 

u... In“ _

L
-
Aux... u... ‘

-..

 

.

1w "=1 m.

II.
4“...

v
_'-¥o~'a—b--J—.‘—.4.. ...

\

 

LIBRARIES
‘ MICHIGAN STATE UNIVERSITY
r; ’ ' EAST LANSING, MICH 48824-1048

This is to certify that the
thesis entitled

Exploring self-efﬁcacy as a possible moderator of the KOhler
discrepancy effect.

presented by
Dong-Heon Seok

has been accepted towards fulﬁllment
of the requirements for the

MA. degree in Psychology

 

 

WW

 

Major Professor’s Signature
07/ '7/ 6 7
F I

Date

MSU is an Afﬁrmative Action/Equal Opportunity Institution

 

-a-.-- —A- -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-,-.-.-.-.-.-,-.-_-._.-.-.-.-

 

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 c:/C|RC/DateDue.pB$-p.15

 

EXPLORING SELF-EFFICACY AS A POSSIBLE MODERATOR OF THE
KOHLER DISCREPANCY EFFECT

By

Dong-Heon Seok

A THESIS

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

MASTER OF ARTS
Department of Psychology

2004

ABSTRACT

EXPLORIN G SELF-EF F ICACY AS A POSSIBLE MODERATOR OF THE
KOHLER DISCREPAN CY EFFECT

By

Dong-Heon Seok

Kohler (1926, 1927) found that when performing a taxing physical persistence
task, less able participants tried harder when working in a team under conjunctive task
demands than when working individually (the Kb'hler motivation gains effect), and that
this increase in effort was greatest when there was a moderate discrepancy in coworker
ability (the Ko'hler discrepancy eﬂect). A recent investigation (Messé, Hertel, Kerr, Lount
& Park, 2002) successfully reproduced both thler’s overall motivation gain and the
Kohler discrepancy effects. In an extension of the successﬁil replication of the thler
discrepancy effect, the present study investigated 1) how the manipulation of self-
efﬁcacy affects motivation gains in conjunctive dyad group performance, 2) whether the
Kohler discrepancy effect would be maintained or altered when participants were given
either high or low self-efﬁcacy information. Results of this study suggested that
participants Showed greater motivation gains when they had low self-efﬁcacy rather than
high self-efﬁcacy, and the impact of self-efﬁcacy was strongest under a moderate level of
perceived discrepancy. Also, the results indicated that the Kohler discrepancy effect was
more likely to occur when participants had the low rather than high self-efﬁcacy
information. Implications of these results for understanding the Kohler discrepancy effect

and directions for future study were discussed.

To my parents, Doo-Ryoung and Jung-Ac, and my wife, J i-Hyeong,

for their unconditional support and love.

iii

ACKNOWLEDGEMENT

Many people helped me to complete this thesis. First of all, I would like to
express my gratitude to my committee: Dr. Norbert L. Kerr, Dr. Lawrence A. Messé, and
Dr. Richard E. Lucas, who gave me plenty of feedback and research ideas throughout the
long process. Without their help, I would never have ﬁnished this thesis. Also, I am
indebted to my old advisor in Korea, Dr. Doug-Woong Hahn, who taught me valuable
lessons about how to study social psychology.

I would also like to thank Ernest S. Park, Brendan M. Baird, Eric J. Sambolec and
Kimdy Le, who gave me a lot of support and encouragement. Finally, I would like to
express my gratitude to Rai-Kyoung Lee and Mai Kikuchi for their help in data

collection.

iv

TABLE OF CONTENTS

LIST OF TABLES ...................................................................................................... vi

LIST OF FIGURES .................................................................................................... vii

CHAPTER 1

INTORDUCTION ....................................................................................................... 1
K6hler Motivation Gain and Discrepancy Effects .......................................... 1
Self-efﬁcacy .................................................................................................... 4
Self-efﬁcacy and the Kiihler Discrepancy Effect ........................................... 7

CHAPTER 2

METHOD .................................................................................................................. 13
Participants ................................................................................................... 13
Design ........................................................................................................... 13
Experimental Task ........................................................................................ 13
Procedure ...................................................................................................... 14

CHAPTER 3

RESULTS .................................................................................................................. 21
Manipulation Checks ................................................................................... 21
Performance Measures ................................................................................. 23
Subjective Experience .................................................................................. 35

CHAPTER 4

DISCUSSION ........................................................................................................... 43
Implications of the Present Study ................................................................ 43
Limitations and Directions for Future Research .......................................... 52

CHAPTER 5

CONCLUSION ......................................................................................................... 61

APPENDICES ............................................................................................................ 63

REFERENCES ........................................................................................................... 73

LIST OF TABLES

Table 3.1 — Means and Standard Deviations of Subjective Ratings ........................... 37

LIST OF FIGURES

Figure 2.1 — An Example of High Self-efﬁcacy Manipulation .................................... 16
Figure 3.1 — Motivation Gain for the Average of Trial 3 and 4 by Condition (Line
graph). ........................................................................................................................... 27
Figure 3.2 — Motivation Gain for Trial 3 by Condition (Line graph) ........................... 28
Figure 3.3 — Motivation Gain for Trial 4 by Condition (Line graph) .......................... 30
Figure 3.4 — Motivation Gain for the Average of Trial 3 and 4 by Condition (Bar
graph) ............................................................................................................................ 32
Figure 3.5 — Motivation Gain for Trial 3 by Condition (Bar graph) ........................... 33
Figure 3.6 — Motivation Gain for Trial 4 by Condition (Bar graph) ........................... 34

Figure 4.1 — Comparison Between the Present Results of Low Self-efﬁcacy
Participants and Messé et al.’s (2002) Results (Motivation Gain for the Average
of Trial 3 and 4) ............................................................................................................. 55

Figure 4.2 — Comparison Between the Present Results of Low Self-efﬁcacy
Participants and Messé et al.’S (2002) Results (Motivation Gain for Trial4) .............. 56

Figure 4.3 — Comparison Between the Present Results of High Self—efﬁcacy
Participants and Messé et al.’s (2002) Results (Motivation Gain for the Average
of Trial 3 and 4) ............................................................................................................ 58

Figure 4.4 - Comparison Between the Present Results of High Self-efﬁcacy
Participants and Messé et al.’s (2002) Results (Motivation Gain for Trial 4) .............. 59

Chapter 1

INTRODUCTION

Kiihler motivation gain and discrepancy effects

Testing members of a Berlin rowing club, K6hler (1926, 1927) found that
members tended to work harder at a demanding physical persistence task when they
performed it as part of a team (in dyads or triads) than when they worked at it
individually — a phenomenon that later was termed the K5hler motivation gain eﬂect
(Hertel, Kerr, Scheffler, Geister, & Messé, 2000). Contemporary social psychologists
have begun to investigate the processes underlying this effect (e. g., Hertel, Kerr, & Messé,
2000; Messé, Hertel, Kerr, Lount & Park, 2002; Stroebe, Diehl, & Abakoumkin, 1996).
Considering the fact that studies on individual’s motivation in groups have been focused
mainly on group motivation losses (e.g., social loaﬁng; Latané, Williams, & Harkins,
1979), the rediscovery of this work was important in that it — along with another line of
motivation gain studies (e.g., social compensation; Williams & Karau, 1991) — helped
rekindle interest in group motivation gains.

In Kiihler’s (1926, 1927) motivation gain study, two features are particularly
noteworthy. First, in the group trials, it was virtually impossible for the stronger coworker
to continue working at the task after the weaker coworker had stopped. In Steiner’s
(1972) perspective on task demands, this type of activity can be classiﬁed as a
conjunctive task because, in performing it, a group can do no better than its least capable

member.

The second noteworthy feature of Kiihler’s (1926, 1927) research is that the
motivation gains were maximal when the discrepancy between the capabilities of the two
dyad members was moderate (i.e., when working individually, the weaker member had
about 70% of ability as that of the stronger member). In contrast, (perhaps due to
coordination problems) motivation losses occurred when group members were nearly
equal, and smaller motivation gains were found when members were very discrepant in
their capabilities. Later, this phenomenon was termed the Ko'hler discrepancy effect
(Hertel, Kerr, Scheffler, et al., 2000).

Inspired by the Kiihler’s (1926, 1927) seminal work on group motivation gains,
several subsequent studies have been conducted. However, these studies have mainly
focused on the Kiihler motivation gain effect rather than K6hler discrepancy effect.
Seven studies (Hertel, Kerr, & Messé, 2000, Experiments 1 and 2; Hertel, Kerr, Schetfler,
etal., 2000, Experiments 1 and 2; Lount, Messé, & Kerr, 2000; Messé, etal., 2002,
Experiments 1 and 2) consistently found signiﬁcant motivation gains for weaker
coworkers when they performed as a team member under conjunctive task demands, with
average performance increases in group compared to individual trials that ranged
between 10 and 50 percent. These studies by successfully replicating the K6hler
motivation gain effect have given researchers a solid foundation for conducting ﬁrture
research designed to uncover the underlying mechanisms that produce this effect.

Of these contemporary investigations, the study of Messé, et a1. (2002) is
noteworthy, especially for the thler discrepancy effect. In Experiment 2, Messé and his
colleagues manipulated discrepancy by having a confederate enact the role of a Slightly,

moderately, or substantially better coworker. Through this procedure, they found that

knowledge of a partner’s ability could produce the thler discrepancy effect — i.e., the
results Showed that the largest motivation gains occurred under moderate discrepancy.
Messé et a1. (2002, pp. 936-93 7) speculated that one or both of two processes
might underlie the Kbhler discrepancy effect: Indispensability of effort (e. g., Karau &
Williams, 2001; Stroebe & Frey, 1982; Vroom, 1964) and social comparison (or goal
setting) (e.g., Locke & Latham, 1990; Stroebe, etal., 1996). With regard to
indispensability, which derives from the widely accepted Instrurnentality X Value
approach to explaining people’s work activity, a weaker worker’s subjective sense of
how important his or her performance is to the group’s success could be a critical factor
in generating the Kiihler discrepancy effect. When the discrepancy in ability is very small,
a less able coworker could readily perceive that only a little extra effort is necessary to
promote the maximum possible group success, given the Slightly stronger coworker’s
presumed capability. When the discrepancy iS very large, the much less able coworker
would be faced with the realization that he or She could not even begin to approach the
very strong partner’s potential performance level. Under conjunctive task demands, in
which the weaker member sets the limit for team performance, the perceived discrepancy
between the likely outcome and what might have been (had the weaker member’s ability
been closer to that of the more able coworker), would likely lower the weaker worker’s
motivation to work at the task. Thus, in either circumstance (i.e., if the discrepancy in
coworkers’ abilities is either Slight or very large), the weaker worker’s impetus to try
harder Should be reduced. In contrast, when the discrepancy in ability is moderate, the
(somewhat) weaker team member could very likely perceive that a concerted and

achievable increase in effort would produce an outcome that is reasonably close to the

maximum possible by their group, given the partner’s ability level. Therefore, moderate
ability discrepancy would be likely to generate the greatest motivation gain.

In terms of the social comparison or goal-setting explanation, it is assumed that
the weaker coworker uses the more able partner’s performance as a goal-comparison
reference. Therefore, for conjunctive task Situations, this explanation always assumes
upward social comparison by the weaker coworker. When the discrepancy in ability is
very small, this small discrepancy could be considered as an easy goal to be reached,
calling for little, if any increased effort. Because of this small goal, people would
generate only a small motivation gain. When the discrepancy in abilities is very large,
consistent with the idea that unrealistic goals can undermine motivation (Hinsz, 1995),
the weaker coworker is likely to reject the much better partner’s performance as a viable
standard of comparison and, thus, increase effort to a much smaller extent. However,
when the difference in ability is moderate, people might consider this performance
difference as a realistic and achievable goal and thus increase their effort accordingly.

Messé and his colleagues’ (2002) investigation was useful in that it provided
convincing evidence that the difference in relative ability underlies the Kiihler
discrepancy effect. However, further research is needed to identify variables that could
moderate the Kéhler discrepancy effect. That is, future work needs to explore what
factors can contribute differentially to participants’ motivation to work harder when they
have various discrepancy information. In the present study, I examined the role of self-
efficacy, which in general is considered to be a critical variable when people perform a

task, as a possible moderator of the Kiihler discrepancy effect.

Self-eﬂicacy

Self-efﬁcacy refers to “people’s judgments of their capabilities to organize and
execute courses of action required to attain designated types of performances” (Bandura,
1986, p. 391). In other words, self-efﬁcacy is conﬁdence in one’S own ability to perform
behaviors required to produce a desired outcome (Bandura, 1977, 1986). A wide range of
empirical evidence has Shown that self-efﬁcacy inﬂuences the kind of behavior in which
a person engages, how much effort he or she expends after the behavior has been initiated,
and how long that effort is sustained in the face of obstacles and aversive experiences
(e.g., Manning & Wright, 1983).

Two decades of empirical research have generated a great number of studies that
demonstrated the positive relationship between self-efﬁcacy and various motivational and
behavioral outcomes in clinical (e.g., Bandura, Adams, Hardy, & Howells, 1980),
educational (e.g., Lent, Brown, & Hackett, 1994; Schunk, 1995), and organizational
settings (e. g., Bandura, 1988; Wood & Bandura, 1989). As more convincing evidence of
the relationship between self-efﬁcacy and performance, a recent meta-analysis by
Stajkovic and Luthans (1998) found a strong positive relationship between self-efﬁcacy
and performance based on 109 studies conducted on work-related tasks and/or in work
settings.

