Article 7 - What we should know about motivation theories - Tin học đại cương Học Viện Phụ Nữ Việt Nam

姝 Academy of Management Review
2004, Vol. 29, No. 3, 388–403.
WHAT SHOULD WE DO ABOUT MOTIVATION
THEORY? SIX RECOMMENDATIONS FOR THE TWENTY-FIRST CENTURY EDWIN A. LOCKE University of Maryland GARY P. LATHAM University of Toronto
We present six recommendations for building theories of work motivation that are
more valid, more complete, broader in scope, and more useful to practitioners than
existing theories. (1) Integrate extant theories by using existing meta-analyses to
build a megatheory of work motivation. (2) Create a boundaryless science of work
motivation. (3) Study the various types of relationships that could hold between
general (trait) and situationally specific motivation. (4) Study subconscious as well as
conscious motivation. (5) Use introspection explicitly in theory building. (6) Acknowl-
edge the role of volition in human action when formulating theories.
The concept of motivation refers to internal
berg’s work by developing a model suggesting
factors that impel action and to external factors
the specific work characteristics and psycholog-
that can act as inducements to action. The three
ical processes that increase employee satisfac-
aspects of action that motivation can affect are
tion and the motivation to excel. All these theo-
direction (choice), intensity (effort), and duration
ries center on the issue of the organization’s
(persistence). Motivation can affect not only the
effect on the individual employee’s “cognitive
acquisition of people’s skills and abilities but growth.”
also how and to what extent they utilize their
Other theories and approaches have focused skills and abilities.
on specific psychological processes, as does
Work motivation has been of interest to
Vroom’s theory. Organizational behavior (OB)
industrial/organizational (I/O) psychologists at
modification (Luthans & Kreitner, 1975), which is
least since the 1930s, stimulated in large part by
not influential today, was derived from Skin-
the famous Hawthorne studies (Roethlisberger &
ner’s behavioristic philosophy that denied the
Dickson, 1939), which focused mainly on the ef-
importance of consciousness. This approach
fects of supervision, incentives, and working
stresses the automatic role of rewards and feed-
conditions. However, it was not until 1964 that
back on work motivation; however, these effects
Vroom made the first attempt to formulate an
are mediated by psychological processes such
overarching theory—namely, a hedonistic cal-
as goals and self-efficacy (Bandura, 1986; Locke,
culus called the “valence-instrumentality-
1977). Goal-setting theory (Locke & Latham, 2002)
expectancy model.” Theory building in the field
and control theory—a mechanistic combination
of work motivation, however, has typically been
of cybernetics and goal theory (Lord & Hanges,
more specialized than Vroom’s overarching
1987)—focus on the effects of conscious goals as model.
motivators of task performance. Attribution the-
Argyris (1957), for example, focused on the
ory’s (Weiner, 1986) emphasis is on ways that the
congruence between the individual’s needs and
attributions one makes about one’s own or oth-
organizational demands. Herzberg and col-
ers’ performance affect one’s subsequent
leagues (Herzberg, Mausner, & Snyderman, 1959)
choices and actions. Social-cognitive theory
focused primarily on sources of work satisfac-
(Bandura, 1986) is very broad in scope—its do-
tion and, within that domain, mainly on ways in
main is much wider than that of work motiva-
which the job could be designed to make the
tion— but Bandura’s core concept of self-efficacy
work itself enriching and challenging. Later,
has been found to have powerful motivational
Hackman and Oldham (1980) extended Herz-
effects on task performance (Bandura, 1997). 388 2004 Locke and Latham 389
Two work motivation theories have a social
When beginning to study the plethora of ex-
emphasis (although Bandura [1986] stresses the
isting work motivation theories, one’s reaction is
motivational effects of role modeling). Adams’
sometimes bewilderment at the enormous vari-
(1963) theory focuses on the motivational effects
ety of concepts and approaches. But, if one looks
of distributive justice, which is based on com-
closely, it is evident that, for the most part, these
parisons between the inputs and outcomes of
theories, though flawed and/or limited in vari-
oneself versus those of comparison others. More
ous respects (see Miner, 2002), do not so much
recently, scholars have extensively researched
contradict one another as focus on different as-
procedural justice (Greenberg, 2000), stressing
pects of the motivation process. Therefore, there
the important effect on employee satisfaction of
is now an urgent need to tie these theories and
the methods or processes by which organiza-
processes together into an overall model, inso-
tional decisions affecting employees are made. far as this is possible.
