Content uploaded by Remus Ilies
Author content
All content in this area was uploaded by Remus Ilies on Jan 16, 2020
Content may be subject to copyright.
The role of self-efficacy, goal, and affect in dynamic motivational self-regulation
q
Myeong-gu Seo
a,*
, Remus Ilies
b
a
Department of Management and Organization, Robert H. Smith School of Business, University of Maryland, College Park, MD 20742-1815, USA
b
The Eli Broad Graduate School of Management, Michigan State University N475 North Business Complex, East Lansing, MI 48824-1112, USA
article info
Article history:
Received 11 October 2006
Accepted 9 March 2009
Available online 29 April 2009
Accepted by John Schaubroeck
Keywords:
Self-efficacy
Goal
Affect
Self-regulation
Motivation
abstract
In this paper, we examined the within-person relationship between self-efficacy and performance in an
Internet-based stock investment simulation in which participants engaged in a series of stock trading
activities trying to achieve performance goals in response to dynamic task environments (performance
feedback and stock market movements). Contrary to the results of several previous studies, we found that
self-efficacy was positively related to effort and performance, and goal level partially mediated the effi-
cacy–performance relationship. We also found that participants’ affective reactions to performance feed-
back, measured as positive affect and negative affect, uniquely contributed to their motivation and
performance either directly or by indirectly influencing their self-efficacy.
Ó2009 Elsevier Inc. All rights reserved.
Almost three decades of extensive research from multiple theo-
retical perspectives has investigated the processes and conse-
quences of motivational self-regulation, in particular, how
individuals set or revise a set of goals and strive for the achieve-
ment of those goals and what is the role of their self-efficacy beliefs
in this process (Bandura, 1991; Bandura & Locke, 2003; Locke & La-
tham, 1990a; Vancouver, 2000). Recently, scholars have begun to
question whether self-efficacy, one’s belief in his or her capacity
to perform (Bandura, 1986), plays the same role in motivational
self-regulation at different levels of analyses (e.g., Chen & Bliese,
2002; Chen et al., 2002; Vancouver, Thompson, & Williams,
2001; Yeo & Neal, 2006). In particular, there have been interesting
scholarly debates between control theorists (e.g., Vancouver, 2005)
and social-cognitive theorists (e.g., Bandura & Locke, 2003) regard-
ing the relationship between self-efficacy and performance at the
within-person level of analysis. The debate was sparked by a series
of experimental studies conducted by Vancouver and his col-
leagues (Vancouver & Kendall, 2006; Vancouver, Thompson, Tisch-
ner, & Putka, 2002; Vancouver et al., 2001), which showed that
self-efficacy is negatively related to performance within individu-
als, and thus challenged the long-standing perspective among
self-regulation theorists that self-efficacy is generally functional
to performance as well as the body of research supporting that
perspective.
The conflicting views on self-efficacy, motivation and perfor-
mance, and the associated evidence in support of both perspec-
tives, raise the question of whether the two perspectives can be
integrated and the differences explained either by substantive dif-
ferences in the underlying psychological processes linking these
constructs or by methodological differences and artifacts. The pri-
mary objective of this study is aimed at offering an explanation for
these differences.
In addition, although self-regulation theorists have discussed
the role of affect-related factors (e.g., self-satisfaction) in the pro-
cesses of self-regulation (e.g., Bandura, 1991; Bandura & Cervone,
1983; Carver & Scheier, 1998; Locke & Latham, 1990a), cognitive
explanations, such as self-efficacy and goal, have been predomi-
nant in understanding motivational self-regulation processes (cf.,
Seo, Barrett, & Bartunek, 2004). Recently, scholars began to directly
investigate the role of basic affective experience in the process of
dynamic motivational self-regulation (e.g., Bagozzi & Pieters,
1998; Cron, Slocum, VandeWalle, & Fu, 2005; Ilies & Judge,
2005). For example, Ilies and Judge (2005) provided evidence that
motivational states (goal level) vary within individuals as a func-
tion of performance feedback, and more importantly that positive
and negative affect partially mediate the within-person relation-
ship between performance feedback and goal level across time.
Yet, most studies investigated the role of affect in the processes
of motivational self-regulation without considering cognitive pro-
cesses, such as those related to self-efficacy, that are simulta-
neously operating in motivational self-regulation. As a result, we
know little about how affective and cognitive processes simulta-
neously explain motivational self-regulation. To address this gap
0749-5978/$ - see front matter Ó2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.obhdp.2009.03.001
q
This research was supported by NSF DRMS#0215509 to Lisa Feldman Barrett
and a Boston College Dissertation Research Grant to Myeong-Gu Seo.
*Corresponding author. Fax: +1 301 314 8787.
E-mail addresses: mseo@rhsmith.umd.edu (M. Seo), ilies@msu.edu (R. Ilies).
Organizational Behavior and Human Decision Processes 109 (2009) 120–133
Contents lists available at ScienceDirect
Organizational Behavior and Human Decision Processes
journal homepage: www.elsevier.com/locate/obhdp
in the literature, our secondary objective is to further explore the
role of affect in the process of motivational self-regulation that is
driven by two essential cognitive properties: self-efficacy and
goals. In doing so, we examine both the indirect mechanism
through which individuals’ positive and negative affective states
influence motivation and performance by affecting self-efficacy
and goal choice, and the direct mechanism that goes beyond the ef-
fects of self-efficacy and goal processes.
To accomplish our objectives, we developed and ran a stock
investment simulation in which stock investors recruited from
investment clubs engaged in trading activities each day for 20
business days in face of the dynamic and uncertain stock market
movements. Building on Vancouver, More, and Yoder (2008),we
explain how and why the efficacy–performance relationship
should be positive in such a task environment, and thus attempt
to reconcile the conflicting views on the efficacy–performance
relationship. Below we review theory and research relevant to
our investigation, and we then formulate hypotheses for the rela-
tionships between the constructs described above.
Self-efficacy, performance, and task environment
Sociocognitive theory suggests that self-efficacy beliefs operate
in concert with goal systems to enhance motivation and perfor-
mance by increasing effort or persistence (Bandura, 1997, 2001).
For example, Locke and Latham (1990b) suggested a high-perfor-
mance cycle in which self-efficacy leads to the adoption of more
difficult goals, and both self-efficacy and goals positively contrib-
ute to motivation and performance. In addition, an overwhelm-
ingly large accumulation of evidence from diverse empirical
settings and methodological strategies has supported the theory
that self-efficacy is positively related to motivation and perfor-
mance (see, Bandura & Locke, 2003, for a review). For example,
there are a number of large-scale meta-analyses that examined
this relationship, and the results were predominantly positive
(e.g., Sadri & Robertson, 1993; Stajkovic & Luthans, 1998). As a re-
sult, it has been a generally accepted view among scholars that
self-efficacy is positively related to motivation and performance
(Vancouver et al., 2001).
Recently, scholars began to consider the relationship between
self-efficacy and performance at two different levels of analysis
(e.g., Chen & Bliese, 2002; Chen et al., 2002), and several studies
examining this relationship at the within-individual level have
provided evidence that contradicts the dominant view that the
relationship is positive (e.g., Vancouver & Kendall, 2006; Vancou-
ver et al., 2001, 2002, 2008; Yeo & Neal, 2006). In particular, based
on control theory (Powers, 1973), Vancouver and his colleagues
(2001, 2002, 2006) predicted that self-efficacy may have complex,
mutually opposing effects on performance at the within-person
level of analysis. On the one hand, consistent with socicognitive
theory, high self-efficacy can motivate individuals to set or adopt
higher levels of goals over time, and as a result, may positively
contribute to performance (e.g., Locke & Latham, 1990a). On the
other hand, however, they argued that self-efficacy would be neg-
atively related to performance. They reasoned that if goal level is
held constant, high levels of self-efficacy may create optimism or
overconfidence regarding the discrepancies between current and
desired states, which in turn, lead to lowering the levels of
resources allocated to a given task and thus to lower levels of
performance (Vancouver & Kendall, 2006).
Consistent with these predictions, Vancouver et al. (2001)
demonstrated in two studies adopting a computerized analytic
task (the Mastermind game) a weak negative relationship between
self-efficacy and performance at the within-person level of analysis
in two different conditions. They also found that personal goals
were positively influenced by self-efficacy but negatively related
to subsequent performance, contrary to the predictions of socio-
cognitive theory. In two additional studies using the same task,
Vancouver et al. (2002) experimentally manipulated self-efficacy
and found the same negative within-person relationship between
self-efficacy and performance. They further showed that self-effi-
cacy increased overconfidence and the chances of committing log-
ical errors, which might explain the negative relationship.
Bandura and Locke (2003) argued that these results are likely to
be methodological artifacts based on an overly simple performance
task (Mastermind), rather than general findings applicable to other
dynamic, real-life, and/or learning contexts. However, Yeo and
Neal (2006) examined the within-person relationship between
self-efficacy and performance in an air traffic control task that al-
lowed growth in both performance and self-efficacy (learning) over
time and found that the relationship was negative. Moreover, Van-
couver and Kendall (2006) investigated the relationship in a real-
life, learning context where a group of undergraduate students en-
rolled in an introductory course repeatedly reported their self-effi-
cacy, goal, planned study time, and actual study time
corresponding to each of the five exams given to them during
the course. They found not only that self-efficacy was negatively
related to exam performance within individuals, but also that the
negative relationship occurred via its negative effect on resource
allocation (planned and actual study time).
Vancouver and colleagues have proposed a conceptual argu-
ment that explains why self-efficacy predicts motivation (or per-
formance) positively in some situations and negatively in others;
and this explanation is best summarized by Vancouver et al.
(2008). These authors argue that distinguishing goal-planning pro-
cesses from goal choice processes plays a crucial role in the deter-
mining the direction of the within-individual effect of self-efficacy
on performance. That is, when participants are planning for ac-
cepted goals, their self-efficacy beliefs relate negatively to perfor-
mance because high self-efficacy involves a higher expectancy of
reaching the goal and thus leads to less resources (i.e., effort and
persistence) allocated to the task, thus lower motivation. Under
the goal choice process, when participants actively select the goals
they are motivated to accomplish, higher self-efficacy leads to
higher goals which, in turn lead to higher motivation – therefore
the relationship between self-efficacy and motivation will be posi-
tive. This argument can be extended to explain the within-individ-
ual relationship between self-efficacy and performance, of course
for most situations (tasks) where motivation (i.e., effort and persis-
tence) does predict performance.
