Content uploaded by Gary P. Latham
Author content
All content in this area was uploaded by Gary P. Latham on Sep 08, 2016
Content may be subject to copyright.
ORGANIZATIONAL BEHAVIOR AND HUMAN DECISION PROCESSES 50,
212-247 (1991)
Self-Regulation through Goal Setting
GARY
P.
LATHAM
University
of
Toronto
AND
EDWIN
A.
LOCKE
University
of
Maryland
The extant literature on goal setting through 1990 has been reviewed and
integrated by Locke and Latham (199Oa). The result was the development of
a theory of goal setting with special emphasis on its practical implications for
the motivation of employees in organizational settings. The purpose of the
present paper is twofold. First, the theory is summarized and updated with
respect to research completed since publication of the 1990 book. Second, the
self-regulatory effects of goal setting are described. Emphasis is given to ways
that people can use goals as a self-management technique.
0 1991 Academic
Press. Inc.
GOAL SETTING THEORY
Goal setting theory (Locke & Latham, 1984, 199Oa) is based on the
simplest of introspective observations, namely, that conscious human
behavior is purposeful. It is regulated by the individual’s goals. Goal
directedness, however, characterizes the actions of all living organisms
including those of plants. Thus the principle of goal-directed action is not
restricted to conscious action.
Binswanger (1990) has shown that goal-directed action is defined by
three attributes: (1) self-generufion: the source of energy is integral to the
organism; (2) value-significance: the actions not only make possible but
are necessary for an organism’s survival; and (3) goal-causation: the
resulting action is caused by a goal. In the case of vegetative action,
goal-directed behavior in the present is caused by past instances of suc-
cessful goal-directed action. For example a person’s heart beats today
because it beat successfully (i.e., facilitated survival) yesterday.
The lowest level of goal-directed action is physiologically controlled
(e.g., plants). The next level, present in the lower animals, entails con-
scious self-regulation through sensory-perceptual mechanisms including
Address correspondence and reprint requests to Gary P. Latham, Faculty of Manage-
ment, University of Toronto, 246 Bloor St., W., Toronto, Ontario, Canada MSS lV4.
212
0749-5978191 $3.ocl
Copyright Q 1991 by Academic Press, Inc.
All rights of reproduction in any form reserved.
SELF-REGULATION THROUGH GOAL SETTING
213
pleasure and pain. Human beings possess a higher form of consciousness,
the capacity to reason. They have the power to conceptualize goals and
set long range purposes (Locke, 1969). Purposeful action in human beings
is volitional (Binswanger, 1991). Thus, people must choose to discover
what is beneficial to their welfare, they must set goals to achieve it, they
must choose the means for attaining these goals, and then they must
choose to act on the basis of these judgments.
The domain of goal setting theory lies within the domain of purposefully
directed action. The theory focuses on the question of why some people
perform better on work tasks than others. If they are equal in ability and
knowledge, then the cause must be motivational. Goal setting theory
approaches the issue of motivation from a first-level perspective; its em-
phasis is on an immediate level of explanation of individual differences in
task performance (Ryan, 1970). The theory states that the simplest and
most direct motivational explanation of why some people perform better
than others is because they have different performance goals.
Goal setting theory, in sharp contrast to control theory, was developed
inductively in that it was based on the accumulated research findings of
literally hundreds of studies which were conducted over the past 25 years
(Locke, in press). The initial research focused on the hypothesis that
goals, given the person has the requisite ability, motivate action. Once
this hypothesis was supported, research proceeded in several different
directions. The generalizability of the initial findings was investivated by
determining whether goal setting worked with different tasks and in dif-
ferent settings. In addition, there were attempts at lateral integration. This
involved connecting goal setting with related concepts at the same level of
abstraction, such as feedback, participation, incentives, self-efficacy, and
satisfaction. Similarly, there were attempts at vertical integration. This
involved tying goal setting to broad concepts such as values and person-
ality. The theory also underwent elaborations through attempts to specify
the mechanisms by which goal setting affects performance. And finally,
attempts were made to identify moderators or boundary conditions for
goal setting.
Goal Attributes
Two attributes of goals have been studied in relation to performance,
namely content and intensity. With regard to content, two aspects have
been the focus of the research to date. The first is specificity.
Goal content can vary on a continuum from vague (“work on this
task”) to specific (“try for a score of 62 correct on this task within the
next 30 minutes”). The second aspect of content that has been studied is
difficulty. (For an analysis of the effects of different operationalizations of
214
LATHAM AND LOCKE
goal difficulty, see Wright, 1990.) Goals can be easy (“try to get 5 prob-
lems completed in the next 30 minutes”), moderate (“try to get 10 . . .“),
difficult (“try to get 15 . . .“), or impossible (“try to get 50 . . .“).
Difficulty is a concept of relationship; it pertains to the relationship be-
tween a person and a task or goal. Thus the same task or goal can be easy
for one person and hard for another depending on the person’s ability and
experience. On the average, however, the higher the absolute level of the
goal the more difficult it is for a person to achieve it.
Approximately 400 studies have examined the relationship of goal at-
tributes to task performance. It has been found consistently that perfor-
mance is a linear function of goal difficulty. Given adequate ability and
commitment to the goal, the harder the goal the higher the performance.
We attribute this finding mainly to the fact that people normally adjust
their level of effort to the difficulty of the task undertaken and thus try
harder for difficult than for easy goals. A scatter-plot based on some of
the earliest studies of goal difficulty (derived from Locke, 1968) is shown
in Fig. 1. This linear function is different in shape than the function in
Atkinson’s (1958) theory which relates task difftculty to performance.
Atkinson’s research showed a performance drop at the highest level of
task difficulty, thus yielding an inverse U function. Knowing task difli-
culty, however, does not reveal the person’s goals and thus makes it
difftcult to predict how well a person will perform the task (Locke &
6- . .
.
. .
4-
.
.
-4
:
.
:
.
-6 1
I
I I 1 I 1 I I , I
100 90 80 70 60 50 40 30 20 10 0
(Obpxtlve Probablhty oi Success)
Goal
Difficulty
FIG.