Similar to the results of the above studies, which were mainly based on
correlational data, the positive relationship between self-efﬁcacy and performance has
also been demonstrated in experimental settings in which self-efﬁcacy was directly
manipulated (e.g., Bouffard-Bouchard, 1990; Miyake & Matsuda, 2002; Sanna, 1992).
For example, Sauna (1992) examined the effects of self-efﬁcacy in the areas of social

facilitation and social loaﬁng using a vigilance task. In his Experiment 1, self-efficacy

was manipulated by providing false performance feedback after participants worked on a
preliminary task: In the high self-efﬁcacy condition, participants were told that they
scored in the upper 80” percentile on established norms for the task. In contrast, in the
low self-efﬁcacy condition, participants were told that they scored in the lower 20th
percentile on established norms for the task. Using this manipulation, Sanna found that
fewer vigilance errors were made by high than low self-efﬁcacy participants. Such
ﬁndings established that the positive relationship between self-efﬁcacy and task
performance could be demonstrated in experimental studies as well as in correlational
studies.

However, not all studies supported this positive relationship between self-efﬁcacy
and future performance. For example, Stone (1994) found that- high self-efﬁcacy led to
overconﬁdence in one’s abilities. In his study, instead of high self-efﬁcacy participants
contributing more of their resources toward the task performance, they contributed less.
Also, high self-efficacy participants were both less attentive and expended less effort than
were their low self-efﬁcacy counterparts. Bandura and Jourden (1991) reported Similar
results. Using normative feedback to induce high or low self-efﬁcacy in a group of
participants, they found that the performance of high self-efﬁcacy participants was not
better overall, and in fact, over time, higher self-efﬁcacy contributed to decrements in
performance. More recently, Vancouver and his colleagues (Vancouver, Thompson,
Tischner, & Putka, 2002) manipulated self-efﬁcacy with feedback about past success and
found that individuals lowered their effort to a greater extent when they were induced to
have high self-efﬁcacy than low self-efﬁcacy. Consistent with Stone’s (1994) position,

Vancouver et a1. (2002) also explained the negative effect of high self-efﬁcacy by

suggesting that it could be a product of complacency. Thus, it would appear that at least
in certain circumstances, people seems to try less hard even though they had high self-
efﬁcacy, contrary to the position advanced by a predominance of the self-efficacy
literature.

AS we can see from the previous discussion, self-efficacy can be a Si gniﬁcant
determinant of how much effort workers put into performing a task. I speculated that this
connection between self-efficacy and effort would also be true of work settings in which
people performed under conjunctive task demands — for instance, the context that yielded
the thler motivation gain and discrepancy effects.

Self-efﬁcacy and the Ko'hler discrepancy effect

In the present study, I examined whether the Kiihler discrepancy effect would be
altered by varying participants’ self-efﬁcacy as well as relative ability information (i.e.,
the discrepancy in ability between coworkers). To this end, I used a 2 (self-efﬁcacy: low
versus high) X 4 (task demand and discrepancy condition: individual no discrepancy
control, conjunctive-small discrepancy, conjunctive-moderate discrepancy, conjunctive-
large discrepancy) between subjects factorial design. In this study, similar to the Sanna’s
(1992) manipulation, self-efﬁcacy was varied by giving participants false feedback about
their task performance based on the two previous individual trials. Also, consistent with
Messé et al.’s (2002) procedure, I manipulated coworker discrepancy by informing
participant of their actual performance score and their partner’s ostensive performance
score (which was calculated based on the participant’s performance), as well as giving

them comments about the performance difference.

Given the mixed effects of self-efﬁcacy on performance in the literature, as I
described in the previous section, it is difﬁcult to predict conﬁdently how the Kéhler
discrepancy effect would be changed by self-efﬁcacy information. Because of the
exploratory nature of the study, I describe possible results in each condition rather than
propose Speciﬁc, omnibus hypotheses. In discussing the possible results in the
conjunctive discrepancy conditions, I utilized the Instrumentality X Value explanation,
the Goal-comparison explanation and the complacency or overconﬁdence effects of high
self-efﬁcacy which I discussed earlier.

First, for the overall effect of self-efﬁcacy, it is unclear whether high self-efﬁcacy
would lead to greater motivation gains than low self-efﬁcacy. Based on the studies (e.g.,
Sanna, 1992; Stajkovic & Luthans, 1998) which dealt with the beneﬁcial effects of self-
efﬁcacy on performance, there is a possibility that participants with high self-efﬁcacy
would perform better than participants with low self-efﬁcacy. However, as noted, there
also is a possibility that high self-efﬁcacy participants would Show a poorer performance
than low self-efﬁcacy participants because high self-efﬁcacy participants might feel
complacency or overconﬁdence for their task performance (Stone, 1994; Vancouver et al.,
2002) and, as a result, they might not increase their effort substantially. Also, another,
related possibility is that low self-efficacy participants might Show greater motivation
gains than their high self-efﬁcacy counterparts to prove their capability. That is, high
self-efﬁcacy participants might not have strong needs to prove their ability if they already
had received very favorable performance feedback via the self-efﬁcacy manipulation (i.e.,
80th percentile in the total population). In contrast, low self-efﬁcacy participants, given

that they did not prove that they are good performers in the initial two individual trials

(i.e., 20th percentile in the total population) might view the upcoming dyad trials as an
opportunity to demonstrate their real ability. Therefore, because of the need to prove their
ability, low self-efﬁcacy participants might Show better performance than high self-
efﬁcacy participants.

Second, in the small discrepancy condition, because participants already know
their partner’s ability is Similar to their own (through the discrepancy manipulation after
A the second trial), they are likely to have an expectation that their partner would end the
trial soon after they ended the trial. Also, participants might think that their trial would be
over anyway even though they were willing to exert extra effort unless their partner were
also willing to put out the same amount of effort. Therefore, participants might not think
their effort is particularly instrumental or indispensable for the group’s success in this
condition. In terms of the Goal-comparison explanation, the goal to be reached is the
partner’s performance level and participants who are in the small discrepancy condition
should set relative low goals. Under this low instrumentality and very small goal-setting,
perception of high or low self-efﬁcacy might have little inﬂuence on their motivation to
reduce this small amount of performance gap. That is, participants might easily keep up
with their partner regardless of their self-efﬁcacy because the performance difference is
very small. Therefore, I thought there might be no difference in performance between
high and low self-efﬁcacy participants when there was a small discrepancy in their
capabilities.

Third, the Instrumentality X Value explanation suggests that weaker workers in
the large discrepancy condition might not perceive their effort as indispensable because

the gap between their performance level and their partner’s is too large to achieve an

outcome that comes anywhere near the maximum possible group success. Also, in terms
of the Goal-comparison explanation, weaker workers in the large discrepancy condition
should perceive approximating their coworker’s performance level as an unrealistically
high goal and, thus, they would be inclined to reject this standard — a decision that could
result in lower effort. Based on these explanations, I thought that low instrumentality
and/or unrealistically high goal-setting effect would work to produce lower motivation
gains irrespective of self-efﬁcacy level, because, with large discrepancy, participants
might perceive the importance of performing well to be very low and/or they might be
overwhelmed by the unrealistically high goal that their coworker’s performance would
impose on them. That is, in this condition, even high self-efﬁcacy participants may doubt
their ability to attain the goal of matching their partner’s excellent performance despite
their expectation of performing well in an absolute sense. Also, low self-efﬁcacy
participants might view their expectation of performing relatively poorly in an absolute
sense as reinforcing their anticipation that their possible contribution to the group
outcome would be disappointing and/or their coworker’s performance would be an
unrealistically high goal for them. AS a result of this consideration, they might think it
would be useless to work hard to prove their ability under this very discouraging situation.
Thus, Similar to the small discrepancy condition, I thought that there might be no
performance difference between low and high self-efﬁcacy participants when perceived
discrepancy is very large.

In contrast, unlike for the small and large discrepancy conditions, I reasoned that
self-efﬁcacy would moderate performance in the moderate discrepancy condition. In this

condition, participants would expect that their partner had ability to continue working at

10

the task moderately longer than they, based on the discrepancy feedback that they were
given. Therefore, the conjunctive nature of the group task could lead the weaker members
to think that, given their effort was indispensable for the group’s success, they could
realistically increase their effort to make a marked difference in the outcome. The goal-
comparison explanation draws the same conclusion, but for a different reason: Because
the participant iS presented with a realistic and obtainable goal (i.e., the moderately better
performance level of their partner) the participant Should increase his/her effort to a
greater degree. Furthermore, I thought that this default pattern could be moderated or
changed by the addition of the high or low self-efﬁcacy information. First, there is a
possibility that beneﬁcial effects of self-efﬁcacy which were described in the previous
section, would heighten motivation gains when the discrepancy was moderate. Compared
with the small and large discrepancy conditions in which these beneﬁcial effects of self-
efﬁcacy were attenuated by, the low instrumentality and very modest or unachievable
goal-setting, I thought that these effects would not be attenuated because there were high
instrumentality and a realistically achievable goal in the moderate discrepancy condition.
Therefore, in the moderate discrepancy condition, there is a possibility that participants
who had high self-efﬁcacy would Show more motivation gains than those who had low
self-efﬁcacy.

It also was possible that low self-efﬁcacy participants would Show heightened
motivation gains in the moderate discrepancy condition. Participants with high self-
efﬁcacy might not increase their effort substantially even though they had high
instrumentality and a realistic and achievable goal because they might feel complacency

or overconﬁdence about their task performance. Also, as I discussed earlier, because low

11

self-efﬁcacy participants had reason to prove their ability, they might Show bi gger
motivation gains than high self-efﬁcacy participants. Therefore, combined with the high
instrumentality and realistic goal-setting in this condition, low self-efﬁcacy participants
might try harder when they believe that good performance could prove their greater
capability.

In summary, in the moderate discrepancy condition, there is a possibility that
participants who had low self-efﬁcacy would Show bigger motivation gains than those
who had high self-efﬁcacy because: 1) high self-efﬁcacy participants might feel
complacency or overconﬁdence about their performance and, thus, Show a lesser degree
of effort exertion than low self-efﬁcacy participants; and 2) low self-efficacy participants
are likely to consider dyad trials as an opportunity to prove their ability and might work
harder than high self-efﬁcacy participants under moderate performance discrepancy,
when they had realistic expectations that trying harder would matter and/or doing better
was a realistic goal.

To investigate these possibilities, I conducted an experiment that independently

manipulated self-efﬁcacy and discrepancy information.

12

Chapter 2

METHOD

Participants

One hundred and ninety four female undergraduate students at Michigan State
University participated in this study to partially satisfying a psychology course research
requirement. Because there were considerably more women in the department subject
pool, for ease of recruitment, only female students participated in this study. Two female
experimenters conducted all sessions.
Design

The experiment utilized a 2 (self-efﬁcacy: low versus high) X 4 (task demand and
discrepancy condition: individual no discrepancy control, conjunctive-small discrepancy,
conjunctive-moderate discrepancy, conjunctive-large discrepancy) between subjects
factorial design. Sessions were conducted with up to four participants. Each session was
randomly assigned to one of the eight experimental conditions. The (performance)
dependent variable was the difference in time that a participant persisted in holding up a
weight between the ﬁrst and the second trials for a given arm — i.e., Trial 3 — Trial 1,
Trial 4 — Trial 2, averaged across the two arms.
Experimental Task

The experiment used a modiﬁed version of the persistence task that was used in
Messé, et a1. (2002). Participants were instructed to hold a 3.19 lb. weight horizontally
above a thin cord that was strung between metal bars for as long as they felt comfortable

in doing so. When participants lowered their arm to the point that their hand hit the cord,

13

this action activated a key on a computer that recorded that the trial was over, as well as
the time (in seconds) that participants had persisted at the task. To prevent using their
other arm to support their performing arm, participants were told that they needed to
press and hold down the “Control” key on the computer keyboard with their free hand
during each trial.

Procedure

On the date of their experiment, up to four female participants arrived for their
session. Prior to the start of their actual participation, the experimenter made sure that
none of the participants had any disabling arm, shoulder, or back injuries. After bringing
them into the preparation room, the experimenter explained and demonstrated the task.
They were told that they would perform this task over a series of trials (the exact number
of which was not disclosed).

After demonstrating the task, the experimenter told participants that other students
were taking part in the experiment at the same time in another building, and the
computers located in their own lab rooms were connected to both each other and to
computers being used in the other laboratory in next building. To reinforce this cover
story, the experimenter pretended to make a phone call to the other laboratory in which
She asked how many participants were there. She then “repeated” out loud a number that,
when added to the participants her own laboratory, made an even number (greater than
four) of participants, total, for the sessions. For example, if there were one or three actual
participants, the “conversation” revealed that there were three more subjects in the other
laboratory. Also, if there were two or four actual participants, the “conversation” revealed

that there were four more subjects in the other laboratory. This procedure accomplished

14

two goals. First, for a few of the dyad sessions there was only one student Signed up. This
meant that without such deception, it would not be possible to complete all of the dyad
conditions. With other supposed participants in another lab, students were led to believe
that they would not be completing the experimental tasks alone. Second, when
participants in the dyad conditions complete the tasks on the computer, the program tells
them the name of their partner. So if students knew all the other participants in their
session, they would see through this deception. By telling them that other students in a
different lab were also participating in this study, they would be more likely to believe
that their experimental partner really existed.

