the substantial break in the data was the clearest

indictor of questionable scores.

14

emphasis do you place on the social aspects of college (e. g., going to parties, getting to
know others?); Byrne, 1971; Appendix B). These items served as the primary similarity
manipulation.

Persistence Ta_s_lg, The persistence task involved having participants hold out their
arm, which had a weighted wristband on it out straight between the two metal poles of a
goal-post-shaped apparatus that had a trip wire approximately 6 inches from the top of
the poles. This trip wire was connected to a mechanism attached to the computer
keyboard. The poles were approximately 4 inches apart so that participants could place
their arm between them. Once the participant’s arm was lowered enough to touch the trip
wire, the mechanism hit the space bar on the computer keyboard and ended the trial.
Participants were instructed that it was important to persist as long as they could at the
task without risking possible injury and undue fatigue. Each participant completed four
trials total. These trials consisted of two practice trials and tWo performance trials, one
with each arm. Participants were instructed to begin each with their dominant arm. The
difference in persistence duration between the practice and performance trials served as
the primary dependent measures.

Post-experimental questionna_irg This questionnaire, as displayed in Appendix C,
contained items assessing participant’s perceptions of their own performance and, when
relevant, that of their co-worker (e.g., How well do you think you did on the practice
trials of the persistence task?) In addition, questionnaire items assessed issues relevant
to the proposed hypotheses (e. g., How much were you concerned that your partner may
evaluate your performance on the performance trials?) and served as manipulation checks

(e.g., How much do you like your partner?; Karau & Williams, 1997). These items were

15

worded to correspond to the appropriate work condition (e.g., partner vs. co-worker).
Each item was rated on a 7-point Likert scale with higher values representing greater
endorsement.

Procedure

Because participants were led to believe that they were paired with a member of
the other sex, all sessions contained both men and women. Participants were told to
anive for the experiment at separate locations. At the designated time, the experimenter
escorted each participant to the lab and into a private workroom in which they were
randomly assigned to one of the six conditions. Because participants were led to believe
that they would be working with another participant, they were able to hear the other
participants as they entered the lab, as well as brief interactions with the experimenter
(e.g., “We have another room set up right there”). Before signing the consent form,
participants were told that they would be participating in an experiment that involved a
variety of tasks, including a persistence task. The rest of the instructions and
questionnaires were given via computer.

Similarity manipulation taala After reading a brief cover story explaining how the
changing global economy had created work environments in which people interact more
readily via electronic means than face-to-face, all participants completed the
demographic and attitude questionnaires. Afier completing the questionnaire and attitude
survey, participants completed a series of procedures that were designed to simulate the
transmission of information between members of the dyad. These procedures included
logging on to and receiving information regarding connection to another computer in the

lab. After participants were “connected” to the other member of the dyad, they received

16

demographic information that supposedly came from the other participant. In all of the
conditions, this other participant had similar demographic questionnaire information as
the actual participant in the study, except that this other participant was of the other sex.
Men were paired with a woman named Stephanie, and women were paired with a man
named Brandon. Furthermore, this alleged participant was single and had recently moved
to the college area.

In addition, participants in the similarity conditions (cells A-D) received
information regarding the responses of this other participant on the attitude items. The
computer program created a profile of attitude responses that was similar to that of the
actual participant, altering a few response numbers in order to alleviate suspicion. To
reinforce the similarity manipulation, after viewing the attitude item responses,
participants in the similarity condition were given an opportunity to communicate with
this other participant by responding to and receiving responses from the other member of
the dyad on an additional four items (See Appendix B.). These responses informed the
participant that the other dyad member liked them and thought they were similar. The
other member of the dyad gave participants the highest rating on three of the four items.
Appropriate time lags between “sending” and “receiving” information (e. g., 5-7 seconds)
were built into the computer program in order to simulate actual transmission.

Persistence Talk; After completing the similarity manipulation portion of the
experiment, the experimenter set up the equipment for the persistence task. For the first
practice trial, participants wrapped the weighted wristband tightly around the wrist of
their dominant arm. This wristband weighed 2 pounds for female participants and 4

pounds for male participants. These weight differences have been pre-tested in other

17

research to correct for gender differences in strength (Lount, Messé, & Kerr, 2000).
Participants were instructed to hold their weighted arm out straight above the trip wire.
As the trial progressed, and as the participant’s arm began to fatigue, participants were
instructed not to raise their wrist above a 4-inch starting point, and, when they lowered
their wrist enough to trip the mechanism, the trial would be over. To insure that
participants did not use their other hand to support the weighted arm, participants used
the hand of their non-weighted arm to press a button that supposedly was necessary to
record their trial time. In addition, to minimize cheating, participants were led to believe
that a small video web camera, located adjacent to the apparatus, would be able to
monitor whether their wrist was properly positioned between the metal poles during each
trial. During each trial, participants remained seated in their chair with both feet flat on
the floor. To insure that the participants understood the instructions, they were asked to
demonstrate the task before beginning their first practice trial.

All participants in a given session began their first practice trial at approximately
the same time and were given a 10 second warning before each trial to position
themselves. After the experimenter started the first practice trial, the remaining portion
of the experiment was conducted entirely by computer. At the end of their first practice
trial, which was performed with their dominant arm, participants had a break period of
130 seconds before performing a second practice trial with their non-dominant arm.
Participants were given one minute to rest, one minute to switch the wristband to their
non-dominant arm and reposition their chair so their weighted non-dominant arm was
lined up with the apparatus, and a 10-second warning period before the second trial

began. Each participant performed one practice trial per arm. These trials served the

18

purpose of orienting participants to the task, establishing a baseline performance to assess
variations in effort across the two sets of performance trials, and setting the stage for
informing participants of the other dyad member’s performance.

At the end of the second practice trial, all participants received instructions
regarding the second set of persistence trials. Participants were informed that they would
be performing the next set of trials with the participant with whom they had previously
shared demographic and attitude information. In the collective condition (cells A, C, &
E), participants were told that the experimenter was interested in the performance of the
dyad and each individual’s performance would be combined with that of their partner. In
addition, participants in the collective condition were told that there would be no way to
monitor their own individual performance on the upcoming performance trials because
only the combined persistence time would be recorded. In the co-active condition (cells
B, D, & F), participants were told that the experimenter was interested in individual
performance, and that the participant’s performance and that of their co-worker would be
evaluated separately on the performance trials.

These instructions presented to participants between the practice and performance
trials were also used to manipulate practice trial performance feedback conditions and set
up the expected discrepancy in performance necessary for social compensation to occur.
Participants were presented with information stating that the amount of feedback
employees receive in the workplace varies. In some work situations employees may
receive regular feedback regarding their performance and the performance of their co-
workers (e. g., performance reviews or recognition of outstanding employees). However,

in other Situations, this feedback may be sparse or non-existent. Participants were told

19

that in order to Simulate these types of work environments, some will receive feedback
regarding their previous performance and other participants would not.

All participants were informed that they had been randomly assigned to receive
feedback regarding their performance on the previous trials; and they received the same
false practice trial performance feedback for themselves and the other member of the
dyad. In addition, all participants were informed that they would be told the final
performance scores on the upcoming trials. In the firll feedback condition (cells A & B),
as well as in the no similarity condition (cells E & F), participants were informed that the
other member of the dyad would also receive this performance feedback. However, in
the partial feedback condition (cells C & D), participants were told that they had been
randomly selected to be the only member of the dyad to receive the practice and
upcoming trials performance information. Finally, all participants were told that they
would have an opportunity to meet their co-worker when the task was completed.
However, participants in the partial feedback condition were told that the experiment
required that they not share any information about the task performance during this
interaction.

The false feedback given to participants stated that they performed, on average,
28 seconds longer on the two practice trials, compared to the average college student who
had previously performed the task, and that their co-worker performed, on average, 36
seconds below average. In addition, participants were informed that their partner’s
performance was relatively stable across the practice trials and, based on the performance

of previous college students, would probably remain stable. However, this false practice

20

trial performance feedback of the participant tended to demonstrate a moderate level of
variability.

To succeed at the upcoming performance trials, participants in the co-active (cell
B, D, & F) condition were informed that they would have to perform above the average
of previous college students who had completed the task. For the collective condition
(cell A, C, & E), the average of the combined scores for the dyad would need to be above
the average of previous college students. The moderate discrepancy in performance, in
conjunction with the variability information, was intended to suggest to participants in
the collective work condition that, with increased effort on the upcoming performance
trials, they could compensate for their poorer performing partner. In addition, any
increase in effort on the upcoming performance trials would primarily need to come from
these participants.