1. The relation of goal difficulty to performance (based on Locke, 1968).
SELF-REGULATION THROUGH GOAL SETTING
215
Latham, 199Oa). Given sufficient ability, goal theory predicts a drop at
high goal difftculty levels only if there is a large decrease in goal commit-
ment. Performance levels out, of course, when the limits of ability are
reached. The drop that may occur with hard goals on complex
tasks
is
discussed at length below.
In an interesting study, Earley and Erez (1990) found that goals and
specific norms function similarly in influencing performance. The norms
were communicated as the normal or average performance of other peo-
ple. If individuals were given specific goals and then different norms a
week later, the latter regulated subsequent behavior. The converse oc-
curred when information on norms preceded the assignment of goals.
Meyer and Gellatly (1988) found that goals and norms affected each other
and performance.
A second consistent finding pertaining to goal content is that specific
and challenging or difficult goals lead to a higher level of performance
than vague but challenging goals such as “do your best,” vague but
unchallenging goals, or the setting of no goals. The specific, difftcult vs do
best goal comparison has been a primary focus of study in goal setting
research. The consistent superiority of the former is attributed to the fact
that vague goals are compatible with many different outcomes, including
ones that are lower than the person’s actual best. For example, Kernan
and Lord (1989) found that individuals with no specific goals generally
evaluated their performance more positively than those with specific,
hard goals in response to varying degrees of negative feedback. Moss-
holder (1980) obtained a similar finding. Similarly, Mento, Locke, and
Klein (1990) found that people with do best goals anticipated more satis-
faction from virtually every level of anticipated performance than did
people with specific, hard goals. Thus maximum effort is not aroused
under a do best goal. This is because the ambiguity inherent in doing one’s
best allows people to give themselves the benefit of the doubt in evalu-
ating their performance. From the standpoint of self-regulation, a specific
hard goal clarifies for the person what constitutes effective performance.
The person is no longer able to interpret a wide range of performance
levels as indicative of excellent performance.
Enumerative reviews of the literature (Latham & Yukl, 1975; Locke,
1968; Locke & Henne, 1986; Locke, Shaw, Saari, & Latham, 1981) found
strong support for these first two sets of findings. These reviews also have
been corroborated by meta-analysis (e.g., Mento, Steel, & Karren, 1987;
Tubbs, 1986).
Evidence for the generalizability of the finding is substantial as well.
For example, Latham and Lee (1986) found that the results generalize
across laboratory and field settings, quantity and quality criteria, soft and
hard criteria, and individual and group goals. Goal setting experiments
216
LATHAM AND LOCKE
have been conducted with 88 different tasks including bargaining, driving,
faculty research, health promoting behaviors, logging, maintenance and
technical work, managerial work, management training, and safety. In
reviewing this literature, Locke and Latham (199Oa) found that although
more total studies have been done in the laboratory than in field settings
(239 vs 156), a greater variety of tasks have been used in field than in
laboratory settings (53 vs 35). These data make clear that laboratory
findings regarding goal setting generalize very well to field settings. The
total number of subjects used in the goal setting studies reviewed by
Locke and Latham (1990a) was nearly 40,000. These people included
males, females, blacks, whites, managers, students, engineers and scien-
tists, and college professors. While the overwhelming number of these
studies were conducted in the United States and Canada, significant find-
ings have been obtained in Australia, the Caribbean, England, Germany,
Israel, and Japan. Thus it would appear that goal setting theory is appli-
cable across cultures.
An exception to the usual findings was obtained by Mitchell and Silver
(1990). They found no difference between specific, hard vs do best group
goals on a task that involved building a tower with blocks. However, the
trials were only 15 s in duration, a very short time for the effects of effort
and persistence to take effect, especially considering that the task entailed
careful balancing and coordination of block placement among members.
Mitchell and Silver also found that group goals were superior to individual
goals in their tower building task, since it required group cooperation.
Using a different task and a longer trial length, Larson and Schaumann
(1990) found that specific, hard goals led to better group performance than
do best goals on a cooperative task so long as the groups cooperated with
one another when it was necessary to do so.
A third content finding, but based on only two studies, is that goal
specificity as such (that is, divorced from difficulty) affects the variability
of performance (Locke, Chah, Harrison, 8z Lustgarten, 1989). Assuming
performance is controllable, people with very specific goals show less
variation in performance than people with vague goals. The reason is the
same as that above. Vague goals allow many possible outcomes as com-
pared with specific goals. For example, the goal to “take a walk” is
compatible with a walk of 10 feet or 10 miles, whereas a goal of “walk one
mile” is explicit for the goal setter.
The second attribute of goals that has been studied extensively is that
of
intensify.
Intensity is a broad term referring to the scope, clarity,
mental effort, etc., involved in a mental process (Rand, 1990). For exam-
ple, in their study of goal intensity, Gollwitzer, Heckhausen, and Ra-
tajczak (1990) found that subjects who thought most intensely and com-
prehensively about how to solve a problem (which involved attaining a
SELF-REGULATION THROUGH GOAL SETTING
217
personal goal) were most likely to become committed to solving it and,
more importantly, were most likely to take action to solve it.
The major aspect of goal intensity that has been studied in depth is
commitment. Commitment refers to the degree to which the individual is
attached to the goal, considers it significant or important, is determined to
reach it, and keeps it in the face of setbacks and obstacles. It must be
stressed, however, that the feeling of commitment does not automatically
lead one to act in accordance with it. As Salancik (1977) noted, the uhi-
mate proof of goal commitment is the action taken to attain it which in
turn reflects the thinking (or lack thereof) which preceded it and the
choice to act on that thinking (see Binswanger, 1991).
Goal commitment can operate both as a direct causal factor and as a
moderator of performance. These effects are shown in Fig. 2. The direct
effect operates when goal difficulty is held constant. Observe in Fig. 2
that when goals are high, high commitment leads to better performance
than low commitment. This is because less committed people give up
their hard goals in favor of easier ones (Erez & Zidon, 1984). When goals
are low, on the other hand, high commitment may restrict performance
because committed people will be loathe to raise their goals, whereas
uncommitted people may set higher goals (perhaps because they want
additional challenge).