Personality-based approaches to motivation,
Locke (1997) made a preliminary attempt at
although in and out of fashion over the past
integrating theories of motivation in the work-
several decades, have always had some strong
place. The model, shown in Figure 1, begins
supporters. McClelland and his colleagues (e.g.,
with an employee’s needs, moves to acquired
McClelland & Winter, 1969) stressed the effect of
values and motives (including personality), then
subconscious motivation—specifically, need for
to goal choice, and thence to goals and self-
achievement— on economic growth. In recent
efficacy. The latter two variables constitute a
years the study of conscious, self-reported traits
“motivation hub” in that they are often the most
has become popular, especially traits such as
direct, conscious, motivational determinants of
conscientiousness, which is fairly consistently
performance. Performance is followed by out-
related to effective job performance (Barrick &
comes, and outcomes by emotional appraisals, Mount, 2000).
such as employee satisfaction and involvement,
All of the above theories have limitations.
that lead to a variety of possible subsequent
None of them are above criticism, and some
actions. (Job satisfaction, of course, may also
have dropped by the wayside in recent years,
affect performance; the precise causal relation-
yet most provide some useful insights into em-
ship between them is not fully known [see Judge,
ployee motivation. Thus, it is clear that the field
Thoreson, Bono, & Patton, 2001].) Job character-
of work motivation has not only progressed but
istics are shown as affecting satisfaction. The
has progressed in multiple directions over the
place where a specific theory applies is shown
last several decades. Nevertheless, our knowl-
by the dotted boxes. This is not a speculative
edge of the subject of work motivation is far from
model. Every connection but one—namely, the
complete. The issue, then, is where should we go
link from needs to values—is based on empiri- from here? cal research.
Our goal in this article is not to offer yet an-
A useful next step would entail identifying the
other theory of work motivation. Rather, our fo-
size or strength of the various relationships
cus is on metatheory—the process or processes
shown in Figure 1. This could be done by com-
through which we can build more valid, more
bining the results of all known meta-analyses
complete, and more practical theories. This pa-
relevant to each path in the model and would
per provides rationales for six categories of rec-
include calculating known mediation effects, as
ommendations for advancing knowledge and
well as known moderator effects. It would also
understanding of employee motivation in the
entail adding pathways based on theories for
twenty-first century. We provide examples of
which there is some empirical evidence but
specific types of studies that might be carried
which are not, as yet, included in the model (e.g.,
out relevant to each recommendation.
Kanfer & Ackerman’s [1989] resource allocation
theory and Weiner’s [1986] attribution theory).
The result could be the first motivation mega- SIX RECOMMENDATIONS
theory in the behavioral sciences derived from
combining different meta-analyses.
Recommendation 1: Use the results of
Using meta-analyses to build theory, which is
existing meta-analyses to integrate
called “mega-analysis,” was originally sug-
valid aspects of extant theories.
gested by Schmidt (1992). He and his colleagues 390
Academy of Management Review July a n tio a tiv Mo rk 1 Wo E f R o U l e IG d F Mo d te ra g te In n A 2004 Locke and Latham 391 e l w re re s. e e g a . e s m th o a a e sk th th ce e n n ice d ct sl , . ta f a o st d o fro is, a n ce tiv n o n n d o tim id m ju a m fle n t M rse io n n o a r o a u ss in e re h ct a ce a g e v s v tic , th g co a in , in a d ice o a s. issi u f rm f e ce d irn d ice st t st rse n s, o o o n n n o iv u c u d tisf rfo ju e e fa a p tio rm e s, e s p d ch l si co o e e lth n sa p ct st e e rce ra in a f m p n A ch e . b n d iv a ly P u n b o e d tio ffe rsi , b . a e , ith p jo o e e a d in ith ro n ie d s e cy w e p ce rce p th m ce w d ce n ct n e p io m d d u x a th rt, a p a su ct ro n e rt e ffe e a p co u ffica lv a st s n e te ffo rm e 1997, lly re in b fo o st n se e th p v e e ir ia , a tisf lf-e e tio d rfo . d d d n d e in st h e n b p e e n ci n e n S u T p e ca a g t . th cy m io a , a issa lv t l. m o rce ct d sp t d o e in n d im e n e n v u ip e 375–412, n n rn b d s a p a ire a ffica ice , e tp sh o b tu . d st n jo in u n s ,” tio a cl b itm s o m a w a h jo cy n ce ity lf-e ju io to o s is in o n x . y tio n se n l ct m d n tiv d th to itie n s a le ra a m se la tio e K o e p s io n x a g e ffica ich ct u ct re a m rm o le d n to h m ct ct d co st s sm a s p o tiv le We h f w ce n lf-e ffe rfo co