Indeed, the goal-planning process seems to have been at work in
the studies that supported a negative within-individual relationship
between self-efficacy and motivation or performance. For example,
although the relationship was examined in some studies (e.g., Van-
couver et al., 2002; Yeo & Neal, 2006) without explicitly considering
and measuring intra-individual variations in personal goals, the
other studies that did allow goal adjustment (i.e., Vancouver & Ken-
dall, 2006; Vancouver et al., 2001) showed that intra-individual
goals were unrelated or negatively related to performance. This re-
sult suggests that goal adjustments in these studies may not have
represented meaningful changes in one’s personal goal level, but in-
stead may have merely reflected his or her past performance (e.g.,
Vancouver et al., 2001). Vancouver and Kendall (2006), for example,
observed little within-individual variation in personal goals (what
grade were respondents aiming for on the upcoming test), which
suggests that participants strived for an overall goal (performance
in the course) and the little adjustment they made did not reflect
participants’ motivational state with respect to the subsequent
performance episode. On the other hand, Vancouver et al. (2001),
in their second study, did find support for a positive relationship
between self-efficacy and performance only in the difficult goal
M.-g. Seo, R. Ilies / Organizational Behavior and Human Decision Processes 109 (2009) 120–133 121
condition, even when controlling for previous performance (which
is excessively conservative due to statistical over-control; Bandura
& Locke, 2003). This result suggests that in certain situations where
goals are inherently meaningful (goal choice process), the negative
within-person relationship between self-efficacy and performance
can be reversed.
According to Bandura and Locke (2003), one possible reason for
the likely dominance of goal-planning process (or the lack of goal
choice process) in the studies supporting a negative efficacy–per-
formance relationship is the nature of the task settings itself. In
particular, the task settings in these studies (e.g., the Mastermind
game and college exams) were relatively simple, static, and dis-
joined in nature; they had relatively simple and invariant rules
and skill sets that participants could quickly learn and apply to in-
crease their performance; and the same task was repeated over
time with little substantive relationship between the task trials
(each task trial, its goal and performance in particular, was defined
in isolation from the previous or the next task trial). In such partic-
ular settings, participants are likely to initially set goals (goal
choice process), but quickly switch to the mode of goal-planning
(Vancouver et al., 2008) in which their goals become a relatively
fixed overall goal (and show little variation) and the small goal
adjustments within individuals over time, although reported as a
form of personal goals, likely reflect previous performance instead
of a real or significant change in goal level (Vancouver et al., 2001).
Moreover, participants who perform well develop high self-effi-
cacy and upwardly adjust their goals somewhat, while at the same
time developing a sense of overconfidence and allocate less re-
sources for the subsequent performance episode, leading to lower
performance (Vancouver & Kendall, 2006; Vancouver et al., 2008).
If the empirical findings that support the negative efficacy–per-
formance relationship are bounded by certain study characteristics
that induce a goal-planning orientation in which case self-efficacy
beliefs negatively predict motivation and performance – then
when would a positive relationship be observed? We believe that
in highly dynamic task environments, where both tasks and their
performance fluctuations reflect continuous processes in response
to changing task conditions towards ultimate (cumulative) perfor-
mance, each task carries different implications of prior perfor-
mance. That is, employees or study participants are encouraged
to actively set and continually adjust meaningful goals in an effort
to ensure high ultimate (cumulative) performance, which may
suppress or overwhelm the goal-planning processes (allocating
motivational resource in order to accomplish a fixed goal in invari-
ant task conditions). Such dynamic task environments may not
only produce meaningful variance in goal levels within individuals
over time, but also do not allow a sufficient interval for individuals
to experience the discontinuity in the motivation processes to
which Vancouver et al. (2008) referred (i.e., shifting resource allo-
cation as a result of overconfidence or tedium). That is, they are
likely too preoccupied with adapting to changing task conditions
to become complacent and thus reduce their motivational re-
sources for a given task.
Given the positive relationship between self-efficacy and goals,
therefore, self-efficacy should positively predict performance with-
in individuals in a dynamic task environment. In this situation,
higher self-efficacy may encourage individuals to continually set
and reset higher levels of meaningful goals (goal choice processes)
without developing overconfidence and/or lowering motivational
resources (goal-planning processes), which in turn, may lead to
higher task performance. On this point, Bandura and Locke
(2003), p. 96 note: ‘‘A second requirement for informative research
on the functional properties of self-efficacy beliefs is the use of a
dynamic rather than a static environment” and further noted the
disjointed nature of the experimental task used by Vancouver
et al. (2001, 2002).
However, the motivational effects of self-efficacy are not likely to
be realized exclusively through goals. For example, self-efficacy can
enhance task performance through helping effective strategy devel-
opment, particularly when the task involves complex decision-mak-
ing (Bandura & Wood, 1989; Wood & Bandura, 1989; Bandura,
Wood, & Bailey, 1990). In complex decision-making environments,
task performance depends on discovery and use of effective task
strategies that typically involve two-phase processes: (1) construct-
ing tentative rules (hypotheses) for how various factors may affect
performance outcomes and (2) testing those rules by varying these
different factors one at a time and assessing how they affect perfor-
mance outcomes (Bandura et al., 1990). It requires a strong sense of
self-efficacy to deploy one’s cognitive resources optimally and to re-
main focused on strategy development in face of repeated difficul-
ties and failures (Wood & Bandura, 1989). For example, individuals
with low self-efficacy are likely to develop various disruptive think-
ing (e.g., self-doubt or blaming situations) in face of failures or neg-
ative feedback, which impairs effective use of feedback information
and/or foster faulty strategies (e.g., choices by random or irrelevant
factors). In sum, in dynamic and complex task environments, self-
efficacy should positively predicted performance directly, in addi-
tion to its effect realized through goals. As we explain below, we be-
lieve we have created such a paradigm for our study, thus we expect
self-efficacy to positively affect performance both directly and indi-
rectly (through goals).
To test our predictions, as noted, we developed and ran a real-
istic stock investment simulation and recruited actual stock inves-
tors who were asked to participate in daily stock trading activities
for twenty business days. We believe that the dynamic nature of
the task, as well as the sequence and methodology for measuring
self-efficacy and goals (to be explained shortly) facilitated goal
choice processes by which investors chose and reported meaning-
ful daily goals that motivated them to perform well. This task and
methodology, because of the complex nature of the task environ-
ment (We explain this in detail in the ‘‘Method” section.), are also
well-suited to capture motivating effects of self-efficacy (i.e., posi-
tive effects on performance) that are realized through other mech-
anisms (not through goals). Therefore, within this dynamic
paradigm, we hypothesize a positive within-person relationship
between self-efficacy and performance:
H1-1: self-efficacy will be positively related to performance
within individuals.
H1-2: the relationship between self-efficacy and performance
will be partially mediated by goal choice (goal level).
The role of affect in the efficacy–performance relationship
Building upon the hypothesized relationship between self-effi-
cacy and performance, we further explore how affective processes
may influence the relationship. From a broader framework of moti-
vational self-regulation, the efficacy–performance relationship can
be delineated as a model depicted as the solid arrows shown in
Fig. 1. In sum, in a dynamic and realistic multi-trial setting, past
performance is an important determinant of self-efficacy (arrow
‘‘a”), and self-efficacy influences effort (motivation) either directly
(arrow ‘‘b”) or indirectly via affecting goal level (arrows ‘‘c” and
‘‘d”). Effort, then, leads to performance (arrow ‘‘f”), which in turn,
becomes past performance (performance feedback) for the next
trial (arrow ‘‘g”). In this model, self-efficacy and goal level are
the two core mechanisms explaining the cognitive processes
underlying motivational self-regulation.
We expand this model to further identify three affective mech-
anisms through which motivation and performance is self-regu-
122 M. Seo, R. Ilies / Organizational Behavior and Human Decision Processes 109 (2009) 120–133
lated within individuals over time. As depicted by the dotted ar-
rows shown in Fig. 1, we predict that positive and negative affect,
triggered by past performance (arrow ‘‘h”), influence motivation
and performance either directly (arrow ‘‘i”) or indirectly via affect-
ing goal level (arrow ‘‘j”) and/or self-efficacy (arrow ‘‘k”). Below we
explain each mechanism and suggest testable hypotheses.
Direct mechanism
A number of scholars, using the frameworks of both socio-cog-
nitive theory (e.g., Bandura, 1991; Bandura & Cervone, 1983; Locke
& Latham, 1990a) and control theory (e.g., Carver & Scheier, 1990,
1998), have suggested that positive and negative affective reac-
tions have important motivational implications during the process
of goal regulation. For example, Bandura and his colleagues (Ban-
dura, 1991, 1997; Bandura & Cervone, 1983, 1986) argued that
affective self-reactions, both projected self-satisfaction with future
performance and self-dissatisfaction with past performance, serve
as important motivators, positively contributing to both motiva-
tion and performance. Consistently, research on affect and emotion
suggests that experience of positive and negative affect may acti-
vate the two motivational systems of promotion and inhibition
(Frijda, 1987; Gray, 1990) or behavioral predispositions to ap-
proach or avoid (e.g., Cacioppo, Gardner, & Berntson, 1999; Wat-
son, Wiese, Vaidya, & Tellegen, 1999), which can generate
behavioral responses outside of conscious awareness (e.g., Winki-
elman, Zajonc, & Schwarz, 1997; Izard, 1993) or independently
from cognitive evaluations (e.g., Loewenstein, Weber, Hsee, &
Welch, 2001; Brehm, 1999). Therefore, positive and negative affect,
by activating promotion (approach) motivation for goal attainment
and prevention (avoidance) motivation for goal failure, will be pos-
itively and directly related to motivation (effort) and performance
(arrow ‘‘i” in Fig. 1).
Unlike positive affect, however, the motivational effect of nega-
tive affect on motivation and performance can be complex, because
a negative effect is also possible. That is, negative affect and its
associated avoidance motivation can be directed towards the task
itself, and in this case, it may directly lead to task disengagement
or withdrawal of effort. Particularly when the task itself is chal-
lenging and/or participants’ level of perceived control over perfor-
mance is relatively low, there may be greater chances that the
negative effect of negative affect is more dominant than its positive
effect. Therefore, since we focus on a complex, challenging, and dy-
namic task in this study, we hypothesize a negative relationship
between negative affect and performance (motivation) and a posi-
tive relationship between positive affect and performance:
H2-1: positive affect will be positively related to performance.