The moderator effect is shown by the slopes of the two curves. When
there is high goal commitment, there is a strong association between goals
Goal Level
FIG. 2. Main and interaction effects of goals and commitment. Reproduced, by permis-
sion of the publisher, from Locke and Latham (l!WOa).
218
LATHAM AND LOCKE
and performance; people are more likely to do what they say they will do.
However, when commitment is low, people do not perform in line with
their goals.
In the typical laboratory experiment as well as in many natural settings
in which people are rewarded for compliance, gaining goal commitment is
rarely a problem. In fact, Bassett (1979) viewed goal commitment as so
routine that he argued that a theory of goal rejection rather than of goal
commitment should be developed. Thus, it is not surprising that assigning
people goals, accompanied by a rationale, leads to as high a level of goal
commitment as having people participate in the setting of their goals.
Nevertheless, there has been considerable controversy in the literature
concerning the effectiveness of assigned versus participatively set goals in
achieving goal commitment and increasing performance on the part of
subordinates. A series of 11 studies by Latham and colleagues generally
showed little or no difference in the effectiveness of the two goal setting
methods. In contrast, several studies by Erez and her colleagues showed
that participatively set goals produced greater commitment than did as-
signed goals. To resolve this disagreement, Latham and Erez, with Locke
as mediator, jointly designed a series of 4 studies in which the effect of
methodological differences between the Latham and Erez studies were
systematically assessed (Latham, Erez, & Locke, 1988). The results re-
vealed that the main reason for the differences in their results was that
Erez assigned goals with curt, brief “tell” instructions, whereas Latham
assigned goals in a more supportive manner and provided a rationale for
them. This “tell and sell” style used by Latham was found to be just as
effective in increasing performance as was participation; and both styles
were significantly more effective than the “tell” style that had been used
by Erez.
More recently, Latham, Winters, and Locke (1991) have suggested that
the key benefits of participation are not due to motivation (e.g., goal
commitment) but rather to cognition (e.g., task strategy development).
Their study found that although participation enhanced a self-report mea-
sure of goal commitment, it was not sufficient to make a difference in
actual performance. In contrast, participation in developing effective task
strategies had substantial effects on performance through the mediating
effects of self-efficacy and the quality of the strategies which the subjects
developed and used.
Factors which have been found to enhance commitment fall into two
broad categories, namely, those which convince people that achieving the
goal is possible and those which convince them that achieving the goal is
important or appropriate (Klein, in press). The first class of factors raise
the individual’s expectancy of success or what Bandura (1982, 1986) has
termed self-efficacy. These include ability, experience, training, informa-
SELF-REGULATION THROUGH GOAL SETTING
219
tion about appropriate task strategies, past success, and internal attribu-
tions (e.g., Earley, 1986a; Hall 8c Foster, 1977; Locke, Frederick, Lee, &
Bobko, 1984; Silver & Greenhaus, 1983).
Managers can play an important role in facilitating goal commitment in
subordinates by persuading them that the goals are both attainable and
important. This can be done by managers asserting their legitimate au-
thority, conveying normative information, showing that the goals provide
opportunities for self-improvement, challenging people to show what they
can do, being physically present at the work site, being supportive and
trustworthy, providing a convincing rationale for the goal, exerting rea-
sonable pressure for performance, being knowledgeable about the task
and job, and serving as a role model for the behavior they desire in the
subordinate (e.g., Earley, 1986b; Likert, 1%7; Mento et al., 1990; Ronan,
Latham, & Kinne, 1973; Podsadoff & Fahr, 1989). For a complete liter-
ature review see Locke, Latham, and Erez (1988) and Locke and Latham
(199Oa).
Goals that are assigned by legitimate authority figures typically influ-
ence peoples’ personal goals. Instructions to try for a certain goal even
carry over to later trials in which people are free to choose whatever goals
they want to attain (Locke, Frederick, Buckner, & Bobko, 1984; Locke er
al.,
1984). These findings are in alignment with Dember (1975), who, after
examining the literature on the cognitive aspects of motivation, concluded
that in certain settings being asked to do something is tantamount to being
motivated to do it. A similar argument has been made by Salancik (1977).
He stated that assigned goals lead to goal commitment because listening
to the assignment without objection is in itself a form of consent. More-
over, assigning the goal implies that the recipient is capable of attaining it
which in turn increases the person’s self-efficacy regarding the task.
It should not be concluded from the above that the persuasive requests
of authority figures compel commitment. Commitment is still a choice
process; it is often easy for the manager to obtain precisely because the
goal assignment
is
appraised as legitimate by the subordinate.
Peers can influence goal commitment by conveying normative informa-
tion, by persuasion, and by serving as role models (Earley & Kanfer,
1985). In addition, they can generate competition.
Agreeing publicly to strive for a goal can also enhance commitment as
compared with agreeing to it only in private (Hollenbeck, Williams, &
Klein, 1989). Finally, rewards can affect goal commitment, but the man-
ner in which these operate is not fully understood. It appears that large
rewards are generally more effective than small ones in this regard; but
rewards also interact with goal difficulty. Rewards offered for moderate
or easy goals appear to raise commitment to those goals but to lower
commitment to impossible goals, perhaps because people resent being
220
LATHAM AND LOCKE
enticed by a bonus which they cannot attain (Mowen, Middlemist, &
Luther, 1981; Wright, 1989). Thus bonuses for goal achievement may only
be effective if the goals are, in fact, reachable (e.g., moderate). Under
piece-rate payment, goals operate in the usual way, with difficult goals
leading to the highest level of performance. This is because under piece-
rate systems people are paid for performance and not for goal attainment
as such (Locke & Latham, 199Oa).
