Relationship of Goal Level to Valence and Instrumentality

Abstract

Eight experiments were conducted to explore the relationships between goal level, valence, and instrumentality. Valence, measured in terms of anticipated satisfaction across a range of performance levels, was strongly but negatively related to goal level. This finding was explained by showing that low goals entail using less stringent standards for self-evaluation than do high goals. Instrumentality was positively associated with goal level. Ss believed that trying for hard goals would be more likely to give them a sense of achievement, develop their skills, and prove them competent than would trying for easy goals. Ss also believed that high goals would lead to more practical (job and life) benefits, as well as more pride and self-respect, than would low goals. (PsycINFO Database Record (c) 2012 APA, all rights reserved)

Full text

Journal
of
Applied Psychology
1992.
Vol.
77, No. 4,
395-405
Copyright
1992
by
Ihe
American Psychological Association,
Inc.
0021-9010/92/$3.00
Relationship
of
Goal
Level
to
Valence
and
Instrumentality
Anthony
J.
Mento
Loyola College
in
Maryland
Edwin
A.
Locke
University
of
Maryland
Howard
J.
Klein
Ohio State University
Eight
experiments
were
conducted
to
explore
the
relationships
between
goal
level,
valence,
and
instrumentality.
Valence,
measured
in
terms
of
anticipated
satisfaction
across
a
range
of
perfor-
mance
levels,
was
strongly
but
negatively
related
to
goal
level.
This
finding was
explained
by
showing
that
low
goals
entail
using
less
stringent
standards
for
self-evaluation
than
do
high
goals.
Instrumentality
was
positively
associated
with
goal
level.
Subjects
believed
that
trying
for
hard
goals
would
be
more
likely
to
give
them
a
sense
of
achievement,
develop
their
skills,
and
prove
them
competent
than
would
trying
for
easy
goals.
Subjects
also
believed
that
high
goals
would
lead
to
more
practical
(job
and
life)
benefits,
as
well
as
more
pride
and
self-respect,
than
would
low
goals.
Locke
and
Latham (1990) recently integrated
the
research
findings
on
goal setting into
a
formal
theory.
In
constructing
the
theory, they attempted
to
integrate goal-setting theory with
other related theories, especially social-cognitive theory (Ban-
dura, 1986)
and
expectancy (VIE) theory. Mento, Cartledge,
and
Locke (1980)
first
attempted
to
integrate goal
and
expec-
tancy
theories
by
treating
VIE
concepts
as
determinants
of
goal
choice
or
commitment
or
both.
Later research confirmed these
relationships
(Klein,
in
press; Locke
&
Latham, 1990). How-
ever,
expectancy
and
self-efficacy
have also been
found
to
have
effects
on
performance that were independent
of
goals (i.e.,
di-
rect
effects;
Locke, Frederick, Lee,
&
Bobko,
1984).
Whereas
goal theory
and VIE
theory
now
seem well recon-
ciled
with
respect
to the
relationship between expectancy
and
performance
(Locke
&
Latham,
1990),
the
same cannot
be
said
with
respect
to
valence
and
instrumentality. Little research
has
been
done
to
date
on the
relationship
of
goal level
to
these
variables.
Garland
(1985)
and
Meyer
and
Gellatly
(1988)
found
goal
level
and
hence performance
to be
negatively
associated
with
average valence, which
was
measured
in
terms
of
antici-
pated satisfaction with
a
range
of
possible performance attain-
ments.
That
is,
people with high goals anticipated less satisfac-
tion
with
each possible performance level than
did
people with
low
goals.
One
purpose
of the
present studies
was to see if
previous
findings
would replicate.
If
they
did,
a
second
pur-
pose
was to
explain
the
reason
for the
negative association.
None
of the
previous studies cited provided
a
full
explanation
for
their results.
With
respect
to any
single goal
or
outcome (such
as
winning),
high
performance
in
relation
to the
goal
or
outcome should lead
to
higher
satisfaction
than should
low
performance; that
is,
peo-
Partial
funding
for
this
research
was
made available
to
Anthony
J.
Mento
through
a
research grant
from
the
Sellinger School
of
Business
and
Management
at
Loyola College
in
Maryland.
Correspondence concerning this article should
be
addressed
to
Anthony
J.
Mento, Department
of
Management
and
Law, Loyola Col-
lege,
4501
N.
Charles
Street,
Baltimore, Maryland 21210.
pie
should like succeeding
better
than
failing
(Locke
&
Latham,
1990).