Finally, to fulfill the requirements necessary for social compensation to occur, the
following measures were taken. Participants were informed that some would be involved
in a high supervision condition in which their practice trial data would be recorded, while
others would be randomly assigned to a low supervision condition in which only their
upcoming performance trials would be recorded. This information was presented within
the context that some companies closely monitor the performance of their employees,
while other companies conduct only random assessments. All participants were led to
believe that they had been randomly assigned to the low supervision condition.
Participants were then instructed to follow a series of steps that would ostensibly erase
their practice trial data and thus eliminate any possibility of the experimenter knowing

that they outperformed the other member of the dyad on the practice trials. In addition,

21

participants in the collective condition (cells A, C, & E) were ostensibly assigned a new
team subject number. Participants in the collective condition were told that the computer
would combine the performance of the dyad members on the upcoming performance
trials and record it under this new number. Finally, as an incentive to increase the
importance of task success, all participants were informed that if they were successful on
the first series of upcoming trials they would have the opportunity to exit the experiment
early. In addition, participants who succeeded at the upcoming trials would have the
opportunity to win $50 ($100 in the collective condition, i.e., $50 per participant) in a
drawing at the end of the semester. These instructions between the practice and
performance trials were presented such that each participant progressed through the
information at their own pace, thus giving him or her the necessary time to understand
and process the information. After reading this information, all participants completed
two additional performance trials (i.e., one trial for each arm).

At the end of the second performance trial, participants completed the post-
experimental questionnaire. Participants were debriefed, thanked for their time, and all
participants were given a raffle ticket for the $50 prize. When data collection was
completed, a raffle ticket was randomly drawn and the $50 prize was awarded to one

participant.

22

RESULTS

Overview of Anih/SCS

Analyses were conducted utilizing planned contrasts. Initially, manipulation
check items were analyzed. Then, analyses were conducted to explore the proposed
hypotheses utilizing the persistence trial data. To further explore the proposed
hypotheses, analyses for perceptions of one’s own performance were conducted, followed
by analyses for perceptions of one’s co-worker and task success. In addition to the
planned analyses, emergent patterns of gender differences on perceptions of co-worker
and task performance were also explored.
Manipulation checks

Relevant responses, as displayed in Table 2a (Appendix D), demonstrated that the
similarity manipulation had an effect on how much participants felt they were similar to
and how much they liked their co-worker. These results demonstrate that, on average,
participants in the similarity condition thought they were more similar to their co-worker,
liked their co-workers more, were more concerned about how their co-worker felt, and
were more likely to be willing to work with their co-worker in the future. In addition,
they were more likely to believe that their co-worker felt the same way and would like to
meet them, compared to participants in the no similarity condition. As expected, further
analyses demonstrated that ratings of similarity and liking did not significantly differ

between the collective (cells A, C, & E) and co-active (cells B, D, & F) work conditions

(all as > .40).

23

Other results, also displayed in Table 2a, demonstrated additional differences
between the similarity and no similarity manipulation conditions that are consistent with
the proposed negative evaluation hypothesis. Compared to participants who were not led
to believe they were similar to their co-workers, participants who were led to believe that
they were similar to their co-worker, and thus liked them more, were more concerned
about being evaluated by their co-worker on the performance trials and believed that their
co-worker was more concerned about being evaluated by them.

Analyses of the similarity manipulation checks also revealed a significant work
condition x similarity condition interaction for willingness to work together in the future
(£(1, 120)=5.66, p<.05). Simple effects analyses summarized in Table 2b (Appendix E),
demonstrated that participants in the no Similar-collective condition, compared to the
similarity-collective and no Similarity-co-active conditions, were less likely to be willing
to work with their partner in the future. These results suggest that participants were less
likely to continue to work with a co-worker who they did not like and was not performing
well on a group task.

As displayed in Table 20 (Appendix F), in regards to the feedback manipulation, a
repeated measures analysis of variance demonstrated that overall participants were aware
that that they had worked harder and performed better on the practice trials compared to
their co-worker. In addition, participants who believed that both they and their co-worker
had been given feedback regarding their practice trial scores were more concerned about
being evaluated by their partner, compared to participants who believed they were the
only one to receive performance feedback. Also displayed in Table 2c, participants in the

partial feedback condition, compared to those participants in the firll feedback condition,

24

liked their co-worker more, believed that their co-worker liked them more also, and
tended to believe their co-worker would be more willing to work with them again in the
future. These results suggest that removing the potential for a negative evaluation
contributed to the feelings of liking among participants above the positive feelings due to
similarity.
Persistence trial data

As previously stated, all participants began both sets of trials with their dominant
arm. Analyses exploring the effects of arm for the practice and performance trials
demonstrated a main effect for arm. Across both the sets of trials, participants persisted
longer with their dominant than non-dominant arm (see Table 3a, Appendix G).
However, no significant arm x condition interactions were found for feedback condition,
work condition, similarity condition, nor gender for either the practice or performance
trials. In addition, no three-way interactions were found for either set of trials and no
four-way interaction was found for the performance trials. However, an anomalous
significant feedback condition x work condition x gender interaction x practice arm was
found (E(2,117)=4.04, p<.05). Examination of the means of this four-way interaction
suggested that all but one of the possible 12 cells demonstrated the main effect for
dominant arm (i.e., persisted longer with the dominant arm). This one cell was for men
in the similar-partial-collective condition (cell C). However, this cell also has the
smallest n and thus can not be considered a reliable estimate of this group. Still, any
feedback x work condition x gender interactions involving persistence scores should be
interpreted with caution. Because the trials using the dominant and non-dominant arm

did not interact with work condition, feedback condition or gender, arm trials were

25

collapsed across each trial set and averaged for the persistence duration. Thus, three
persistence task scores were calculated for each participant. The practice trial average
was the average time for the dominant and non-dominant arm on the first set of
persistence trials. A similar average score was calculated for the performance trials.
Finally, an average motivation change variable was calculated by subtracting the practice
trial average from the performance trial average, whereas positive scores demonstrated an
increase in motivation (i.e., persisted longer on the performance trials) and a negative
motivation gain score would demonstrate a decrease in motivation.

As expected, analyses of the practice trial data revealed no significant differences
on the average practice trial persistence time between work conditions, feedback
conditions, similarity conditions, or gender. One significant feedback x work condition
interaction was found, however (E0, 120)=5.11, p<.05). Simple effects tests
demonstrated that, on average on the practice trials, participants in the full feedback-co-
active condition persisted significantly less time (cell B; M=106.24, _S_D=38.50),
compared to participants in the partial feedback-co-active condition (cell D; M=140.21,
S_D=58.51, E(1,117)=4.04, p<.05), and tended to persist less time than participants in the .
full feedback-collective condition (cell A; M=123.74, §Q=62.85, E(1,117)=3.13, p<.08).
No additional statistically significant or marginally significant effects were found. Given
that performance analyses also utilize the average score for the performance trials and the
increase in persistence (i.e., the difference score between the practice and performance
trials) to determine social compensation effects, this difference between the full
feedback-co-active condition and the other conditions should have had minimal effects

on the differences between work conditions and subsequent analyses.

26

As displayed in Table 3b (Appendix H), even though no statistically significant
nor marginal differences were found on the performance trials for gender or gender x
work condition (as > .20), marginal differences were found between work groups on the
performance trials. Consistent with the social compensation effect, participants in the
collective condition tended to persist longer, on average, on the performance trials
compared to participants in the co-active condition. However, this difference was not
sufficient to demonstrate a significant difference between work conditions in motivation
gain between the practice and performance trials. In addition, participants in the
collective condition took significantly longer between the end of the second practice trial
and the beginning of the first performance trial to read the work and feedback condition
manipulation instructions, compared to participants in the co-active work condition.
However, auxiliary analyses demonstrated that controlling for the amount of time
between sets of trials had minimal effect on the results of the Subsequent persistence task
analyses, so this difference in time to read the instructions was not analyzed further.
Proposed hypotheses

Differences in the length of persistence time between the practice and second set
of performance trials for all conditions are displayed in Table 4a (Appendix 1). One of
the primary purposes of this study was to examine whether the previous research results
which have demonstrated a lack of the social compensation when working with a poorer
performing fiiend could be explained better by a concern over a negative evaluation for
outperforming one’s fiiend or by determining one’s performance by comparing with this
friend’s expected performance. To remind the reader, if concern over a negative

evaluation is driving the matching effect, then participants in the similarity conditions

27

should demonstrate a decrease in motivation on the performance trials (i.e., match the co-
worker’s previous and expected performance) across work conditions (cells A & B) when
the co-worker has received feedback regarding their performance, but should show the
standard social compensation effect (i.e., demonstrate a motivation gain) (cell C greater
than cell D) when the co-worker does not have feedback regarding the participant’s
performance, which should be similar to the no similarity manipulation condition (cell E
greater than cell F). Furthermore, it is speculated that once the evaluative concern is
removed, this increase in motivation in the partial-collective condition (cell C) may be
even greater than the motivation gain demonstrated in the no similarity-collective
condition (cell E).

However, if the previous matching effects are better explained by a social
comparison explanation, then participants in the similarity condition should Show a
decrease in effort on the performance trials across work conditions regardless of whether
their co-worker is aware of the dyad’s practice and performance scores or not (cells A &
B and cells C & D). In addition, participants in the similarity-collective condition should
demonstrate a greater decrease in motivation compared to participants in the no-
similarity-collective condition (cell E greater than cell A or C, cells E greater than AC).

An initial examination of the mean differences in persistence between the practice
and performance trials (see Table 4a) demonstrates little change, on average, across the
work and feedback conditions. However, a great deal of variability in motivation was
found across each condition. lrritially, to assess whether social compensation occurred in

the no similarity condition, a planned contrast was conducted between work conditions

28

(cells E & F). However this comparison failed to reach statistical significance
(1(1 l6)=.40, p > .65).