In an innovative study, Earley, Shalley, and Northcraft (in press) found
that commitment/rejection processing time was longer for moderately dif-
ficult goals than for easy or hard goals. When subjects were given the
choice, impossible goals were hastily rejected, whereas easy goals were
readily accepted. In contrast, subjects required more thought to make a
decision regarding commitment to moderate goals.
Goal Choice
The factors that affect goal choice are similar to those that affect goal
commitment. The probability of choosing a given goal is increased if the
individual thinks that it can be attained either because of ability or past
success. People with high self-efficacy are more likely to choose difficult
(high) goals than those with low self-efficacy (Locke et al., 1984).
Choice is also affected by the person’s belief that a given goal is ap-
propriate or desirable. This can occur when a person is provided with
normative information (Meyer & Gellatly, 1988), role models (Rakestraw
& Weiss, 1981), competition (Mueller, 1983), or pressure (Andrews &
Fart-is, 1972). However, the most direct method of influencing choice is
simply for an authority figure to assign the goals. Not only do subordi-
nates usually consider goal assignment to be legitimate, as noted earlier,
but authority figures usually have the power to reward and punish em-
ployees for accepting or rejecting the assigned goals. The correlation
between assigned and (subsequently) self-set goals is around SO, indicat-
ing that goal assignment does affect choice although it obviously does not
totally determine choice (Locke & Latham, 199Oa). In real life, choosing
high goals, if they lead to high performance, is more likely to be rewarded
than choosing low goals which lead to low performance and reward
(Mento et al., 1990).
Goals, Self-Efficacy, and Performance
We have shown that personal goals affect performance and that as-
signed goals influence personal goals. It remains to integrate these two
concepts with that of self-efficacy. Self-efficacy, a key concept in Bandu-
ra’s (1986) social-cognitive theory, refers to task-specific self-confidence.
It is broader in meaning than effort-performance expectancy in expec-
tancy theory in that self-efficacy includes all factors that could lead one to
SELF-REGULATION THROUGH GOAL SETTING
221
perform well at a task (e.g., adaptability, creativity, resourcefulness, per-
ceived capacity to orchestrate complex action sequences). Self-efficacy is
measured by asking subjects whether they believe thay can attain each of
a graded series of performance levels (self-efficacy magnitude) and by
asking them to rate their degree of confidence in attaining each level
(self-efficacy strength). It should be noted that some studies of expectan-
cy theory have measured performance expectancy (E,) in a similar way
(Ilgen, Nebeker, & Pritchard, 1981).
It has been shown consistently that self-efficiency has powerful, direct
effects on performance (Bandura, 1986). This finding holds when goals
are manipulated as well. Thus both goals and self-efficacy have direct,
independent effects on performance. In addition to affecting performance
directly, self-efficacy can affect it indirectly by affecting personal goal
choice and commitment to assigned goals. Finally, assigning goals influ-
ences self-efficacy in that people who are assigned challenging goals are
more likely to have high self-efficacy than those who are assigned low
goals since assigning high goals is in itself an expression of confidence
(Salancik, 1977).
The above relationships are summarized in Fig. 3. We have added
ability to this figure, because it has been found that it has independent
effects on both self-efficacy and performance (Locke
et al., 1984).
A
recent series of studies by Earley and Lituchy (in press) showed consid-
erable support for the model in Fig. 3. In two of the three studies, how-
ever, self-efficacy, while showing a significant first order
r,
did not add a
significant increment to the performance relationship beyond that pro-
vided by goals. In one of these studies (grade performance), the self-
efficacy correlation was approximately the same magnitude as that found
by Wood and Locke (1987) who had used the same task. However, Wood
and Locke’s self-efficacy scale was more elaborate and they used a larger
number of subjects. In their studies, the self-efficacy increment was sig-
nificant. There is a possible explanation for the null result in the third
study.
The task was complex and the self-efficacy measure was taken
after only two practice trials. A measure of self-efficacy may not be
meaningful this early in the learning process on such a task.
Ability --+ Self-Efficacy
AsI. x 1
Goal - Personal Goal
Performance
FIG.
3. Relation of ability, self-effkacy, goals, and performance
222
LATHAM AND LOCKE
Keman and Lord (1990) claimed that expectancies do not affect per-
formance on single goal tasks; however, they measured expectancy of
goal achievement rather than expectancy of attaining each of a number of
performance levels. Locke, Motowidlo, and Bobko (1986) have shown
that the former is a poor method of measuring expectancy, because the
referent for answering the item is different for each goal. Using the latter
(performance-anchored) method, the results are significant and quite con-
sistent (Locke 8z Latham, 199Oa).
In contrast to self-efficacy and personal goal level, we have not found
subjective goal difficulty to be useful in predicting performance (e.g.,
Yukl & Latham, 1978). The reason, we believe, is that measures of sub-
jective goal difficulty are confounded. On the one hand, they are posi-
tively associated with goal level (which would imply a positive association
with performance) and, on the other hand, they are negatively associated
with self-efficacy (which would imply a negative association with perfor-
mance). To the degree that the two associations cancel one another out,
the net ability to predict performance is small (Locke & Latham, 199Oa).
Goals, Valences, and Instrumentalities
Garland (1985) reported a negative relationship between goal level and
valence measured as expected or anticipated satisfaction with attaining
each of a number of performance levels. This finding was replicated by
Klein (in press) and in a series of eight studies by Mento et al. (N90). The
explanation for this finding is that goals are at the same time targets to
shoot for and standards for evaluating one’s performance (Bandura,
1986). This is shown graphically in Fig. 4. If one views one’s goals as
minimally acceptable levels of performance (Locke & Bryan, 1%8), we
can see that a person with low goals will be satisfied with reaching a low
level of performance and thus even more satisfied with attaining more
than this minimal level. A person with high goals, on the other hand, will
be minimally satisfied only with reaching the high goal and thus will be
quite dissatisfied with reaching the low goal. The person with moderate
goals will be between the other two. Thus goals affect the “calibration”
of the satisfaction scale, raising it when the goals are high and lowering it
when the goals are low. Self-satisfaction, therefore, is harder to attain
when goals are hard than when they are easy.