ffe ce ffe sp m o t g o e a n w tici e a t. ct re lm a u t se cy p ro o sk , tisf n co a a m o u tio ro ir p rs, ffe n o p e in , e u d e n ct ta ce sa s, o a y rm ck Wh lth e th n n n th : w ce a ffica ) e itm ct e a rld a A x n ffe w a ce cy k o rfo rk ro e h in ). d ls e g a n m fa m is s. g s lici w n a lf-e u , (lo rm a o m r ta -p n ci n e l e Wo s u ics. a o se h it d re rfo n p ffica n a cy w D io e ro a ro th g lty co th s d n e e ’s o l o io re a , to ct lu e e m ct n . th a p s n lf-e ct re sh w n e a e a r m f e a , r a e e t o n v n ffica u o ffe lty ce ifficu te tio ith se th o fa tio n n e st a ls n d fo u w , ct th n sl a rk d a a ility s tio , l-e so a co le In o iza lf-e o l b e g p ffe le . a re in M tiv l re a n o w a co g ifficu rm a k ttrib iza n a p se lity d o a . a a n a o lo m ce g , t ., sa a f a ct l g t, n rg t. a a a m n e n p .g M u w n rfo n m n o re b : x lso e d o a e o a a ce e n ffe o e e io jo e rg t, d e a x th (e ca h p lly e o f tio x h , th a n rso a g a n ct n e d m e te r o n rm o p e rt e g , e itm o a th e s o r o “T d tio p s ct ci s te f lv n F ca o rfo s ce s, p , th in e lth e m . la lly e ce . d ffe n tisf o n o itm s F e se x u p e f e e a la n e d tiv a a m v e . p e o o n e re n sp io a m b d ly su ck h a n ci tio re ss e e co sa e in h m lu e to a sci o d ll-b y l e te s cy , u d ct n ld itte a n n d L e e e n th e ce sp ls, ce n a b e co u m v ra a o . w th ta ck u irn o o cy ite E n jo lu in e a a ttrib a n d d o -co rch d t e to d ic a , d ffica o b a e fa tisf s n sh n b ith b a n a ro io re a m lim 10, h m s V n e G d d e g a la w p n e sa ce ct l a e ffica a su se a . lity a . e e l. l w f in d n a n s t, o a n th s iv ct a d n u e ly o re a fe e ce lle io lity y h rta cy n a v g h o w a lly lf-e d b iv ice n y rd a tisf th lu ce e o a b a rce ffe n ct ig a m n fu se e t, lly in a a rro h rv a n p rso ls ch e e f n ce n v e ffica e ch a S t rm d d w l p , n is a rso o o o n its r e su is p l o u re t. tisf l a o h e e m m e a G ce cl ta ice io n a th n d o e irica le sci h s ’s T lf-e o rfo n d n P ct e S f sa . se s rm v p p e T st u o a . o se ik a in n e u re a d L g s. p s. , . a . n a ca ie m e n h a h n io . s M n ju itm n s ca , rfo e ism p th d rd n . n io e e a b n ce e s. ch t o io n io cy ct ism to a e n ss n io tisf m e ct ry th p n u A st a a b ct a n d s, w a ct tio io ct tiv sa m m a o a a ffe re rg a a e th d . a to ls a re a cce d ct u in b ita so (1994) tio l ilitie le o e rm u co tisf th t a ith n g e tisf a cy ffica ch to t. e e ism s s S a tisf rib Jo l d n l p n b a e e o cy t a w in n a tiv sa g m n ct ct rfo . t. n lim a e a ce o s n th th e lf-e a e n tisf sa ist n e l issa o s. itica d o f n x M im ffica w to to ffe d e se to ch ffe P io to e ra rlo d itm g e cr n m rie tio is o a ffica a sa tio e e . ct e f a a s s, m itm e rie , ic. s o cl y d d E ls s, l cy e v v o m m o n re u e is ts h a to e e o lity lity n n a m e a iza m e m e w m a o tiv h n o l in w a a a a ls. o l lv n m S ce fro o a ills th o T e a a G tisf lici lici o a . a co -th n m s n n g ics o fa e ill sk sci e M ct ls ls o ics. o lici ffica o v co tu l d b sh y n d s. n a a g d g o e sa p p rg rk p e a t d sk ir u in e ire rso rso o o ct p rs. rist in o d o si t rre so l su e e n o o g g to to e n th a o d e co cl s lu u a to a se p p to a ta is, ’s w g n e to to a rce g n n th n a q co y to r n ra ct n r e in s s, a o in ce v re t re d d b cy cy o e ce ce th n n n o s tin re t n e n n s a ra tio -p ct e ss s d n n a e s tio d a f io io io b e rie e tic, o a to e rtly e g th a a th o a a o o p s a n v te ie iza ct ct ct jo tiv ffe e d se d s iv a f s s tiv ffica ffica g iza m rm rm ch n a a a sse a a m ct st s A d ct p o e e ia n ls, n l a lts e itiv th le a cu e e s lu lu d te a a a rk o tisf tisf th n rn ly co ffe is w fo e je e lf-e lf-e rfo rfo tisf g s le d m b a n a a e ce e e ra o rg o e e rg su a a a f d l u f e o l N o w o V V m In S e l S o p st G o G P P Wo O re S S S o co lte e n e a te d o e fro b k rio r v o . e d d 1. 2. 3. 4. 5. . , 7. 8. 9. it u e ld o o d 10. 11. 12. 13. 14. 15. n p o a b rsi m o d v m 16a io d m e h m te n lim w ct n e u cu h ility o e a d e b e rin n o h h T a u T n R T to A h T h p r. 6. a • • • • • •T e ie R v 16. a lse E 392
Academy of Management Review July
used it on a small scale in the field of human
Weingart & Jehn, 2000, for some preliminary
resources management by tying together empir-
findings). Social loafing is another potent group
ical studies of the relationships among job ex-
motivation phenomenon that is not part of ex-
perience, ability, knowledge, and performance
tant work motivation theories (Karau & Wil-
on work samples, as well as in the workplace
liams, 2001). A separate megamodel might have
(Schmidt, Hunter, & Outerbridge, 1986). However,
to be constructed to explain team motivation.