This
is
because goals
or
values serve
as
stan-
dards
for the
evaluation
of
oneself
and
one's performance (Ban-
dura, 1988).
The
seeming anomaly
of the findings of
Garland
(1985)
and
Meyer
and
Gellatly (1988)
was
that mean valence decreased
across goal levels, that
is, as
goal level increased. These results,
as
a
group, suggest that
the
distinction between valence
across
goal
levels
and
valence within
a
goal level
is a
crucial
one be-
cause
the
valence-outcome correlations were
in
opposite
direc-
tions
in the two
cases.
In the
present experiments,
we
were
concerned with relationships across goal levels.
As we
show,
the
two
sets
of findings are
fully
reconcilable.
A
third purpose
of the
present studies
was to
explore
the
relationship between instrumentality
and
goal level. Although
previous
studies have
found
that instrumentality
predicts
goal
choice
(e.g.,
Dachler
&
Mobley,
1973),
only
one
previous study
examined
the
relationship between assigned goals
and
subse-
quently
rated instrumentality. Matsui, Okada,
and
Mizuguchi
(1981)
asked subjects
to
rate
the
desirability
of
each
of five
intu-
itively
derived outcomes that could
be
attained
by
working
to-
ward
a
goal:
(a) a
sense
of
achievement,
and
increased self-con-
fidence
with
respect
to (b)
general
competence,
(c)
ability
to
concentrate,
(d)
clerical
ability,
and (e)
persistence.
The Va-
lence
X
Instrumentality
product
was
significantly
higher
for the
high
goals than
for the low
goals
on a
clerical task. Because only
one set of
valence ratings
was
obtained,
the
Valence
X
Instru-
mentality
differences
were necessarily
due to the
instrumenta-
lity
ratings. Thus,
we
hereinafter
refer
to
these
as
instrumenta-
lity
ratings.
Matsui
et
al.'s
(1981)
findings
indicate that there
is a
positive
association
between instrumentality
and
goal level.
If
both
the
valence
and
instrumentality
findings
replicated
in our
studies,
then
we
expected
the two
types
of
ratings
themselves
to be
negatively
correlated.
The
present article reports
the
results
of
eight studies
de-
signed
to
explore
the
relationship between valence, instrumen-
tality,
goals,
and
performance. However, because some
of
these
395

396
A.
MENTO,
E.
LOCKE,
AND H.
KLEIN
were pilot
or
replication studies, only
four
are
reported
in de-
tail.
Study
1
For
this study,
as
well
as the
following
four,
we
used
the
"list-
ing
uses
for
common objects" task that
has
been used
in
many
other studies (Garland, 1982; Locke,
1966,1982).
Subjects
who
were
assigned goals were asked
to
list
4, 7, or 12
uses
for
each
object
in 1
min.
We
predicted that,
as the
difficulty
of the as-
signed
goal increased, subjects would report higher
self-effi-
cacy
but
lower valence,
set
higher personal goals,
and
perform
at a
higher
level.
Method
Subjects
The
subjects were
114
students
in
undergraduate business courses.
The
experiment
was
conducted during weekly class meetings within
a
one-week
time frame.
Task
The
task
was to
list
uses
for
common objects (one practice trial
and
two
experimental trials).
On the
basis
of
data
obtained
by
Locke
(1966),
the
objects
for the
experimental trials (rubber tire
and
book)
were chosen
to be of
equal
difficulty.
The
order
of the
objects
was
counterbalanced within each goal condition
for the two
experimental
trials.
Goals
The
experiment used three specific goal levels, which ranged
from
easy
to
difficult.
Everyone within
a
course section
was
assigned
the
same goal,
and
goals were assigned
to
sections
at
random.
All of the
sections were taught
by the
same instructor
and
were part
of the
same course.
All
data
were collected dur-
ing
the
same semester.
The
total
number
of
subjects
for
each
goal group
was as
follows:
4
uses
(n
=
23);
7
uses
(«
=
34);
12
uses
(n
=
31).
A "do
your
best"
group
(«
= 26) was
also included
in
which
no
specific
goals were assigned.
Procedure
After
the
task
was
explained,
the
subjects were given
a 1
-min practice
trial
(object: wire
coat
hanger) followed
by two
1-min
experimental
trials.
On the
practice trial, subjects were instructed
to
list
as
many
uses
as
they could
for the
object
in 1
min. Before
the first
experimental
trial
(Tl)
the
goal-setting groups were assigned goals
(4, 7, or 12
uses)
and
were instructed
to
reach
those
goals during
the 1
-min
period.