H2-2: negative affect will be negatively related to performance.
Indirect mechanism via goal level
The hypothesized relationship between affect and performance
can be partly mediated by goal choice. That is, positive and nega-
tive affect may first influence goal choice (arrow ‘‘j” in Fig. 1),
which in turn, will affect effort and performance (arrows ‘‘d” and
‘‘f”). We argue that the promotion and prevention motivation acti-
vated by positive and negative affect may not only influence the le-
vel of effort devoted to achieve a given goal, but also alter goal level
itself, the level of performance that one chooses to strive. In partic-
ular, following the same theoretical logic that we used to predict
the affect–effort relationship, we hypothesize that positive affect
will be positively related to upward goal adjustment, whereas neg-
ative affect will be negatively related to downward goal
adjustment.
Indeed, Ilies and Judge (2005) found consistent support that po-
sitive affect was positively related to goal level in all six samples
studied. However, they found weak support regarding the negative
effect of negative affect on goal level; it was significant in only one
of these samples. In contrast, Cron and colleagues (2005) recently
found that negative affect was significantly and substantially re-
lated to goal level in their longitudinal field study. We speculate
that these inconsistent results might come from different research
designs; the effect of negative affect on goal regulation might be
stronger in a field setting where threat-related stimuli are more
salient than in an experiment setting and thus negative affect
and its related inhibition motivation could operate more strongly.
H3-1: positive affect will be positively related to goal level.
H3-2: negative affect will be negatively related to goal level.
Indirect mechanism via self-efficacy
Affective states that individuals experience at a given moment
and the cognitive judgments that they make while experiencing
such affective states are intimately related to each other. This fact
has been supported by several distinctive streams of research
including basic neuropsychological research on the interdepen-
dence between affect and cognition (see, for example, Damasio,
1994, 1998; LeDoux, 1996), basic psychological research on affec-
tive influences on judgment (e.g., Forgas, 1995), and experimental
Performance
a
Past
Performance
Self-efficacy Effort
Goal Level
Affect
b
c d
f
g
h
i
k
j
Fig. 1. Extended model of affective and cognitive processes explaining goal regulation.
M.-g. Seo, R. Ilies / Organizational Behavior and Human Decision Processes 109 (2009) 120–133 123
research on the relationships between affect and expectancy moti-
vation (Erez & Isen, 2002). We explore one possibility in this paper,
which is strongly supported in the literature both conceptually and
empirically. This is, positive and negative affect are likely to influ-
ence self-efficacy (arrow ‘‘k” in Fig. 1), which in turn, will affect ef-
fort and performance (arrow ‘‘b” and ‘‘f”).
Affective states, or moods, are thought to influence judgments
and evaluations through mood congruency effects (Bower, 1981;
Forgas, 1995; Forgas & Bower, 1987). When people experience
heightened positive affect, for example, positive material is acti-
vated in memory which makes positive cognitions more likely
(e.g., Isen, Shalker, Clark, & Karp, 1978). It follows that positive af-
fect should lead to more positive beliefs regarding one’s capacity to
perform, and negative affect should be associated with less positive
beliefs (e.g., Kavanagh & Bower, 1985; also see George & Brief,
1996). Indeed, Erez and Isen (2002), in their laboratory studies,
found evidence that experimentally induced positive affect influ-
enced one’s performance expectations. Thus, affect should influ-
ence self-efficacy ratings via a congruency effect; higher positive
affect should be associated with higher self-efficacy and higher
negative affect should be associated with lower self-efficacy
ratings.
H4-1: positive affect will be positively related to self-efficacy.
H4-2: negative affect will be negatively related to self-efficacy.
Method
Task
To test the hypotheses in a realistic, dynamic, and complex task
environment, the first author developed and ran an internet-based
stock investment simulation for 20 consecutive business days (4
weeks). The participants were initially given hypothetical cash of
$10,000. During the simulation, they were allowed to invest the
entire or part of this hypothetical cash on any of the 12 anonymous
stocks selected from the national stock market for this simulation.
Once a day during the simulation period, participants logged on to
the stock investment simulation website, viewed current market
and stock information that were updated daily from the national
stock market, checked their current investment performance, and
finally made their investment decisions about which and how
many shares of stocks to buy or sell for the day. Just before making
their investment decisions for the day, they reported their current
affect as well as their self-efficacy beliefs about their investment
performance, and set their performance goal for the simulation.
To make the performance in this simulation more personally
relevant and inherently motivating to the participants, they were
monetarily rewarded for their participation at the end of the sim-
ulation. Rewards were distributed as a function of performance rel-
ative to the local stock index (%), ranging from $100 to $1000, and
the detailed monetary remuneration plan was clearly and formally
informed to participants before the simulation (also presented in
Appendix A).
This simulation involves two critical aspects, which may distin-
guish this study from the previous studies supporting a negative
efficacy–performance relationship. First, the nature of the task in-
volved in this simulation is highly dynamic in a sense that the re-
sults of the task in a given day – both the investment decision
(stock portfolio) and its performance – are directly linked to the
task condition in the next day. Instead of constructing an entirely
new stock portfolio each day, participants see and work on the
stock portfolio that they have developed until the last day (once
logged in, the website automatically pulls out this information
and presents it to each participant for each round). Their current
investment performance is also directly reflected in the amount
of resource (capital) available for investment in the next day. Thus,
participants are likely to perceive their daily tasks as a series of
interrelated and continuous processes instead of a mere repetition
of independent and disjoined tasks. Moreover, the information crit-
ical for developing effective investment strategies also evolves over
time as more information regarding the past trends of each stock
as well as the stock market accumulates after each round. In partic-
ular, participants’ own choice at each round and its performance
outcomes observed in the next (few) rounds become the essential
feedback information for effective strategy development (Wood &
Bandura, 1989). This interactive process of strategy acquisition also
adds an important dynamic element to the simulation. As noted,
this continuous and cumulative nature of the task and task perfor-
mance is different from the disjointed nature of the task trials used
by Vancouver et al. (2001, 2002) and, as we explained in the intro-
duction (see also Bandura & Locke, 2003), is a necessary condition
for the manifestation of positive motivational effects of self-
efficacy.
Second, the task used in this study is a complex and difficult
task. Participants in this study had to decide the number of shares
to be invested in each of the twelve stocks in constructing their
own stock portfolio for each day, which had almost an infinite
number of possible combinations at a given time. This is in contrast
to the tasks used in the previous studies which required relatively
simple rules and skill sets. In such simple tasks, outcomes become
highly predictable and controllable and therefore participants’ le-
vel of self-efficacy can inversely predict the motivational resources
allocated for goal accomplishment, as suggested by Vancouver
et al. (2008). In contrast, in the stock trading task of this study,
there are no simple rules or skill sets that can be applied to pre-
cisely predict the price changes in the twelve stocks, which are
completely determined by the national stock market.
Yet, task performance is not so uncertain or uncontrollable to be
viewed as a task with random performance outcomes. Instead, as
recommended by financial theorists (e.g., Bodie, Kane, & Marcus,
2001), participants can cope with this level of uncertainty and thus
improve their investment performance over time in several ways.
For example, they can actively manage the overall level of risk in
their stock portfolio (e.g., taking neither too high nor too low in
risk) and by appropriately reacting to the short-term trends in
the stock market and individual stocks (e.g., neither over-reacting
nor under-reacting to the stock market movements in determining
the amount of trading). In addition, participants could also keep
their current level of performance (e.g., when they were perform-
ing well) unchanged in the next rounds of the simulation (and its
corresponding cash reward) by diversifying their current asset
equally into the 12 stocks (in this case, the value of a given stock
portfolio changes the exactly same rate as the local market index
changes). Consistently, our sample based on 1870 observations
showed significant (p< .001) inverted ‘‘U” relationships between
performance and the overall level of risk in one’s stock portfolio
(computed as the averaged beta coefficients – a standard volatility
indicator – of the selected stocks) and between performance and
the amount of trading in response to stock-price change (these ex-
plained 6.2% and 3.5% of the variance in performance, respectively).
The use of diversification (holding) strategy was also positively and
significantly (p< .001) related to performance (explaining 4.8% of
the variance in performance).
Participants
We recruited 118 private stock investors for an investment sim-
ulation from six investment clubs located in the northeast United
States. Their ages ranged from 18 to 74 (Mean = 24.7; SD = 13.2),
and as is typical in most investment clubs, the majority of the par-
ticipants (80%) were male. Their average investment experience
124 M. Seo, R. Ilies / Organizational Behavior and Human Decision Processes 109 (2009) 120–133
was 4.3 years (SD = 7.4). Among the 118 investors recruited for this
study, 108 participants completed the stock investment simulation
task (91.5%). They generated 2059 case-level data (each case in-
cluded all measures generated by one participant going through
one investment session per day).We dropped seven participants
due to non-compliance with instructions (the data from these par-
ticipants showed a strong pattern of random response in affect
measures), and eliminated an additional 63 cases (3%) due to re-
ported interruptions during the sessions (57 cases) or data transfer
errors (6 cases). As a result, we used 1870 cases of data completed
by 101 participants for data analysis.
Measurement
Performance
Participants’ daily performance was assessed by the partici-
pant’s overall and cumulated investment return determined by
the next day stock-prices based on the national stock market
(t+ 1) after she or he made his or her investment decision for a gi-
ven day (t); that is, it equaled the percentage of the initial capital
($10,000 of hypothetical cash) that was earned or lost up to the
next day, which was further adjusted by the next day local market
return (index) across all participants in order to neutralize the per-
formance factors attributable to the general fluctuations of the lo-
cal stock market. However, daily performance alone may not
meaningfully reflect the performance of one’s investment decision
in a given day for two reasons. First, due to a high degree of fluctu-
ations in individual stock’s daily price, the performance of a given
stock is rarely assessed on a daily basis, but instead, typically as-
sessed based on an averaged price for a certain period of time
(e.g., Bodie et al., 2001). Second, the participants in this simulation
seemed to make their investment decisions not only for the next
day performance, but also for the overall performance for at least
the next couple of rounds. For example, in about one third (33%)
out of a total 1870 cases, participants did not make any change
in their existing stock portfolio for the next round, indicating that
they made one decision for at least next two or more rounds. For
about additional one third of the total cases (31%), participants
made change of only less than 10% of their previous investment
decision (stock portfolio). On average, participants maintained
about 86% of the previous stock portfolio in the next round. There-
fore, we measured one’s performance by constructing a forward
looking 3-day moving average of participants’ daily performance,
an approach often used in analyzing daily financial performance
in the finance and economics literature (e.g., Caginal & Laurent,
1998; Horton, in press). We measured past performance by calcu-
lating a backward looking three-day moving average of partici-
pants’ daily performance.