a mega-analysis of extant work motivation the-
Motivation also should be studied within the
ories would be on a much wider scale and
realm of decision making. For example, Schnei-
would integrate an enormous amount of data
der and Lopes (1986) have argued that level of
into a comprehensible framework that would be
aspiration (i.e., goals) needs to be incorporated
useful to both theorists and practitioners. The
into prospect theory. Along this line, Knight,
model could be expanded, of course, as new
Durham, and Locke (2001) have found that goals discoveries were made.
affect the degree of risk people take when mak-
ing decisions. Personality theory has implica-
Recommendation 2: Create a bound-
tions for prospect theory as well. For example,
aryless science of work motivation.
those high in extroversion may assess risk quite
Jack Welch coined the term boundaryless or-
differently from those high in neuroticism.
ganization when he was CEO of General Elec-
Within the field of personality, an issue that
tric (GE), as a result of his frustration over
needs to be addressed is the extent to which
knowledge that was being ignored rather than
certain traits are stable aspects of the person
shared and embraced among the myriad divi-
versus readily manipulable motivational states.
sions of GE. Similar dysfunctional behavior had
For example, Dweck and her colleagues (e.g.,
been referred to within the Weyerhaeuser Com-
Dweck & Elliott, 1983) have argued that goal
pany as the “not invented here” mindset—a
orientation is a relatively stable disposition. Yet
mindset that prevented managers within one
there is a paucity of studies that have assessed
region of the company from building on the
its test-retest reliability (e.g., VandeWalle, Cron,
knowledge gained by managers in other re-
& Slocum, 2001). Moreover, the empirical re- gions.
search suggests that goal orientation is readily
This implies two things. First, work motivation
malleable. Dweck herself has even acknowl-
theory needs to be extended into and further
edged this in the field of educational psychol-
developed within areas other than isolated task
ogy (e.g., Elliott & Dweck, 1988; Mueller & Dweck,
performance settings. Second, motivation theo-
1998). In the OB field, Seijts, Latham, Tasa, and
rists should consider using concepts developed
Latham (in press) found that when people were
in fields outside OB and I/O psychology.
given do-your-best instructions, Dweck’s (1986)
For example, motivation could be studied fur-
predictions regarding the goal orientation trait
ther in the realm of team effectiveness. There
were supported. But when a specific difficult
are processes affecting teams that do not arise
learning goal was set, it masked the effect of
when the focus is on the individual’s motivation,
this trait. A learning goal, as is the case with an
such as the specific ways in which team mem-
outcome goal (Adler & Weiss, 1988), was shown
bers motivate and demotivate one another. For
to be a strong variable that mitigates the effects
instance, team members might encourage one
of this individual-difference variable (trait). Re-
another through building efficacy by means of
search is needed to see under what conditions
persuasion or the offering of useful ideas. They
situationally induced motives negate trait
might undermine one another through belittle- effects.
ment and insults. Extending motivation re-
Motivation theory can be better incorporated
search into the realm of teams would lead to the
into macrotheories, particularly organization
exploration of such issues as conflicts among
theory. For example, there is little doubt that
personalities, values, and/or goals that are not
degree of centralization and decentralization
yet a part of extant work motivation theories.
has motivational consequences, as appears to
Although team cohesion has been studied, less
be the case with span of control (Donaldson,
attention has been paid to the sources, content,
2000). Firms that have subsidiaries in different
and effects of team conflict and how these spe-
countries inevitably run into the issue of value
cifically influence team motivation (but see
differences (Erez, 2000). Hence, more knowledge 2004 Locke and Latham 393
is needed about how value differences actually
Dysfunctional thoughts lead people to evalu-
operate. For example, are goal setting, partici-
ate information inappropriately, thus leading to
pation in decision making, performance ap-
negative emotional states. Beck and his col-
praisal, and so forth differentially effective as a
leagues developed methods of consciously cor-
consequence of value differences, or are they
recting dysfunctional thought processes. Clients
simply used in a different form— or both?