Sub-
jects
in the
remaining condition were instructed
to do
their
best.
Following
the
instructions,
all
subjects were asked
to
rate their antici-
pated satisfaction (valence) with generating
2-13
uses
on the
next trial.
Subjects
responded
to
each
of the 12
possibilities
on a
scale ranging
from
1
(incredibly
dissatisfied)
to 9
(incredibly
satisfied).
Similarly,
all
subjects
were requested
to
rate their
self-efficacy
for
generating
2-13
uses
on the
next trial. Subjects responded
to
each
of
these
12
possibili-
ties
on a
scale ranging from
10
(absolute
certainty)
to 0
(no
chance
at
all).
On
the
second experimental trial, subjects were allowed
to
choose
their
own
goal
(self-set
goal condition).
Results
To
examine
the
effects
of
goal level
on the
variables
of
inter-
est,
we
performed one-way univariate analyses
of
covariance
(ANCOVAs)
for
each
of the
dependent variables, followed
by
post
hoc
cell-means analysis with
the
Newman-Keuls
test.
Per-
formance
of
subjects
on the
practice trial served
as the
ability
covariate. Table
1
summarizes
the
results
of
these analyses.
In
this
and all
following
studies
in
which analysis
of
variance
was
used
as an
analytic technique,
u>
2
—an
estimate
of
treatment
magnitude—was
calculated according
to
formulas provided
by
Keppel
and
Zedeck
(1989).
On
Trial
1
(top half
of
table),
as
goal
level
increased,
self-efficacy
increased
nonsignificantly,
self-set
goals
and
performance increased
significantly,
and
valence
de-
creased (for
the
effect
of
assigned goal
on
personal goal,
va-
lence,
and
performance,
<o
2
=
.35,
.15,
and
.12, respectively).
Estimates
of the
internal consistency
(Cronbach's
alpha)
of the
self-efficacy
and
valence measures were
.91 and
.93, respec-
tively.
With respect
to the
do-your-best condition
in the as-
signed
goal treatment,
the
valence
level
was
similar
to the va-
lence
levels
of the
easy-
and
moderate-goal groups
but was
sig-
nificantly
higher than
the
valence level
of the
difficult-goal
group.
As
usual, do-your-best subjects performed less
well
than
subjects
assigned
specific,
difficult
goals.
The
results
for
Trial
2
were similar.
The
self-set goals
in
Trial
2
were
significantly
affected
by the
assigned goals
in
Trial
1
(see
bottom
half
of
Table
1),
in
line with previous
findings
(Locke
&
Latham,
1990).
Effect
sizes
(co
2
)
were
.26 for
personal
goal,
.06
for
valence,
and .05 for
performance. Reliability estimates
for
both
self-efficacy
and
valence were .93.
The
intercorrelation matrix
of the
Trial
1
variables (without
the
do-your-best subjects)
is
presented
in
Table
2.
Assigned goal
level
was
positively related
to
self-efficacy,
personal goal level,
and
performance. Valence, however,
was
negatively related
to
goal level and,
as a
result,
was
negatively related
to
perfor-
mance.
The
negative relationship also
was
revealed
by the va-
lence
functions
for
each goal
level,
shown
in
Figure
1.
Also
depicted
in
Figure
1 are the
valence ratings
for
subjects
told
to do
their best.
As
shown
in
Table
1,
the
anticipated satis-
faction
ratings
of the
do-your-best subjects were
very
similar
to
those
of the
easy-goal subjects,
indicating
that subjects trying
to
do
their best
did not
have
a
very stringent criterion
for
what
their best was. This suggests that
a
do-your-best (i.e., vague) goal
is
compatible
with
a
wide range
(i.e.,
easy
to
difficult)
of
perfor-
mance
levels
(Locke
&
Latham.
1990).
Study
2
Although
the
goals were assigned
to
course sections
at
ran-
dom in
Study
1,
and the
sections were comparable
and
taught
by
the
same instructor, subjects were
not
assigned
to the
sec-
tions
at
random.
To
preclude
the
small chance that there could
have
been
a
Goal
X
Section interaction,
we
repeated Study
1
(without
the
do-your-best condition), using
a
within-subjects
design
(n =
25).
The
same task, goals,
and
measures were used,
and the
results were
virtually
identical.
Because
our first two
experiments replicated
the
negative
association between goal
level
and
valence,
it now
remains
to
explain
the
association.