In addition, in order to evaluate the proposed negative evaluation hypothesis,
planned contrasts were conducted within the fill] and partial feedback conditions between
work conditions. Consistent with this hypothesis, no statistical difference was found
between work conditions for the full feedback condition (cells A & B, 1(1 l6)=.29, p >
.75). In addition, as displayed in Table 4a, also consistent with the negative evaluation
hypothesis, participants in the partial feedback-collective condition tended to show an
increase in motivation (i.e., compensate), whereas those in the partial feedback-co-active
condition tended to decrease their effort between the practice and performance trials.
However, participants in the partial feedback-collective condition (cell C) failed to
demonstrate greater motivation gain compared to participants in the no similarity-
collective condition (cell E), as anticipated (1(1 l6)=.48, p > .60).

Furthermore, additional planned contrasts were conducted to assess the social
comparison hypothesis. As previous analyses demonstrated, participants did tend to
match the performance of their co-worker across work conditions in the full feedback
condition, however the same matching was not found in the partial feedback condition
between work conditions. In addition, as previously stated, no significant difference was
found between the no similarity-collective (cell E) and partial feedback-collective
condition (cell C), nor was a difference found for the full feedback-collective condition
(cell A, 1(1 16)=l.01, p >.30). Furthermore planned contrasts demonstrated no significant
differences when comparing the no similarity-collective condition (cell E) with the two

similarity-collective conditions (cells AC, t(1 l6)=.23, p_ >.80). Thus, the social

29

comparison hypothesis was not supported. All of the planned contrasts for the proposed
hypotheses were also conducted separately by gender, however, no statistical nor
marginal results were found (all as >20, see Table 4b, Appendix J).
Pemwtioasfiof own perforrnflaa

Overall, participants were relatively accurate in their perceptions of how hard they
worked on both the practice and performance trials. Statistically significant zero-order
correlations demonstrated participants who actually worked harder (i.e., persisted longer)
also reported working harder for both the practice (r=.24, p<.01) and performance (5.29,
p<.01) trials. HGwever, displayed in Table 5 (Appendix K), participants who believed
they were similar to their co-worker tended to believe they worked harder on the practice
trials, compared to participants who did not receive the similarity manipulation.
Interestingly, also displayed in Table 5, participants who believed that they were the only
one of the dyad members to receive feedback believed they worked harder on the practice
trials, compared to participants who thought both members received this feedback. In
addition, planned contrasts analyses demonstrated significant differences between the
collective and co-active work conditions that are consistent with the social compensation
effect. Participants in the collective work condition believed that they worked harder on
the performance trials, and were more likely to have used their co-worker’s practice trial
feedback as a gauge for the performance trials, compared to participants in the co-active
condition.
Percaptions of task successfi and co-wgker’s performw

Further analyses of participants’ perception of their partners’ performance and the

probability of succeeding at the task demonstrated main effects for similarity condition

30

and work condition, as well as a similarity x gender and similarity x work condition x
gender interaction. As displayed in Table 6a (Appendix L), with regard to responsibility
for success on the performance trials, planned contrasts demonstrated that, consistent
with social compensation (Williams & Karau, 1991), participants in the collective work
condition, compared to those in the co-active condition, took more responsibility for
success on the performance trials. However, participants in the collective condition,
compared to those in the co-active condition, also believed that their partner was more
responsible for success. In addition, compared to participants in the no Similarity
condition, participants who were led to believe that they were similar to their co-worker
took more responsibility for success, as well as failure, on the performance trials.
Furthermore, a significant similarity x gender interaction (_E(1, 120)=4.71, p<.05),
displayed in Table 6b (Appendix M), demonstrated that men in the similarity condition
were significantly more likely than men in the no similarity condition, and marginally
more likely than women in the similarity condition to take responsibility for failure.
Finally, a three-way similarity x gender x work condition interaction (E(l , 120)=6.50,
p_<.05) firrther qualified the relation between gender and responsibility by demonstrating
that similar-collective women tended to feel that their partner was more responsible for
the success of their team, compared to no similarity-collective women. Men in the
collective conditions demonstrated a non-significant trend in the opposite direction with
men in the no similarity-collective condition stating that their partner was more
responsible for success, compared to women in the no similarity-collective condition.
Differences were also found between conditions for perceptions of work

performance for the other member of the dyad. AS displayed in Table 60 (Appendix N), a

31

marginal main effect for feedback was found for the perceptions of how well one’s co-
worker performed on the practice trials, with participants in the full feedback condition
stating that their co-worker performed better, compared to participants in the partial
feedback condition. The main effect of feedback was further qualified by significant and
marginal feedback x gender interactions for perceptions of how well one’s co-worker
performed on the practice trials (E(1, 120)=5.52 , p<.05), as well as how hard one’s co-
worker worked (E (1, 120)=6.66 , p<.05) demonstrating that men in the full feedback
condition thought their co-worker performed better and worked harder on the practice
trials, compared to men in the partial feedback condition and women in the full feedback
condition. However, a marginal feedback x gender interaction for perceptions of how
hard one’s co-worker worked on the performance trials (E (1, 120)=3.6l, p<.07)
demonstrated that women, who thought their co-worker had knowledge of their score on
the practice trials (i.e., full feedback condition), thought these co-workers worked less
hard on the performance trials, compared to women in the partial feedback condition.
Furthermore, as displayed in Table 6d (Appendix 0), similarity and work
condition also interacted with gender to affect perceptions of one’s co-worker’s
performance. First, a marginal similarity x gender interaction (EU, 120)=2.82, p_<.10)
indicated that men in the similar condition believed their co-worker worked harder on the
practice trials, compared to men in the no similarity condition and women in the similar
condition. In addition, marginal and statistically significant similarity x gender x work
condition interactions were found for perceptions of one’s co-worker on the performance
trials. Specifically, for perceptions of how well one’s co-worker did on the performance

trials (E(1, 120)=3.57, p<.07), men in the no similarity-co-active condition believed their

32

partner performed less well, compared to men in the no similarity-collective and
similarity-co-active conditions. Moreover, also displayed in Table 6d, for perceptions of
how hard one’s co-worker worked on the performance trials (E(1, 120)=4.58, p<.05), a
Similar pattern was found. Men in the no similarity-co-active condition believed that
their co-worker worked less hard, compared to men in the no similarity-collective
condition and men in the similarity-co-active condition. Men in the similarity condition
did not statistically differ in their ratings between work conditions. However, the
opposite pattern was found for women. Women in the no similarity condition did not
differ on how hard they thought their co-workers worked, however, women in the
similarity condition believed that their partner worked harder in the collective condition
than their co-worker in the co-active condition.
Gender effects

Additional analyses demonstrated some interesting gender differences. In
general, the performance of men, across the 6 conditions, tended to be affected more
when working with women, compared to women working with men. As displayed in
Table 7a (Appendix P), men, compared to women, were more concerned about how their
co-worker felt, wanted to meet their co-worker more, and were more likely to feel that
their co-worker felt the same way. Furthermore, also displayed in Table 7a, marginal and
Significant similarity x gender interactions for concern over being evaluated by one’s co-
worker (EU, 120)=3.24, p<.08), perception of how concerned one’s co-worker is over
being evaluated (EU, 120)=3. l 7, p<.08), perceptions of how concerned one’s co-worker
is about how the participant feels (E(l , 120)=6.50, p<.05) reinforced that male

participants were more concerned about their opposite gender co~workers, compared to

33

female participants. Simple effects comparisons demonstrated that men who felt that the
women in their dyad were similar to them were more concerned over being evaluated, as
well as thought their co-worker was also more concerned about being evaluated and
concerned about how they felt, compared to men who did not feel similar to their co-
worker and women who felt similar. For the previous analyses, female participants did
not significantly differ between similarity conditions. Finally, as displayed in Table 7b
(Appendix Q) feedback also had an effect on perceptions of how concerned one’s co-
worker was in regards to how the participant felt (E(l , 120)=2.89, p<.10) demonstrating
that women in the partial feedback condition thought their male co-worker was less
concerned about how they felt, compared to men in the partial feedback condition and
women in the full feedback condition. In addition, when participants believed that their
co-worker also received feedback regarding the practice trial performance, men,
compared to women, thought their co-worker was more concerned about how they felt.
As displayed in Table 7c (Appendix R), gender differences were also found for
perceptions of the participant’s own performance. Overall, men believed they worked
less hard on the performance trials compared to women. A work x gender interaction
(lj(1, 120)=4.71, p<.05) indicated that men in the co-active condition felt they worked
less hard on the performance trials, compared to men in the collective condition and
women in the co-active condition. Finally, as displayed in Table 7d (Appendix S), a
marginal work condition x gender x feedback interaction (£(1, 120)=3.69, p_<.06) further
qualified this main effect and interaction. Simple effects comparisons demonstrated that
men in the partial-collective condition and women in the partial-co-active condition

believed that they worked harder on the performance trials, compared to men in the

34

partial-co-active condition. However, women in the partial-co-active condition believed
that they worked harder on the performance trials, compared to women in the full-co-
active condition.