1
We adopted the differential
(backward–forward) procedure so that (1) the two measures do
not overlap and (2) our independent variables predict a performance
measure that only included current or future scores.
Self-efficacy
Self-efficacy beliefs were measured by four questionnaire items.
Following the procedure described by Klein’s (1991) and also
adopted by Vancouver et al. (2001, 2002, 2008), we asked partici-
pants to indicate on a five-point Likert scale (1 – no chance at all
(0%), 2 – a slight chance (25%), 3 – a 50/50 chance (50%), 4 – a good
chance (75%), 5 – completely certain (100%)) the subjective proba-
bility of obtaining four specific performance outcomes: (1) beat the
market by more than 10%, (2) beat the market by more than 5%, (3)
beat the market (above 0%) and (4) go below the market return
(below 0%/scores were reversed). The items were averaged to pro-
duce a single index with a higher value indicating higher task-spe-
cific self-efficacy. The probabilities represent performance
outcome expectancies (Klein, 1991), and by averaging multiple
expectancy scores for a range of performance outcomes is equiva-
lent to task-specific self-efficacy (Klein, 1991). The reliability (al-
pha) of this index was .74 on average across the 20 rounds of the
simulation.
Goal level
We asked participants to indicate their overall target perfor-
mance outcome for the simulation – a specific investment return
compared to the local market return in percentage. There were
12 performance goal levels in a pull-down box for choice in a
descending order; the highest level is ‘‘more than 30% above the
market return” and the lowest level is ‘‘no less than 5% below
the market return.” We coded goal level scores by taking the
investment return (%) specified by the respondents (out of 12
choices). For example, we coded 30% if a participant selected ‘‘I will
obtain an investment return that is more than 30% above the mar-
ket return” as their performance goal for a given day. Because the
market return was the same for all participants, the goals set daily
by participants concerned their own performance and did not in-
volve comparisons with others’ performance (i.e., they were not
relative goals; see Ilies & Judge, 2005).
Effort
Following Vancouver and Kendall (2006), we measured effort by
two indicators. First, we measured ‘intended time,’ the amount of
time that a participant intended to spend for his or her stock trading
task for the day by having participants indicate in a pull-down box
the maximum amount of time (minutes) up to which they were will-
ing to devote to their investment decision-making. In the pull-down
box, participants saw eight response choices in a descending order
(‘‘more than 1 h” to ‘‘less than 10 min”). Second, we measured ‘actual
time,’ directly through programming the Internet web page in such a
way that both the start and the end time of investment decision-
making could be automatically and precisely (in seconds) measured
whenever participants moved into and out of the web page desig-
nated for their investment decisions. The difference (minutes) be-
tween the start time and the end time in the decision-making web
page was calculated and used to indicate the actual amount of time
devoted for making an investment decision.
Positive and negative affect
Positive affect was measured with five affect items that repre-
sent the positive affect dimension in the affect circumplex (Barrett
& Russell, 1998): ‘‘excited,” ‘‘joyful,” ‘‘enthusiastic,” ‘‘proud,” and
‘‘interested.” Participants were asked to report their current feel-
ings (‘‘how are you feeling right now?”) on a five-point Likert scale
(0 – not at all, 1 – a little, 2 – moderately, 3 – quite a bit, 4 – extre-
mely so) for each of the five affect items. The items were averaged
to produce a single index of positive affect and its reliability (alpha)
was .87 on average across the 20 rounds of the simulation (ranging
between .92 and .79). Similarly, negative affect was measured by
the mean score of the following adjectives: ‘‘irritated,” ‘‘afraid,”
‘‘angry,” ‘‘nervous,” and ‘‘frustrated.” Its reliability (alpha) was .80
on average (ranging between .86 and .68).
Control variables
In this study, we controlled three variables that might system-
atically influence the results. First, there could be a trend in the
data as the participants learned more about the stock investment
simulation over time, which could systematically influence the re-
sults. Thus, we created, and controlled for, a time variable (1–20)
representing the amount of experience that participants have
1
The results of this study are not sensitive to the number of days used to
calculating the moving average or the ways to treat the tails (e.g., dropping versus
overweighting the first and last two cases).
M.-g. Seo, R. Ilies / Organizational Behavior and Human Decision Processes 109 (2009) 120–133 125
had in the simulation before participating in a given session. Sec-
ond, we controlled for daily stock market movement, which could
systematically influence the participants’ affect, beliefs, and invest-
ing behaviors. In measuring daily stock market movement, we
used the local market index, the composite index of the 12 stocks
that had been randomly selected from the national stock market
for use in the simulation. Third, due to the cumulative nature of
the task as well as a high within-person correlation between per-
formance and past performance (r= .38), it is possible that self-effi-
cacy and goal are automatically correlated with performance via
their spurious relationships with past performance. Thus, we con-
trolled for past performance in all of our analyses.
Moreover, as discussed below, we used a within-subject meth-
odology in analyzing the data, which allows us to control for any
individual-specific effects that might be systematically related to
the endogenous variables in the hypothesized model (Griliches &
Mairesse, 1998), for example, the effects of participants’ age, gen-
der, or personality traits, etc.
Procedure
Each day for 20 consecutive business days (4 weeks) partici-
pants visited an Internet website once between 6:00 p.m. and
9:00 a.m. the next morning. Once they accessed the website, they
saw a log-in page. After they logged in (using a code name and
password), participants were presented with information on the
national daily stock market, including the daily changes and the
past-five-day trends of the three major market indices (the Dow
Jones, NASDAQ, and S&P500), as well as on the local market index
for the simulation, the composite index of the 12 anonymous
stocks that had been randomly selected from the national stock
market on the basis of varying degrees of risk and profitability
and of various industry and company size. The local market index
had been tracked over several months prior to the simulation and
found to be highly correlated with the national market indexes
(r> .8). During the 20-day period of the simulation, the local mar-
ket index maintained a relatively good balance of ups and downs
(14 ups and 6 downs). The individual stock names were repre-
sented (e.g., Stocks A, B and C) in such a way that participants could
not identify the real names.
On the next web page, participants received daily updated
information on several key aspects of the twelve individual stocks:
current daily price, daily percentage price change, 1 year stock per-
formance, beta coefficient, price per earning ratio, company size
and company industry. The following web page contained a report
that summarized participants’ most updated investment perfor-
mance and expected reward so far. All participants began the sim-
ulation with a certain amount of monetary reward ($200, real
cash), but they were given an indication of their current standing
with regard to winning (up to $800, real cash) or losing money
(up to $100, real cash) each day depending on their current invest-
ment performance, which was determined by their overall and
cumulative investment return – the percentage amount that they
earned or lost by investing their initial capital ($10,000, hypothet-
ical cash). All participants’ investment returns were consistently
adjusted by the current local market index (%) to discount simple
gains and losses incurred by the stock market fluctuations.
2
The
performance information that each participant received daily con-
cerned mostly his or her own performance and provided only limited
information regarding the other participants’ performance (the par-
ticipants’ averaged performance and the top performer’s perfor-
mance). Appendix B provides the actual screenshot shown to
participants with regard to their current performance information.
On the next page, participants rated their current affective state
(positive affect and negative affect). Moving to the next page, they
reported their subjective beliefs regarding the likelihood of obtain-
ing several specific performance outcomes (self-efficacy). Then they
indicated their performance goal (goal level).
On the subsequent page, participants made their investment
decisions for the day – which stocks to sell and which to buy. Each
participant were allowed to invest all or a part of the initial hypo-
thetical cash of $10,000 on any of the 12 stocks in the local market
as long as the cash balance did not go below zero, and were also
allowed to trade those stocks freely, with no transaction costs.
The web page had been designed in such a way that it automati-
cally performed all mathematical calculations required for invest-
ment decision-making, and instantly checked for mistakes (e.g.,
overinvestment). Before logging out, participants were presented
with their investment summary (in a table) and were asked to re-
port whether, when, and how long they had experienced any type
of interruptions during the tasks for the day. This process was re-
peated daily for the twenty business days.
3
Data structure and data analysis strategy
To enable a direct comparison between the results of this study
and the results of the past studies in different task environments,
we followed the analytical approaches adopted by Vancouver
and Kendall (2006). In particular, we used Hierarchical Linear Mod-
eling (HLM; Bryk & Raudenbush, 1992) to analyze the data. HLM
allows within-subject (lower-level) analyses while taking into ac-
count the non-independence in the level-1 data at level-2 (i.e.,
the case-level data were hierarchically nested within individuals).
We used HLM for a series of within-individual regression analyses
(at level-1) to test each set of the hypotheses proposed in Fig. 1.In
specifying the HLM models, we centered the predictor scores rela-
tive to each individual’s mean scores in order to obtain estimates
that are based exclusively on within-individual variance (the cen-
tering eliminated all between-individual variance in the predictor
scores). At level-2, we estimated the pooled parameters describing
the within-individual relationships of interest.
Results
Table 1 presents the means and standard deviations of the key
variables considered in this study, as well as the correlations be-
tween them, computed both between individuals and within indi-
viduals (using averages). Consistent with the previous studies, self-
efficacy was positively correlated with performance at the between
level of analysis (r= .49, p< .001). However, unlike the results of
other previous studies (e.g., Vancouver et al., 2001, 2002), self-effi-
cacy was also positively correlated with performance within indi-
viduals (r= .23, p< .001). In Table 1, we also show how the
variance in the daily scores on the endogenous variables is parti-
tioned across the two levels of analysis. These results show that
a substantial amount of the total variance in these scores was
due to within-individual variation across days (between 35% and
57%), which indicates that within-individual analyses predicting
these construct scores are appropriate. In contrast to other studies
that observed little variance in goal level within individuals over
2
Participants were initially informed that their investment return was not
determined by the simple increase or decrease of their stock portfolio value per se,
nor was it affected by other participants’ performance on the simulation, so that
participants would not be motivated to simply capitalize on market fluctuations or to
compete with each other.