The findings
make
perfect
sense
if one

GOAL LEVEL,
VALENCE,
INSTRUMENTALITY
397
Table
1
Effect
of
Assigned
and
Self-Set Goals
on
Personal Goals,
Achievement
Valence,
Self-Efficacy,
and
Performance (Study
1)
Goal
Variable
Easy
Moderate
Difficult
Do
your
best
ANCOVA
df
Trial
1
(assigned goals)
Personal goal
M
SD
Self-efficacy
M
SD
Achievement
valence
M
SD
Performance
M
SD
5.36
a
1.96
5.07
a
1.16
6.34
a
0.71
5.66,
1.77
7.31
b
1.44
5-38
a
1.07
6.32
a
0.52
6.83
b
1.75
8.73
C
2.22
5.71,
0.96
5.63
b
0.79
7.51
b
1.96
5.29
a
0.93
6.4
l
a
0.89
5.83
a
1.44
30.61**
1.91
7.30**
9.47**
2,84
3, 103
3, 102
3,
109
Trial
2
(self-set
goals)
Personal goal
M
5.83,
SD
1.87
Self-efficacy
M
5.34
a
SD
1.33
Achievement
valence
M
6.29.
SD
1.16
Performance
M
6.41
a
SD
1.97
7.54
b
1.70
5.60
a
1.15
6.14.
0.66
7.43
b
1.48
8.78
C
2.13
5.98
a
0.98
5.63
b
0.85
7.96
b
1.91
5.21.
1.08
6.26
a
1.03
7.23
b
2.24
19.46*
2,84
2.67
3, 109
3.63**
3, 109
3.39*
3, 109
Note.
N =
114. Means
are
adjusted
to
control
for
pretest covariate
(ability).
Within rows, means with
different
subscripts
differ
significantly
from
one
another according
to the
Newman-Keuls
test. Partici-
pants
in the
easy-goal group
(n = 23)
attempted
to
name
4
uses
for an
item
within
1
min;
participants
in the
moderate-goal group
(n = 34)
attempted
to
name
7
uses;
and
participants
in the
difficult-goal
group
(n
=
31)
attempted
to
name
12
uses. Participants
in the
do-your-best group
set no
personal goal
and
simply
named
as
many uses
for an
item
as
they could
think
of in 1
min.
ANCOVA
=
analysis
of
covariance.
*p<.05.
**/?<.01.
grasps
that
the
same
principle
that
operates
within
a
single
goal
level
also
operates
across
goal
levels
(as
suggested
briefly
by
Garland,
1985).
That
is, a
goal
is
at the
same
time
a
target
to aim
for
and a
standard
by
which
to
evaluate
the
adequacy
of
one's
Table
2
Zero-Order Correlations
Among
Variables—
Specific
Goal Groups Only (Study
1)
Variable
1
1.
Ability
—
-.13
2.
Assigned
goal
—
3.
Personal goal
4.
Self-efficacy
5.
Achievement
valence
6.
Performance
Note.
N = 88.
*p<.05.
**p<.01.
.37**
.21*
.53**
.19*
—
.51**
—
-.02
-.39**
-.51**
-.42**
.49**
.29**
.64**
.42**
-.31**
—
performance
(Locke
&
Latham,
1990).
Bandura
(1988)
noted
that
--==-
goals
specify
the
conditional requirements
for
positive
self-evalua-
tion.
By
making self-satisfaction conditional
on
matching
adopted goals, people
give
direction
to
their actions
and
create
self-incentives
to
persist
in
their
efforts
until
their performances
match their
goals,
(p.
41).
Let us
apply
this
principle
to the
case
in
which
three
individuals
have
goals
of 4
(easy),
7
(moderate),
and 12
(difficult)
uses,
re-
spectively.
If
each
person
fully
accepts
the
assigned
goal,
the
person
with
a
goal
of 4
uses
will
be
minimally
satisfied
with
a
performance
attainment
of 4,
will
be
increasingly
satisfied
with
attainments
above
this
level,
and
will
be
less
satisfied with
at-
tainments
below
this
level.
In
contrast,
the
person
with
a
goal
of
12
uses
will
be
minimally satisfied only with
an
attainment
of
12
and
will
be
increasingly
dissatisfied
with
attainments
below
that.
Thus,
if
this
person
lists
only
4
uses,
he or she
will
be
extremely
dissatisfied,
in
contrast
with
the
individual
with
a

398
A.
MENTO,
E.
LOCKE,
AND H.