These results, in conjunction with the other gender interactions previously
discussed, suggest that at least for this particular persistence task, men, who outperform
their female co-workers, are more concerned about how they may appear toward them.
However, when women performed better than men, these women believe the men are not
trying or are not concerned with how the women may feel. Given these gender
differences, it is plausible that the proposed negative evaluation hypothesis may be
supported for men but not for women. Specifically, men may be more concerned over
evaluation and thus be more affected by the feedback manipulation. However, as
displayed in Table 4b, no significant differences were found for gender for any of the

planned hypotheses contrasts.

35

DISCUSSION

In summary, some support was found for both the social compensation effect and
the proposed negative evaluation hypothesis. In regards to the standard social
compensation effect demonstrated by Karau and Williams (1991), when considering the
performance trials, participants in the collective work condition felt that they tended to
work harder (i.e., persist longer) and took more responsibility for success, compared to
participants who worked co-actively with another participant. Moreover, participants in
the collective condition did tend to persist longer, on average, over the two performance
trials, compared to participants in the co-active condition.

Furthermore, evidence was found offering limited support for the negative
evaluation hypothesis. Specifically, participants who believed they were similar to their
co-worker, and liked them more, were more concerned about being evaluated by them
and were willing to take more responsibility for success as well as failure on the
performance trials. In addition, when evaluation was a concern (i.e., in the similarity-full
feedback condition), participants were less likely to say that their co-worker performed
worse and they themselves performed better on the practice trials, compared to
participants who believed that their co-worker did not have knowledge of their
persistence trial scores. In addition, a statistically marginal difference in motivation gain
was found between the collective and co-active conditions in the no feedback condition.
These results suggest that, when working with someone who is liked and is similar to
themselves, participants were willing to compensate in the collective condition when they

believed that the poorer performing partner could not evaluate their performance.

36

Evidence to support the social comparison hypothesis would involve differences
between the similarity and no similarity conditions. Co-workers who are seen as similar
to the participants should be more readily used as a comparison standard to assess one’s
performance. Specifically, participants in the similar condition should have been more
likely to endorse using their co-worker’s practice trial performance as a gauge, should
have been influenced more by their co-worker’s practice trial performance, should have
felt they worked less hard on the performance trials, and demonstrated a decrease in
motivation between the practice and performance trials, compared to participants in the
no similarity manipulation condition. However, no evidence of these effects was found.
In fact, the only evidence that participants in the similarity condition were more likely to
compare themselves with their co-worker was that they felt they worked harder on the
practice trials, compared to participants in the no similarity condition, and that
participants in the collective condition were more likely to use their partner’s practice
trial performance as a gauge for their own performance. This suggests that participants in
the Similarity condition may have used the feedback regarding their co-worker’s
performance on the practice trials to evaluate their own practice trial performance and
thus realized they worked harder and may be required to do the same on the upcoming
performance trials. F urthermore, participants in the collective condition may have been
more likely to compare their performance with that of their partner in order to assess how
hard they would have to compensate. In addition, in support of the social comparison
hypothesis, it is plausible that, when solely analyzing the performance data for the full
feedback condition, participants in this condition were using their co-workers as a

comparison standard for their performance on the performance trials. The lack of

37

statistical difference between the full feedback—collective condition and the no similarity-
collective condition could be due to the lack of social compensation (i.e., no increase in
effort) in the latter condition. Finally, it is unclear how much participants utilized their
own performance on the practice trials as a gauge for how much effort to put into the
performance trials. This may be especially true for participants in the co-active condition
given that they were less likely to use their co-worker’s performance as a gauge,
compared to participants in the collective condition. Perhaps their own practice trial
feedback offered participants the necessary comparison information to calculate the
amount of effort required on the second set of performance trials.

Results also demonstrated gender differences that were consistent with prescribed
gender roles. In short, men were more concerned about being evaluated by their female
co-workers than were female participants by their male co-workers. In addition, and
maybe because of this evaluation concern, male participants tended to be more accepting
toward their female co-workers, especially when male participants felt Similar to these
co-workers and believed that they also had received performance feedback. However,
when men felt that they were not similar to their co-worker, they had a more accurate
view of their co-worker’s performance, especially in the co-active conditions, when their
co-worker’s performance had no effect on their own success. Ironically, for male
participants, this also had an effect on perceptions of their own performance. Male
participants working co-actively with a female tended to think they themselves worked
less hard as well. Even though men Should have worked harder when working beside a
female participant especially on a task that could be considered masculine (Johnson &

Staffieri, 1971), it is possible that, because men had already outperformed their female

38

co-worker on the practice trials, and would probably do so on the performance trials, the
situation was less threatening. Because of this, men may not have felt the need to overly
“outshine” their female co-workers (Lount et al., 2000). However, this explanation
doesn’t address why these men would also be more concerned about how their female co»
workers felt. Perhaps, in this situation, men may be more motivated than women to
protect the feelings of their co-worker (Comer, 1995). Given that these female co-
workers are Single, similar to them, and they will have the opportunity to meet, these men
may have been motivated to “put their best foot forward” and present themselves
favorably to their female co-worker, possibly in hopes of getting to know them better.

In contrast, female participants thought that their poorer performing male co-
workers were less concerned about how they felt, specifically when the men had no
feedback regarding their performance scores. In addition, female participants tended to
feel that their poorer performing male partners were more responsible for success in the
collective conditions. Again, consistent with role expectations on masculine tasks, these
results suggest that female participants believed that male participants would work harder
on the performance trials and “pull their own weight” if and when the men were aware
that it was necessary. These results are consistent with Kerr and MacCoun (1985) who
found that both men and women performing a bulb-squeezing task which required
strength, felt that their male partners were more capable at performing the task and thus
were less likely to carry this male partner when they felt he was “free-riding,” compared
to men and women with female partners. Furthermore, for the female participants, it may
have been more important to express their competence at the task (Kerr & Sullaway,

1983). So, even though women believed that their male co-workers were just as

39

responsible for the success for the team, they also stated that they worked hard to
contribute to the success of the team as well.

Finally, the results of this study should be interpreted with caution given several
limitations of this study. First, as previously discussed, cohesion has been
operationalized in many different ways (Brawley, 1990; Carron, 1988; Forsyth, 1999).
However, as in this study, cohesion is often manipulated by means of similarity with and
liking for other members of one’s group (Lott & Lott, 1965). In addition, mixed-sex
dyads were used in this study. This was done in part to replicate the design used in other
social compensation studies (Karau & Williams, 1997; Experiment 2) and also to
increase the level of attraction between dyad members. Participants who felt similar to
their co-workers and thus were more attracted should have been more willing to
compensate for them (Clark & Mills, 1979). However, this mixed-sex dyad manipulation
may have generated more concern about oneself or the other member of the dyad, instead
of feelings of connectedness with the group. In this way, the level of cohesiveness
between dyad members may be better understood as an individual phenomenon instead of
a group phenomenon, as cohesion has generally been considered.

Other limitations should also be considered. For example, some of these results
reach only marginal statistical significance. In addition, there are condition x gender
cells with very few participants, specifically the cells comprised of men. These small cell
sizes may lead to non-significant results because of a lack of statistical power.
Furthermore, even though some of these comparisons involving small cell sizes are
statistically significant, the small cell sizes may lead to incorrectly interpreted

conclusions because there are too few participants to make accurate generalizations.

40

Moreover, most of the results presented are based on self-reports that are not fully
supported by the behavioral data. Finally, the lack of statistical significance between
conditions for the persistence task may be due in part to the high degree of variability on
performance scores. One unforeseen property of the persistence task was the large
degree of variability across participants. It is unclear whether this variability may have
contributed to the lack of social compensation effects in the current study.

Future research assessing social compensation should consider these limitations
as well as allow for the following improvements. Previous research that has
demonstrated the social compensation effect may be due in part to participants “loafing”

in the co-active work conditions. A better assessment of motivation differences in the

 

collective and co-active work conditions may be to compare participants in the work
conditions with participants working completely alone. Future research assessing
individual performance is suggested as a more accurate comparison given that
participants in the co-active condition would most likely be influenced by the
performance of their co-worker.

In addition, future research on social compensation should address other relevant
variables that may moderate the social compensation effect. For example, if there is an
assumption or expectation by the poorer-performing friend that the stronger member will
compensate, then the stronger member may be more concerned with not meeting the
expectations of the friend, compared to concerns over making the friend look bad, and
thus try even harder. Furthermore, that which makes the task meaningful may influence
the level of motivation gain on the task. Previous studies (e.g., Karau & Williams, 1997)

have made the task meaningful by implying that performance is associated with

41

intelligence. However, in this study, performing well was meaningful in that participants
could win money and leave the experiment early. Although both outcomes should be
meaningful and lead to motivation gains, it may be possible that they are psychologically
distinct (i.e., demonstrating low intelligence may have more negative consequences).
Future research should explore whether these differences moderate the social
compensation effect.