After the phone call, participants were placed in separate lab rooms and sat down
in ﬁ'ont of the computers that ostensibly were connected with each other. The
experimenter left the lab room, so from then on, all instructions, task trials, and
questionnaires were administered via the computer. First, participants were asked to take
off any bracelets and watches that they were wearing and place them in the black paper
bag located in the desk for that purpose. This procedure was included to ensure that the
participants could not time themselves or wear something that would interfere with task
performance. Then, the task was explained in more detail by the computer, including
calling to the participants’ attention that a video camera was located to the Side of the
keyboard. It was explained that the camera would be monitoring a participant’s activities
to ensure, for example, that She was pressing the Control button with her other hand
(which are not supposed to be used for task performance on a particular trial). The
camera’s angle was low so that only a participant’s hands could be viewed. (To see the

verbatim instructions, refer to Appendix A.)

15

To increase the meaningfulness of the task, it was explained that the 2-person
team could earn money through its performance on the persistence task. One such team
would be randomly picked at the end of the experiment and would receive ﬁve cents for
every second of the total group performance (up to a maximum of $ 50). Moreover, to
further increase interest in doing well, computer instructions informed participants that
we were having a little competition between students taking part in this experiment at the
University of Michigan and those performing it here at Michigan State. Then they were
told that we would compute the average Total Score for each school, and award the
winning persons from the school with the stronger teams an additional 15%. In addition,
we reminded every participant that even if She were not picked, her good performance at
this task could possibly beneﬁt other MSU students.

After performing the ﬁrst and second trial as individuals, participants’ self-
efﬁcacy was manipulated by giving them ﬁctitious information about the likelihood of
performing well during the upcoming trials accompanied via a chart of a Likert-type

scale which indicated their chance of doing well (for an example, refer to Figure 2.1, and

 

Appendix B).
Likelihood of Performing Well
You are here
i
Notatall 1 1 1 1 1 1 1 n, 1 1 namely
likely likely

Figgre 2.1 An Example of High Self-efﬁcacy Manipulation

16

In the low self-efﬁcacy conditions, participants were given the following
information: “The computer has compared your performance on the previous two trials
with data which were collected earlier (approximately 849 previous female participants in
6 studies). Based on our comparisons, the computer has calculated that if you work hard,
the likelihood that you will perform well during the upcoming trials is 20%. In other
words, it is not very likely that you can perform well during the upcoming trials.” In the
high self-efﬁcacy conditions, participants were given the following information: “The
computer has compared your performance on the previous two trials with data which
were collected earlier (approximately 849 previous female participants in 6 studies).
Based on our comparisons, the computer has calculated that if you work hard, the
likelihood that you will perform well during the upcoming trials iS 80%. In other words,
it is very likely that you can perform well.”

After giving the self—efﬁcacy information, Participants in the conjunctive dyad
conditions were told that for the remainder of the trials the computer would randomly
choose their partner from the current participants and informed them of their partner’s
name. Actually, this name was pre-programmed to be chosen randomly from the four
pre-selected names (i.e., Anne Roberts, Erin Anderson, Susan Turner, and Steph Taylor).

Then, the discrepancy information was manipulated by presenting each
participant’s actual performance score, her partner’s ostensible performance score (which
was calculated based on the participant’s score), and comments about their performance
difference. Before the manipulation, the computer screen acknowledged that the

computer received these performance data and comments from experimenter via a LAN

l7

connection. Then, the screen Showed the box which contained information about the
participant’s performance score (in seconds) and their partner’s performance score (in
seconds) averaged across the two individual trials. Below this information, there was
another box which contained the experimenter’s ostensible comments about these scores.
In the small discrepancy condition, partner’s score was calculated by multiplying the
participant’s real performance data by 1.05, and the participant was told that the two of
them had performed just about the same. In the moderate discrepancy condition, partner’s
score was calculated by multiplying the participant’s real performance data by 1.42, and
the participant was told that the coworker had done moderately better than She had. In the
large discrepancy condition, partner’s score was calculated by multiplying the
participant’s real performance data by 2.05, and the participant was told that the
coworker had done substantially better than She had. The reason why the value 1.42 was
used to create the artiﬁcial partner’s performance data in the moderate condition was to
give the participant the belief that her ability was 70% as good as her partner’s, which
was the performance discrepancy that Ktihler found associated with the greatest
motivation gain. The levels of small and large discrepancies were Similarly selected
based on Kéhler’s original ﬁndings. However, I gave slight variation in the discrepancy
scores to make the feedback more realistic (i.e., 1.05 for the small discrepancy condition
instead of 1.00; 2.05 for the large discrepancy condition instead of 2.00). To see an
example of the discrepancy manipulation, refer to Appendix C.

After these manipulations, participants in the conjunctive dyad conditions
received information about how the remaining trials (i.e., 3rd and 4th trial) would be

conducted. Participants were informed that for the next few trials they would be

18

following a somewhat different procedure than before. During these trials they would be
paired with their partner at the task to create a two-women team. Each participant would
receive the same score. That score would be the time that the person whose arm dropped
ﬁrst had lasted. They were told that during the trials two boxes would appear on the
screen. Each box would have one of the team member’s names in it. The two boxes
would remain green as long as both persons’ arm remained above the string. As soon as
one person’s arm touched the string, her corresponding box would turn red (on both
computer screens). This change would Signal the end of the trial and the other person
should lower her arm. AS described earlier, the participants were led to believe this
partner would be in a booth either next to (or near) them or in the other lab. In actuality
the (ﬁctitious) partner’s box would never turn red. Thus, real participant always would be
the one who quit ﬁrst. Also, participants were informed that for the next few trials we are
interested in obtaining both group scores and the identity of the person who ended a trial
by hitting the string ﬁrst. For this reason, the equipment has been wired so that the
master-computer could record and save both the group score for each trial and which
person ended the trial. Therefore, the experimenter would know how well the team had
performed and which group member ended each trial. The computer program followed
this procedure for two dyad trials, one with participant’s dominant arm and one with the
non-dominant arm.

In the individual no discrepancy condition, participants performed four trials,
receiving either low or high self-efﬁcacy information after Trial 2 according to the
condition to which they had been assigned, but not any discrepancy information (Since

these participants never had a partner). To hold the rest time constant across conjunctive

l9

and individual conditions during the manipulation of discrepancy, participants in the
individual condition had to wait a comparable amount of time between trials as it took
participants in the dyad conditions to receive instructions and rest.

Before each trial, participants answered a question about the pre-trial self-efﬁcacy
on a 9-point scale: “How well do you think you can perform in the upcoming trial?”(1 =
not very well, 9 = very well). Also, before the third and fourth trial in all conjunctive
dyad conditions, participants answered a question on performance expectation: “How
likely is it that you will do as well or better than your partner?”(1= very likely I will do
less well, 9 = very likely I will do better).

After each trial, participants completed a Short questionnaire comprised of ﬁve
scales: “How much effort did you put into this last trial?”(1 = not very much, 9 = very
much), “How much did you enjoy performing on the last trial?”(l = not very much, 9 =
very much), “How important was it to you to perform well in the last trial?”(l = not at all,
9 = extremely), “During the last trial, how well do you think that you performed?”(1 =
not very well, 9 = very well), “How much were you satisﬁed on the last trial?” (1 = not
very much, 9 = very much). In addition, after third and fourth trial, participants in the
conjunctive dyad conditions answered an additional question: “How large of a gap did
you think there was between you and your partner's performance on the last trial?”(l =
very small, 9 = very large).

After the fourth trial of the persistence task, in all conditions participants
answered manipulation check questions. After participants completed this ﬁnal task, the

experimenter debriefed, thanked, and dismissed them.

20

Chapter 3

RESULTS

Manipulation checks

In order to ascertain if the manipulation of self-efficacy was successﬁrl,
participants were asked their likelihood of performing well, based on the scale which was
presented to them between the second and third trial (7-point scale; 1 = Much less well
than average, 4 = Average, 7 = Much more than average). Differential levels of self-
efﬁcacy were successfully manipulated: Participants in the high self-efﬁcacy condition
responded that they had above average likelihood (M = 5.39, s = 1.28), whereas
participants in the low self-efﬁcacy condition reported having a below average likelihood
(M = 1.97, s = 1.07). The mean difference between these two conditions was signiﬁcant
(F(1, 193) = 405.87, p < .001).

For a discrepancy manipulation check, participants in dyad conditions were asked
to indicate just what information they had received about the performance difference
between them and their partners (5-point Scale; 1 = My partner had performed
substantially worse than I did, 2 = My partner had performed moderately worse than I did,
3 = We had performed about the same, 4 = My partner had performed moderately better
than I did, 5 = My partner had performed substantially better than I did). As an indication
of successful discrepancy manipulation, all participants responded with a 3, 4, or 5
because they always were the weaker members of a team. Also, the percentages whose
response exactly matched the performance discrepancy information that they had

received for the manipulation were very high: 80.4% for the small discrepancy condition;

21

82.0% for the moderate discrepancy condition; and 94.2% for the large discrepancy
condition. It should be noted that some participants might have misinterpreted the
discrepancy between their score and their partner’s score: Participants in the small
discrepancy condition might choose option 4 because their partner’s score was always a
little bit higher than their score even though they were told that they had performed about
the same. In the same vein, some participants in the moderate and large condition might
have depended more on their subjective perceptions of the discrepancy scores provided
than on the given statements. In addition, Newman-Keuls pairwise post-hoe tests
revealed that the means in all three conditions were Signiﬁcantly different from one
another (Small discrepancy, M = 3.22, s = .46; Moderate discrepancy, M = 4.18, s = .39;
Large discrepancy, M = 4.92, s = .33; p < .05). Therefore, the discrepancy manipulation
also was successful.

In order to test whether self-efﬁcacy and discrepancy manipulations were
independent, a 2 (self-efﬁcacy: low versus high) X 3 (discrepancy: conjunctive-small,
conjunctive-moderate, conjunctive-large) AN OVA was conducted on each manipulation
question. For the self-efﬁcacy manipulation question, the main effect of self-efﬁcacy was
Signiﬁcant (F (1 , 147) = 286.64, p < .001), while neither the main effect of discrepancy
nor the interaction effect of these two variables was (both Fs < 1). For discrepancy
manipulation question, the main effect of discrepancy was signiﬁcant (F (2, 147) = 235.22,
p < .001 ), while neither the main effect of self-efﬁcacy nor the interaction effect of these
two variables was (both Fs < 1). Thus, results indicate that participants did not perceive
the manipulation of one variable differently in each condition of another variable because

the interaction effect was not Signiﬁcant in either analysis.

22

Performance measures

As mentioned earlier, participants performed a persistence task by holding up one
of their arms as long as possible for four trials, alternating between their dominant and
non-dominant arms. In order to ascertain and later adjust for fatigue effects in later trials,
and to test whether there was any differential inﬂuence of self-efﬁcacy, a 2 (self-efﬁcacy:
low versus high) X 2 (trial block: ﬁrst two trials vs. last two trials) X 2 (performing arms:
dominant vs. non-dominant) analysis of variance (AN OVA) with repeated measures on
the last two factors was conducted on performance scores in the individual control
condition. The trial block factor Simply contrasts the ﬁrst two trials (i.e., Trial 1 and 2)
with the last two trials (i.e., Trial 3 and 4), while the performing arms factor compares
two trials in which participants used their dominant arm (i.e., Trial 1 and 3) with two
trials in which they used their non-dominant arm (i.e., Trial 2 and 4).

A signiﬁcant trial block main effect emerged (F (1 , 39) = 8.59, p < .01) indicating
that participants experienced effort reducing fatigue, boredom, etc. Participants in the
individual control condition persisted longer in the ﬁrst two trials with each arm (M =
74.265, 3 = 43.91) than in their last two trials with each arm (M = 62.288, s = 29.25).

There was also a Signiﬁcant performing arms main effect (F (1, 39) = 26.10, p <
.001), indicating that these participants persisted longer with their dominant arms (M =
73263, S = 35.36) than with their non-dominant arms (M = 63.283, s = 35.67). However,
the trial block X performing arms interaction was not Signiﬁcant (F < 1), which
suggested that the fatigue effect was Similar across arms (i.e., this result provided the
evidence that one arm did not fatigue faster than the other arm). Finally, there were no

main or interaction effects with the self-efﬁcacy factor, all F S < 1, suggesting that the

23

observed fatigue effects were comparable for low or high self-efficacy condition.
Therefore, I concluded that different fatigue corrections for low and high self-efﬁcacy
conditions were not necessary in the conjunctive dyad conditions.

Based on the above analysis, as in past studies (e.g., Messé et al., 2002), scores on
the third and fourth trial were corrected for fatigue in order to analyze the performance of
the participants in the conjunctive dyad conditions. First, the scores ﬁom the second trial
of each arm were subtracted from the ﬁrst trial of each arm in the individual condition.
By averaging these individual difference scores, two means were generated: 12.02S for
Trial 1 — Trial 3 and 11945 for Trial 2 — Trial 4. The average of these two means (M =
11.985, 3 = 25.99) reﬂected the fatigue effect. Second, to correct fatigue, I added this ﬁnal
mean to the performance scores of the third and fourth trial in each conjunctive dyad
condition.