KLEIN
4
6 8 10
Level
of
Performance
(number
of
uses)
Figure
1.
Achievement valence
functions
for the
four
goal groups
in
Study
1.
goal
of 4
uses,
who
will
be
minimally satisfied with that same
level
of
attainment.
In
contrast,
if the
low-goal person attains
a
score
of
12,
he or she
will
be
ecstatic because this score will
be
far
above
the
minimum, whereas
the
individual with
a
goal
of
12
will
be
only minimally satisfied with that level.
The
evalua-
tions
of the
person with
a
moderate goal
of 7
will
fall
between
those
of the
other two.
Thus,
the
valence measure
is
negatively rather than positively
related
to
performance because
it
represents
the
other side
of
the
goal coin:
the
fact
that
for
high goals, higher performance
is
required
to
produce satisfaction than
is
required
for low
goals.
These results
and the
rationale behind them reveal
why
Gar-
land
(1985)
found
that partialling
out
valence vitiated
the
per-
sonal goal-performance relationship.
In
partialling
out va-
lence,
Garland
was in
effect
partialling
out the
goals them-
selves.
The
type
of
valence function
we
have discussed
so
far, which
is
measured
in
terms
of
anticipated satisfaction when
no
extrin-
sic
incentives
are
offered
for
performance,
may be
called
the
achievement
valence function. Note that
the
achievement
va-
lence
function
within
any
given goal
level
is
positive;
the
higher
the
performance
in
relation
to the
goal,
the
more favorable
the
appraisal
(see each separate
function
in
Figure
1).
Thus, individ-
uals
prefer
to
exceed rather than
to
fall
short
of
their goals.
Raising
the
goal
level
simply shifts
the
valence function
to a
different
plane;
the
individual must
do
more
to get the
same
degree
of
satisfaction.
Study
3
The
next study
was
similar
in
design
and
results
to
Study
2
but
also
introduced instrumentality questions.
We
conducted
posttask interviews
to ask
subjects
in
detail
why
they
accepted
or
agreed
to try for the
assigned goals.
In
general,
four
types
of
answers emerged. Subjects agreed
to try for the
goals
(a)
because
they
were told
to do so; (b) so
that they could improve their
skills,
for
example, their ability
to
concentrate;
(c)
because try-
ing
for the
goals gave them
a
sense
of
achievement;
and (d) so
that they could prove themselves (prove that they were compe-
tent,
not
quitters,
etc.).
Study
4
Study
4
utilized
the
same design
as
Studies
2 and 3 and en-
tailed further refinements
of the
instrumentality items.
The
same results
for
goals
and
valence were
found as
were
found in
Studies
1-3.
In
this study
(as in
Study
3),
self-efficacy
was
posi-
tively
and
significantly
associated with goal level.
The
total
in-
strumentality scale
was
also positively
and
significantly asso-
ciated with goal
level,
the
opposite
of the
direction
for
valence.
The
instrumentality items, however,
did not
factor clearly.
Study
5
Like
Studies 2-4, this
was a
within-subjects design with
as-
signed goal
levels
of
4,7,
and
12
uses. Study participants were
60
graduate students
of
business administration
who
were concur-
rently
enrolled
in an
organizational behavior course. This study
involved
three modifications
of
Study
4.
First,
the
wording
and
context
of the 16
outcomes were further revised
and
clarified.
Second,
a yes or no
item
was
added
to
each performance level
on
the
self-efficacy
scale.
The sum of the yes
responses
(how
many
levels
a
subject thought
he or she
could reach) reflects
self-efficacy
magnitude
in
Bandura's (1986) terminology,
whereas
the
confidence ratings (used
in the
previous studies
and
revised slightly
in
this one) reflect what
he
called
self-effi-
cacy
strength. Third,
a new
goal measure
was
developed.
Method
After
a
practice trial that served
as a
pretest covariate
(as in all of the
studies
reported
here), subjects were asked
to
rate
16
outcomes:
1.
Increasing
my
self-confidence.
2.
Developing
my
ability
to
concentrate.
3.
Developing
my
creative ability.
4.
Developing
my
ability
to
persist
on a
hard task.
5.
Getting
a
sense
of
achievement.
6.
Making
the
task interesting
and
challenging.
7.
Seeing
a
sense
of
progress.
8.
Showing improvement.
9.
Performing
as
expected.
10.
Doing what
the
experimenter asked
me to do.
11.
Doing what
I am
requested
to do.
12.
Doing what
I am
told.