It might be assumed from this that people, therefore, should set only
low goals in life because that would produce more satisfaction with less
effort. However, there is another set of factors involved in choosing a
goal, As noted above, in the real world, additional rewards typically come
to the person who sets and achieves high rather than low goals. Thus high
goals are more instrumental in gaining practical as well as psychological
benefits than are low goals. Mento et al. (1990), for example, asked MBA
SELF-REGULATION THROUGH GOAL SETTING
223
Low Medium High
Performance Level Attained
FIG.
4.
Idealized
valence functions for subjects at three goal levels. Reproduced, by
permission of the publisher, from Locke and Latham (199Oa).
students what the benefits would be to them of getting an A vs a B vs a
C average. Their responses indicated that the higher the grade point av-
erage, the greater the anticipated benefits with respect to personal pride,
school, future job, and life outcomes.
Figure 5 plots both the mean anticipated satisfaction and instrumental-
ity with attaining each GPA. Observe that valence and instrumentality are
related to goal level in opposite directions. The results for pride suggest
how these seemingly contradictory results can be conceptually inte-
grated. People who make pride contingent upon attaining high goals will
be more motivated to attain them than to attain easy goals if their “higher
level” purpose is to attain self-satisfaction. Such people will have to
achieve more to be satisfied than a person who sets low goals, and com-
mitment to high goals is reinforced by additional practical benefits. The
desire for such benefits also may be tied to higher level values such as life
or career ambition and their associated outcomes.
In view of the above, it could be asked why everyone does not set high
goals. One answer is suggested in a previous section of this article. People
consider not just what they want when setting goals, but what they think
they can attain. People with low self-efficacy are unlikely to pursue goals
beyond their perceived capability. Second, striving for high goals has
costs with respect to effort, time, and other values. Thus goal choice
reflects an integration of two types of considerations: those pertaining to
224
LATHAM AND LOCKE
low medium
Goal or Perlormence Level
high
FIG. 5. The relation of valence and instrumentality to goal level.
what is possible and those pertaining to what, among the total array of
possibilities, one wants.
Goals and Feedback
Few concepts in psychology have been written about more uncritically
and incorrectly than that of feedback. In organizational settings the aph-
orism “what gets measured gets done” describes cogently the positive
halo surrounding feedback. Actually, feedback is only information, that
is, data, and as such has no necessary consequences at all. Like any fact,
its effect on action depends on how it is appraised and what decisions are
subsequently made with respect to it. Studies of the effects of feedback
typically show positive effects (Kopelman, 1986), but this is because peo-
ple often set improvement goals when given information about their past
performance. The only way to isolate the effects of feedback as such is to
give it in such a form that it cannot be used to set goals (e.g., vary the
length of each work period so that the subjects cannot directly compare
their performance from one trial to the next). When this is done, feedback
has no motivational effect on performance (Locke & Latham, 199Oa).
Even more intriguingly, a field experiment showed that even when engi-
neers and scientists were urged to do their best, their subsequent perfor-
mance was not significantly different from that of a control group. This
SELF-REGULATION THROUGH GOAL SETTING
225
occurred despite the fact that they received the same amount of feedback
as those people who were in the specific goal conditions (Latham, Mitch-
ell, & Dossett, 1978). Feedback that does not lead to the setting of and
commitment to specific difficult goals does not increase motivation to
increase one’s performance. Figure 6 from Locke and Bryan (1969a) il-
lustrates this point. Goal subjects in this study did have feedback about
their progress in relation to goals, but feedback subjects did not have
goals. Feedback alone did not affect performance. Thus with respect to
feedback as a motivator, goal setting is a
mediator
(cause) of its effects on
performance.
This relationship is easiest to envision when considering the case where
the individual receives multiple types of information. In such cases, the
individual cannot act, at a given time, on all of it and thus must select
which feedback elements to attend to and act upon. Goals single out for
attention one or more elements by providing a standard indicating
whether the feedback is good or bad; the elements with “value
significance” will be those accompanied by goals which serve as stan-
dards of evaluation (e.g., Nemeroff & Cosentino, 1979). In real-life, of
course, people are bombarded with information of every sort, but they act
only in response to a small segment of it, namely that segment which they
decide is relevant to their own life interests and goals.
On the other side of the same coin, goal setting is not very effective
without feedback (Erez, 1977). Thus feedback
moderates
the effect of
goals on performance.
Integrating the above results leads to the conclusion that goals and
‘2 1
Hard
GO.9
124
No KR
KR
Trial No.
Pw 1 2 3 4
Trial No.
FIG.
6. The effects of goals vs feedback on performance (from Locke & Bryan, 1%9a)
226
LATHAM AND LOCKE
feedback together are more effective in motivating high performance or
performance improvement than either one separately. Table 1 summa-
rizes the results of 33 studies compiled by Locke and Latham (199Oa)
which compared the effects of goals plus feedback, versus either one
alone. Nearly all of these studies support the hypothesis.
It remains to discuss the actual role played by each component and the
mechanisms by which performance improvement occurs. The goal is the
object or outcome one is aiming for as well as the standard by which one
evaluates one’s performance. Feedback provides information to the indi-
vidual as to the degree to which the standard is being met. If performance
meets or exceeds the standard, performance is typically maintained (al-
though eventually the goal may be raised). If performance falls below the
standard, subsequent improvement will occur to the degree that: (a) the
individual is dissatisfied with that level of performance and, more impor-
tantly, expects to be dissatisfied with it in the future; (b) the individual has
high self-efficacy, that is, confidence in her ability to improve; and (c) the
individual sets a goal to improve over her past performance. The joint
effect of these three factors is shown in Fig. 7, based on research by
Bandura and Cervone (1986).
Positive feedback normally raises self-efficacy, but it should not be
concluded from this that such feedback always enhances performance.