It Should be mentioned that to correct for fatigue in this study, I did not utilize an
individual control condition in which no self-efﬁcacy information was provided, because
I wanted to correct fatigue by using individual control conditions that were most Similar
to the experimental conditions in this study. However, to determine whether providing
self-efﬁcacy information (vs. providing no such information) had any effect on the
magnitude of the fatigue effect, I compared the fatigue correction in the individual
control condition of a previous study (Messé, Kerr, & Seok, in preparation) with the
individual control condition of this study. The previous study (Messé et al., in
preparation) was identical to the present one in all but one respect—viz, introduction of
the self-efﬁcacy manipulation between Trial 2 and Trial 3. To test whether there was any

trial block or performing arms effect, a 2 (trial block) X 2 (performing arms) analysis of

24

variance (AN OVA) was conducted on the performance data in this condition. In this
analysis, both variables were repeated measure variables. The trial block factor simply
contrasts the ﬁrst two trials with the last two trials while the performing arms factor
compares participants’ dominant versus non-dominant arm. A signiﬁcant trial block main
effect emerged (F (1 , 18) = 11.37, p < .01) indicating that those in the individual control
condition persisted longer in the ﬁrst two trials with each arm (M = 82.005, 5 = 33.68)
than on their last two trials with each arm (M = 65.505, 3 = 25.45). There was also a
Signiﬁcant performing arm main effect (F(1, 18) = 6.17, p < .05) indicating that these
participants persisted longer with their dominant arms (M = 78.135, 3 = 30.60) than with
their non-dominant arms (M = 69.375, 3 = 27.13). However, the trial block X performing
arms interaction was not Signiﬁcant (F < 1), which suggested that the fatigue effect was
Similar across arms. Besides the Similarity of the pattern of fatigue data, it is noteworthy
that the overall fatigue effect observed in the present study (viz. 11.985) was not
signiﬁcantly different ﬁ'om the corresponding effect observed in the study of Messé et al.
(in preparation) (viz. M = 82.05 — 65.55 = 16.55, s = 21.33), t(58) < 1. These results
indicate that the magnitude of the fatigue effect was comparable when participants
received either no self-efﬁcacy information (i.e., Messé et al., in preparation) or received
(high or low) self-efﬁcacy information (i.e., the present study)

Fatigue corrected scores were analyzed separately for Trial 3, Trial 4 and the
average of these two trials. This was done because there was a difference in believability
of discrepancy manipulation between before Trial 3 and before Trial 4. That is, even
though participants had already received the manipulation on discrepancy information

before Trial 3, their believability was likely to be maximal in Trial 4 because they now

25

(after Trial 3) had conﬁrming evidence that their partner was better at the task. Moreover,
past work (Messé et al., 2002) suggested the Kiihler discrepancy effect was more likely to
occur in Trial 4 than in Trial 3.

The second trial — ﬁrst trial (i.e., Corrected Trial 3 — Trial 1; Corrected Trial 4 —
Trial 2; the average of these two scores) difference in how long each participant persisted
at the task for each arm was used as the index of performance change (1055 or gain).
Therefore, any score above zero represents a motivation gain because the result of the
fatigue correction equation eliminates fatigue and arm differences. All of the following
analyses on the conjunctive dyad conditions were based on these corrected data.

The primary focus of this study was to examine whether self-efﬁcacy had
differential effects in each conjunctive dyad condition. In order to examine whether the
trial factor interacted with either the self-efﬁcacy or discrepancy factors, I conducted a 2
(self-efficacy: low versus high) X 3 (discrepancy: conjunctive-small, conjunctive-
moderate, conjunctive-large) X 2 (trial: T3 — T1 vs. T4 — T2) analysis of variance, with
repeated measure on the last factor. The results revealed that there was a Signiﬁcant main
effect of the trial factor (F (l , 147) = 18.43, p < .001) indicating that participants’
performance increases were greater when they worked in a dominant arm trial (M =
40.295, 3 = 36.10) than in a non-dominant arms trial (M = 27.565, s = 33.97). However,
this trial factor did not interact with either the self-efﬁcacy or discrepancy factors (all F s
< 1). However, given the fact that there was a qualitative difference in the amount of
information participants had about their partner’s likely ability when beginning Trial 3 vs.
Trial 4, I also did a trial by trial analysis, besides analyzing the average score of the last

two trials.

26

First, in order to test whether self-efﬁcacy had differential effects on each
conjunctive dyad condition across the third and fourth trials, a 2 (self-efﬁcacy) X
3(discrepancy) ANOVA was performed on the fatigue corrected overall difference score
[{(Trial 3 — Trial 1) + (Trial 4 — Trial 2)} / 2]. Means for all conditions on these overall
difference scores are presented in Figure 3.1. Results revealed a signiﬁcant main effect of
self-efﬁcacy, F (l , 147) = 5.10, p < .05, while the main effect of discrepancy (F (2, 147)
= .01 , us) was not signiﬁcant. However, the interaction effect of these variables was
marginally Signiﬁcant (F (2, 147) = 2.59, p < .08). Because it was of particular theoretical
interest to determine if perceived self-efﬁcacy moderated the relationship between inter-
member ability discrepancy and motivation, planned contrast tests were also conducted to
examine whether self-efficacy had differential effect in each discrepancy condition. The

results revealed that, as predicted, those in the moderate discrepancy condition responded

 

60 ~
" l
05 :
A, 50 1
a l ”we“-..
I: 40 7 :+High Self-f
.5 l 1 Efﬁcacy
: 30 I l
'3 l I—I—LOW Self- i
‘9 20 i Efﬁcacy 1
.5 I
g 10 -,
.5 l
g 0 .1 ~ v~ --« we ._

Small Moderate Large
Ability Discrepancy

Figure 3.1 Motivation Gain for the Average of Trial 3 and 4 by Condition (Line

graph).

27

differently to the self-efﬁcacy information, compared with the small and large
discrepancy conditions: When participants were given moderate ability discrepancy, low
self-efﬁcacy participants (M = 50.905, 5 = 51.16) Signiﬁcantly outlasted high self-efﬁcacy
participants (M = 27.355, s = 31.07; t(147) = 2.81, p < .01), while those in the small and
large discrepancy conditions did not Show Similar performance differences (both ts <
1.46).

Second, a 2 (self-efﬁcacy) X 3(discrepancy) AN OVA was performed on the
fatigue corrected Trial 3 - Trial 1 scores. Means for all conditions on the Trial 3 — Trial 1
scores are presented in Figure 3.2. Results revealed a marginally Signiﬁcant main effect
of self-efﬁcacy, F (1, 147) = 3.02, p < .09, while neither the main effect of discrepancy
(F (2, 147) = .04, ns) nor the interaction effect of these variables (F(2, 147) = 1.38, ns)
was signiﬁcant. Planned contrast tests were conducted to examine whether self-efﬁcacy

had differential effect on each discrepancy condition. Similar to the results on the overall

 

60 a
‘2 l
E 50 l
1..
c l 1---... __.
'E 40 "l 1—0—High Self-l
'- i 1 1
8 30 2 Efﬁcacy l
C = 1+LOW Self- 1
.8 20 é-___Efﬁ<_>es=_v_,
3’ 2
s 10 ‘1
= .

o - — . m— —---—-

Small Moderate Large
Ability Discrepancy

Figgre 3.2 Motivation Gain for Trial 3 by Condition (Line graph).

28

difference scores, these results revealed that those in the moderate discrepancy condition
responded differently to the self-efﬁcacy information compared with the small and large
discrepancy condition: When participants were given moderate ability discrepancy, low
self-efﬁcacy participants (M = 56.295, 3 = 56.59) Signiﬁcantly outlasted high self-efﬁcacy
participants (M = 32.855, 3 = 37.88; t(147) = 2.30, p < .05) but those in the small and
large ability discrepancy conditions did not Show any performance difference (both ts <
1).

As stated earlier, participants may have been more certain about the superiority of
their partner for Trial 4 than Trial 3 because they had received feedback that their partner
had outperformed them in Trial 3. In order to test whether self-efﬁcacy had differential
effects on each discrepancy condition under the more realistic Situation, a 2 (self-
efﬁcacy) X 3(discrepancy) ANOVA was performed on the fatigue corrected Trial 4 —
Trial 2 scores. Means for all conditions on the Trial 4 —- Trial 2 scores are presented in
Figure 3.3. Results revealed a Signiﬁcant main effect of self-efﬁcacy, F (l, 147) = 4.53, p
< .05, while, once again, the main effect of discrepancy (F (2, 147) = .07, ns) was not
Signiﬁcant. However, in this analysis, the interaction effect of these variables was
marginally Signiﬁcant (F (2, 147) = 2.94, p < .06). Planned contrast tests were conducted
to examine whether self-efﬁcacy had a differential effect across discrepancy conditions.
Similar to the results of the overall difference scores and Trial 3 - Trial 1 scores, those in
the moderate discrepancy condition responded differently to the self-efﬁcacy information
compared with the small and large discrepancy condition in Trial 4: When participants
were given moderate ability discrepancy, low self-efﬁcacy participants (M = 45.505, 5 =

51.03) Signiﬁcantly outlasted high self-efﬁcacy participants (M = 21.855, 5 = 31.64;

29

 

 

60 1
3 .
.2 5O 1
P I ' ﬁ'H—T E
g 30 1 1 Efﬁcacy
: 1 |-l—Low Self- 1
.3 20 —1 L, 1 Efﬁcacy 1
«I 1 “—“m
>
"3 1° "1
E l

0 re- —— ~ e -—~~——w -— a—

Small Moderate Large
Ability Discrepancy

Figure 3.3 Motivation Gain for Trial 4 by Condition (Line graph).

t(147) = 2.50, p < .05) but those in the small discrepancy condition did not Show any
performance difference (t < 1). Different from the previous results, in the large
discrepancy condition, however, low self-efficacy participants lasted 17.565 longer than
high self-efﬁcacy participants; this difference was marginally signiﬁcant (t(147) = 1.90, p
= .06).

Previous research (e.g., Messé et al., 2002) suggested that among those given no
self-efﬁcacy feedback but well aware of the discrepancy between themselves and their
partner, discrepancy in ability and motivation are characterized by an inverted-U function.
To examine the moderating effect of self-efﬁcacy information on this function, I
decomposed the 3-level discrepancy factor into two orthogonal trend contrasts. The
contrast weights for the test of linear trend were (-1, 0, +1), whereas the contrast weights
for the quadratic trend were (-1, +2, -1). Then I redid the 2 (self-efﬁcacy) X 3

(discrepancy) ANOVAS with the latter factor decomposed into the two trend contrasts.

30

The analysis of the overall difference scores resulted in a marginally signiﬁcant self-
efﬁcacy X quadratic interaction effect (t(147) = 1.87, p = .06). As Figure 3.1 shows, low
self-efﬁcacy participants Showed inverted-U Shape trend (as found by Messé et a1. with
no self-efﬁcacy feedback) while high self-efﬁcacy participants Showed an opposite, U-
shaped trend. Also, a non-Signiﬁcant self-efﬁcacy X linear interaction effect on the
overall difference scores (t(147) = 1.30, p = .20) suggested that it would be appropriate to
interpret the results as more consistent with a quadratic pattern than a linear pattern. The
only effect of interest to emerge ﬁ'om parallel analyses of separate examinations of the
last two trials was a marginally signiﬁcant self-efﬁcacy X linear interaction on the Trial 4
— Trial 2 difference score (t(147) = 1.85, p = .07), indicating that low self-efﬁcacy
participants showed increasing linear trend and high self-efﬁcacy participants Showed
decreasing linear trend as the discrepancy become larger (see Figure 3.3).

In addition, to investigate whether there were Si gniﬁcant motivation gains on
these difference scores from the baseline mean of zero, I conducted t-tests for all
conjunctive dyad conditions. First, I examined the overall fatigue corrected difference
score. The overall difference scores Showed Signiﬁcant motivation gains in all
conjunctive dyad conditions. Means and 95% conﬁdence intervals for all conditions are
presented in Figure 3.4. When participants received the low self-efficacy information, the
mean of the conjunctive-small discrepancy condition (M = 32.575, s = 18.37) was
Signiﬁcantly greater than the baseline mean of zero in the individual control condition,
t(65) = 11.41 , p < .001, as were the corresponding means in the conjunctive-moderate
discrepancy condition (M = 46.375, 5 = 51.16), t(63) = 5.84, p < .001, and the

conjunctive-large discrepancy condition (M = 39.805, 3 = 25.66), t(64) = 9.98, p < .001.

31

801

l [:1 Low Self— ‘
Efﬁcacy

‘ [:1 High Self— .
Efﬁcacy

Motivation gain (Trial 3 & 4)

 

 

   

, __7 *1

Small Moderate large

Discrepancy

Figure 3.4 Motivation Gain for the Average of Trial 3 and 4 by Condition (Bar

graph).

The mean for participants who received high self-efﬁcacy feedback in the conj unctive-
small discrepancy condition (M = 35.725, s = 16.32) was signiﬁcantly greater than the
baseline mean of zero in the individual control condition, t(64) = 14.08, p < .001, as were
the corresponding means in the conjunctive-moderate discrepancy condition (M = 22.825,
5 = 31.07), t(65) = 4.73, p < .001, and the conjunctive-large discrepancy condition (M =
27.775, 5 = 22.95), t(66) = 7.78, p < .001.

Second, I examined the fatigue corrected Trial 3 — Trial 1 scores. Means and 95%
conﬁdence intervals for all conditions are presented in Figure 3.5. When participants
received low self-efﬁcacy feedback, the mean of the conjunctive-small discrepancy
condition (M = 41.825, 5 = 24.80) was Signiﬁcantly greater than the baseline mean of zero

in the individual control condition, t(65) = 10.85, p < .001, as were the corresponding

32

 

:3 70
1...
1:, 60 ~
L:
c 50 ,
'5 DLow Self— 1
g 40 . Efﬁcacy
3;: 30 ElHigh Self-
-% 20 , Efﬁcacy
E 10 .