13.
Proving that
I am
competent.
14.
Proving that
I can
respond
to a
challenge.
15.
Proving that
I am not a
quitter.
16.
Proving that
1
will
keep trying when things
get
tough.

GOAL
LEVEL,
VALENCE,
INSTRUMENTALITY
399
The
instrumentality
rating
instructions
were
as
follows:
"Use
any
whole
number
between
0 and
10.
Rate
the
degree
to
which
reaching
or
trying
for the
goal
of [4, 7, or 12]
will
actually
lead
to
each
out-
come."
Our
measure
of
achievement
valence
was the
same
as in
Studies
1-4.
Subjects
were
first
asked
whether
or not
they
could
generate
a
number
of
uses
that
ranged
from
2 to
13.
A yes or no
response
was
required
for
each
performance
level.
Self-efficacy
strength
was
measured
as
before
(on
a
scale
ranging
from
0 to
10).
The
exact
instructions
for the new
personal goal measure
were
as
follows:
"Your
personal
goal
is
denned
as the
lowest
number
of
uses
that
you
would
be
minimally
satisfied
with
generating."
Achievement
valence,
self-efficacy,
and
personal
goal
were
all
measured
three
times,
once
after
each
goal
assignment.
The
usual
three
goal
levels
(4,7,
and
12)
were
used,
in
sequence.
Results
As
previously, ANCOVAs were conducted
to
assess
the ef-
fects
of
assigned goal
on our
dependent variables. Cell means
and
significant
differences
are
shown
in
Table
3.
As
expected, assigned goals exerted
a
significant
effect
on
personal goal,
both
self-efficacy
measures, achievement
va-
lence, instrumentality,
and
performance.
It is
relevant
to ask
whether there
was a
learning
effect
in the
within-subjects stud-
ies
(Studies 2-5). There
are two
arguments against this. First,
the
within-subjects studies replicated
the
results
of the be-
Table
3
Effect
of
Assigned
Goal
on
Personal
Goal,
the
Strength
and
Magnitude
of
Self-Efficacy
Ratings, Achievement
Valence,
Instrumentality,
and
Performance
(Study
5)
Assigned
goal
ANCOVA
Variable
Easy
Moderate
Difficult
F(
1,118)
Personal
goal
M
SD
Self-efficacy
Magnitude
M
SD
Strength
M
SD
Achievement
valence
M
SD
Instrumentality
M
SD
Performance
M
SD
6.48
a
2.01
1.49
a
0.15
5.38
a
1.13
6.44.
0.94
5.27
a
2.01
5.40
a
1.32
7.12
b
1.90
1.53
b
0.17
5.52
b
1.23
5.92
b
0.94
6.33
b
1.83
7.02
b
1.48
8.98
C
2.67
1.59.
.18
6.09
C
1.30
5.27
C
1.16
6.67
b
2.03
8.10
C
1.93
38.18
17.96
19.23
45.69
21.87
65.4
Note.
Repeated-measures
design,
N=60.
Means
are
adjusted
to
con-
trol
for
task
ability.
Within
rows,
means
with
different
subscripts
differ
significantly
from
one
another
according
to the
Newman-Keuls
pro-
cedure
for
repeated
measures.
All Fs are
significant
at p <
.01.
Easy
goal
= 4
uses;
moderate
goal
= 7
uses;
and
difficult
goal
= 12
uses
named
within
1
min.
ANCOVA
=
analysis
of
covariance.
tween-subjects study (Study
1).
Second, Locke (1964)
found
no
improvement
in
performance over
the first
three experimental
trials
in any of his
three studies, which used similar
or
identical
tasks.
As
in the
previous
four
experiments, achievement valence
was
related inversely
to
goal level.
The
achievement valence
function
(not shown)
was
similar
to
those
in
previous experi-
ments.
In
contrast, instrumentality
was
positively related
to
goal level.
Effect
sizes
(o>
2
)
were
.38 for
personal goal (indicating
that
the new
goal measure worked
as
well
as the old
one),
.21
for
self-efficacy
magnitude,
.22 for
self-efficacy
strength,
.42 for
achievement valence,
.25 for
instrumentality,
and
.51
for
perfor-
mance.
Our
self-efficacy
magnitude measure
was
dichoto-
mous,
either
yes or no.
Responses were
coded
0 or 1, so the
variance
and
mean
scores
across goal levels
are
necessarily
small.