In addition, some methodological changes should be made in future studies. First,
the persistence task utilized in this study was not an efficient measure of motivation
gains. Although the task could prove useful in future research, the current method proved
to allow for too much individual variability. Perhaps, this could be corrected if the
weights were heavier, possibly twice the weight utilized in this study, or if there were
more persistence trials. Second, perhaps the relatively anonymous nature of one’s partner
was not enough to motivate participants to compensate in the collective condition. Future
studies could allow participants to briefly meet their partner or allow them to see their
partner via a webcam before beginning the task. In each case, the other participant could
be a confederate so this other person would be similar across conditions.

Although the pr0posed hypotheses were not unequivocally supported, this
research does offer indirect evidence as to the nature of evaluative concerns from a
poorer performing friend when working collectively on an additive task. These results
suggest that further research is needed to explore possible moderators to the effect of

social compensation.

42

APPENDICES

43

 

APPENDIX A

Table 1. Experimental design and corresponding cell reference letters.

Similarity Manipulation Conditions

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Work Feedback
Condition Full Partial
Collective A C

(n=23) (n=21)
Co-A ctive B D
(n=21) (n=22)
No Similarity Control
Conditions
Work Feedback
Condition Full
Collective E
(n=23)
Co-A ctive F
(n=22)

 

 

APPENDIX B

Attitude items:
As they usually function, how do you feel about fraternities and sororities?

VERY MUCH AGAINST 1 2 3 4 5 6 7 VERY MUCH IN FAVOR
In regards to religion, I consider myself:

NOT RELIGIOUS AT ALL 1 2 3 4 5 6 7 EXTREMELY RELIGIOUS
I feel that most professors are very much concerned about the needs of students

NOT CONCERNED AT ALL 1 2 3 4 5 6 7 EXTREMELY CONCERNED
In general, how do you feel about people smoking in areas where you work or live?

IDONOTLIKEITATALL 1 2 3 4 5 6 7 ILIKEIT

In college, how much emphasis do you place on the social aspects of college (e.g., going
to parties, getting to know others)?

NOTMUCHATALL 1 2 3 4 5 6 7 ’ QUITEALOT
How do you feel about the university grading system as it now exists?
IT NEEDS TO BE CHANGED l 2 3 4 5 6 7 IT WORKS WELL
I consider my political Views generally to be:
CONSERVATIVE l 2 3 4 5 6 7 LIBERAL
In regards to abortion, I am:
AGAINSTIT l 2 3 4 5 6 7 INFAVOROFIT
I believe that two people should get to know each other very well before having sex.
CASUAL SEX IS OKAY l 2 3 4 5 6 7 CASUAL SEX IS NOT OKAY
I would most like to spend my time:

DOING SOMETHING ALONE 1 2 3 4 5 6 7 BEINGWITH OTHERS

45

During my free time, I would rather:
STAY IN/WATCH TV 1 2 3 4 5 6 7 GET OUT/DO SOMETHING ACTIVE
I would rather spend a Saturday night:

ONAROMANTICDATE 1 2 3 4 5 6 7 ATAROWDYPARTY

Responses to partner:
In regards to working with my partner on a task, I believe I would:
NOTLIKEITATALL l 2 3 4 5 6 7 LIKEITVERYMUCH
In regards to spending time with my partner, I believe I would:
NOTLIKEITATALL 1 2 3 4 5 6 7 LIKEITVERYMUCH
The attitudes of my partner are:
HIGHLY DISSIMILAR TO MINE 1 2 3 4 5 6 7 HIGHLY SIMILAR TO MINE
I believe my partner is:

VERY UNINTELLIGENT 1 2 3 4 5 6 7 VERY INTELLIGENT

46

APPENDIX C

Collective work condition

Using the scales provided, please answer the following items, pertaining to the task that
you just completed, by circling the one number that best describes you.

1)

2)

3)

4)

5)

6)

7)

8)

9)

How well do you think you did on the practice trials of the persistence task?
NOT WELL AT ALL 1 2 3 4 5 6 7 VERY WELL
How well do you think you did on the team trials of the persistence task?

NOTWELLATALL l 2 3 4 5 6 7 VERYWELL

How hard did you work on the practice trials of the persistence task?

NOTHARDATALL 1 2 3 4 5 6 7 VERYHARD

How hard did you work on the team trials of the persistence task?
NOTHARDATALL 1 2 3 4 5 6 7 VERYHARD

How well do you think your partner did on the practice trials of the persistence task?
NOT WELL AT ALL 1 2 3 4 5 6 7 VERY WELL

How well do you think your partner did on the team trials of the persistence task?
NOTWELLATALL 1 2 3 4 5 6 7 VERYWELL

How hard do you think your partner worked on the practice trials of the persistence
task?

NOTHARDATALL 1 2 3 4 5 6 7 VERYHARD

How hard do you think your partner worked on the team trials of the persistence task?
NOTHARDATALL 1 2 3 4 5 6 7 VERYHARD

How likely was it that you and your partner could succeed on the team trials?

NOT LIKELY AT ALL 1 2 3 4 5 6 7 VERY LIKELY

47

 

10) How much were you concerned that your partner may evaluate your performance on
the team trials?

NOT AT ALL CONCERNED 1 2 3 4 5 6 7 VERY MUCH CONCERNED

11) How much do you think your partner was concerned that you may evaluate his or her
performance on the team trials?

NOT AT ALL CONCERNED l 2 3 4 5 6 7 VERY MUCH CONCERNED

12) How similar do you feel you are to your partner in your attitudes and beliefs, and in
personal characteristics (e.g., age, major)?

NOT AT ALL SIMILAR l 2 3 4 5 6 7 VERY SIMILAR
13) How similar do you think your partner feels he or she is to you?
NOT AT ALL SIMILAR 1 2 3 4 5 6 7 VERY SIMILAR
14) How much do you like your partner?
DO NOT LIKE AT ALL 1 2 3 4 5 6 7 LIKE VERY MUCH
15) How much do you think your partner likes you?
DONOT LIKE ATALL l 2 3 4 5 6 7 LIKE VERY MUCH
16) How much would you be willing to work with your partner again in the future?
NOT WILLING AT ALL 1 2 3 4 5 6 7 VERY WILLING

17) How much do you think your partner would be willing to work with you again in the
future?

NOT WILLING AT ALL 1 2 3 4 5 6 7 VERY WILLING
18) How much would you like to meet your partner?

NOTATALL I 2 3 4 5 6 7 VERYMUCH

19) How much do you think your partner would like to meet you?

NOTATALL l 2 3 4 5 6 7 VERYMUCH

48

20) How concerned are you about what your partner feels about you?
NOT AT ALL CONCERNED l 2 3 4 5 6 7 VERY MUCH CONCERNED
21) How concerned do you think your partner is about how you feel about him or her?
NOT AT ALL CONCERNED l 2 3 4 5 6 7 VERY MUCH CONCERNED
22) How important was it to you personally to succeed at this task?
NOTATALL l 2 3 4 5 6 7 VERYMUCH
23) How important was it to you personally to succeed in other aspects of your life?
NOTATALL 1 2 3 4 5 6 7 VERYMUCH
24) How responsible did you feel for the success of your team?
NOT AT ALL RESPONSIBLE l 2 3 4 5 6 7 VERY MUCH RESPONSIBLE
25) How responsible do you feel your partner was for the success of your team?
NOT AT ALL RESPONSIBLE 1 2 3 4 5 6 7 VERY MUCH RESPONSIBLE
26) How responsible would you have felt if your team failed?
NOT AT ALL RESPONSIBLE l 2 3 4 5 6 7 VERY MUCH RESPONSIBLE
27) How responsible do you feel your partner would have been if your team failed?
NOT AT ALL RESPONSIBLE l 2 3 4 5 6 7 VERY MUCH RESPONSIBLE

28) How much do you think your performance on the team trials was influenced by the
expected performance of your partner?

NOT AT ALL INFLUENCED l 2 3 4 5 6 7 VERY MUCH INFLUENCED

29) How much did you use your work partner’s performance on the practice trials as a
gauge for your own performance on the team trials?

NOTATALL l 2 3 4 5 6 7 VERYMUCH

Co-active work condition

Using the scales provided, please answer the following items, pertaining to the task that
you just completed, by circling the one number that best describes you.

49

1)

2)

3)

4)

5)

6)

7)

8)

9)

How well do you think you did on the practice trials of the persistence task?
NOT WELL AT ALL 1 2 3 4 5 6 7 VERY WELL

How well do you think you did on the persistence task when you worked with a co-
worker?

NOT WELL AT ALL 1 2 3 4 5 6 7 VERY WELL

How hard did you work on the practice trials of the persistence task?
NOTHARDATALL l 2 3 4 5 6 7 VERYHARD

How hard did you work on the persistence task when you worked with a co-worker?
NOTHARDATALL l 2 3 4 5 6 7 VERYHARD

How well do you think your co-worker did on the practice trials of the persistence
task?

NOTWELLATALL l 2 3 4 5 6 7 VERYWELL

How well do you think your co-worker did when working with you on the second set
of persistence trials?

NOT WELL AT ALL 1 2 3 4 5 6 7 VERY WELL

How hard do you think your co-worker worked on the practice trials of the
persistence task?

NOTHARDATALL 1 2 3 4 5 6 7 VERYHARD

How hard do you think your co-worker worked when working with you on the second
set of persistence trials?