0 1

Small Moderate large

Discrepancy

Figr_rre 3.5 Motivation Gain for Trial 3 by Condition (Bar graph).

means in the conjunctive-moderate discrepancy condition (M = 51.775, s = 56.59), t(63) =
5.89, p < .001, and the conjunctive-large discrepancy condition (M = 43.065, s = 29.41),
t(64) = 9.42, p < .001. When participants received high self-efﬁcacy information, the
mean of the conjunctive-small discrepancy condition (M = 41.385, 5 = 21.33) was
signiﬁcantly greater than the baseline mean of zero in the individual control condition,
t(64) = 12.48, p < .001, as were the corresponding means in the conjunctive-moderate
discrepancy condition (M: 28.325, 5 = 37.88), t(65) = 4.81,p < .001, and the
conjunctive-large discrepancy condition (M = 36.575, s = 36.16), t(66) = 6.50, p < .001.
Finally, the fatigue corrected Trial 4 — Trial 2 difference scores in the conjunctive
dyad conditions also showed signiﬁcant motivation gains. Means and 95% conﬁdence
intervals for all conditions are presented in Figure 3.6. The mean for participants who

received low self-efﬁcacy information in conjunctive-small discrepancy condition (M =

33

23.325, 5 = 22.97) was signiﬁcantly greater than the baseline mean of zero in the
individual control condition, t(65) = 6.53, p < .001, as were the corresponding means in
the conjunctive-moderate discrepancy condition (M = 40.985, 5 = 51.03), t(63) = 5.17, p
< .001, and the conjunctive-large discrepancy condition (M = 36.545, s = 26.06), t(64) =
9.02, p < .001. When participants received the high self-efﬁcacy information, the mean in
the conjunctive-small discrepancy condition (M = 30.065, 3 = 20.32) was signiﬁcantly
greater than the baseline mean of zero in the individual control condition, t(64) = 9.52, p
< .001, as were the corresponding means in the conjunctive-moderate discrepancy
condition (M = 17.325, 3 = 31.64), t(65) = 3.52, p < .01, and the conjunctive-large

discrepancy condition (M = 18.985, 5 = 38.97), t(66) = 3.13, p < .01.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

so 1
1

E 70 1

60 -l 1"
é, .
c 50 1 1_ , ..... _
'8 1 ‘ EJLow Self-1
2’ 40 1 1' -_ 1‘ Efﬁcacy l
o 1 1
g 301 1 1 1 “ 1DHighSeif-'
> 1
=3 20 1 r. I 1 Efﬁcacy
. 1 1 WW
2 10 1 1i-

0 rW. 1 T‘ -_ _______

Small Moderate large
Discrepancy

Figure 3.6 Motivation Gain for Trial 4 by Condition (Bar graph).

34

These ﬁndings are consistent with the results of previous studies (e. g., Hertel et
al., 2000; Messé et al., 2002) which found Signiﬁcant motivation gains under conjunctive
task demands compared with the zero baseline of the individual control condition.

In summary, the preceding analyses suggested the following conclusions. First,
overall, participants Showed greater motivation gains when they had low self-efﬁcacy
compared to when they had high self-efﬁcacy. Second, across the three performance
scores (third trial, fourth trial, and combined trials), the effect of self-efﬁcacy consistently
was strongest under a moderate level of performance discrepancy. Combined with the
trend analyses which were summarized earlier, this result provides evidence that the
Kéhler discrepancy effect is moderated by self-efﬁcacy. Third, in the analysis of
performance in Trial 4 in which participants had more believability on feedback, the
result revealed that low self-efﬁcacy participants also Showed greater motivation gain
than high self-efﬁcacy participants when their relative ability discrepancy was large, even
though this effect did not quite reach conventional statistical Signiﬁcance (p = .06). This
last result suggests the possibility that low self-efﬁcacy participants would Show more
motivation gains than high self-efﬁcacy participants once they passed a certain
discrepancy threshold and were conﬁdent of their partner’s superiority.

Subjective experience

As noted earlier, participants subjectively rated several aspects of their experience
before and/or after each trial. The subjective rating of self-efﬁcacy was administrated
before each trial. In addition, questions on effort, enjoyment of the task, perceived

importance of own performance, evaluation of their performance and satisfaction were

35

asked after each trial. Participants also were asked about performance expectation before
Trial 3 and 4, and the question on perceived discrepancy was asked after Trial 3 and 4.

In the following analyses, I used the relative difference score (RDS) for the Six
questions that were administered for all 4 trials. This score was computed by subtracting
the sum of ratings in the ﬁrst two individual trials (i.e., Trial 1 and 2) — before the self-
efﬁcacy and discrepancy manipulations were administered — from the sum of ratings in
the last two dyad trials (i.e., Trial 3 and 4) — after the manipulations were administered.
Thus, this RDS reﬂected the change in subjective ratings from before to after participants
received the self-efﬁcacy and discrepancy manipulations. Because I used a 9-point scale,
the possible range of RDS score was from —16 to +16. For the two questions which were
asked only before or after Trial 3 and 4 (because the questions were partner-related), I
used the average of these ratings for analysis. The possible range of this average score.
was from 1 to 9. All means and standard deviations on these questions are presented in
Table 3.1.

To examine the question on the subjective rating of self-efﬁcacy, which asked
how well they thought they could perform in the upcoming trial, a 2 (self-efﬁcacy) X
3(discrepancy) ANOVA was conducted on the relevant RDS. The results revealed that
the main effect of self-efﬁcacy was Signiﬁcant (F(1, 147) = 10.97, p < .01), while the
main effect of discrepancy and interaction of these two variables were not signiﬁcant
(both Fs < 1.16). In contrast to the low self-efﬁcacy condition in which participants felt
less self-efﬁcacious in the last two dyad trials compared with the ﬁrst two individual
trials (M = -l .00, s = 2.46), in the high self-efﬁcacy condition participants reported that

they felt more self-efﬁcacious in the last two trials compared with the ﬁrst two trials (M

36

Table 3.1

Means and Standard Deviations of Subjective Ratings

 

 

 

 

Self-efﬁcacy
Low High
Discrepancy
Measures Small Moderate Large Small Moderate Large

 

Differences of Subjective Ratings in the (Dyad) Group and in the Individual Trials (RDS)

Self-efﬁcacy -.35 (2.12) -154 (2.59) -1.16(2.59) .36 (2.27)

Effort 1.00 (2.08) .83 (1.81) 1.68(1.82) .68 (2.46)

.35 (1.44) -.04 (2.65)

.62 (2.61) .33 (2.62)

Enjoyment -.85 (2.98) -150 (2.11) 4.00 (2.93) .36 (2.45) -1.96 (2.58)-1.37 (2.31)

Importance 1.96 (2.75) .58 (2.47) 1.60 (2.58) .60 (3.03)

.19 (2.88) .26(2.38)

Evaluation -.54 (2.97) -1.63 (2.75) -1.28 (2.59) 1.32 (2.82) -1.08 (3.16) -1.19(2.68)

Satisfaction -.27 (2.78) -1.38 (2.16) -1.52 (3.23) .80 (2.69) 2120.25) 4.44 (3.09)

 

Average Scores of Subjective Ratings in the Trial 3 and 4

Expectation. 5.04(1.32) 4.27(1.96) 3.98 (2.01) 5.30(1.00) 4.44(1.54) 4.46 (2.08)

Discrepancy 5.19(1.44) 6.58(1.35) 6.80(1.13) 4.64(1.25) 6.54(1.09) 6.69(1.44)

 

Note. RDS (Relative Difference Score) indicated the change in subjective ratings after
participants received the self-efﬁcacy and discrepancy manipulations.

37

= .22, s = 2.17). It should be noted that participants in the high self-efficacy and large
discrepancy condition (M = -.04, s = 2.65) felt about the same level of self-efﬁcacy in the
last two trials as they felt in the ﬁrst two trials, perhaps because the positive belief in their
absolute competence was offset by the large, negative discrepancy in ability between
themselves and their partners.

For the question which asked how much effort they put into the last trial, a 2 (self-
efﬁcacy) X 3(discrepancy) AN OVA was conducted on the relevant RDS. Results
revealed a marginally Signiﬁcant main effect of self-efﬁcacy (F (1 , 147) = 2.93, p < .09),
while the main effect of discrepancy and interaction of these two variables were not
signiﬁcant (both Fs < l). The marginally Si gniﬁcant main effect of self-efficacy indicated
that participants in both the low and high self-efﬁcacy condition put out more effort in
the last two dyad trials compared to their ﬁrst two individual trials, but the amount of
their perceived effort was tended to be somewhat greater when they had low self-efﬁcacy
(M = 1.17, s = 1.92) than when they had high self-efﬁcacy (M = .54, s = 2.54).

To examine whether there was any difference in task enjoyment between
conditions, a 2 (self-efﬁcacy) X 3 (discrepancy) ANOVA was conducted on the relevant
RDS. The results revealed that the main effect of discrepancy was Signiﬁcant (F (2, 147)
= 4.30, p < .05), while the main effect of self-efﬁcacy and interaction of these two
variables were not Si gniﬁcant (both Fs < 1.68). Examination of means suggested that
participants in all discrepancy condition perceived leSS enjoyment in the last two dyad
trials compared to their ﬁrst two individual trials, but the level of their perceived task
enjoyment differed by the discrepancy conditions. To examine whether the means in each

discrepancy conditions were Signiﬁcantly different ﬁom each other, a Newman-Keuls

38

pairwise post-hoe test was conducted. This test result revealed that those in the moderate
discrepancy condition (M = -1.74, s = 2.35) perceived Signiﬁcantly less enjoyment on the
task than those in the small discrepancy condition (M = - 0.25, s = 2.77, p < .05). Also,
those in the large discrepancy condition (M = -1.19, s = 2.61) perceived less enjoyment
on the task than those in the small discrepancy condition (M = - 0.25, s = 2.77), but the
signiﬁcance was marginal (p < .07). There was no reliable difference between the
moderate and large discrepancy conditions.

For the question, that asked about the importance of performing well after each
trial, a 2 (self-efﬁcacy) X 3 (discrepancy) ANOVA was conducted on the relevant RDS.
The results revealed that the main effect of self-efﬁcacy was signiﬁcant (F (1, 147) =
5.61 , p < .05), while the main effect of discrepancy and interaction of these two variables
were not signiﬁcant (both Fs < 1.41). In the low self-efficacy condition, participants
reported that they felt it was more important to perform well in the last two trials than the
ﬁrst two (M = 1.40, s = 2.64) to a greater degree than participants in the high self-efﬁcacy
condition did (M = .35, s = 2.74). It would be worthwhile to discuss why low self-
efﬁcacy participants perceived their performance to be more important compared to high
self-efﬁcacy participants. As discussed earlier, when participants received the high self-
efﬁcacy information, supposedly based on the performance in the ﬁrst two individual
trials, they might have experienced feelings of complacency or overconﬁdence, which
could contribute negatively to task performance. Therefore, they might not have any
additional need to prove their ability because they already had received very favorable
feedback. As such, the dyad trials would be less important to them. On the other hand, the

participants who received low self-efﬁcacy information, based on feedback about the ﬁrst

39

two individual trials, might have experienced a feeling of failure, and realize that there
was something wrong with their task performance. With such personal and social
uncertainty about their ability, participants might have perceived the upcoming (dyad)
trials as a “second-chance” opportunity to prove their ability. Such an orientation could
have served to heighten the importance of these new work periods.

To explore the question of the evaluation of their performance, which asked how
well they thought that they performed during the last trial, a 2 (self-efﬁcacy) X 3
(discrepancy) ANOVA was conducted on the relevant RDS. The results revealed a
marginally signiﬁcant main effect of self-efﬁcacy (F (1, 147) = 3.30, p = .07) and a
Signiﬁcant main effect of discrepancy (F (2, 147) = 5.99, p < .01), while the interaction of
these two variables did not reach statistical Signiﬁcance (F < 1.34). The marginally
signiﬁcant main effect of self-efﬁcacy suggested that the tendency to perceive their
performance as successful in the last two trials, compared to the ﬁrst two trials was
weaker when participants had low self-efﬁcacy (M = -1.l3, s = 2.78) than high self-
efﬁcacy (M = -.35, s = 3.08). Also, the signiﬁcant main effect of discrepancy suggested
that those in the small discrepancy condition (M = .37, s = 3.02) Showed a more positive
evaluation on their performance in the last two trials compared to the ﬁrst two trials, but
those in the moderate (M = -1.34, s = 2.95) or large (M = -l .23, s = 2.61) discrepancy
condition Showed a negative evaluation on their performance in the last two trials
compared to the ﬁrst two trials. The mean differences between the small and moderate or
large discrepancy conditions were Signiﬁcant (t(99) = 2.88, p < .01; t(101) = 2.89, p < .01,
respectively). There was not a Signiﬁcant difference between the moderate and large

discrepancy conditions (t < 1). This pattern of result suggested that those in the small

40

discrepancy condition did not perceive the performance result of quitting earlier than
their partner as indicating that they had failed.

For the question that asked about participants’ satisfaction with their task
performance after each trial, a 2 (self-efﬁcacy) X 3 (discrepancy) ANOVA was
conducted on the relevant RDS. The results revealed that the main effect of discrepancy
was Signiﬁcant (F (2, 147) = 7.21, p < .01), while neither the main effect of self-efﬁcacy
nor interaction of these two variables was Signiﬁcant (both F5 < 1.23). The Signiﬁcant
main effect of discrepancy suggested that those in the moderate (M = -1.76, s = 2.78;
t(99) = 3.65,p < .001) and large (M: -1.48, s = 3.13; t(101) = 2.98,p < .01) discrepancy
condition felt more dissatisfaction with their task performance on the last two trials than
the ﬁrst two trials, but, if anything, those in the small (M = 0.25, s = 2.76) discrepancy
condition felt more satisfaction in the last two trials compared to the ﬁrst two trials.