Coefficient
alpha reliabilities were
.92 for
outcome
va-
lence,
from
.89 to .94 for
achievement valence,
from
.79 to .84
for
self-efficacy
magnitude,
from
.92 to .93 for
self-efficacy
strength,
and .94 for
instrumentality
for all
levels
of
assigned
goals.
The
16-item
outcome
valence measure
was
factor analyzed
to
determine
the
underlying factor structure.
A
clear four-factor
solution emerged. These
four
factors corresponded exactly
to
our a
priori categories developed
from
the
Study
3
pilot inter-
views.
Factor
1
represented skill development (items
1,2,3,
and
4);
Factor
2
represented proving oneself (items
13,14,15,
and
16);
Factor
3
represented
obedience
(items
9,10,11,
and
12);
and
Factor
4
represented achievement (items
5, 6, 7, and 8).
Index
scores were obtained
for the
instrumentality ratings
(one
for
each goal
level)
by
summing
the
ratings
for the
four
items
in
each
factor.
The
instrumentality factor indexes were
generally
related
to
goal level
in a
positive direction; however,
only
three
of the
instrumentality
indexes
were significant: skill
development,
F(2,118)
=
39.12,
<o
2
=
.38,
p <
.01;
achievement,
F(2,118)
=
30.06,
w
2
=
.32,
p <
.01;
and
proving oneself
F(2,
116)
=
11.42,
o>
2
=.
14,
p <
.01.
(The sample sizes vary because
of
missing
data.) Obedience
was
endorsed
as a
reason
for
trying
for
all
goals
but did not
change
as a
function
of
goal level,
F(2,118)
=
.76,
ns.
These
first five
studies revealed
two
kinds
offerees
lead peo-
ple to
work harder
for
difficult
goals than
for
easy goals.
Diffi-
cult
goals,
on the one
hand, require people
to
accomplish more
to
attain
self-satisfaction
and,
on the
other hand,
are
associated
with
more
beneficial
outcomes than easier goals. Both issues
were
explored
in a
more
real-life
setting
in the
next
two
studies.
Study
6
In
this pilot study, subjects
from
graduate-level business
ad-
ministration classes
(JV=
106)
were randomly assigned
a
course-
grade goal
(for
the
purpose
of filling out the
questionnaire
only)
of
either
A, B, or C in a
between-subjects
experimental design.
Then subjects were asked
to
respond
to
16
potential outcomes.
The
specific
instructions were:
"To
what
degree
do you
think
a
grade
of
[A,
B, or
C
]
would
bring
you or
help
to
bring
you
each
of
the
following
outcomes?" Participants were asked
to
rate each
of
16
outcomes
on a
Likert scale ranging
from
1 (to no
degree
at
all)
to 5 (to a
very
high
degree).
There
was a
strong positive

400
A.
MENTO,
E.
LOCKE,
AND H.
KLEIN
association between
the
overall instrumentality rating
and as-
signed
goal
level.
Study
7
The
purpose
of
this study
was to
keep
our
research
in a
real-
world setting while
refining
the
measure
of
instrumentality
pi-
loted
in
Study
6. The
sample
in
this between-subjects design
study
was
251
graduate students
in a
business administration
program. Subjects were again asked
to
respond
to
instrumenta-
lity
and
valence questions.
Method
In
contrast
to the
pilot study,
in
which course-grade goals were
as-
signed,
the
assigned goals
in
this study pertained
to an
undergraduate
grade point average (GPA)
of A
(4.00,
the
difficult
goal),
B
(3.00,
the
moderate
goal),
or C
(2.00,
the
easy goal).
These
were assigned
to
sub-
jects
at
random.
Exact
instructions
for the
instrumentality
question
were
as
follows:
For the
purpose
of
answering
this questionnaire, assume that
as
an
undergraduate student, your
GPA
had
been close
to [A
(4.00),
B
(3.00),
or
C
(2.00)].
1.
To
what degree
do you
think
a
grade average
of [A, B, or C]
would
bring
you or
help
to
bring
you
each
of the
following
out-
Subjects
were next requested
to
rate
19
outcomes
on the
relationship
between
undergraduate
GPA and the
outcomes
on a
5-point Likert
scale ranging from
1 (to
770
degree
at
all)
to 5 (to a
very
high
degree).
The
19
outcomes were
1.
Self-respect.
2.
Pride
in
myself.
3.
Pride
in my
performance.
4. The
respect
of my
peers
in
class.
5. The
respect
of my
closest friends.
6.
The
respect
of the
teachers
in my
courses.
7.