NOTHARDATALL l 2 3 4 5 6 7 VERYHARD

How likely was it that you and your co-worker could succeed on the second set of
trials?

NOT LIKELY AT ALL 1 2 3 4 5 6 7 VERY LIKELY

10) How much were you concerned that your co-worker may evaluate your performance

on the second set of trials?

NOT AT ALL CONCERNED l 2 3 4 5 6 7 VERY MUCH CONCERNED

50

l 1)How much do you think your co-worker was concerned that you may evaluate his or
her performance on the second set of trials?

NOT AT ALL CONCERNED 1 2 3 4 5 6 7 VERY MUCH CONCERNED

12) How similar do you feel you are to your co-worker in your attitudes and beliefs, and
in personal characteristics (e. g., age, major)?

NOT AT ALL SHVIILAR 1 2 3 4 5 6 7 VERY SIMILAR
13) How similar do you think your co-worker feels he or she is to you?
NOT AT ALL SIMILAR l 2 3 4 5 6 7 VERY SIMILAR
14) How much do you like your co-worker?
DONOTLIKEATALL 1 2 3 4 5 6 7 LIKEVERYMUCH
15) How much do you think your co-worker likes you?
DONOTLIKEATALL 1 2 3 4 5 6 7 LIKEVERYMUCH
16) How much would you be willing to work with your co-worker again in the future?
NOT WILLING AT ALL 1 2 3 4 5 6 7 VERY WILLING

17) How much do you think your co-worker would be willing to work with you again in
the future?

NOT WILLING AT ALL 1 2 3 4 5 6 7 VERY WILLING

18) How much would you like to meet your co-worker?

NOTATALL l 2 3 4 5 6 7 VERY MUCH
19) How much do you think your co-worker would like to meet you?

NOTATALL 1 2 3 4 5 6 7 VERYMUCH
20) How concerned are you about what your co—worker feels about you?

NOT AT ALL CONCERNED l 2 3 4 5 6 7 VERY MUCH CONCERNED

21) How concerned do you think your co-worker is about how you feel about him or her?

NOT AT ALL CONCERNED l 2 3 4 5 6 7 VERY MUCH CONCERNED

51

22) How important was it to you personally to succeed at this task?
NOTATALL 1 2 3 4 5 6 7 VERYMUCH
23) How important was it to you personally to succeed in other aspects of your life?
NOTATALL 1 2 3 4 5 6 7 VERYMUCH
24) How responsible did you feel for the success of you and your co-worker?
NOT AT ALL RESPONSIBLE 1 2 3 4 5 6 7 VERY MUCH RESPONSIBLE
25) How responsible do you feel your co-worker was for the success of both of you?
NOT AT ALL RESPONSIBLE l 2 3 4 5 6 7 VERY MUCH RESPONSIBLE
26) How responsible would you have felt if you and your co-worker failed?
NOT AT ALL RESPONSIBLE l 2 3 4 5 6 7 VERY MUCH RESPONSIBLE

27) How responsible do you feel your co-worker would have been if you and your co-
worker failed?

NOT AT ALL RESPONSIBLE l 2 3 4 5 6 7 VERY MUCH RESPONSIBLE

28) How much do you think your performance on the second set of trials was influenced
by the expected performance of your cc-worker?

NOT AT ALL INFLUENCED 1 2 3 4 5 6 7 VERY MUCH INFLUENCED

29) How much did you use your co-worker’s performance on the practice trials as a
gauge for your own performance on the second set of trials?

NOTATALL l 2 3 4 5 6 7 VERYMUCH

52

 

APPENDIX D

Table 2a. Means for sirrrilarity/liking manipulation check and ancillary post-experimental questionnaire

items.

 

Similarig Manipulation

Sirrrilar (n = 87)

cells A-D

No Sirrrilar (n = 45)

cells E-F

1(120)

 

Manipulation check items

How similar you are to co-worker?

How similar co-worker is to you?

How much do you like your co-worker?

How much does co-worker like you?
Concerned how co-worker feels about you?
Co-worker concerned about how you feel?
You willing to work with co-worker in future?
Co-worker willing to work with you in future?
Would you like to meet your co-worker?

How much would co-worker like to meet you?

5.89 (5.12 = 0.95)

5.71 (_S_D = 0.93)
5.26 (a = 1.03)
5.13 (S_D = 1.02)
3.54 (g = 1.78)
3.62 (a = 1.71)
5.43 (S_D = 0.93)
5.43 (S_D = 0.95)
5.34 (pp = 1.26)

5.16 (E = 1.21)

4.42 (SD = 1.14)

3.93 (at; = 1.03)
4.49 (S_D = 0.82)
4.27 (_S_D = 0.72)
2.73 (s1; = 1.62)
3.31 (s1; = 1.43)
4.47 (S_D = 0.89)
4.49 (S_D = 0.89)
4.98 (s1; = 01.2)

4.51 (_S_Q = 0.99)

6.26"

8.13"

3.81"

4.32"

2.66"

2.41"

4.15"

4.34"

1.46

2.75"

 

Ancillag items

Co-worker will evaluate your performance?

Co-worker concerned about evaluation?

3.53 (st; = 1.98)

4.05 (_sp = 1.69)

2.98 (E = 1.88)

3.69 (S_D = 1.59)

2.31“

2.04“

 

Note: All response items on a 7-point Likert scale with higher numbers reflecting greater endorsement.
Refer to Table l for cell information. ‘p<.05, I""‘p<.01.

53

APPENDIX E

Table 2b. Means for willingness to work with co-worker in the future for work condition x similarity

 

 

interaction.
Work Condition Similarig Manipulation
Similar Not Similar
Collective cells A & C cell E
_M_ = 5.82. M. = 45711.6
E = 0.92 S12 = 0.59
n = 44 n = 23
Co-Active cells B & D cell F
M = 5.47 _M = 4.95,
E = 0.83 E = 1.17
n = 43 n = 22

 

Note: Response item on a 7-point Likert scale with higher numbers reflecting greater endorsement. Refer
to Table l for cell information. Means with identical subscripts represent significant simple effects
comparisons; 'E(l,120)=l9.01, p<.01, t’£(l,120)=4.04, p<.05.

54

APPENDIX F

Table 2c. Means for feedback manipulation check and ancillary post-experimental questionnaire items.

 

Manipulation check items

How well perform on practice trials?

How hard work on practice trials?

Co—worker will evaluate your performance?

Co-worker concerned about evaluation?

Participant
5.11 (SQ= 1.06)

5.30 ($2 = 1.29)

Co-worker
3.52 (S_D = 1.28)

3.91 (sp= 1.41)

Ema

Full (11 = 44)
cells A-B
4.14 (SD = 1.86)

4.30 (E = 1.58)

Partial (n = 43)
cells CD
2.91 (E = 1.91)

3.79 (E = 1.78)

EU, 131)

105.74”

84.52"

£020)
2.86"

1.05

 

Ancillagg items

How much do you like your co-worker?
How much does co-worker like you?

Co-worker willing to work with you in future?

5.02 (S_D = 1.07)
4.86 ($2 = 1.05)

5.27 (SD = 1.00)

5.51 (S_D = 0.94)

. 5.40 (_S_D_ = 0.93)

5.58 (SD = 0.88)

-2.l6*
-2.59*

-1.80*

 

Note: All response items on a 7-point Likert scale with higher numbers reflecting greater endorsement.
Refer to Table l for cell information. fp<.08, ‘p<.05, "p<.01.

55

APPENDIX G

Table 3a. Mean persistence duration in seconds on practice and performance trials for dyad participants.

 

 

Dominant Arm Non-dominant Arm E
Practice Trials (n=129) M = 134.66 M = 113.85 18.74"
_S_Q = 67.39 S1; = 53.86
Performance Trials (11:13]) M = 129.24 M = 122.74 375*
_S_D_ = 60.51 S1; = 57.55

 

Note: Times for all four trials were not recorded for four participants. Ip<.06, df=( 1,1 19); "p<.01, M
=(1,1 17).

56

 

APPENDIX H

Table 3b. Means for persistence trial duration in seconds for work conditions and by gender.

 

Average for performance trials

Work condition

 

Collectiveb Co-Active
cells A, C & E cells B, D, & F
G_crLde_r
Men (11 = 12) Women (11 = 54) Men (11 = 20) Women (11 == 45)

147.42 (S_D = 43.55) 129.57 (SD = 61.61) 111.95 (E = 47.59) 122.22 (_S_D = 54.54)

 

Work condition

Collective (n = 67) Co-Active (n = 65) t(120)'
cells A, C & E cells B, D, & F
First performance trial 135.99 (S_D_ = 62.46) 123.63 (S_D = 58.78) 1.54
Second performance trial 130.86 (S1; = 64.13)” 114.49 (§I_) = 49.13) 1.95*
Average for performance trials 132.81 (S2 = 58.86)b 119.06 (SD = 52.35) 181'
Reading time between trial sets 342.12 (S_D = 86.03)c 263.23 (S=12 = 78.00) 4.51"

 

 

Note: ‘‘Three times were not recorded, so degrees of freedom for second performance trial and average
for performance trial = 119, degrees of freedom for reading time = 118. n=66, cn=65. Ip<.09, Mp<.01.