3
To ensure that all the variables were measured in an order that is consistent with
the hypothesi zed causal directi ons among the variables, the web pages we re
programmed in such a way that participants could not skip certain web pages or go
back to the previous pages to change their original responses. They also could not re-
enter the web site in the same day.
126 M. Seo, R. Ilies / Organizational Behavior and Human Decision Processes 109 (2009) 120–133
time (e.g., Vancouver & Kendall, 2006), we observe that 97% of the
participants changed their goal level during the simulation and
about 35% of the total variance in goal level came from within indi-
viduals. In addition, the intra-individual correlation between past
performance and goal level was .34 (p< .001), indicating that
although past performance is an important factor affecting goal le-
vel, goal changes are unlikely to be a mere reflection of past perfor-
mance since more than 87% of its within-person variance is
determined by factors other than past performance. This reflects
the dynamic nature of the motivational process examined in this
study, in which the participants actively engaged in setting and
adjusting their goals over time in response to the changing task
environments.
Table 2 presents the results of our HLM analysis examining the
within-person relationship between self-efficacy, goal, effort and
performance. The results show that trial was positively related to
performance (
c
= .038, p< .001), but negatively related to self-effi-
cacy (
c
=.025, p< .001) and goal level (
c
=.14, p< .001) within
individuals over time (the study involved substantial downward
adjustments in their self-efficacy and goal level). Also considering
Table 2
Within-person analysis of self-efficacy, goal, effort, and performance using HLM.
a
Step Goal level not controlled Goal level controlled
c
SE % Var
D
% Var
c
SE % Var
D
% Var
DV: Self-Efficacy
1. Trials 0.025
***
0.002 7.9 7.9
2. Market 0.016
*
0.008 8.0 0.2
3. Past performance 0.129
***
0.004 39.6 31.5
DV: Goal Level
1. Trials 0.140
***
0.012 7.3 7.3
2. Market 0.160
***
0.045 7.9 0.6
3. Past Performance 0.450
***
0.029 19.1 11.2
4. Self-Efficacy 2.208
***
0.152 27.7 8.5
DV: Intended Time
1. Trials 0.037
***
0.003 7.2 7.2 0.037
***
0.003 7.2 7.2
2. Market 0.008 0.038 7.2 0.0 0.008 0.038 7.2 0.0
3. Past performance 0.038
***
0.008 8.2 1.1 0.038
***
0.008 8.2 1.1
4. Goal level – – – – 0.047
***
0.007 10.6 2.4
5. Self-efficacy 0.259
***
0.046 9.8 1.6 0.174
***
0.048 11.2 0.6
DV: Actual Time
1. Trials 0.073
***
0.010 3.0 3.0 0.073
***
0.010 3.0 3.0
2. Market 0.046 0.036 3.1 0.0 0.046 0.036 3.1 0.0
3. Past performance 0.161
***
0.025 5.3 2.2 0.161
***
0.025 5.3 2.2
4. Goal level – – – 0.063
**
0.021 5.5 0.2
5. Self-efficacy 0.279
*
0.139 5.5 0.2 0.158 0.147 5.5 0.0
5. Intended time 0.604
***
0.071 8.8 3.5 0.588
***
0.072 8.7 3.2
DV: Performance
1. Trials 0.038
***
0.009 1.0 1.0 0.038
***
0.009 1.0 1.0
2. Market 0.148
***
0.033 2.1 1.1 0.148
***
0.033 2.1 1.1
3. Past performance 0.500
***
0.019 29.2 27.1 0.500
***
0.019 29.2 27.1
4. Goal level – – – – 0.025 0.016 29.4 0.2
5. Self-efficacy 0.605
***
0.107 30.4 1.2 0.617
***
0.113 30.4 1.0
5. Intended time 0.014 0.056 29.0 0.1 0.025 0.057 29.1 0.1
5. Actual time 0.044
**
0.018 29.4 0.2 0.042
*
0.018 29.4 0.2
a
c
= unstandardized HLM coefficient, SE = standard error. All predictor scores were centered at individuals’ means to eliminate between-individual variance. Results are
based on 101 participants who provided 1870 daily response sets. Step indicates order of entry into HLM. Each variable entered in isolation, so more than one variable at a
single step indicates the variable was entered without the other variables in that step, but with the variables from the previous steps.
*
p< .05.
**
p< .01.
***
p< .001.
Table 1
Means, standard deviations, variance proportions, and correlations.
a
Variables Mean SD Variance proportion 1 2 3 4 5 6 7 8
Within (%) Between (%)
1 Past performance 0.45 3.09 54.1 45.9 0.53
**
0.25
**
0.07
**
0.23
**
0.73
**
0.23
**
0.02
2 Self-efficacy 2.46 0.48 44.3 55.7 0.45
**
0.61
**
0.12
**
0.18
**
0.48
**
0.31
**
0.10
**
3 Goal level 3.40 2.33 35.1 64.9 0.34
**
0.53
**
0.14
**
0.16
**
0.23
**
0.16
**
0.06
**
4 Intended effort 1.50 0.70 40.6 59.4 0.03 0.19
**
0.24
**
0.29
**
0.06
**
0.14
**
0.05
*
5 Actual effort 2.49 1.96 57.1 42.9 0.13
**
0.18
**
0.18
**
0.29
**
0.21
**
0.09
**
0.06
**
6 Performance 0.49 3.36 50.6 49.4 0.38
**
0.23
**
0.16
**
-0.03 0.05
*
0.29
**
0.01
7 Positive affect 1.07 0.57 41.9 58.1 0.15
**
0.28
**
0.24
**
0.14
**
0.15
**
0.25
**
0.02
8 Negative affect 0.77 0.49 51.4 48.6 0.04 0.14
**
0.05
*
0.00 0.03 0.18
**
0.35
**
Means, standard deviations, and correlations below the diagonal were computed for each individual across rounds and then averaged across individuals (averaged within-
individual correlations) whereas correlations above the diagonal were computed across individuals for each round and then averaged across rounds (averaged between-
individual correlations).
a
N= 1870 (101 participants with 20 rounds).
*
p< .05.
**
p< .01.
M.-g. Seo, R. Ilies / Organizational Behavior and Human Decision Processes 109 (2009) 120–133 127
the average past performance that is negative on average (.45),
the results clearly indicate that although the participants learned
the task and thus improved their performance over time, the task
was inherently challenging and difficult. This is another apparent
difference in the nature of the task from other previous studies that
showed either a non-significant (e.g., Vancouver & Kendall, 2006)
or a positive relationship (e.g., Yeo & Neal, 2006) between trial
and self-efficacy.
Testing within-person relationship between self-efficacy and
performance
Consistent with the results of the previous studies, our results
show that past performance positively predicted self-efficacy
(
c
= .129, p< .001). In support of H1-1, but contrary to the results
of the previous studies (e.g., Vancouver & Kendall, 2006), self-effi-
cacy was positively related to intended time (
c
= .259, p< .001), ac-
tual time (
c
= .279, p< .05), and performance (
c
= .605, p< .001)
even after controlling for past performance, explaining 1.6%, 3.5%
and 1.2%
4
of the within-person variances of these variables, respec-
tively. These positive relationships between self-efficacy and in-
tended time and between self-efficacy and performance remained
significant after goal level was additionally entered and controlled
(
c
= .174, p< .001;
c
= .617, p< .001, respectively). Only the relation-
ship between self-efficacy and actual time became non-significant
when goal level was controlled.
In support of H1-2, self-efficacy was positively related to goal
level (
c
= 2.208, p< .001), which in turn, positively contributed to
intended time (
c
= .047, p< .001) and actual time (
c
= .063,
p< .01). Based on Sobel (1982) tests, these indirect paths mediated
by goal level from self-efficacy to intended time and to actual time
produced significant mediated effects (p< .01) even after account-
ing for the direct effects of self-efficacy. However, the direct rela-
tionship between goal level and performance was not significant,
suggesting that goal level was only indirectly related to perfor-
mance via its positive association with intended and actual time.
Therefore, our results did not support or only weakly supported
our H1-2 that goal level partially mediates the efficacy–perfor-
mance relationship.
5
To further examine whether the positive efficacy–performance
relationship found in this study is systematically affected by the
degree to which individuals changed their performance goals dur-
ing the simulation, we first created an individual-level (level-2)
variable, goal change, by measuring the overall degree to which
each participant positively (upwardly) or negatively (downwardly)
adjusted their performance goals during the simulation period; we
first subtracted the goal level at the previous round from the goal
level at the current round and then averaged the resulting scores
across the 20 rounds for each participant. It ranged between
1.61 and 0.83 with a mean of 0.22 and a standard deviation
of 0.38. Thus, a higher score (one or two standard deviations above
the mean) indicates a positive (upward) goal adjustment during
the simulation, while a lower score indicates a downward adjust-
ment. Then, we developed and examined a ‘slope-as-outcomes’
HLM model in which we regressed the efficacy–performance slope
estimates obtained from level-1 analyses (reported in Table 2)on
goal change at level-2. In doing so, we controlled trial, stock market,
past performance, and goal level at level-1 as well as averaged per-
formance (daily performance scores averaged across the 20 rounds
for each individual) and averaged goal level (goal levels averaged
across the 20 rounds for each individual) at level-2.
6
The results showed that goal change significantly and positively
moderated the relationship between self-efficacy and performance
(
c
= 0.962, p< .001) even after controlling for the effects of past
performance and goal level at both within-person (level-1) and be-
tween-person (level-2) levels, explaining 17% of the variance in the
efficacy–performance slope. Similarly, goal change also positively
moderated the relationships between self-efficacy and intended
time (
c
= .298, p< .01) and between self-efficacy and actual time
(
c
= 1.000, p< .01). As illustrated in Fig. 2, these results indicate
that self-efficacy was more strongly related to motivation and per-
formance for those individuals who made more positive (upward)
adjustments in their goals during the simulation regardless of their
overall levels of goal and performance in the simulation.