report their automatic thoughts through intro-
Motivational issues are also important for
spection (an issue to be dealt with at length
strategic management. For example, strategic
below), and then the psychologists discuss with
management frequently involves change, and
the clients the rationality of such beliefs. For
the phenomenon of resistance to change is well
example, a depressed client might claim, “Pat
known (Beer, 2000). When firms decide that they
has left me; therefore, I am worthless.” The psy-
will employ a certain strategy (e.g., low cost),
chologist might then ask, “Is that really true?
they may differ radically in how well they im-
What do you base that on?” Gradually, clients
plement it (e.g., Wal-Mart versus K-Mart). In part,
come to see that their implicit conclusions or
this is an issue of knowledge and skill, but it is
“automatic thoughts” are not rational and that a
also related to motivation. Resistance to change
different perspective is more in line with reality.
is discussed routinely within the field of organ-
By challenging dysfunctional thoughts as they
izational development, but the motivational is-
arise and correcting them consciously, the cli-
sues involved are not directly included in tradi-
ents’ automatic or subconscious processing
tional motivation theories. At best, they are
changes and, thus, their negative emotions are
addressed by implication; for example, resis-
mitigated (Haaga, Dyck, & Ernst, 1991).
tance to change may imply refusal to commit to
Such clinical methods have practical utility in
certain goals and may be motivated by low self-
the realm of work motivation. Millman and
efficacy, low instrumentality, and/or negative
Latham (2001) found that they were able to train
valences. This issue needs to be studied explic-
unemployed individuals to engage in functional
itly. Of course, there are other aspects of strate-
thinking—that is, positive self-talk—and that
gic management that entail motivation—for ex-
such training significantly improved their
ample, decision choice and competitiveness—
chances of finding a new, well-paying job.
requiring further study as well.
Cognitive methods could be used to teach em-
Finally, motivation theory in the realm of work
ployees the principle of reframing dysfunctional
needs to draw on findings from other fields. Both
thoughts in work settings. For example, when
the science and practice of OB have already
individuals encounter difficulties during train-
benefited from theory in social (e.g., Bandura,
ing, they can reframe a self-demeaning state-
1986) and educational psychology (e.g., Dweck,
ment like “I can’t stand always being so stupid”
1986). In the study of motivation, findings by
as “It is normal to make mistakes when I am first
non-I/O scholars in clinical psychology must not
learning to perform a task.” Reframing self-
be overlooked (Latham & Heslin, 2003). Two ex-
deprecating statements in constructive ways
amples include research by Beck and by Selig-
can have a positive effect on motivation and can
man (and their respective colleagues).
sustain a person’s self-efficacy (Bandura, 1997).
Beck and his colleagues (Beck, 1967; Beck,
Similarly, employees might be taught to deal
Rush, Shaw, & Emery, 1979) focused on the rela-
with stress through thought retraining. Stress is
tionship between depression and “automatic
a response to the appraisal that one is being
thoughts,” by which they mean thoughts held in
psychologically or physically threatened. But
the subconscious that affect emotional re-
threat appraisals are not always rational, and
sponses. These researchers examined what they
even when they are, employees can be trained
call “dysfunctional thinking” with respect to
to engage in problem-focused thinking so as to
both content and process. Examples include
develop methods that enable them to mitigate
overgeneralization (e.g., “If I do something bad,
the threats they confront (Lazarus & Folkman,
it means that I am a totally bad person”), (irra-
1984). For example, employees faced with the
tional) perfectionism (e.g., “If I am any good at
possibility of layoffs could be trained to identify
all, I should be able to excel at everything I try”),
the exact nature of the perceived threats (e.g.,
and dependence on others (e.g., “I do things to
financial and/or psychological) and to generate
please other people rather than please myself”). plans to cope with them. 394
Academy of Management Review July
Irrational beliefs may adversely interact with
Optimists attribute their failures to causes
feedback provided by others. Rational beliefs
that are temporary rather than stable, specific to
can mediate the effect on performance feedback
the attainment of a particular goal rather than
from authority figures (e.g., a supervisor). Train-
all their goals, and see the problem as a result of
ing in ways to replace irrational with rational
the environment or setting they are in, rather
beliefs would also appear to be applicable to
than inherent in themselves. Setbacks and ob-
employees whose desire for inappropriate per-
stacles are seen as challenges (Seligman &
fectionism is preventing them from completing
Csikszentmihalyi, 2000). Thus, optimists are usu-
job assignments in a timely fashion.
ally resilient in the face of failure.
Managers and business leaders can engage
Seligman (1998b) found that optimism can be
in dysfunctional thinking, not only when the
learned, using a method similar to that em-
business is doing badly but also when it is do-
ployed by Beck. Step 1 requires the clinician to
ing well (e.g., “We are growing at 40 percent per
help clients identify self-defeating beliefs they
year and will always grow at that rate; thus,
may be unaware of. Step 2 involves gathering
there is no need to change our strategy”). Over-
information to evaluate and dispute the accu-
confidence leads managers to engage in poor
racy and implications of these self-defeating be-
decision making (Audia, Locke, & Smith, 2000).
liefs that are triggered by environmental events.