57

APPENDIX I

Table 4a. Mean difference in seconds between practice and performance persistence trials for all

experimental conditions.

 

 

 

Work Condition Similarifi Manipulation
Similarity No similarity
Feedback
Full Partial
Collective cell A cell C cell E
M = 4.63 M = 12-34. M = 8.29
$=29 18 _SQ=42.81 §Q=39.63
n = 23 = 19* n = 21 *
Co-Active cell B cell D cell F
M = -8.74 M = -9. 14, M = 7.59
§Q=28.80 §Q=38.6I _S_D=26.18
n = 21 n = 22 n = 22

 

Note: *Times for all four trials were not recorded for four participants, so no motivation gain score could
be calculated for these participants. Means with identical subscripts represent marginal planned

comparisons, ‘t(l l6)=l .73, p<.09.

58

APPENDIX J

Table 4b. Mean difference in seconds between practice and performance persistence trials for all

experimental conditions by gender.

 

Work Condition Gender Similarigl Manipulation

Similarity No similarity
M9215
Full Partial
Collective Men M = -20.38 M = 19.25 M = 7.10
S_D = 30.68 S_D = 99.35 SD = 28.91
11 = 4 n = 2" n = 5
Women M = 9.90 M = 11.53 M = 8.66
_S_D_ = 26.76 _S_Q = 37.93 S_D = 43.25
n=19 n=17" n=l6*
Co-Active Men M = 943 M = ~16.00 M = -7. l 7
S_D = 22.94 _S_D_ = 48.70 _S_D = 17.61
n=7 n=7 n=6
Women M = -8.40 M = -5.93 M = 13.13
SD = 32.14 S_D = 34.42 SD = 27.14
11 = 14 n = 15 n = 16

 

Note: *Times for all four trials were not recorded for four participants, so no motivation gain score could
be calculated for these participants.

59

 

APPENDIX K

Table 5. Means for perceptions of own performance on post-experimental questionnaire items for

 

similarity, feedback and work conditions.

 

How hard did you work on practice trials?

Similarly manipulation

 

 

 

Similar (n = 87) No Similar (n = 45) 1(120)
cells A-D cells E-F I
5.44 ($2 = 1.30) 5.04 (at; = 1.24) 173* ,
Feedback
Full (11 = 44) Partial (n = 43) $020)
cells A-B cells C-D “L.
5.l6(_S_D_= 1.40) 5.72 (SD: 1.14) 224*

 

Work condition

Collective (n = 67) Co-Active (n = 65) 1(120)
cells A, C & E cells B, D, & F
How hard on performance trials? 6.33 (E = 0.81) 5.77 ($12 = 1.36) 3.49"
Used co-worker as a gauge? 4.76 (52 = 1.84) 4.25 (S_D = 1.71) 2.30“

 

Note: All response items on a 7-point Likert scale with higher numbers reflecting greater endorsement.
Refer to Table 1 for cell information. Ip<.09, ‘p<.05, "p<.01.

 

60

 

APPENDIX L

Table 6a. Means for perceptions of responsibility for success and failure on the performance trials for work

condition and similarity main effects.

 

 

Work Condition
Collective (n = 67) Co-Active (n = 65)
cells A, C, & E cells B, D, & F 1(120)
Participant responsible for success 5.48 (SQ = 1.13) 5.02 (S_D = 1.33) 2.78"
Co-worker responsible for success 5.10 (S1; = 1.07) 4.45 (S_D = 1.06) 3.27"

 

Similarly manipulation
Similarity (n = 87) No Similarity (n = 45)

cells A-D cells E-F
Participant responsible for success 5.46 (S_D = 1.10) 4.84 (S_D = 1.43) 2.97"
Participant responsible for failure 4.98 (S_D = 1.34) 4.53 (SD = 1.50) 2.59“

 

Note: All response items on a 7-point Likert scale with higher numbers reflecting greater endorsement.
Refer to Table l for cell information. ‘p<.05, I"“p<.01.

61

APPENDIX M

Table 6b. Means for perceptions of responsibility for success and failure on the performance trials for

similarity x gender and similarity x gender x work condition interactions.

 

Similarity Manipulation
Similarity (cells A-D) No Similarity (cells E-F)
Work Condition Gender

Men Women Men Women

Participant responsible for failure

M = 5.43”, M =4.83(, M = 4.008 M = 4.71
_S_D_=0.98 S_I_)_=1.41 §Q=2.19 S_D=1.19
n=21 n=66 n=ll n=34

 

Co-worker responsible for success

Collective cells A & C cell E
M = 4.86 M = 5.27c M = 5.60,, M = 4.72,...
_S_D =1.46 fl = 0.99 SD = 1.34 _S_Q = 0.96
n = 7 n = 37 n = 5 n = 18
Co-active cells B & D cell F
M = 4.43 M = 4.59 M = 3.50 M = 4.56
§Q=0.85 S_D=1.12 §Q=1.52 S_D=0.81
n=14 n=29 n=6 n=16

 

Note: All response items on a 7-point Likert scale with higher numbers reflecting greater endorsement.
Refer to Table 1 for cell information. Means with identical subscripts represent significant or marginal
simple effects comparisons; 'E(l,120)=7.40, p<.01, l’E(1,120)=2.86, p<.10, cE(1,120)=3.39, p<.07,
‘E(l,120)=2.82, p<.10.

62

APPENDIX N

Table 6c. Means for perceptions of co-worker’s performance for feedback main effect and feedback x

gender interactions.

 

Feedback
Full (11 = 44) Partial (n = 43) E(1,120)
cells A-B cells C-D

How well did co-worker perform on practice trials?

 

 

M=3.70 M=3.51 317*
S_D = 1.29 SD = 1.35
Feedpack x Gender
M01 = 11) Women (11 = 33) Men (11 = 10) Women (11 = 33)
M = 4.64._b M = 3.391, M = 3.30,| M = 3.58
_S_D=0.92 §D=L25 S_D_=l.42 S_D=1.35

 

How hard did co-worker work on practice trials?
M = 5.27“, M = 3.58., M = 3.90c M = 4.00

S_D=0.79 _S_D=l.50 §2=1.37 _S_D=1.52

 

How hard did co-worker work on the performance trials?
M = 5.64 M = 5.27e M = 5.20 M = 5.73,

fl=092 S_D=1.18 _S_D=ll3 §Q=0.88

 

Note: All response items on a 7-point Likert scale with higher numbers reflecting greater endorsement.

 

 

Refer to Table 1 for cell information. Means with identical subscripts represent significant simple effects

comparisons; 'F_(l,120)=5.48, p<.05, bl_=_(l,120)=7.13, o<.01, ‘l_=_(1,120)=4.97, p<.05, ‘E(I,120)=12.04,
p<.01, °E(1,120)=4. 12, p<.05. *p<. 10.

63

 

 

APPENDIX 0

Table 6d. Means for perceptions of co-worker for similarity x gender x work interactions.

 

Similarig manipulation
Similar (cells A-D) No similarity (cells E-F)
Work condition gm
Men Women Men Women

How hard did co-worker work on practice trials?

M = 4.62“, M = 3.79., M = 3.55a M = 3.82
S_D=1.28 S_D=1.51 §Q=L44 S_D=1.14
n=21 n=66 n=ll n=34

 

How well did co-worker do on performance trials?

Collective cells A & C cell E
M= 5.14 M=5.08 M=5.00c M=4.83
§D=l.07 S_D=1.32 §Q=l.41 S_D=0.79
n = 7 n = 37 n = 5 n =18
Co-active cells B & D cell F
M = 4.86,, M = 4.86 M = 3.67“, M = 5.38
fl=l.l7 S_D=1.13 S_D=l.03 §D=096
n=l4 n=29 n=6 n=l6

 

How hard did co-worker work on performance trials?

Collective M = 5.43 M = 5.73g M = 5.40c M = 5.39
Q=079 _S_D=I.IO §Q=L34 M2=098
Co-active M = 5.43; M = 5.218 M = 4.17,; M = 5.50
_S_D=IJZ S_D=0.94 _S_D=1.l7 _S_=1.03

 

Note: All response items on a 7-point Likert scale with higher numbers reflecting greater endorsement.
Refer to Table l for cell information. Means with identical subscripts represent significant or marginal
simple effects comparisons; ‘E( l,120)=3.39, p<.07, I’E(1,120)=4.58, p<.05, °_E(1,120)=3.76, p<.06,
‘E(1,120)=4.58, p<.05, ‘E(1,120)=3.84, p<.06, fE(1,120)=6.20, p<.05, l‘E(l,120)=4.41, p<.05.

64

 

APPENDIX P

Table 7a. Means for liking and concern over evaluation post-experimental questionnaire items for gender

and for similarity x gender interactions.