For the performance expectation question, which asked how likely participants
thought it was that they would do as well or better than their partner before each dyad
trial (i.e., before the Trial 3 and 4), a 2 (self-efﬁcacy) X 3(discrepancy) ANOVA was
performed on the average score of trial 3 and 4. Results revealed a signiﬁcant main effect
of discrepancy (F (2, 147) = 4.65, p < .05), while the main effect of self-efficacy and
interaction of these two variables were not Signiﬁcant (both Fs < 1.23). The Signiﬁcant
effect of discrepancy suggested that those in the small discrepancy condition (M = 5.17, s
= 1.17) reported a higher expectation of good performance than those in the moderate (M
= 4.36, s = 1.74; t(99) = 2.75, p < .01) and large discrepancy conditions (M = 4.23, s =
2.04; t(101) = 2.85, p < .01). There was not a reliable difference between the moderate

and large discrepancy conditions (t < l).

41

For the question which asked about the perceived discrepancy between their
performance and the performance of their partner after each dyad trial (i.e., after the Trial
3 and 4), a 2 (self-efﬁcacy) X 3(discrepancy) ANOVA was conducted on the average
score of Trial 3 and 4. The analysis revealed that the main effect of discrepancy was
Signiﬁcant (F(2, 147) = 30.93, p < .001), while the main effect of self-efﬁcacy and
interaction of these two variables were not Signiﬁcant (both Fs < 1.29). A signiﬁcant
main effect of discrepancy suggested that those in the small discrepancy condition (M =
4.92, s = 1.37) reported signiﬁcantly smaller perceived discrepancy than those in the
moderate (M = 6.56, s = 1.21; t(99) = 6.38, p < .001) and large discrepancy conditions (M
= 6.74, s = 1.29; t( 101) = 6.95, p < .001). There was no mean difference between the
moderate and large discrepancy conditions (t < 1). This last result suggested that
participants perceived about the same Size of discrepancy once they had realized a certain

amount of discrepancy.

42

Chapter 4

DISCUSSION

Implications of the present study

The results of this study suggested that participants Showed greater motivation
gains when they had low self-efﬁcacy than when they had high self-efﬁcacy; and, this
effect was strongest under a moderate level of perceived discrepancy. However, given the
fact that there were signiﬁcant motivation gains in all the high self-efﬁcacy conditions as
well as all the low self-efﬁcacy conditions, it would be more accurate to say that low self-
efﬁcacy participants exerted extra effort when they had a moderate discrepancy (and, to a
lesser degree, when they had a large discrepancy) rather than saying that high self-
efﬁcacy impaired task performance in this condition. Therefore, one implication for the
literature on self-efﬁcacy that the present results provide is that these ﬁndings should not
be taken to indicate that high self-efﬁcacy iS a detrimental factor of task performance.
Rather, the results suggest that low self-efﬁcacy could have motivating effects on task
performance.

There are likely to be a few reasons why high self-efﬁcacy participants did not
work as hard as low self-efﬁcacy participants, or alternately, why low self-efﬁcacy
participants Showed extra effort, particularly in the moderate discrepancy condition. First,
high self-efficacy participants must have perceived the high self-efﬁcacy feedback as a
success. AS indicated by previous studies (Bandura & J ourden, 1991; Stone, 1994;
Vancouver et al., 2002), this feeling of success might have elicited complacency for their

previous performance and/or overconﬁdence in their ﬁrture success at the task. This

43

conclusion is consistent with the result of the analysis of the question that asked
participants to report their self-efﬁcacy level before the third trial. Because this question
was administered after the manipulation of self-efﬁcacy and before the third trial,
participants’ ratings were not affected by the performance results of the third trial. In this
analysis, participants who had received the high self-efﬁcacy manipulation thought that
they could perform signiﬁcantly better (M = 5.92, s = 1.63) than participants who had
received the low self-efﬁcacy manipulation (M = 5.35, s = 1.63), t(151) = 2.19, p < .05.
This favorable expectation might have caused complacency or overconﬁdence in high
self-efﬁcacy participants. Therefore, there is a possibility that participants with high self-
efﬁcacy might not have worked very hard in subsequent (dyad) trials, even though, as the
weaker dyad member, they still had high level of instrumentality and a realistic and
achievable goal in the moderate discrepancy condition.

Second, whether participants put an extra effort in the conjunctive dyad trials
might hinge upon the participants’ perceived importance of performing well on these
trials. High self-efﬁcacy participants might not have as strong a need to prove once again
their ability —- to themselves, to the experimenter, to their partner — because they already
had received very favorable and public performance feedback. This ‘ﬁvarm glow of
success” might have caused them to think that demonstrating good performance on the
dyad trials was leSS important and/or necessary. Hence, they might not have as much
reason to expend maximum effort. On the other hand, low self-efﬁcacy participants were
led to believe that they did not prove themselves to be good performers in the initial two
trials. Given this, it is reasonable to expect that they would view the upcoming dyad trials

a “second-chance” opportunity to prove their ability to themselves, the experimenter,

44

and/or their partner. Therefore, low self-efﬁcacy participants might have perceived that
the task performance in the dyad trials was more important to them, and, given this
chance, they expended greater to Show they were, in fact, better at the task than their
original performance had indicated. This interpretation was supported by their responses
to the question that asked about the importance of performing well. Compared to the
initial two trials, those in the low self-efﬁcacy condition perceived that it was
Signiﬁcantly more important for them to perform well in the latter two trials (M diff. = .70,
s = 1.32; t(74) = 4.60, p < .001.), while high self-efﬁcacy participants did not Share this
perception (M diff. = .17, s = 1.37; t(77) = 1.13, ns). Also, when I compared these two
mean differences, the mean in the low self-efﬁcacy condition was Signiﬁcantly greater
than the mean in the high self-efﬁcacy condition (t(151) = 2.42, p < .05). Moreover, it is
reasonable that low self-efﬁcacy participants would actually try hardest in the moderate
discrepancy condition, because knowmg that their coworker performed only somewhat
better than they might suggest to them that they have a realistic chance of improving their
performance. In contrast, feedback that they and their partner had performed about the
same would provide no evidence that they (on their partner, for that matter) could do
better. And, learning that their partner had substantially outperformed them could be
taken as a reason to be pessimistic about any chance to Show improvement.

Finally, to examine the reason why low self-efﬁcacy participants exerted extra
effort especially in the moderate discrepancy condition, it is useful to consider Wright
and Kirby’s (2001) integrative effort analysis of cardiovascular (CV) responses (e. g.,
heart rate, systolic blood pressure, diastolic blood pressure) in motivated performance

situations. In this analysis, the authors ﬁrst took the proposition of Christ (1976, 1981)

45

that the sympathetic nervous system’s inﬂuence on the CV system is proportional to
effort or task engagement (i.e., Changes in CV responses reﬂect effort exertion or
striving). Then, they utilized and extended Brehm’s motivational intensity theory (Brehm
& Self, 1989) to Specify when people should manifest greater or lesser degrees of effort-
related CV response. Through this procedure, Wright and Kirby proposed four effort-
related and Six ability-related implications, and Showed evidence for these implications.
Among these, I would like to discuss a few implications which are particularly related to
the results of the current study.

First, they demonstrated that sympathetic CV responses (and hence, effort) tended
to be greater for low than high ability people where demand is modest (i.e., if it is
possible and worthwhile to meet the challenge). Given the fact that the self-efﬁcacy
manipulation of the current study includes normative ability information (i.e., 20 or 80
percentile in the total population), it would be plausible to consider low or high self-
efﬁcacy participants as seeing themselves as having low or high ability. Also, in terms of
the Social comparison/Goal setting explanation, a “possible and worthwhile challenge”
could be considered to be a realistic and reachable goal (i.e., moderate discrepancy in
conjunctive task demands). Therefore, Wright and Kirby’s (2001) result is consistent
with the current result in that low self-efﬁcacy participants Showed bigger motivation
gain than high self-efﬁcacy participants in the moderate discrepancy condition.

Second, Wright and Kirby (2001) demonstrated that sympathetic CV responses
(i.e., those associated with effort) were low for both ability groups if demand was
particularly high. In other words, when a challenge was especially difficult to meet, effort

level became minimal, irrespective of the perceived capacity to perform. This pattern is

46

also consistent with the current result, which Showed no signiﬁcant performance
difference between high and low self-efﬁcacy participants in the large discrepancy
condition (although there was marginally signiﬁcant difference on the Trial 4 - Trial 2
dependent measure).

Third, Wright and Kirby (2001) demonstrated that sympathetic CV responses to a
possible challenge were moderated by the perceived importance of success. That is, effort
level was proportional to the difﬁculty of challenge when the importance of success was
high enough to make a possible challenge worthwhile, but effort was minimal
irrespective of demand when importance of success was sufficiently low. Here, the high
efﬁcacy participants put low importance on success and were relatively insensitive to the
difﬁculty of the challenge (i.e., the discrepancy between their own and their partner’s
ability). But low efficacy participants placed a high premium on success. Wright and
Kirby’s analysis predicts that a) low efﬁcacy/ability people Should Show greater effort
(compared to their high efﬁcacy/ability counterparts), and b) they Should increase effort
as the degree of challenge increases. However, 0) very high levels of difﬁculty could be
seen as challenges that were impossible to overcome, and d) it would be at this point that
people would reduce their effort. This was essentially the pattern of results that occurred
in the current study.

An important research question addressed in this study was whether the thler
discrepancy effect would be maintained or altered by the high or low self-efﬁcacy
information. In the Results section, a marginally Signiﬁcant self-efﬁcacy X quadratic
interaction was described. When participants received the high self-efﬁcacy information

(i.e., they perceived that they are a good performer and the experimenter and her partner

47

knew this), they tended to exert less effort as discrepancy increased from small to
moderate, but they tended to put out Slightly more effort when discrepancy increased
from moderate to large. The Pattern was a (non-Signiﬁcant) U-shaped function.
Considering this pattern, it is clear that there was no hint of the inverted-U thler
discrepancy effect when participants had high self-efﬁcacy. On the other hand, when
participants received low self-efﬁcacy information (i.e., they perceived that they are a
poor performer and the experimenter and her partner knew this), they consistently
worked harder in the higher discrepancy condition. By Trial 4, the combined Moderate +
Large discrepancy conditions differed marginally (t(147) = 1.91, p < .06) from the Small
discrepancy condition. And overall, the quadratic, inverted-U trend was marginally
Signiﬁcant (t(147) = 1.87, p < .07) for these participants (compared to the high self-
efﬁcacy participants). Thus, in the low self-efﬁcacy condition, we observed the basic
inverted U-shape pattern that K6hler originally found, with a drop in effort between
moderate and large discrepancy but one that was somewhat less dramatic than had been
found in other studies (viz. Messé et al., 2002).

It is also useful to discuss the results of this study in terms of the impression
management concerns. Previous studies (Hertel, Kerr, Schefﬂer, et al., 2000; Kerr, Messé,
Park, & Sambolec, under review; Lount et al., 2000) suggested that impression
management concerns might be a basis for generating motivation gains. Recently, Kerr
and his colleagues (Kerr et al., under review) tested a few theoretical possibilities (i.e.,
indispensability of effort, social comparison/ goal setting and impressions management
explanation) involved in motivation gain paradigm by systematically varying the nature

of the performance feedback and identiﬁability of the weaker performer. In their results,

48

motivation gains decreased as the availability of performance-relevant information to
one’S partner decreased; compared with their Full and End-of-trial feedback conditions
(in which participant’s status as the less capable group member was immediately evident
as soon as a participant quit), participants in their Delayed feedback condition (who could
know only who failed ﬁrst but not when their partner failed) Showed attenuated (but still
Signiﬁcant) motivation gain. More interestingly, participants in the No feedback
condition (in which no one could tell who quit ﬁrst or how long either had persisted) did
not Show any motivation gain, even though they were aware that one of them was
indispensable to their team’s success. Thus, merely knowing that one person — the
“weaker link” — was responsible for how well or poorly the team did (without being able
to identify who this “weaker link” was) was not sufﬁcient information to generate
motivation gains. Kerr and his colleagues speculated that this result could stem from the
impression management concerns; it might be more crucial for participants to avoid being
known to be the weaker member rather than actually being the weaker member. In the
present study, participants may have been concerned with managing a favorable
impression to two targets (i.e., the experimenter and their partner). They were told that
the experimenter would know who quit ﬁrst and also, they could expect that their partner
would see their results on her computer screen (and, since the trial ended as soon as the
participant quit, it was evident to all who quit ﬁrst). If participants’ concern was in
maximizing group scores, participants might have felt equal amount of indispensability in
each self-efﬁcacy condition. However, if the participants’ goal was to create a favorable
(or avoid an unfavorable) impression among others (i.e., proving their capabilities) rather

than maximizing group score, we would expect to see more motivation gains under

49

conditions where participants had low self-efﬁcacy (i.e., low ability). Also, this
impressions management motive might be greater in the moderate or large discrepancy
conditions because participants might have viewed these as contexts where one could
realistically prove their capabilities. Outlasting another person of equal ability would not
enhance one’S reputation. Thus, the current results -— that the strongest motivation gain
occurred in the moderate (and possibly large) discrepancy condition when participants
had low self-efﬁcacy — are consistent with the interpretation of the ﬁndings of previous
studies that impression management could play a role in generating the Kiihler effects.

In addition to the impression management explanation, some motivation gain
researchers (e.g., Hertel et al., 2000; Kerr et al., under review; Messé et al., 2002, in
preparation) have considered the Instrumentality X Value (i.e., indispensability of effort)
explanation and the Social comparison (or Goal-setting) explanation as plausible and
useful explanations of the Kiihler motivation gain effects, as I already described in the
introduction. Recently, to examine the Kéhler discrepancy effect under different task
demands, Messé and his colleagues (Messé et al., in preparation) manipulated relative
ability discrepancy (i.e., small, moderate and large discrepancy between a participant and
her partner) under conjunctive or additive task demands.