Such feedback tells one that one’s performance is “ok” and thus yields
little incentive to improve (Matsui, Okada, & Inoshita, 1983). Bandura
and Jourden (1990) found that providing subjects with normative infor-
mation showing their performance to be consistently superior to that of
their peers led to the setting of lower personal goals than was the case of
subjects told at first that their performance was inferior to that of other
subjects and later that it was matching and then surpassing the perfor-
mance of others. The latter, “progressive mastery” group outperformed
the “superior” group on the task.
Thus the key to performance improvement, as noted, seems to be that
the person be dissatisfied with his or her present performance and (or will
TABLE 1
GOALSPLUSFEEDBACKVSEITHERONEALONE~
Goals plus feedback:
vs goals only
vs feedback only
G&F> G&Fc
G or F only G or F only
17 1
20 2
37 3
o Based on Locke & Latham (MOa, Fig. 8-4). In this table contingent results are classi-
fied as failures rather than as “half” successes.
SELF-REGULATION THROUGH GOAL SETTING
227
Low Self-efficacy High Sell-efficacy
Low Goals Mixed High Goals
Satisfied Combinations Dissatisfied
FIG. 7. Joint effects on performance improvement of self-efficacy, goals, and anticipated
satisfaction. Reproduced, by permission of the publisher, from Locke and Latham (1990a).
be so in the future) yet confident that performance can be improved, thus
leading to the setting of goals above the level of previous performance.
Goal Mechanisms
There are at least three attributes of motivated action, namely direc-
tion, intensity, and duration. These are precisely the mediators or causal
mechanisms by which goals regulate performance. First, a goal
directs
activity toward actions which are relevant to it at the expense of actions
which are not goal-relevant. In prose learning, for example, giving readers
learning objectives leads them to pay more attention to content which is
relevant to those objectives and less attention to the remainder (Rothkopf
& Billington, 1979). The same selective function is revealed in the mul-
tiple feedback situations discussed above in which goals single out from
an array of information those fed back scores to be acted upon (e.g.,
Locke & Bryan, 1%9b). Further, a specific goal can affect the manner in
which information is processed (Cohen & Ebbeson, 1979). Another as-
pect of the direction of action is the automatic arousal of previously
acquired skills which are perceived as relevant to goal accomplishment.
228
LATHAM AND LOCKE
Second, a specific goal regulates effort or energy expenditure (i.e.,
intensity) in that people adjust their effort to the difIiculty level of the task
or goal. This is the core explanation of the goal difftculty effect. The
positive effect of goal difficulty on effort holds when effort is measured in
terms of physical exertion (Bandura & Cervone, 1983), rate of work
(Bryan & Locke, 1967), subjective ratings (Brickner & Bukatko, 1987-
Study I), effort ratings by third parties (Terborg & Miller, 1978), and
physiological indicators (Sales, 1970).
Third, a goal affects persistence (i.e., duration) in situations where
there are no time limits imposed on people. When time limits are imposed,
difficult goals induce people to work faster or harder. Without time limits,
such goals induce people to work longer (LaPorte & Nath, 1976).
Whether the longer duration of work will be accompanied by a faster rate
of work is problematic in these situations because there is a natural but
not inevitable trade-off between intensity and duration. In some cases,
difftcult goals lead to more effort per unit of time and more prolonged
effort (Cannon-Bowers & Levine, 1988; Latham & Locke, 1975) whereas
in other cases people adjust their effort to the time allowed (Bryan &
Locke, 1967).
An aspect of.persistence is tenacity-the refusal to quit, despite obsta-
cles, until the goal is reached. Certainly commitment is one goal attribute
which would affect tenacity, although this has rarely been studied. Goal
difficulty, however, also affects it. Huber and Neale (1987), for example,
found that subjects assigned hard goals in a bargaining task bargained
“harder,” that is, for better deals, than subjects assigned do best or easy
goals. Other studies suggest that the better deals are the result of the
additional time spent bargaining (e.g., Neale, Northcraft, & Earley,
1987).
The above three mechanisms are relatively direct and automatic con-
sequences of goal-directed activity. However, there are times when these
three mechanisms, including the use of previously learned skills, are in-
sufficient to attain a goal. In such circumstances, the individual must
develop or discover new strategies if the goal is to be achieved. This is
especially the case on complex tasks where effort and attention are of
limited usefulness in themselves if the individual is not using an appro-
priate plan or strategy.
The following summarizes what is known or hypothesized to date re-
garding the relationship of goals and task strategies:
(1) When given specific, challenging goals, people spontaneously for-
mulate plans and task strategies to help reach the goals (Latham &
Baldes, 1975; Latham & Saari 1982). Specific, challenging goals stimulate
more planning in general (Earley, Wojnaroski, & Prest, 1987; Weldon,
Martzke, & Pradhan, 1990) and often higher quality planning (Shapiro &
SELF-REGULATION THROUGH GOAL SETTING
229
Hollenbeck, 1990; Smith, Locke, & Barry, 1990) than is the case with do
best goals. Group goal setting may require planning for the explicit pur-
pose of coordinating member activities so that the goal can be achieved.
When such planning is needed, goal setting is effective if coordinated
planning actually occurs (Larson & Schaumann, 1990).
(2) When a difficult, quantity goal is assigned, people may lower work
quality as an implicit strategy to attain it (Bavelas & Lee, 1978). Deem-
phasizing quality for quantity under quantity goals is most likely when
people are not highly confident of their task ability (Erez, 1990). To en-
sure performance quality, of course, goals must be set for quality.
(3) On complex tasks:
(a) Goals are more strongly related to performance when subjects
utilize suitable task strategies than when they do not. This is illustrated in
Table 2, based on Chesney and Locke (1991). Challenging goals increase
the likelihood that known strategies will be used (Earley, Lee, & Lituchy,
1989).
(b) Strategies tend to be more strongly related to performance than
are specific goals (Chesney & Locke, 1991).
(c) There may be a time lag before the goals affect performance; this
lag may occur because individuals are learning which task strategies are
effective (Smith et
al.,
1990; Weldon et
al.,
1989).