 

Gan—chit
Men (11 = 32) Women (n = 100) E (1, 120)
Concerned how co-worker feels? 3.94 (_S_Q = 1.81) 3.05 (SD = 1.70) 655*
Like to meet your co-worker? 5.75 (SD = 1.02) 5.05 (E = 1.27) 7.84"
Co-worker concerned you feel? 4.28 (§I_) = 1.73) 3.27 (S_D_ = 1.51) 11.29"
Co-worker like to meet you? 5.50 (_SQ = 1.05) 4.76 (S_D_ = 1.16) 11.89"

 

Slrmlarity Manipulation

 

Similarity (cells A-D) No Similarity (cells E-F)
91m

Men (n = 21) Women (11 = 66) Men (11 = l 1) Women (11 = 34)
Co-worker will evaluate your performance?

M = 4.38,, M = 3.26., M = 2.732 M = 3.06

52:1.43 S_D=2.06 fl=210 §Q=L82
Co-worker concerned about evaluation?

M = 4.62“, M = 3.86,, M = 3.36c M = 3.79

S_D=l.24 S_D_=l.78 S_D=180 _S_D=IS3
Co-worker concerned about how you feel?

M=4.81ey M=3.24f M=3.27c M = 3.32

§Q=l.40 S12=1.63 S_D_=1.90 §Q=l.27

 

Note: All response items on a 7-point Likert scale with higher numbers reflecting greater endorsement.
Refer to Table 1 for cell information. Means with identical subscripts represent significant or marginal
simple effects comparisons; 'E(1,120)=5.52, p<.05, l’_1_7"_(1,120)=4.45, p<.05, T_(l,120)=4.80, p<.05,
”_E(1,120)=3.13, p<.08, °E(1,120)=8.01, p<.01, fE(1,120)=16.73, p<.01. ‘p<.05, “p<.01.

 

65

 

APPENDIX Q

Table 7b. Means for co-worker concern over evaluation post-experimental questionnaire items for

feedback x gender interaction.

 

Feedback
Full (cells A-B) Partial (cells C-D)
Qflld—fl
Men (11 = 11) Women (11 = 33) Men (11 = 10) Women (11 = 33)
Co-worker concerned about how you feel?
M = 4.73c M = 3.701,, M = 4.90,I M = 2.79.),
§Q=l.35 $=L53 S_D=1.52 S_D=l.62

 

Note: All response items on a 7-point Likert scale with higher numbers reflecting greater endorsement.
Refer to Table l for cell information. Means with identical subscripts represent significant or marginal
simple effects comparisons; 'E(1,120)=15.60, p<.01, hE(1,120)=5.95, p<.05, °_E(1,120)=3.06, p<.09.

66

 

APPENDIX R

Table 7c. Means for perceptions of how hard participant worked on performance trials for gender and

work x gender interaction.

 

SEC—Mg!
Men (11 = 32) Women (11 = 100) E (l, 120)
M = 5.53 M = 6.22 4.84*
E = 1.46 S1; = 0.97

 

Work condition

Collective (cells A, C, & E) Co-Active (cells B, D, & F)
_Qgrflgt
Men (11 = 12) Women (11 = 55) Men (11 = 20) Women (n = 45)
M = 6.25a M = 6.35 M = 5.10“ M = 6.07(,
§Q=0.97 _S2=0.78 S_D=1.55 S_D=1.16

 

Note: Response items on a 7-point Likert scale with higher numbers reflecting greater endorsement. Refer
to Table 1 for cell information. Means with identical subscripts represent significant or marginal simple
effects comparisons; 'E(1,120)=10.37, p<.01, t’E(l,120)=11.76, p<.01. *p<.05.

67

 

 

 

APPENDIX S

Table 7d. Means for perceptions of how hard participant worked on performance trials for work x gender x

feedback interaction.

 

Work condition

 

Collective Co-Active
9mg:
Men Women Men Women
Feedback

Full cell A cell B

M = 6.00 M = 6.63 M = 5.29 M = 5.50c

$=0 82 $=050 _S_D=l.60 $=L51

n = 4 n = 19 n = 7 n = 14
Partial

cell C cell D

M = 7.00, M = 6.56 M = 5.14,, M = 6.40,,

SQ=0.00 _S_D=Ojl §l_)=l.21 _S__12=1.06

n = 3 n = 18 n = 7 n = 15

 

Note: Response items on a 7-point Likert scale with higher numbers reflecting greater endorsement. Refer
to Table l for cell information. Means with identical subscripts represent significant or marginal simple
effects comparisons; '£(l,120)=6.66, p<.05, bE(l,120)=6.92, p<.05, cE_(l,120)=5.38, p<.05. *p<.05.

68

REFERENCES

69

 

1F

REFERENCES

Brawley, L. R. (1990). Group cohesion: Status, problems, and future directions. Mamal
of Sport and Exercise Psychology: 10, 199-379.

Byme, D. (1971). The a_tt_r_action pafldigm. New York, NY: Academic Press.

Carron, A. V. (1988). Group dynamics in sport. London, Ontario: Spodym Publishers.

Clark, M. S., & Mills, J. (1979). Interpersonal attraction in exchange and communal
relationships. Journal of Persofllity and Social Psychology: 37, 12-24.

Comer, D. R. (1995). A model of social loafing in real work groups. Human Relations:
48, 647-666.

Festinger, L. (1954). A theory of social comparison processes. Human Relations: 7,
117-140.

Forsyth, D. R. (1999). Group dynamics. Belmont, CA: Wadsworth Publishing
Company.

Jackson, J. M., & Harkins, S. G. (1985). Equity in effort: An explanation of the social
loafing effect. Journal of Persoflyand Social Psychology: 49, 1199-1206.

Johnson, P. S., & Staffieri, J. R. (1971). Stereotypic affective properties of personal
names and somatotypes in children. Developmental Psychology: 5, 176.

Karau, S. J ., & Hart, J. W. (1998). Group cohesiveness and social loafing: Effects of a
social interaction manipulation on individual motivation within groups. G_ropp
Dulamics: Theory. Researchfind Practice: 2,185-191.

Karau, S. J., & Williams, K. D. (1993). Social Loafing: A meta-analytic review and
theoretical integration. Jourpal of Persopality a_n_d Social Psychology: 65, 681-

706.

70

 

 

 

 

Karau, S. J ., & Williams, K. D. (1997). The effects of group cohesiveness on social
loafing and social compensation. Group Dmamics: Theory. Research. aLd

Practice: 1, 156-168.

 

Kerr, N. L. (1983). Motivation losses in small groups: A social dilemma analysis.
mart Personality and Social Psychology: 45, 819-828.

Kerr, N. L., & MacCoun, R. J. (1985). Role expectations in social dilemmas: Sex roles
and task motivation in groups. Journal of PersonMity and Social Psychology: 49,
1547-1556.

Kerr, N. L., & Sullaway, M. E. (1983). Group sex composition and member task

 

motivation. Sex Roles: 9, 403-417.

h..‘ H.— — ‘

thler, O. (1926). Kraftleistungen bei Einzel- und Gruppenarbeit. Industrielle
Psychotechnik: 3, 274-282.

Latane, B., Williams, K., & Harkins, S. (1979). Many hands make light the work: The
causes and consequences of social loafing. Jourgrl of Personality aid Social
Psychology: 37, 822-832.

Levine, J. M., & Moreland, R. L. (1998). Small groups. In D. T. Gilbert, S. T. Fiske, &
G. Lindzey (Eds), The handbook of social psycholtagy (Vol.2) (pp. 415-469).
New York: Oxford University Press.

Lott, A. J ., & Lott, B. E. (1965). Group cohesiveness as interpersonal attraction.
Psychological Bulletin: 64, 259-309.

Lount, R. B., Messé, L. A., & Kerr, N. L. (2000). Trying harder for different reasons:
Conj unctivity and sex composition as bases for motivation gains in performing

groups. Zeitschrift fitr Sozialpsychologie: 31, 221-230.

 

71

 

Messé, L. A., Hertel, G., Kerr, N. L., Lount, Jr., R. B., & Park, E. S. (in press)
Knowledge of Partner's Ability as a Moderator of Group Motivation Gains:
Exploration of the KOhler Discrepancy Effect. Jauanal of PersoMlity and Socia_l
Psychology.

Mudrack, P. E. (1989). Defining group cohesiveness: A legacy of confusion? Mnall
Group Behavior: 20, 37-49.

Pleban, R., & Tesser, A. (1981). The effects of relevance and quality of another’s
performance on interpersonal closeness. Social Psychology Quarterly: 44, 278-

285.

 

Steiner, I. D. (1972). Group process and productivity. New York, NY: Academic Press.

‘F‘fl' ‘1' r- Jaw).

Stroebe, W., Diehl, M., & Abakoumkin, G. (1996). Social compensation and the KOhler
effect: Toward a theoretical explanation of motivation gains in group
productivity. In E. Witte & J. Davis (Eds), Understanding group behavior: SLna_ll
group processes and interpersonal rela_tio_n§. Mahwah, NJ: Erlbaum.

Tesser, A. (1999). Toward a self-evaluation maintenance model of social behavior. In R.
F. Baumeister (Ed), The self in social psychology. Philadelphia, PA: Psychology
Press.

Williams, K. D., & Karau, S. J. (1991). Social loafing and social compensation: The
effects of expectations of co-worker performance. Journal of Personality an_d

Social Psychology: 61, 570-581.

 

72