The investigators expected greater motivation gains in the conjunctive than
additive task condition because they thought both indispensability and social comparison
processes would additively contribute to produce performance gains in the conjunctive
task condition while only social comparison would contribute to gains in the additive task

condition.

50

As expected, the results revealed that participants Showed signiﬁcant motivation
gains in both task conditions compared to the individual control condition, but those in
the conjunctive condition Showed Signiﬁcantly greater motivation gains than those in the
additive condition. Also, as Kbhler originally found, there was a clear pattern of inverted-
U in the conjunctive task condition (but not the additive task condition) when
participants’ actual perception of the discrepancy between their own and their partner’s
past task performance was the independent variable. These results suggest that both
indispensability and social comparison processes are working additively under the
conjunctive task demands to generate both thler effects.

Based on these results, it could be useful to think about whether participants in the
present study were affected to a different degree by indispensability or social comparison
processes, depending on their self-efﬁcacy manipulations. Considering the task demand,
it seems that the inﬂuences of indispensability are the same for all participants regardless
of their self-efﬁcacy information, because the present study was conducted only under
the conjunctive task condition —- and it is only in this condition that the weaker member is
indispensable for dyadic performance. However, different ﬁ'om the inﬂuences of
indispensability, it seems likely that the impact of a social comparison motive change
depending upon the self-efﬁcacy manipulation.

Stroebe and his colleagues (Stroebe, Diehl, Abakoumkin & Arnscheid, 1990)
suggested that when task accomplishment is unimportant or not valued by group
members, there would be a downward bias in their social comparison process — the more
capable members would adjust their goals downward to be more in line with the less

capable members. Conversely, if task accomplishment is important or valued by group

51

members, the bias should be in the opposite direction — that is, those performing less well
Should set goals closer to the performance levels of the more capable group members. AS
I discussed earlier, low self-efﬁcacy participants considered the dyad trials as more
important than high self-efﬁcacy participants. The higher importance perception of low
self-efﬁcacy participants might have led them to engage more in upward social
comparison than high self-efﬁcacy participants. As a result of this difference in social
comparison tendency, those who had low self-efﬁcacy should have Showed bigger
motivation gains to reduce the performance discrepancy between them and their partner.
Since they always were the less capable dyad members, the high self-efﬁcacy participants
who placed less importance on good task performance in the dyad trial may Simply have
settled for their own, lower performance level as an acceptable goal.
Limitations and Directions for Future Research

There are some limitations in this study that are worthwhile to discuss,
particularly as possible directions for future research. First, to manipulate large
discrepancy, participants received a partner’s score that was twice as good as their own.
This level was chosen to rrrirror prior manipulations (e.g., Messé et al., 2002). However,
there may be ambiguity or differences across people in the concept of unachievable
discrepancy. How large does an actual discrepancy have to be for it to be perceived by
the weaker worker as large enough subjectively to be viewed as creating a situation
where additional effort is perceived as being far from optimally useful and /or
appropriate? In the Wright and Kirby’s (2001) analysis, they demonstrated that
participant’s perception of this threshold depends upon the importance of success. That is,

as the importance of success increases, the range of perceived achievable performance

52

increases. Maybe this connection iS why there was a non-Signiﬁcant and modest drop in
effort by the low self-efﬁcacy participants in the large discrepancy condition of the
present study, compared to the reliable quadratic trend in Messé et al.’S (2002) study,
where self-efﬁcacy information was not provided (i.e., where it was uncertain). That is,
getting low self-efﬁcacy (i.e., own ability) information (in the present study) may have
increased the importance of good performance relative to those (in Messé et al., 2002)
who received no self-efﬁcacy feedback. This Speculation suggests that one might have to
use a much higher “High” discrepancy to replicate the full inverted-U function among
participants who consider good performance very important.

Second, the results of this study for the low self-efﬁcacy participants were similar
to those in a previous study (Messé et al., 2002) which replicated the K6hler discrepancy
effect. Besides the explicit self-efﬁcacy (i.e., own ability) feedback in the present study
(vs. the absence of any such feedback in Messé et al., as noted above), another potentially
important difference between these two studies was whether participants could see each
other during performing the persistence task. Participants in Messé et al.’s (2002) study
performed the task with their partner (i.e., a confederate) in the same room — so they
could watch each other - while the participants in the present study brieﬂy met one
another before they performed the task (i.e., at a point, when they did not know who their
partner would be) and they performed the task without seeing each other. With these
differences in the experimental settings in mind, I compared the results of the low self-
efﬁcacy participants in the present study with the Messé et al.’S (2002) to examine how
the self-efﬁcacy feedback and the face-to-face interaction might have affected task

performance. In order to examine whether there were differences in motivation gains

53

between these two studies and whether there were overall linear trend, overall quadratic
trend or study X quadratic interaction, I conducted contrast analyses in which 3-level
discrepancy factor was decomposed into linear and quadratic trends. These comparisons
are presented in the Figure 4.1 and 4.2 (The Figure 4.1 utilized the average scores of Trial
3 and 4, and the Figure 4.2 utilized the scores of Trial 4). Because there was no difference
on the scores of Trial 3 between these two studies, I do not present the results of the Trial
3. AS presented in the Figure 4.1, for the average scores of Trial 3 and 4, the participants
in the Messé et al.’s (2002) study showed greater overall motivation gains than the low
self-efﬁcacy participants in the present study, but this difference was only marginal
(t(119) = 1.84, p < .07). Also, while there was no signiﬁcant overall linear trend (t < 1),
an overall quadratic trend was marginally signiﬁcant (t(I 19) = 1.80, p < .08). Similarly,
for the scores of Trial 4, which are presented in the Figure 4.2, the same pattern was
found, but the effects were stronger. As was with the average scores, the participants of
the Messé et al.’s (2002) study Showed greater overall motivation gains than the low self-
efﬁcacy participants of the present study (t(1l9) = 4.33, p < .001). Also, an overall
quadratic trend was Signiﬁcant (t(119) = 2.49, p < .05) while an overall linear trend was
not (t < 1). In addition, study X quadratic interactions for both dependent variables (i.e.,
the average scores of Trial 3 and 4, and the scores of Trial 4) were not Signiﬁcant (both ts
< 1.22). In summary, 1) the motivation gain was greater in Messé et al.’s (2002) study,
although the same inverted-U pattern basically occurred in both the studies, as well as in
thler’s original research. These results are consistent with this possibility that direct
concerns with impression management in the face-to-face group could have produced

bigger motivation gains (i.e., extra effort exertion). 2) As presented in both ﬁgures, the

54

participants in the Messé et al.’s (2002) study Showed greater performance decreases

from the moderate to large discrepancy condition compared to the low self-efﬁcacy

participants of the present study. Perhaps the visual feedback of an unfatigued, high

discrepancy partner that the procedure afforded participants in Messé et al.’S (2002) study

convinced them that it was unrealistic to attempt to match her level of performance.

100 “-1
90 1
80 1
7o 1
a 1
50 1

l
l

40 1
30
20 l

101

1

Motivation Gain in Trial 3 8. 4

0 .;___.___ ‘

rW W. ..______.._____.____ W

1 + Participants in
Messé et al.'s

/\ ' (2002) Study
1’ a

+ Low self-efficacy :
participants in the.E

present study 1
_ ___.l

 

Small Moderate Large
Ability Discrepancy

Figr_rre 4.1 Comparison Between the Present Results of Low Self-efﬁcacy

Participants and Messé et al.’s (2002) Results (Motivation Gain for the Average of Trial

3 and 4).

55

+Participants in 1
70 J 1 Messé et al.’s f
60 _1 1 (2002) study 1

1 l l
l l

40 1 I+Low self-efﬁcacy 1
' participants in the 1
30 1 1 present study 1
l
l

l. __ ____._1_-___ _ﬂ, 1

Motivation Gain in Trial 4
01
O

o -— —-— _ 4 .
Small Moderate Large

 

Ability Discrepancy

Figr_rre 4.2 Comparison Between the Present Results of Low Self-efﬁcacy

Participants and Messé et al.’S (2002) Results (Motivation Gain for Trial 4).

Through the above analyses, 1 found that the participants who had low self-
efﬁcacy in the present study Showed Similar results to participants who were uncertain
about their ability (i.e., Messé et al.’S study). To further compare the results of these two
studies, it would be useﬁrl to compare the high self-efﬁcacy participants of the current
study with participants who did not receive any information on their ability. To do so, I
compared the present results of the high self-efﬁcacy participants with the results of the
Messé et al.’s (2002) participants by conducting the same contrast analyses, reported
above, that I did for the low self-efﬁcacy participants (i.e., contrast analyses on the study
main effect, overall linear trend, overall quadratic trend and study X quadratic interaction
effect). These comparisons are presented in the Figure 4.3 and 4.4 (The Figure 4.3

utilized the average scores of Trial 3 and 4, and the Figure 4.4 utilized the scores of Trial

56

4; Again, because there was no difference on the scores of Trial 3 between these two
studies, I do not present the results for Trial 3). As presented in the Figure 4.3, for the
average scores of Trial 3 and 4, the participants of the Messé et al.’s (2002) study Showed
greater overall motivation gains than the high self-efﬁcacy participants of the present
study (t(122) = 3.79, p < .001). Also, while there were no Signiﬁcant overall linear and
overall quadratic trends (both ts < 1), a study X quadratic interaction was marginally
signiﬁcant (t(122) = 1.98, p < .06). For the scores of Trial 4, which are presented in the
Figure 4.4, the same basic results were found. AS was the case for the average scores, the
participants of the Messé et al.’S (2002) study showed greater overall motivation gains
than the high self-efﬁcacy participants of the present study (t(122) = 6.01 , p < .001). Also,
while there were no Si gniﬁcant overall linear and overall quadratic trends for the scores
of Trial 4 (both ts < 1.52), the study X quadratic interaction was Signiﬁcant (t(122) = 2.40,
p < .05). In summary: 1) Similar to the comparison between the participants of Messé et
al.’s study and the low self-efﬁcacy participants of the present study, overall motivation
gains were greater for the participants of the Messé et al.’s study than the high self-
efﬁcacy participants of the present study; and, 2) a Signiﬁcant study X quadratic
interaction suggested that the participants of the Messé et al.’s study showed inverted-U
pattern while the high self-efﬁcacy participants in the current study showed U-Shape
pattern. Different from the previous comparison with the low self-efﬁcacy participants,
this last result suggested that high self-efﬁcacy participants Showed quite different
sensitivity to partner discrepancy compared to the participants who were uncertain about
their ability. AS noted earlier, because the experimental settings of these two studies were

different, it is a bit risky to compare the results. Therefore, future work needs to examine

57

within the same experimental design and procedure whether the Kéihler discrepancy
effect would be pronounced (i.e., a stronger effect) if face-to-face groups were examined
and if the actual large discrepancy in ability more uniformly generates equivalent

perceptions in group members.

 

 

 

 

v 100 *

°5 90 1

2 80 l T ”WT—WT l
.9 l 1+Participants in 1
ﬁ 70 ~ ' Messé etal.’s 1
.5 so ~ 1 (2002) study 1
5 50 ~ 1 l
8 40 _ l-o—High self-efficacy i
r: 1 participants in thel
13 30 g -\'/. 1 present study 1
g 20 ~ L___ l
“3 1o 4’

5 1

0 wH-L Err-W- _- _._-_-----..-.-.

Small Moderate Large
Ability Discrepancy

Figgre 4.3 Comparison Between the Present Results of High Self-efﬁcacy
Participants and Messé et al.’S (2002) Results (Motivation Gain for the Average of Trial

3 and 4).

58

100 *1

 

v 901

: 8,1 1 .. .-
1: + Partrcnpants m

c 701 1 Messé etal.’s 1
'2 60 1 (2002) study 1
'5 1 1 l
(D 50 1 -—o— High self-efﬁcacy 1
5 40 1 1 participants in the1
g 201

5 101

01-_ﬂf4_s_wws

Small Moderate Large
Ability Discrepancy

Figgre 4.4 Comparison Between the Present Results of High Self-efﬁcacy

Participants and Messé et al.’s (2002) Results (Motivation Gain for Trial 4).

In an effort to generalize the present results, it also would be useful to investigate
these effects under different task demands and/or in different situations. For example, it
would be interesting to examine whether low versus high self-efﬁcacy would elicit
different magnitudes of motivation gain in a coactive task in which social comparison
and competition motives are thought to be more salient. In addition to this, future work
should investigate the effect of self-efﬁcacy on motivation gains with other and more
realistic group tasks.

In this study, participant’s perception of the importance of performing well was a
crucial variable affecting whether or not participants put forth extra effort. Based on these
results, it is likely to be fruitful for ﬁiture work to directly manipulate the importance of

performing well as a means of examining how this importance perception could affect the

59

relationship between discrepancy and performance. To manipulate importance, there are
several feasible options. For example, to increase importance, researchers could give
participants extrinsic incentives for their good performance or they could give them
information that statements linked task performance with some valued personal ability
(e.g., IQ). Conversely, to decrease the importance, researchers could use some methods to
derogate the task itself (e. g., allege that performance was of little interest to the

experimenters).