Training in metaprinciples of how to think ra-
Step 3 involves replacing maladaptive beliefs
tionally should be beneficial to people at all
with constructive, accurate ones based on the organizational levels.
data collected in the second step.
Based on over twenty-five years of program-
The ASQ may prove useful for identifying peo-
ple in organizations who suffer from learned
matic research in the laboratory and in the
helplessness. Seligman and Schulman (1986)
clinic, Seligman (1968, 1998a,b) established a
have provided evidence suggesting the value of
causal relationship between a person’s pessi-
ASQ for OB. They found that salespeople with
mistic explanatory style and subsequent de-
an optimistic explanatory style sold 35 percent
pression, on the one hand, versus an optimistic
more insurance than did those whose explana-
explanatory style and a person’s creativity, pro-
tory style was pessimistic. Moreover, people
ductivity, and overall sense of well-being, on the
with a pessimistic style were twice as likely to
other. Drawing on attribution theory, Seligman
quit their job in the first year than those with an
and his colleagues (Peterson et al., 1982) devel-
optimistic style. Similarly, Schulman (1999)
oped the Attribution Style Questionnaire (ASQ),
found that those who scored high on optimism
which assesses a person’s explanatory style
outsold those who scored as pessimists by 20 to
with regard to the locus, stability, and globality
40 percent across a range of organizations (e.g.,
of attributions. Locus refers to the extent to
auto sales, telecommunications, real estate, and
which a noncontingency between one’s actions
banking). Strutton and Lumpkin (1992) found that
and the consequences experienced is attributed
the mediator of the two attribution styles on
primarily to either oneself or to factors in the
employee performance is strategy. Salespeople
environment. Stability is the extent to which the
who scored high on optimism used problem-
lack of a response outcome is temporary or is
solving techniques, whereas those who scored
likely to persist into the future. Globality is the
high on pessimism focused on ways of seeking
extent to which noncontingent outcomes are social support.
perceived as either domain specific or likely to
Seligman’s training technique may provide a
undermine many areas of one’s life.
framework for mentors, coaches, and trainers to
Learned helplessness results from setbacks
predict, understand, and influence a person or
that are considered long lasting (stable), under-
team who has given up trying to attain goals
mining the attainment of most if not all of one’s
because of repeated failures. No one as yet has
goals (global), and caused by personal deficien-
shown whether the ASQ has general applica-
cies (internal) rather than situational con-
tions to the workforce. We also need to deter-
straints. The resulting low outcome expectancy
mine whether learned optimism is basically
causes deficits in future learning, as well as
equivalent to trait-level self-efficacy and
motivational disturbances such as procrastina-
whether optimism effects are mediated by situ-
tion and depression (Seligman, 1998a).
ationally specific self-efficacy (Bandura, 1997). 2004 Locke and Latham 395
Recommendation 3: Identify how gen-
situations, people may feel less free to act as
eral variables such as personality get
they want or “really are” as compared to when
applied to and are mediated by task-
they are in “weak” situations. However, this
and situationally specific variables,
likely occurs because people appraise situa-
how they are moderated by situations,
tions partly in terms of what they can and
and how they affect situational choice
should do in them. Furthermore, what has yet to and structuring.
be studied is the other side of the strong versus
weak situation coin—namely, the possibility of
A problem that must be overcome in combin-
“strong” versus “weak” personalities. Strong
ing motivation theories is how to integrate the
personalities should be less constrained by sit-
general with the specific. For example, a Big
uations than weak ones. For example, hyper-
Five personality trait such as conscientiousness
competitive people might look for ways to com-
is, by definition, general. It reflects action pat-
pete everywhere—not only in sports or business
terns that cross tasks and situations. Typically,
but also in social and personal relationships.
trait measures correlate about 0.20 with action
Thus, they would construe every situation as an
in specific settings. This mean correlation is bet-
opportunity to demonstrate their superiority.
ter than chance, but it does not answer such
Finally, we must not overlook the fact that
questions as: How do traits actually operate?
people are not merely the passive victims of
How can we make better predictions?
situations. For example, employees choose the
A partial answer to these questions becomes
jobs they apply for and quit those they dislike.
evident when we recognize that there is no such
They may restructure jobs to make a better fit
thing as action in general; every action is task
with their own talents and proclivities. They
and situationally specific. Specific measures, if
may also work with others to change situations
chosen properly, virtually always predict action
they dislike. They can choose what new skills to
better than general measures. However, general
develop and what careers to pursue. Going fur-
measures predict more widely than do specific
ther afield, they can also choose (in most free ones (Judge et al., 2002).
countries) whom they marry, where they live,
A general value or motive must presumably
how many children they have, how they spend
be “applied,” consciously or subconsciously, to
their money, whom they want as their friends,
each specific task and situation. It follows that
and what off-the-job activities they engage in.
situationally and task-specific knowledge, as-
As Bandura (1986, 1997) has noted, people are not
sessments, and intentions should be affected by
simply dropped into situations; they themselves
such motives and that these assessments, in
create, choose, and change situations. We need
turn, should affect actions taken in the situation.
to study how traits affect these processes.