(4) Trying for specific, challenging goals may actually hurt performance
in certain circumstances. The circumstances appear to include a combi-
nation of the following;
(a) In the early stages of learning a new, complex task when no
strategy training is given (Kanfer & Ackerman, 1989). Once initial learn-
ing has taken place, however, the introduction of specific, challenging
goals can facilitate performance (Kanfer & Ackerman, 1989);
TABLE 2
GOAL-PERFORMANCE CORRELATIONS AS A FUNCTION OF DEGREE OF USE OF
SUITABLE STRATEGY
Time High use of most
period” suitable strategy Low use of most
suitable strategy
1 .47** .49**
2 .69** .27*
3 .69** -.22 __--.-
Note. From Chesney & Locke (1991).
L1 Computer simulation business game was divided into three time periods:
1 (Weeks l-3),
2 (Weeks &7), and 3 (Weeks g-10). Students played game once per week; scores were
cumulative.
* p < .05.
**p-c .Ol.
230
LATHAM AND LOCKE
(b) When the task is heuristic and no strategy training is given (Hu-
ber, 1985; Earley, Connolly, & Ekegren, 1989);
(c) When there is pressure to perform well immediately and no
strategy training is given.
Wood and Locke (1990) have discussed the issue of goal and strategies
on complex tasks in some detail from a theoretical perspective (see also
Locke 8z Latham, 199Oa, Chap. 13).
Other Moderators
In addition to commitment, feedback, and task complexity, there are at
least two additional moderators of the goal-performance relationship.
The first is ability. Battle (1966) found that the goal-performance rela-
tionship is somewhat stronger among high than among low ability sub-
jects, especially insofar as the goals are moderate to challenging (Battle,
1966). However, Kanfer and Ackerman (1989) found that goal effects can
be stronger for low than high ability subjects on complex tasks when
implemented after some initial learning has taken place. Motivation in the
absence of ability is unlikely to affect performance positively unless peo-
ple are working below capacity. And obviously, ability limits the goal-
performance relationship at very high (i.e., impossible) goal levels be-
cause such goals exceed the reach of virtually all people (Locke, 1982).
An intriguing study by Wood and Bandura (1989) showed that beliefs
regarding one’s intellectual ability affect performance. In a management
simulation exercise, the people who viewed their ability as an acquirable
skill that could be enhanced through practice set challenging goals en-
gaged in effective problem solving strategies, and subsequently attained
high performance. Those people who viewed intellectual ability as a more
or less fixed capacity viewed their errors as indicative of the fact that they
were indeed not intelligent. Consequently, they set low goals, their prob-
lem solving strategies deteriorated, and they subsequently performed
poorly.
The second additional moderator involves situational constraints. Pe-
ters, Chassie, Lindholm, O’Connor, and Kline (1982) found that goal level
was significantly associated with performance when situational con-
straints were low rather than high. If the situation can be managed, of
course, high goals could motivate a person to overcome obstacles, espe-
cially if the person has high commitment and self-efficacy. Nevertheless
there are limits to a person’s ability to change situations.
There is little evidence that factors such as race, age, education, gen-
der, or tenure moderate the goal-performance relationship. Nor is the
evidence clear with respect to such factors as personality and culture.
However, only a limited number of studies have been conducted on these
latter variables (Locke & Latham, 1990a).
SELF-REGULATION THROUGH GOAL SETTING
231
Goals and Affect
The basic model for understanding the relationship between goals and
affect comes from Locke’s (1976) satisfaction theory (based on Rand,
1964) which states that emotional responses are the result of automatic,
subconscious value appraisals (Locke & Latham, 199Oa).
As noted earlier, goals are at the same time the aim of action and a
standard by which people evaluate their performance. Goals are valued or
desired outcomes. Thus one would expect that the greater or more fre-
quent the degree of success experienced, the greater the degree of satis-
faction with performance. This is precisely what is found. Locke and
Latham (199Oa) found a mean correlation between degree of success and
satisfaction of .51 across 16 studies that reported such correlations.
Table 3 shows the results from a study reported in Locke, Cartledge,
and Knerr (1970) which involved striving for an end goal across a number
of trials. On any given trial, satisfaction with performance was a joint
function of (a) the goal-performance discrepancy for that trial and (b) the
perceived instrumentality of performance on that trial for attaining the
(overall) end goal. Discrepancy and instrumentality are themselves re-
lated, in that small, discrepancies are usually more instrumental for long
term success than are large ones.
The precise degree of satisfaction experienced in a given case is also
affected by other factors including the importance of the goal and causal
attributions of success. The more important the goal, the stronger the
positive affect experienced after success, and the stronger the negative
affect experienced after failure (Locke, 1976).
There is evidence that goals may also increase task interest and reduce
boredom, at least on those tasks that are initially boring (Latham &
Kinne, 1974; Locke & Bryan, 1967; Mossholder, 1980). Goals may also
reduce role conflict and ambiguity.
Field experiments on goal setting have not shown consistent effects on
satisfaction, probably because goals can lead to disvalued as well as val-
TABLE 3
RELATIONSHIPOFGOAL-PERFORMANCEDISCREPANCY AND INSTRUMENTALITY
TO SATISFACTION
Instrumentality Satisfaction
Goal-performance discrepancy
Instrumentality in reaching
end goal
- .57* - .61*
.72**
Note. From Locke, Carledge,
&
Knerr (1970,
Study 2).
* p < .05.
** p i
.Ol.
232
LATHAM AND LOCKE
ued consequences (e.g., stress, failure, punishment, job insecurity, pres-
sure, conflict). The typical field experiment result is no change in satis-
faction (e.g., Latham et al., 1978) possibly because any positive conse-
quences (e.g., role clarity, satisfaction with success) were offset by
negative consequences.
Correlational field studies typically show positive associations between
various positive attributes or concomitants of goal setting programs (e.g.,
clarity, participation, supervisory supportiveness, feedback, rewards for
goal attainment, communication) and satisfaction with the job or some
aspects of it. In contrast, various negative attributes of goals or goal
setting programs (e.g., stress, failure, overload, punishment, conflict) are
negatively associated with satisfaction (Lee, Bobko, Earley, & Locke, in
press). The reason for the clear-cut results of the correlational in contrast
to the experimental field studies is that these former studies distinguished
valued from disvalued attributes of goal programs. The results make it
clear that such programs can have very different affective consequences
depending on how they are implemented.