Testing the effects of positive affect and negative affect
Table 3 presents the HLM results testing the hypothesized effects
of positive and negative affect on the key constructs linking self-effi-
cacy and performance. To minimize overlapping information, we
did not report the information that is presented in Table 2.Aspre-
dicted (arrow ‘‘h” in Fig. 1), the results show that past performance
was positively related to positive affect (
c
=.036,p< .001). However,
past performance was not significantly related to negative affect.
In support of H2-1, positive affect was positively related to in-
tended time (
c
= .096, p< .01), actual time (
c
= .323, p< .001) and
performance (
c
= 0.742, p< .001), and continued to be significantly
related to actual time (
c
= .097, p< .01) and performance (
c
= .073,
p< .001) after controlling for self-efficacy and goal level. The rela-
tionship between positive affect and intended time was no longer
4
The relatively low percentage of within-person variances in performance
explained by self-efficacy (1.2%) was largely due to the control of past performance.
Without controlling for past performance, self-efficacy explained 16.6% of the within-
person variances in performance.
5
We also note that this non-support or weak-support of H1-2 was due to the
control of past performance. Without past performance, goal level significantly and
directly influenced performance explaining 7.5% of its variance.
6
These specifications are represented by the equations below:Level 1: Performance
=b
0
+b
1
X
1
+b
2
X
2
+b
3
X
3
+b
4
X
4
+b
5
Self-efficacy + r
Level2 :b
0
¼
c
00
þ
c
01
C
1
þ
c
02
C
2
þ
c
03
Goal Change þU
0
b
1
¼
c
10
þU
1
b
2
¼
c
20
þU
2
b
3
¼
c
30
þU
3
b
4
¼
c
40
þU
4
b
5
¼
c
50
þ
c
51
C
1
þ
c
52
C
2
þ
c
53
Goal Change þU
5
where X
1
is trial (level-1), X
2
is market index (level-1), X
3
is past performance (level-1), X
4
is
goal level (level-1), C
1
is averaged performance (level-2), and C
2
is averaged goal level (level-
2). In specifying the HLM models, we centered the level-1 predictor scores relative to the
participants’ group mean score while centering the level-2 predictor scores relative to the
grand mean score.
Fig. 2. The moderating effect of goal change on the relationship between self-
efficacy and performance.
128 M. Seo, R. Ilies / Organizational Behavior and Human Decision Processes 109 (2009) 120–133
significant when self-efficacy and goal level were controlled. These
results suggest that positive affect uniquely and positively contrib-
ute to actual effort and performance, in addition to the effects of
self-efficacy and goal level (e.g., explaining additional 3.4% of the
within-person variance of performance). Also in support of H2-2,
negative affect was negatively and significantly related to perfor-
mance both before (
c
=.619, p< .001) and after (
c
=.570,
p< .01) controlling for self-efficacy and goal level. However, nega-
tive affect was not significantly related to either intended time or
actual time both with and without controlling for self-efficacy
and goal level. These results suggest that negative affect may neg-
atively contribute to performance by other means than reduced ef-
fort mobilization (e.g., by affecting risk-taking).
As predicted in H3-1, positive affect was significantly and pos-
itively related to goal level (
c
= .391, p< .001), but the relationship
became completely non-significant when self-efficacy was entered
and controlled. This finding indicates that the effect of positive af-
fect on goal level was mostly mediated by self-efficacy. H3-2 was
not supported since negative affect was not significantly related
to goal level.
7
Finally, both H4-1 and H4-2 were strongly supported. Positive
affect was significantly and positively related to self-efficacy
(
c
= .151, p< .001) while negative affect was significantly and neg-
atively related to self-efficacy (
c
=.099, p< .001) even after con-
trolling for past performance (
c
= .072, p< .001). We summarized
the results of this study in Fig. 3.
Discussion
In this study, we examined the within-person relationship be-
tween self-efficacy and performance in a dynamic stock investment
simulation. Our results suggest that in a highly dynamic and complex
task environment, self-efficacy is positively relatedto motivation and
performance, contrary to the results of several recent studies (e.g.,
Vancouver & Kendall, 2006; Vancouver et al., 2001, 2002; Yeo & Neal,
2006) that showed a negative relationship.We found that when indi-
viduals had higher self-efficacy beliefs, they intended to spend and
actually spent more time on the task, and partly as a result, achieved
higher performance. Also contrary to previous findings (e.g., Vancou-
ver & Kendall, 2006; Vancouver et al., 2001), our results showed that
performance goals uniquely and positively contributed to both the
intended and actual amount of time spent for the task, which in turn,
led to higherperformance. In addition, we found that the within-per-
son relationship between self-efficacy and performance was
strengthened by the extent to which individuals made more positive
changes in their performance goals during the task.
7
Although this relationship became positive and significant when self-efficacy was
controlled (
c
= .244, p< .05), this explains little within-person variance in goal level
(0.1%). In addition, the relationship was negative and non-significant when negative
affect was entered alone. Thus, this result is likely to be an artifact due to the presence
of other strong predictors in the same HLM equation such as past performance and
self-efficacy.
Table 3
Within-person analysis of affect and its effects using HLM.
a
Step Self-efficacy and goal not controlled Self-efficacy and goal controlled
c
SE % Var
D
% Var
c
SE % Var
D
% Var
DV: Positive Affect
1. Trials 0.014
***
0.002 1.8 1.8
2. Market 0.024
**
0.009 2.1 0.3
3. Past performance 0.036
***
0.006 3.6 1.7
DV: Negative Affect
1. Trials 0.003 0.002 0.1 0.1
2. Market 0.013 0.008 0.1 0.0
3. Past performance 0.004 0.006 0.1 0.0
DV: Self-Efficacy
1. Trial, market, past performance
b
39.6
2. Positive affect 0.151
***
0.016 42.5 2.9
2. Negative affect 0.099
***
0.018 40.6 1.0
DV: Goal Level
1. Trial, market, past performance
b
19.1 19.1
2. Self-efficacy – – – – 2.208
***
0.152 27.7 8.5
3. Positive affect 0.391
***
0.109 19.6 0.5 0.060 0.106 27.6 -0.1
3. Negative affect 0.021 0.121 19.0 0.1 0.244
*
0.116 27.8 0.1
DV: Intended Time
1. Trial, market, past performance
b
8.2 8.2
2. Self-efficacy, goal level
b
– – – – 11.2 3.0
3. Positive affect 0.096
**
0.048 8.6 0.4 0.056 0.032 11.2 0.0
3. Negative affect 0.035 0.035 8.1 0.1 0.052 0.034 11.2 0.0
DV: Actual Time
1. Trial, market, past performance
b
5.3 5.3
2. Self-efficacy, goal level
b
– – – 5.5 0.2
3. Positive affect 0.323
***
0.094 5.8 0.5 0.292
**
0.097 5.8 0.2
3. Negative affect 0.031 0.105 5.2 0.1 0.016 0.106 5.4 0.1
DV: Performance
1. Trial, market, past performance
b
29.2 29.2
2. Self-efficacy, goal level
b
– – – 30.4 1.2
3. Positive affect 0.742
***
0.071 33.3 4.1 0.690
***
0.073 33.8 3.4
3. Negative affect 0.619
***
0.080 31.5 2.4 0.570
***
0.080 32.3 2.0
a
c
= unstandardized HLM coefficient, SE = standard error. All predictor scores were centered at individuals’ means. N= 1870 (101 persons). Step indicates order of entry
into HLM. More than one variable at a single step indicates the variable was entered without the other variables in that step, but with the variables from the previous steps.
b
These variables were entered together. The specific results of each variable were summarized in Table 2.
*
p< .05.
**
p< .01.
***
p< .00.
M.-g. Seo, R. Ilies / Organizational Behavior and Human Decision Processes 109 (2009) 120–133 129
Moreover, our findings suggest that individuals’ affective states
play an additional role in the processes of goal regulation by un-
iquely contributing to motivation and performance above and be-
yond what is explained by self-efficacy and goal choice. In
particular, we found that when individuals experienced more posi-
tive affect, they tended to spend more time in a given task and
achieve a higher level of performance, whereas experiencing more
negative affect led to achieving lower levels of performance. These
effects occurred both directly, regardless of individuals’ levels of
self-efficacy and goal, and indirectly via affecting self-efficacy:
experiencing more positive affect led to higher levels of self-effi-
cacy, but experiencing negative affect lowered self-efficacy.
Theoretical implications
The findings of this study provide several theoretical implica-
tions to the motivational self-regulation literature. First, the results
of this study enhance our understanding of the within-person rela-
tionship between self-efficacy and performance, which has been
extensively debated between socio-cognitive theorists (e.g., Ban-
dura & Locke, 2003) and control theorists (Vancouver & Kendall,
2006; Vancouver et al., 2001, 2002). In particular, the positive rela-
tionship between self-efficacy and performance found in this
study, together with its several features that are critically different
from those of other previous studies showing a negative efficacy–
performance relationship, provide several possible theoretical
explanations by which the opposite views can be reconciled. One
explanation is that the nature of the efficacy–performance rela-
tionship can be determined by the relative degrees to which indi-
viduals engage in goal-choice versus goal-planning during
motivational self-regulation. This explanation is based on a discon-
tinuous model recently offered by Vancouver et al. (2008), which
suggest that in the process of goal choice, self-efficacy may create
a positive discontinuity in motivation by directing disproportion-
ately high resources towards a goal, whereas it is likely to be neg-
atively related to motivation and performance once a goal is
accepted (in the process of goal-planning) as self-efficacy reduces
resources allocated to achieve an accepted goal. Thus, a positive
relationship is more likely in a dynamic task environment where
individuals continuously engage in goal choice processes, while
focusing relatively less on goal-planning processes. Clearly, our re-
sults strongly support this explanation because substantial within-
person variance in goal level was observed in this study (compared
to other previous studies) and also because these within-person
goal changes partially mediated the efficacy–performance
relationship.
However, our results also showed that the efficacy–perfor-
mance relationship remained positive and significant within indi-
viduals even after statistically controlling for the goal effects.