A person’s goals, as well as self-efficacy, have
Recommendation 4: Study subcon-
been found to partly or wholly mediate the ef-
scious as well as conscious motivation
fects of some personality traits, as well as the
and the relationship between them.
effects of various incentives (Locke, 2001). These
traits include conscientiousness, competitive-
The concept of the subconscious is not a “hy-
ness, Type A personality, general (trait) efficacy,
pothetical construct” but a fully objective one. It
need for mastery, and self-esteem. VandeWalle
refers to information that is “in consciousness”
et al. (2001) found that goals and efficacy medi-
but not, at a given time, in focal awareness.
ate the effects of the trait of goal orientation on
Psychologists have shown that people can only
performance. The mediation hypothesis is im-
hold about seven separate (disconnected) ele-
plicit in Figure 1, in that values and personality
ments in focal awareness at the same time
are shown to work through goals and efficacy.
(Miller, 1956). The rest of one’s knowledge, to use
Nevertheless, it is possible that some trait ef-
the usual computer analogy, is “stored in mem-
fects are direct and, thus, not mediated at all. If
ory.” We validate the concept of the subcon-
so, it will be necessary to discover when and
scious by observing that we can draw knowl- why this occurs.
edge out of memory without any additional
The identification of personality trait media-
learning. Typically needed information is pulled
tors does not preclude the study of person-
out automatically, based on our conscious pur-
situation interactions. In “strong” or constrained
pose (e.g., when we read a book, the meanings 396
Academy of Management Review July
of the words and our knowledge of spelling and
land & Winter, 1969) believed that subconscious
grammar are automatically engaged). We can
motives are differentially aroused by different
also observe that certain events and experi-
situations and operate differently than con-
ences (e.g., early childhood memories) are scious motivation. harder to recall than others.
Failure to specify the effect of the subcon-
It is undeniable that people can act without
scious on action is a limitation of goal-setting
being aware of the motives and values underly-
theory (Locke & Latham, 2002)—not to mention
ing their behavior. This assertion does not re-
other motivation theories. Yet, over a century
quire the positing of an unconscious that is
ago, the Wurzburg school in Germany showed
made up of primitive instincts devoid of any
that goals that are assigned to people can affect
access to, or contact with, the conscious mind, as
their subsequent behavior, without their being
Freud asserted. Nor does acknowledging the
aware of it. In this century, Wegge and Dibblett
subconscious require a leap to the unwarranted
(2000) have shown that high goals automatically
conclusion that all actions are governed by un-
increase the speed with which information is
conscious forces (Wegner & Wheatley, 1999).
cognitively processed. Locke (2000b) has argued
Such a claim would clearly be arbitrary. This
that goals may arouse task-relevant knowledge
assertion only requires acknowledgment that
automatically, but almost nothing is known
the subconscious is a storehouse of knowledge
about how and when this occurs.
and values beyond what is in focal awareness
Studying the subconscious is difficult pre-
at any given point in time (Murphy, 2001) and
cisely because people, including laboratory par-
that accessibility to this stored information dif-
ticipants and employees, cannot always directly
fers within and between people.
provide the needed information stored there.
McClelland, Atkinson, Clark, and Lowell
Thus, indirect measures are required. Projective
(1953) claimed that the achievement motive,
measures may be useful (see Lilienfeld, Wood, &
which they asserted to be related to entrepre-
Garb, 2000), but they are riddled with such diffi-
neurship, was a subconscious motive. Thus,
culties as low internal reliability and the effect
they argued, it had to be measured with a pro-
of choice of pictures (in the case of the TAT). In
jective test—namely, the TAT—which involves
the realm of achievement motivation, a 2 (high/
people telling stories in response to pictures.
low projective measure) ⫻ 2 (high/low conscious
This claim may be true, but to the present au-
self-report) factorial design might reveal
thors’ knowledge, no self-report measure of
whether responses to these two measurement
achievement motivation has been designed
techniques—subconscious and conscious—
with items that match exactly the type of TAT
assuming they are actually referring to the
story content that is indicative of high need for
same concept, interact or work additively. The
achievement. Thus, TAT-measured achievement
same type of study could be conducted in rela-
motivation may or may not be assessing a con-
tion to other traits. The Big Five, for example,
cept different from self-reported achievement
might be measured projectively as well as motivation measures. through self-reports.