6O

Chapter 5

CONCLUSION

In this study, I investigated the potential moderation effect of self-efﬁcacy on the
Kohler discrepancy effect (i.e., whether the Kohler discrepancy effect would be
maintained, enhanced, or diminished depending on the self-efﬁcacy information that
participants were given). Self-efﬁcacy on the task was manipulated by giving false
feedback about one’s percentile rank in the population. Results revealed that participants
showed greater motivation gains performing under conjunctive task demands when they
had low self-efficacy than high self-efﬁcacy, and that this effect was strongest with a
moderate level of perceived discrepancy in own and coworker’s ability. Also, the results
suggested that the Kohler discrepancy effect seemed to be maintained when participants
had low self-efﬁcacy but not when they had the high self-efﬁcacy information. In
interpreting the present results, I discussed a number of psychological processes that are
likely to be involved in this effect (e.g., complacency or overconﬁdence of high self-
efﬁcacy participants, social comparison or goal comparison, perceived indispensability of
effort, participants’ perceived importance of task performance, striving to create a
favorable impression to others)

Until now, more empirical evidence relevant to identifying the underlying
psychological processes for the Kohler’s simple motivation gains effect has been
accumulated than for the Kohler discrepancy effect. The present study was an attempt to
increase our understanding of the latter by examining if self-efﬁcacy moderates the

Kohler discrepancy effect. Findings suggested the present research could prove to be the

61

initial step in a fruitful line of inquiry, particularly if future work pursues some of the
interesting clues about the Kohler discrepancy effect that the present study yielded. In
addition, future research needs to examine other possible moderating variables and

underlying processes that determine the conditions that promote group motivation gains.
Once we accumulate such evidence we will have achieved a ﬁrmer basis for

understanding the nature of the Kohler discrepancy effect and related phenomena.

62

APPENDICES

63

APPENDIX A

Instructions on the computer program

Screen 1.

Welcome to the Holding Hands experiment, "Participant ’s Name". Our research is
following other research which looks at the persistence of people working at a physical
task. In a moment, you will be performing a simple endurance task. Please read through

all of these instructions carefully before you begin in order to prepare for this task.

The instructions for performing this task are as follows. First, for your own comfort,
please take off all bracelets and watches that you may be wearing. Please place whatever
you removed in the black paper bag located to the left of your keyboard. You may do this

now.

64

Screen 2.
At this time, grasp the weight, slipping your hand through the grip provided with your
dominant hand. Your dominant hand is usually the hand that you write with. So, if you

write with your right hand, your dominant hand is your right hand.

Next, you should see two metal poles near the computer keyboard with a string between
them. Move your chair so that the outside of the shoulder of your dominant arm is lined
up between the metal poles; you should be sitting so that when you hold your dominant

arm straight out in front of you, it is right in the middle of the two metal poles with your

wrist over the string.

When a trial begins, you will raise your dominant arm between the two metal bars so that
your wrist is between the top of the two poles and your wrist is over the string. The
purpose of the task will be to hold your arm out straight with your elbow locked, for as

long as you can.

When you are done reading this page, please click NEXT button.

65

Screen 3.

When you cannot hold your arm up any longer, you may let your arm drop. This motion
will bring your wrist down to the string that is attached to a mechanism. This mechanism
will press the spacebar on your keyboard. When the spacebar is pressed, your trial will

be over.

Again, the goal of this task is to hold your arm up for as long as you can after your trial
has begun. During the trial, you must always keep your wrist over the string, your arm
straight, and your arm between the two metal poles. In addition, when you bring your
wrist down don't do it too fast or with too much force; this could cause damage to the

equipment.

While you are performing this task with one arm, you must also keep your other hand on
one of the Control buttons located at the bottom of your keyboard (they have a red sticker
attached to help you ﬁnd them). During your trial, if you do not have your non-dominant
hand pressing the most accessible Control button, your time will not be recorded and,

thus, you will not get credit for the work you do.

When you are done reading this page, please click NEXT button."

66

Screen 4.

In addition, during each trial, you must remain seated in your chair with both of your feet
ﬂat on the ground. You are NOT allowed to use your non-dominant arm to support your
dominant arm in any way during the trial. Remember, you must keep the button pressed
with your non-dominant hand while you are performing the task. There is a video camera

located to the side of the keyboard to make sure all of these rules are followed.

We do want to emphasize that this is a tiring task. The longer you hold the wrist weights
up, the more tired your arm will become. WE DO NOT want anyone to hold up their arm
so long that they experience undue pain or risk injury to their arm or shoulders. Rather,
we expect that as your arm becomes too tired or uncomfortable you will lower it so the

mechanism will press the spacebar and end the trial.

We do want to do as well at the persistence task as you can, but only as well as you can
do consistent with your own physical well being and comfort. In summary, do your best,
but do not attempt to do more than you can do--don't risk injury or undue pain. How well
people do at this task depends on a pair of factors--their physical strength and their
tolerance for discomfort. People seem to differ a lot on both, and particularly on the latter
factor. That is, some people ﬁnd even a short period of holding up the weight to make
them too uncomfortable to continue. Others can hold the weight for quite a while without

experiencing noticeable discomfort.

When you are done reading this page, please click NEXT button.

67

Screen 5.
You can earn money through your performance at the persistence task. Here's how. We
will sum up all the scores on all the trials today to come up with a Total Score. Then, at

the end of the study, we will randomly pick two persons taking part in the study.

Every person, no matter how well or poorly s/he does, has an equal chance of being
picked and winning. We will award the winning persons 5 cents for every point of their
Total Scores. Depending upon how well each of the winning persons performs, they

could earn up to $50.

As an additional incentive to perform your best, we are having a little competition
between students taking part in this experiment at the University of Michigan and those
performing it here at Michigan State. We will compute the average Total Score for each
school, and award the winning persons from the school with the strongest teams an
additional 15%. So even if you are not picked, your good performance at this task will

possibly beneﬁt another MSU student.

When you are done reading this page, please click NEXT button.

68

Screen 6.

OK, we're now ready to begin the ﬁrst trial.

Before and after each trial, you will be given a few questions. After responding the
questions, you will see a countdown (10, 9, 8,etc.). When the countdown gets near the
end (around 2 or 1), you should put your wrist above the string, arm straight and elbow
locked, and press the Control key with your non-dominant hand. The trial will begin at
the end of the countdown (the computer screen will say 'START'). You should continue
to keep your arm above the string as long as you can. When your wrist touches the string,

it will end the trial.

After the trial ends simply follow the instructions on your computer screen. If you have

any questions, please ask the experimenter now.

If you understand the instructions and are ready to respond the question and begin the

ﬁrst trial, please click 'NEXT' button. Be ready to start the ﬁrst trial at the end of the

countdown.

69

APPENDIX B

Self-efficacy manipulation screen

(Low self-efficacy)

r? "
1%

17.; ‘-.‘ :1.

 

70

(High self-efﬁcacy)

 

.

- " 2"; " w ."2 5M '- ' has; -.- Wren; .ws.
ﬁWEEJ Mam“: , “-‘ééé’z’élyz
." '5‘ >r ' i«:~-‘ rl "'r/l:‘
1%; ' r 1" (A. .151" ._

 

71

APPENDIX C

Discrepancy manipulation screen

(An example of the large discrepancy condition)

..4- .JM- , .
Our data show that your putter. Erin Anderson
did substantially better Iran you did.

 

72

REFERENCES

73

REFERENCES

Bandura, A. (1977). Self-efﬁcacy theory: Toward a unifying theory of behavioral change.
Psychological Review, 84, 191-215

Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory.
Englewood Cliffs, NJ: Prentice-Hall.

Bandura, A. (1988). Organizational applications of social cognitive theory. Australian
Journal of management, 13, 137-164.

Bandura, A., Adams, N. E., Hardy, A. B., & Howells, G. N. (1980). Test of the generality
of self-efﬁcacy theory. Cognitive Therapy and Research, 4, 39-66.

Bandura, A., & J ourden, F. J. (1991). Self-regulatory mechanisms governing the impact
of social comparison on complex decision making. Journal of Personality and
Social Psychology, 60(6), 941-951.

Bouffard-Bouchard, T. ( 1990). Inﬂuence of self-efﬁcacy on performance in a cognitive
task. The Journal of Social Psychology, 130(3), 353-363.

Brehm, J. W., & Self, E. (1989). The intensity of motivation. In M. R. Rozenweig & L.
W. Porter (Eds.), Annual review of psychology (pp. 109-131). Palo Alto: Annual
Reviews, Inc.

Hertel, G., Kerr, N. L., & Messé, L. A. (2000). Motivation gains in performance groups:
Paradigmatic and theoretical developments on the Kohler effect. Journal of
Personality and Social Psychology, 79, 580-601.

Hertel, G., Kerr, N. L., Schefﬂer, M., Geister, S., & Messé, L. A. (2000). Exploring the
Kohler motivation gain effect: Impression management and spontaneous goal
setting. Zeischrzft fur Sozialpsychologie, 31, 204-220.

Hinsz, V. B. (1995). Goal setting by groups performing an additive task: A comparison
with individual goal setting. Journal of Applied Social Psychology, 25, 965-990.

Karau, S. J ., & Williams, K. D. (2001). Understanding motivation in groups: The
collective effort model. In M. E. Turner (Ed.), Groups at work: Theory and
research. Applied social research (pp. 113-141). Mahwah, NJ: Erlbaum.

Messé, L. A., Kerr, N. L., & Seok, D. H. (in preparation). Examination of the Kohler

discrepancy effect in conjunctive and additive task demands. Unpublished
manuscript, Michigan State University.

74

Kerr, N. L., Messé, L. A., Park, E. S., & Sambolec, E. J. (under review). Identiﬁability,
performance feedback and the Kﬁhler effect. Group Processes and Intergroup
Relations.

Kohler, O. (1926). Kraftleistungen bei Einzel-und Gruppenabeit[Physical performance in
individual and group situations]. Industrielle Psychotechnik, 3, 274-282.

Kohler, O. (1927). Uber den Gruppenwirkungsgrad der menschlichen Korperarbeit und
die Bedingung optimaler Kollektivkraftreaktion[On group efﬁciency of physical
labor and the conditions of optimal collective performance]. Industrielle
Psychotechnik, 4, 209-226.

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

Lent, R. W., Brown, S. D., & Hackett, G. (1994). Toward a unifying social cognitive
theory of career and academic interest, choice, and performance. Journal of
Vocational Behavior Monograph, 45, 79-122.

Locke, E. A., & Latham, G. P. (1990). A theory of goal-setting and task performance.
Englewood Cliffs, NJ: Prentice Hall.

Lount, R. B., J r., Messé, L. A., & Kerr, N. L. (2000). Trying harder for different reasons:
Conjunctivity and sex differences as bases for motivation gains in performing
groups. Zeischriftﬁir Sozialpsychologie, 31, 221-230.

Manning, M. M., & Wright, T. L. (1983). Self-efﬁcacy expectancies, outcome
expectancies, and the persistence of pain control on childbirth. Journal of
Personality and Social Psychology, 45, 421-431.

Messé, L. A., Hertel, G., Kerr, N. L., Lount, R. B., Jr., & Park, E. S. (2002). Knowledge
of partner’s ability as a moderator of group motivation gains: An exploration of

the Kohler discrepancy effect. Journal of Personality and Social Psychology,
82(6), 935-946.

Miyake, M., & Matsuda, F. (2002). Effects of generalized self-efﬁcacy and negative
social comparison feedback on speciﬁc self-efﬁcacy and performance.
Psychological Reports, 90, 301-308.

Obrist, P. A. (1976). The cardiovascular-behavioral interaction as it appears today.
Psychophysiology, 13, 95-107.

Obrist, P. A. (1981). Cardiovascular Psychophysiology: A perspective. New York:
Plenum.

Sanna, L. J. (1992). Self-efﬁcacy theory: Implications for social facilitation and social
loaﬁng. Journal of Personality and Social Psychology, 62(5), 774-786.

Schunk, D. H. (1995). Self-efﬁcacy and education and instruction. In J. Maddux (Ed),
Self-eﬂicacy, adaptation, and adjustment: Theory, research, and application (pp.
281-303). New York: Plenum Press.

Stajkovic, A. D., & Luthans, F. (1998). Self-efﬁcacy and work-related performance: A
meta-analysis. Psychological Bulletin, 124, 240-261.

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

Stone, D. N. (1994). Overconﬁdence in initial self-efficacy judgments: Effects on
decision processes and performance. Organizational Behavior and Human
Decision Processes, 59, 452-474.

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: Consensual action
by small groups (V 01. 2, pp. 37-65). Mahwah, NJ: L. Erlbaum.

Stroebe, W., Diehl, M., & Abakoumkin, G., & Amscheid, R. (1990, Octobrer). The
Ko'hler eﬂect: Motivation gains in group performance. Paper presented at the
annual meeting of the Society of Experimental Social Psychology, Buffalo, NY.

Stroebe, W., & Frey, B. (1982). Self-interest and collective action: The economics and
psychology of public goods. British Journal of Social Psychology, 21, 121-137.

Vancouver, J. R, Thompson, C. M., Tischner, E. C., & Putka, D. J. (2002). Tow studies
examining the negative effect of self-efﬁcacy on performance. Journal of Applied
Psychology, 87(3), 506-516.

Vroom, V. H. (1964). Work and motivation. New York: Wiley.

Williams, K. D., & Karau, S. J. (1991). Social loaﬁng and social compensation: The
effects of expectations of co-worker performance. Journal of Personality and
Social Psychology, 61, 570-581.

Wood, R. E., & Bandura, A. (1989). Impact of conceptions of ability on self-regulatory
mechanisms and complex decision making. Journal of Personality and Social
Psychology, 56, 407-415.

Wright, R. A., & Kirby, L. D. (2001). Effort determination of cardiovascular response:
An integrative analysis with applications in social psychology. In M. Zanna (Ed.),

Advances in experimental social psychology (Vol. 33, pp. 255-307). San Diego,
CA: Academic.

76

   

111111111111111