Since a major factor causing satisfaction with goal setting is goal suc-
cess, a certain dilemma is posed for applied goal setting programs. Since
goal success is increasingly more frequent as goals become easier, it
means that the greatest degree of satisfaction is experienced when goals
are easy. On the other hand, it was noted earlier that the highest degree
of performance was attained when goals were difficult, that is, hard to
achieve. The dilemma, then, is how to balance the two outcomes. Since
satisfaction is based on both internal and external rewards which are also
typically based on success, an associated dilemma is how to reward per-
formance under a goal setting program. If we maximize productivity, we
minimize satisfaction and rewards and vice versa. There are several pos-
sible solutions to this dilemma:
(1) Satisfice by setting moderate goals and rewarding success, so that
the net total of satisfaction and productivity is maximized.
(2) Give credit for partial goal attainment; that is, give credit and re-
wards for performance rather than for success as such.
(3) Follow the Japanese principle of Kuizen or constant improvement
(Imai, 1986). Make goals reachable at any given time, but strive for con-
tinual increments above this initial level by constantly raising the goals by
small amounts. (This does not necessarily imply working harder; it can
also be done by “working smarter”.)
(4) Use multilevel goal and reward structures, so that some reward is
provided for reaching a minimum goal, more is provided for reaching a
more challenging goal, and maximum reward is given for achieving
“stretch” goals.
Each of the above procedures has its pro’s and con’s. There is no
SELF-REGULATION THROUGH GOAL SETTING
233
“right” way to decide among them (and among additional, not-yet-
conceived structures) without further experimental study.
The High Performance Cycle
The integrated goal setting model has been described elsewhere as the
high performance cycle (Locke & Latham, 199Oa, 199Ob, 1990~). The
model starts with high challenge in the form of specific, difftcult goals. If
there is commitment to these goals, adequate feedback, high self-efficacy
(and ability), and suitable task strategies, high performance will result. If
high performance leads to desired rewards (including self-rewards in the
form of self-satisfaction) high satisfaction will result. Job satisfaction is, in
turn, highly associated with commitment to the job
(r
= .64, based on 11
studies summarized in Locke & Latham, 199Oa), although the causal
relationship between these two variables is not definitively established.
High commitment in turn is associated with an increased propensity to
stay on the job (Mobley, 1982; Mowday, Porter, & Steers, 1982). People
who are satisfied and stay on the job are then ready and willing to accept
new challenges. Thus the cycle repeats itself. Deviations from the re-
quirements of the cycle (e.g., low challenge, dissatisfaction) lead to a low
performance cycle.
SELF-REGULATION
Self-regulation is implicit in goal setting theory because, as noted
throughout this paper, the setting of goals and their translation into action
is a volitional process. However, most goal setting experiments have not
emphasized self-regulation explicitly because goals were assigned in or-
der to ensure sufficient variation in goal level and type.
A review of the literature (Locke & Latham, 1990a) revealed that self-
set goals are as effective as, but not more effective in increasing perfor-
mance than, goals that are assigned or are set participatively. This finding
is the basis for training people in effective self-regulation skills.
Self-regulation occurs through goal setting because the setting of a goal
is first and foremost a discrepancy-inducing process (Locke, in press). To
quote Bandura (1988, p. 47):
Human self-motivation relies on
discrepancy production
as well as
discrepancy
reduction.
It requires
feedforward
control as well as
feedback
control. People
initially motivate themselves through feedforward control by adopting performance
standards that create a state of disequilibrium and then mobilizing their effort on
the basis of anticipatory estimation. Feedback control comes into play in subse-
quent adjustments of effort expenditure to achieve desired results. After people
attain the standard they have been pursuing, they generally set a higher standard
for themselves. The adoption of further challenges creates new motivating discrep-
ancies to be mastered. Similarly, surpassing a standard is more likely to raise
aspiration than to lower subsequent performance to conform to the surpassed
234
LATHAM AND LOCKE
standard. Self-motivation thus involves a dual cyclic process of disequilibrating
discrepancy production followed by equilibrating discrepancy reduction.
In short, goal setting facilitates self-regulation in that the goal defines
for the person what constitutes an acceptable level of performance. Ac-
tions that fall short of a described goal level result in a negative perfor-
mance evaluation. Such negative appraisals usually lead to problem solv-
ing and subsequent action plans for eliminating the source of the dissat-
isfaction, such as improving subsequent performance. Actions that attain
or exceed desired ends lead to a positive performance evaluation. If a
positive appraisal is followed by the anticipation that subsequent attain-
ment of the same goal will lead to a neutral or negative appraisal, the
person is likely to set a higher goal. Thus the self-regulatory behavior
sequence is one that aligns the person to current and future behaviors
with some criterion that permits the person to evaluate progress toward a
specific goal (F. Kanfer, 1986).
Once the person chooses to strive for a goal, the three direct mecha-
nisms of effort, persistence, and direction, described earlier, are brought
into play more or less automatically. Where the task is hard, not because
it requires a great deal of effort, but rather because it requires a high level
of the person’s knowledge and skill, training in self-regulation emphasizes
the discovery or learning of appropriate task strategies. Thus the task
strategies are the indirect result of the goal or goals that were set.
Teaching people self-regulatory skills is based in large part on the work
of F. Kanfer, a clinical psychologist. Self-control situations are defined as
situations in which a person is faced with the task of engaging in or
stopping behaviors that are initially less motivated, less enjoyable, and
may be less skilled than the automatically processed acts that are carried
out easily from moment to moment (F. Kanfer, 1986). In this training
program, each person sets a goal that is difficult but attainable in order to
minimize the probability of failure (Kanfer & Gaelick, 1986). They engage
in self-control by making decisions and generating their own personal
incentives. Usually they must overcome concurrent social or internal,
aversive cues to get the