Self-report measures of achievement motiva-
Projective tests do not have to be confined to
tion are typically uncorrelated with projective
the TAT. Other projective measures may be
measures, even though both types of measures
equally if not more useful. An example is the
are significantly associated with entrepreneur-
incomplete sentence blank (ISB), used exten-
ial action (Collins, Hanges, & Locke, in press).
sively by Miner (e.g., Miner, Smith, & Bracker,
Need for achievement, measured projectively,
1994). Different projective methods should be
also appears to be unrelated to conscious per-
compared for agreement, when the same al-
formance goals (e.g., Tracy, Locke, & Renard,
leged concepts or motives are measured, as well
1999). Similarly, A. Howard (personal communi- as for predictive validity.
cation) found that, in a reanalysis of her twenty-
Another way to examine subconscious effects
five-year AT&T study with Bray, conscious goals
is through “priming.” Priming involves giving
for promotion had no relationship with a set of
people information that is apparently unrelated
projective measures that had been designed by
to the task at hand but that can affect an indi-
McClelland to predict managerial progress (see
vidual’s subsequent responses, without being
Locke & Latham, 2002). McClelland (e.g., McClel-
aware of the effect. In two experiments Earley 2004 Locke and Latham 397
and Perry (1987) used priming to influence the
practically, but it is psychologically trivial if the
task strategies that subjects used to attain
basis for the behavior is not explained. If traits
goals. Priming could be used in many other
are more than just behavioral regularity, they
types of motivation studies. Bargh, Gollwitzer,
must be caused by underlying motives. We can
Lee-Chai, Barndollar, and Troetschel (2001)
only learn about the nature of these motives by
found that primed goals for performance and
having people with varying levels of trait scores
cooperation had significant effects on these two
engage in introspection. With regard to the
outcomes. Research should be conducted com-
above discussion of projective versus self-report
paring the effect sizes of, and possible interac-
measures of traits, such as need for achieve-
tions between, consciously assigned versus sub-
ment, people who are highly effective versus consciously primed goals.
ineffective at introspection could be studied to
see if the two types of measures predict differ-
Recommendation 5: Use introspection
ently within each type of person. In addition,
explicitly as a method of studying and
people can be trained in introspection
understanding motivation.
(Schweiger, Anderson, & Locke, 1985). Research
Few methodologies in the history of the be-
is needed to determine whether training would
havioral sciences have been more controversial
produce greater convergence between con-
than introspection. Introspection was used ex-
scious and subconscious measures of the same
tensively by Titchner, an influential psycholo-
concept. Motive “constructs” (i.e., concepts) in
gist in the early twentieth century, but it was
OB are often defined statistically, as a conglom-
subsequently rejected by his followers because
eration of measures or of items. They are seldom
they found his view of psychology to be unduly
defined experientially. This is especially true of
narrow. Freud and his followers also rejected
so-called high-order constructs, which may have
introspection because they believed that moti-
little or no psychological reality. For example,
vational dynamics were in the unconscious, not
the Big Five personality dimensions are statis-
the subconscious— or, as they called it, the “pre-
tical conglomerations of a number of related
conscious”—and, thus, inaccessible to direct
subdimensions. But little is known about how
awareness or observation. Drive reductionists,
people with high scores on traits such as extra-
such as Hull and Spence, agreed with this inac-
version actually experience themselves and the
cessibility argument because they believed that
world. Such an understanding should enable
motivation was strictly physiological. The be-
researchers to develop better measures.
haviorists, especially Watson and Skinner, re-
(2) Increasing accuracy. The conditions under
jected introspection because they believed the
which self-reports of psychological states are
subject matter— consciousness—was irrelevant
more versus less accurate need to be identified.
to understanding human behavior. Neverthe-
Ericcson and Simon (1980) have described the
less, it is self-evident that motivational states
conditions under which introspective reports are
exist in consciousness; thus, introspection must
most reliable. The evidence suggests that the
be used to study it. Psychological concepts (e.g.,
more immediate and specific the information
desire, self-efficacy, purpose, satisfaction, be-
requested, the more accurately the respondent
lief) could not even be formulated or grasped
is able to introspect and, thus, to report the in-
without introspection. Furthermore, question-
formation accurately. It is usually difficult for
naire studies in OB have always relied on intro-
respondents to formulate broad abstractions
spection by the respondents, even though all
about themselves, especially personality traits
people are not equally good at it. The use of
or broad values. It is even harder for them to
introspection, as an accepted methodology in
formulate accurate and comprehensive state-
OB, will provide at least six important benefits
ments about the causes of their own and others’
for advancing our understanding of employee
actions. A major reason Herzberg used his men-
motivation. These are as follows.
tor Flanagan’s (1954) critical incident technique
(1) Understanding traits and motives. In the
to collect data was to avoid the problems asso-
field of personality, it is often unclear whether
ciated with asking people to introspect in order
researchers are describing behavior or an un-
to answer such abstract questions. Rather, he
derlying motive that causes the behavior. Pre-
used very specific questions, such as the follow-
dicting behavior from behavior may be helpful
ing: “Tell me a time when you were very satis-