Certainly, this requires additional explanations. We argue that in
a highly complex decision-making task where effective strategy
development is crucial for performance, self-efficacy may foster
discovery and use of effective analytic strategies, for example, by
helping individuals remain focused on task strategies and less af-
fected by disruptive thinking in face of repeated difficulties and
failures (Wood & Bandura, 1989), which in turn, may increase per-
formance. In this case, the effect of self-efficacy on performance is
not necessarily mediated by goals. However, our results showing
that the within-person relationship between self-efficacy and per-
formance is significantly higher for individuals who positively
changed their goals during the simulation suggest that goal setting
moderates the relationship, perhaps by further facilitating effective
strategy development processes (Smith, Locke, & Barry, 1990).
Thus, performance is likely to be highest when high self-efficacy
and positive goal changes are combined.
Second, our findings in this study extend our understanding of
the role of individuals’ affective experience in motivational self-reg-
ulation, particularly in relation to the cognitive processes repre-
sented by self-efficacy and goal. First of all, we found a close
association between cognitive and affective processes, particularly
between self-efficacy beliefs and state affect. Consistent with a large
accumulation of previous studies (see George & Brief, 1996; Forgas &
George, 2001; Seo et al., 2004; for reviews), our results showed that
positive and negative affect experienced during the process of moti-
vational self-regulation was significantly associated with self-effi-
cacy beliefs above and beyond what is explained by performance
feedback. Also consistent with the results of Ilies and Judge’s
(2005) experimental studies, positive affect, but not negative affect,
was positively related to goal choice. However, this affect-goal asso-
ciation disappeared when self-efficacy was introduced and con-
trolled, indicating that affective states are related to goal choice
only indirectly via their association with self-efficacy.
On the other hand, we found that positive affect and negative
affect uniquely contributed to motivation and performance within
individuals over time, unmediated by self-efficacy or goal. This
finding supports a distinctive perspective in the emotion literature
that affect can activate certain motivational systems or behavioral
predispositions (e.g., Cacioppo et al., 1999; Frijda, 1987; Gray,
1990; Watson et al., 1999) beyond cognitive evaluations or con-
scious awareness (e.g., Winkielman et al., 1997; Izard, 1993; Loe-
wenstein et al., 2001; Brehm, 1999). Moreover, our results
showed that the effects of positive and negative affect on perfor-
Performance
.56
Past
Performance
Self-
Efficacy
Intended
Effort
Goal
Level
Positive
Affect
.18
.38
.11 f
.14 .18
.08
Negative
Affect
Actual
Effort
.20 .05
.15
-.10 .07 .19 -.14
Fig. 3. Supported model and summary of results (All path coefficients are statistically significant at p< .01. Path coefficients are the standardized beta coefficients in HLM
regressions entered with controls (
c
adjusted by Sx/Sy).).
130 M. Seo, R. Ilies / Organizational Behavior and Human Decision Processes 109 (2009) 120–133
mance were virtually unaffected by either intended time or actual
time spent for the given task. This finding suggests that the affect-
performance link in the process of motivational self-regulation is
more complex than what can be explained by a simple mediating
mechanism of effort. For example, affect can influence perfor-
mance via directly affecting individuals’ certain behavioral tenden-
cies, for example, the degree of risk-taking (e.g., Au, Chan, Wang, &
Vertinsky, 2002; Seo & Barrett, 2007). Future research needs to
determine precise mechanisms though which affect leads to per-
formance in the process of motivational self-regulation.
Practical implications
The theoretical implications of this study directly inform prac-
tice. First, the negative within-person relationship between self-
efficacy and performance found in several recent studies (Vancou-
ver & Kendall, 2006; Vancouver et al., 2001, 2002; Yeo & Neal,
2006) suggests that self-efficacy, at least in certain conditions,
can hinder performance within individuals over time. In contrast,
the results of this study suggest that self-efficacy can positively
contribute to performance within individuals in other conditions,
particularly in a dynamic task environment in which individuals
actively and continuously engage in a series of goal choice process
by setting and adjusting their performance goals. Therefore, task
environments with certain characteristics (the degree of dyna-
mism in particular) need to be considered in implementing effi-
cacy-enhancing programs in organizations. Moreover, in a
dynamic task environment, goal choice (goal change) uniquely
and positively contributed to motivation beyond the effect of
self-efficacy within individuals. Thus, performance improvements
can be obtained by encouraging or formally asking employees to
set and periodically adjust their goals in the process of performing
a dynamic task. Finally, our findings suggest that individuals’ affec-
tive states experienced during motivational self-regulation signifi-
cantly influence their motivation and performance either directly
or indirectly via affecting their self-efficacy beliefs. Thus, manage-
rial practices aimed to directly increase their positive affective
experience (e.g., pleasant music or humor) or reduce negative feel-
ings (e.g., information sharing to alleviate uncertainty or anxiety)
can be useful means to further improve performance.
Limitations and future research directions
We propose several future research directions that address lim-
itations of this study and should advance understanding of dy-
namic goal-regulation. First, this study employed several unique
task characteristics, such as enhanced dynamism and high com-
plexity as we explained above, which may fundamentally differen-
tiate this study from other previous studies, and thus may provide
evidence that certain task characteristics might change or reverse
the efficacy–performance relationship. However, our study design
does not allow us to completely eliminate alternative explanations
of our results or identify the precise underlying mechanisms
through which our results were produced. For example, it is also
possible that the challenging nature of the task as well as the
strong external rewards associated with task performance might
trigger strong achievement motivation (Atkinson, 1957) or ap-
proach motivation (Cacioppo et al., 1999; Gray, 1990), which in
turn, may produce a positive efficacy–performance relationship
by increasing individuals’ regulatory focus on a specific target of
the achievement or approach motivation that is being activated,
and/or by minimizing possible distractions of other task demands
(e.g., Schmidt & DeShon, 2007). Future studies need to disentangle
the precise task characteristics that determine the direction of the
efficacy–performance relationship, as well as the precise underly-
ing mechanisms through which such effects occur.
Second, this study is based on a correlational research design,
which makes it impossible to determine the precise causal direc-
tions among the key variables. We designed the Internet web pages
so we could measure the variables in an order consistent with the
predicted causal direction, but we could not completely eliminate
the possible reverse causal effects, and past research shows that af-
fect and cognitive judgments are so closely related to each other
that mutual influence is possible at any given moment (cf., Dama-
sio, 1994; Forgas, 1995; LeDoux, 1996). To precisely determine the
causal relationships among the key variables, this study should be
supplemented by future studies in which key variables are exper-
imentally induced.
Third, although the stock investment simulation enabled us to
examine the hypothesized framework in a dynamic task setting
with naturalistic components, this approach has several features
that limit its generalizability to actual stock investing decisions
and behaviors. For example, participants could not thoroughly re-
search the companies selected, made their investment decisions
based on the limited information provided, faced with only limited
gain or risk, and changed their stock portfolio only in limited times
(e.g., once a day in the evening for 20 days). Future studies that add
more natural elements to this study would increase the external
validity of the results of this study in the domain of stock investing
behavior.
Fourth, the expected return for participating in this simulation
was highly skewed towards the gain side ($100–$1000 cash remu-
nerations), while it prevented participants from any real loss as the
worst performer still getting paid by $100. This imbalance between
possible gains and losses in the remuneration structure may have led
participants to be more sensitive to rewards and less sensitive to po-
tential losses, which may explain why negative affect did not have as
much effects on motivation and performance as positive affect.
Additional research is required to examine whether negative affect
plays a significant role in goal regulation in a more loss-sensitive
setting.
Next, the nature of task behavior involved in this study, simu-
lated stock investing, differs from other work settings. For exam-
ple, task performance is less effort-dependent (but more
strategy-dependent) in stock investing than in most other tasks.
In addition, people’s behaviors in most other settings are continu-
ously influenced by complex group and organizational factors (e.g.,
rules and norms, colleagues, bosses, etc.) whereas this simulation
is a purely individual-based task with a simple reward structure
and minimum rules. Moreover, this simulation involves a series
of cumulative tasks with a definite end-time as is the case in some
work situations (e.g., project work), but such a definite end-time
does not characterize many other day-to-day work situations.
Thus, future studies should empirically test the effects observed
in this study within other task environments, particularly in team
or work-group settings, to examine the extent to which these ef-
fects generalize.
Finally, there are other possible situational and individual exog-
enous factors that may also influence motivational self-regulation
and thus could have influenced the results of this study: situational
factors such as multiple vs. single goal environments (e.g., Kernan
& Lord, 1990), cumulative vs. intermediate goal environments, and
individuals factors such as participants’ goal orientations (e.g.,
Cron et al., 2005), achievement motivation (e.g., McClelland &
Burnham, 1976) and causal attributions (e.g., Donovan & Williams,
2003; Ilgen & Davis, 2000). A valuable future research direction is
to theorize and empirically examine the extent to which the effects
documented in this study are moderated by individual and situa-
tional factors like these.
To conclude, we believe this research contributes to the moti-
vation literature by providing results that help reconcile the two
opposing perspectives regarding the role of self-efficacy in the
M.-g. Seo, R. Ilies / Organizational Behavior and Human Decision Processes 109 (2009) 120–133 131
process of motivational self-regulation within individuals. In
doing so, our results as well as the unique task settings that pro-
duced the results place a particular emphasis on the role of task
environment (context) in understanding the nature of the effi-
cacy–performance relationship. Furthermore, by examining affec-
tive processes within a broader context of cognitive processes
operating in motivational self-regulation, our study demonstrated
the unique effects of affective experience on motivation and per-
formance, which may occur either in coordination with or inde-
pendent from the core cognitive processes.
Appendix A. Participants monetary rewards distribution
Participants’ investment return
to the local market return
Above 7% Above 5%
below 7%
Above 3%
below 5%
Above 1%
below 3%
Between 1%
and 1%
Below 1%
above 3%
Below 3%
Participants’ reward Top: $1000
Second: $700 $350 $300 $250 $200 $150 $100
Third: $500
Rest: $400
Appendix B
132 M. Seo, R. Ilies / Organizational Behavior and Human Decision Processes 109 (2009) 120–133
References
Atkinson, J. W. (1957). Motivational determinants of risk-taking behavior.
Psychological Review, 64, 359–372.
Au, K., Chan, F., Wang, D., & Vertinsky, I. (2002). Mood in foreign exchange trading:
Cognitive processes and performance. Organization Behavior and Human
Decision Processes, 91, 322–338.
Bagozzi, R., & Pieters, R. (1998). Goal-directed emotions. Cognition & Emotion, 12,
1–26.
Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory.
Englewood Cliffs, NJ: Prentice Hall.
Bandura, A. (1991). Social cognitive theory: An agentic perspective. Annual Review of
Psychology, 52, 1–26.
Bandura, A. (1997). Self-efficacy: The exercise of control. New York: W. H. Freeman.
Bandura, A. (2001). Social cognitive theory: An agentic perspective. Annual Review of
Psychology, 52, 1–26.
Bandura, A., & Cervone, D. (1983). Self-evaluative and self-efficacy mechanisms
governing the motivational effects of goal systems. Journal of Personality and
Social Psychology, 45, 1017–1028.
Bandura, A., & Cervone, D. (1986). Differential engagement of self-reactive
influences in cognitive motivation. Organizational Behavior and Human
Decision Processes, 38, 92–113.
Bandura, A., & Locke, E. A. (2003). Negative self-efficacy and goal effects revisited.
Journal of Applied Psychology, 88, 87–99.
Bandura, A., & Wood, R. (1989). Effect of perceived controllability and performance
standards of self-regulation of complex decision making. Journal of Personality
and Social Psychology, 56, 805–814.
Bandura, A., Wood, R., & Bailey, T. (1990). Mechanisms governing organizational
performance in complex decision-making environments. Organizational
Behavior and Human Decision Processes, 46, 181–201.
Barrett, L. F., & Russell, J. A. (1998). Independence and bipolarity in the structure of
current affect. Journal of Personality and Social Psychology, 74, 967–984.
Bodie, Z. A., Kane, A., & Marcus, A. J. (2001). Essentials of investments. McGraw-Hill.
Bower, G. H. (1981). Mood and memory. American Psychologist, 36, 129–148.
Brehm, J. W. (1999). The intensity of emotion. Personality and Social Psychology
Review, 3, 2–22.
Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models: Applications and
data analysis methods. Thousand Oaks, CA: Sage.
Cacioppo, J. T., Gardner, W. L., & Berntson, G. G. (1999). The affect system has
parallel and integrative processing components: Form follows function. Journal
of Personality and Social Psychology, 76, 839–855.
Caginal, P. G., & Laurent, H. (1998). The predictive power of price patterns. Applied
Mathematical Finance, 5, 181–205.
Carver, C. S., & Scheier, M. F. (1990). Origins and functions of positive and negative
affect: A control-process view. Psychological Review, 97, 19–35.
Carver, C. S., & Scheier, M. F. (1998). On the self-regulation of behavior. New York:
Cambridge University Press.
Chen, G., & Bliese, P. D. (2002). The role of different levels of leadership in predicting
self- and collective efficacy: Evidence for discontinuity. Journal of Applied
psychology, 87, 549–556.
Chen, G., Webber, S. S., Bliese, P. D., Mathieu, J. E., Payne, S. C., Born, D. H., & Zaccaro,
S. J. (2002). Simultaneous examination of the anticedents and consequences of
efficacy beliefs at multiple levels of analysis. Huaman Performance, 15, 381–409.
Cron, W. L., Slocum, J. W., VandeWalle, D., & Fu, Q. (2005). The role of goal
orientation on negative emotions and goal setting when initial performance
falls short of one’s performance goal. Human Performance, 18, 55–80.
Damasio, A. (1994). Descartes’ error: Emotion, reason, and the human brain. New
York: Avon Press.
Damasio, A. (1998). Emotion in the perspective of an integrated nervous system.
Brain Research Review, 26, 83–86.
Donovan, J. J., & Williams, K. J. (2003). Missing the mark: Effects of time and causal
attributions on goal revision in response to goal-performance discrepancies.
Journal of Applied Psychology, 88, 379–390.
Erez, A., & Isen, A. M. (2002). The influence of positive affect on the components of
expectancy motivation. Journal of Applied Psychology, 87, 1055–1067.
Forgas, J. P. (1995). Mood and judgment: The affective infusion model (AIM).
Psychological Bulletin, 117, 39–66.
Forgas, J. P., & Bower, G. H. (1987). Mood effects on person perception judgments.
Journal of Personality and Social Psychology, 53, 53–60.
Forgas, J. P., & George, J. M. (2001). Affective influences on judgments and behavior
in organizations: An information processing perspective. Organizational
Behavior and Human Decision Processes, 86, 3–34.
Frijda, N. H. (1987). Emotion, cognitive structure, and action tendency. Cognition
and Emotion, 1, 115–143.
George, J. M., & Brief, A. P. (1996). Motivational agendas in the workplace: The
effects of feelings on focus of attention and work motivation. In B. M. Staw & L.
L. Cummings (Eds.), Research in organizational behavior (pp. 75–109).
Greenwich: CT: JAI Press.
Gray, J. A. (1990). Brain systems that mediate both emotion and cognition. Cognition
and Emotion, 4, 269–288.
Griliches, Z., & Mairesse, J. (1998). Production functions: The Search for
identification in practicing econometrics. In Z. Griliches (Ed.), Essays in method
and application (pp. 232–411). Northampton, MA: Elgar.
Horton, M. J. (in press). Stars, crows, and doji: The use of candlesticks in stock
selection. The Quarterly Review of Economics and Finance.
Ilgen, D. R., & Davis, C. A. (2000). Bearing bad news: Reactions to negative
performance feedback. Applied Psychology: An International Review, 49, 550–565.
Ilies, R., & Judge, T. A. (2005). Goal regulation across time: The effects of feedback
and affect. Journal of Applied Psychology, 90, 453–467.
Isen, A. M., Shalker, T. E., Clark, M., & Karp, L. (1978). Affect, accessibility of material
in memory and behavior: A cognitive loop? Journal of Personality and Social
Psychology, 36, 1–12.
Izard, C. E. (1993). Four systems for emotion activation: Cognitive and non-
cognitive processes. Psychological Review, 100, 60–69.
Kavanagh, D. J., & Bower, G. H. (1985). Mood and self-efficacy: Impact of joy and
sadness on perceived capabilities. Cognitive Therapy and Research, 85, 507–525.
Kernan, M. C., & Lord, R. G. (1990). Effects of valence, expectancies, and goal–
performance discrepancies in single and multiple goal environments. Journal of
Applied Psychology, 75, 194–203.
Klein, H. J. (1991). Further evidence on the relationship between goal setting and
expectancy theories. Organizational Behavior and Human Decision Processes, 49,
230–257.
LeDoux, J. (1996). The emotional brain.The mysterious underpinnings of emotional life.
New York: Simon and Schuster.
Locke, E. A., & Latham, G. P. (1990a). A theory of goal setting and task performance.
Englewood Cliff, NJ: Prentice Hall.
Locke, E. A., & Latham, G. P. (1990b). Work motivation and satisfaction: Light at the
end of the tunnel. Psychological Science, 1, 240–246.
Loewenstein, G. F., Weber, E. U., Hsee, C. K., & Welch, N. (2001). Risk as feelings.
Psychological Bulletin, 127, 267–286.
McClelland, D. C., & Burnham, D. H. (1976). Power is the great motivator. Harvard
Business Review, March–April, 100–110.
Powers, W. T. (1973). Behavior: The control of perception. Chicago: Aldine.
Sadri, G., & Robertson, I. T. (1993). Self-efficacy and work-related behavior: A review
and meta-analysis. Applied Psychology: An International Review, 42, 139–152.
Schmidt, A. M., & DeShon, R. P. (2007). What to do? The effects of discrepancies,
incentives, and time on dynamic goal prioritization. Journal of Applied
Psychology, 92, 928–941.
Seo, M., Barrett, L., & Bartunek, J. M. (2004). The role of affective experience in work
motivation. Academy of Management Review, 29, 423–439.
Seo, M., & Barrett, L. F. (2007). Being emotional during decision making—Good or
bad? An empirical investigation. Academy of Management Journal, 50, 923–940.
Smith, K. G., Locke, E. A., & Barry, D. (1990). Goal setting, planning, and
organizational performance: An experimental simulation. Organizational
Behavior and Human Decision Processes, 46, 118–134.
Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effects in structural
equation models. In S. Leinhardt (Ed.), Sociological methodology (pp. 290–312).
Washington, DC: American Sociological Association.
Stajkovic, A. D., & Luthans, F. (1998). Self-efficacy and work-related performance: A
meta analysis. Psychological Bulletin, 124, 240–261.
Vancouver, J. B. (2000). Self-regulation in industrial/organizational psychology: A
tale of two paradigms. In M. Boekaerts, P. R. Pintrich, & M. Zeidner (Eds.),
Handbook of self-regulation (pp. 303–341). San Diego, CA: Academic Press.
Vancouver, J. B. (2005). The depth of history and explanation as benefit and bane for
psychological control theories. Journal of Applied Psychology, 90, 38–52.
Vancouver, J. B., & Kendall, L. N. (2006). When self-efficacy negatively related to
motivation and performance in a learning context. Journal of Applied Psychology,
91, 1146–1153.
Vancouver, J. B., More, K. M., & Yoder, R. J. (2008). Self-efficacy and resource
allocation: Support for a nonmonotonic, discontinuous model. Journal of Applied
Psychology, 93, 35–47.
Vancouver, J. B., Thompson, C. M., Tischner, E. C., & Putka, D. J. (2002). Two studies
examining the negative effect of self-efficacy on performance. Journal of Applied
Psychology, 87, 506–516.
Vancouver, J. B., Thompson, C. M., & Williams, A. A. (2001). The changing signs in the
relationships among self-efficacy, personal goals, and performance. Journal of
Applied Psychology, 86, 605–620.
Watson, D., Wiese, D., Vaidya, L., & Tellegen, A. (1999). The two general activation
systems of affect: Structural findings, evolutionary considerations, and
psychobiological evidence. Journal of Personality and Social Psychology, 76,
820–838.
Winkielman, P., Zajonc, R. B., & Schwarz, N. (1997). Subliminal affective priming
resists attributional interventions. Cognition and Emotion, 11, 433–465.
Wood, R., & 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.
Yeo, G. B., & Neal, A. (2006). An examination of the dynamic relationship between
self-efficacy and performance across levels of analysis and levels of specificity.
Journal of Applied Psychology, 91, 1088–1101.
M.-g. Seo, R. Ilies / Organizational Behavior and Human Decision Processes 109 (2009) 120–133 133