The
belief that
I
will
do
well
in
other courses.
8.
A
strong sense
of
competence from
my
performance.
9.
Some tuition reimbursement from
my
employer.
10.
Confidence
to try
other intellectual challenges.
11.
A
good
job
offer.
12.
A
high
starting salary
in my first
job.
13.
Immediate responsibility
in my first
job.
14.
Being
put on the
"fast
track"
for
advancement
in my first
job.
15.
Acceptance into
an MBA or MA
(MS)
or PhD
program.
16.
A
fellowship
or
assistantship
in
graduate school.
17.
A
prize
or
award from
my
school
for
scholarship.
18.
Ultimate success
in my
career.
19.
Ultimate success
in my
life.
Our
achievement valence measure
was as
follows:
Assume
in
answering this question, that your personal goal,
in
terms
of the
undergraduate grade average that
you
wanted
and
considered satisfactory
was
also
[A,
B, or C, the
same
as the as-
signed
goal
in the
instrumentality
question].
Please indicate
be-
low,
given this goal,
how
satisfied
you
would
be if you
actually
got
each
of the
following
grades.
The
same Likert scale used
to
measure valence
in
Studies
1 -6
was
used, ranging from
1
(incredibly
dissatisfied,
totally dis-
gusted)
to 9
(incredibly
satisfied,
delirious with joy).
Results
Reliability
estimates were
.95 and
.71
for
instrumentality
and
achievement valence, respectively.
The
valence function
was
plotted showing satisfaction
as a
function
of
assigned goal level
(GPA
of
either
A, B, or C). The
valence
(anticipated
satisfaction)
function
for the
difficult
assigned-goal
group
(n
= 89) was
lower
at all
performance levels than anticipated satisfaction
for the
moderate assigned-goal group
(n =
85), which
in
turn
was
uni-
formly
lower
at all
performance levels than
the
valence
func-
tion
for the
easy assigned-goal group
(n =
77), F(2, 245)
=
176.63,
p <
.01,
a;
2
= .58
(mean differences were
verified
with
the
Newman-Keuls
procedure).
These
results were very similar
to
those presented
in
Figure
1.
We
factor analyzed
the 19
instrumentality items, using vari-
max
rotation.
An
analysis
of the
scree plots
and the
rotated
factor
matrix suggested that
a five-factor
solution
was the
clear-
est for
interpretation.
We
used
four
of the five
factors
because
the fifth
factor loaded
on
only
one
item. Factor
1
represented
pride, respect,
and
confidence (items
1-8
and
10);
Factor
2
rep-
resented
job
benefits
(items
11-14);
Factor
3
represented school
benefits
(items
15-17);
and
Factor
4
represented career
and
life
benefits
(items
18
and
19). Factor indexes were obtained
for
each assigned-goal condition
by
summing
the
relevant items
and
were plotted
as a
function
of
assigned goal.
Plots
for the
relationships between satisfaction
and
valence
and
between satisfaction
and
instrumentality
are
presented
in
Figure
2. The
achievement valence curve represents
the
mean
anticipated satisfaction score
for
each goal (that
is, as
if
one
took
the
means
for
each curve
in
Figure
1 and
plotted them
as
points).
Figure
2
shows
in
graphic terms
the
opposite slopes
of
the two
measures
with
respect
to
goal level.
The
instrumenta-
lity
factor indexes were
uniformly
higher
at
each
GPA
goal
level,
the
opposite
of the
pattern
found
for
achievement
va-
lence.
All
factor index
means
were
highest
for the
difficult
goal
(an
undergraduate
GPA of
4.00,
or A),
which were
uniformly
higher
than
the
factor means
for the
moderate goal
(an
under-
graduate
GPA of
3.00,
or B),
which were
uniformly
higher than
the
factor means
for the
easy goal
(an
undergraduate
GPA of
2.00,
or C). For
Factor
1,
F(2,248)
=
222.93,
p <
.01,
a.
2
=
.64;
all
means were
significantly
different
from
each other.
For
Factor
2,
F(2,
248)
=
54.802,
p <
.01,
o>
2
=
.30;
all
means were
signifi-
cantly
different
from
each other.
For
Factor
3,
F(2, 248)
=
192.60,
/; <
.01,
co
2
=
.60:
all
means were
significantly
different.
For
Factor
4,
F(2, 248)
=
10.44,
p <
.01,
u
2
=
.07;
the
means
for
the
difficult
and
moderate goals were
significantly
different
from
the
mean
for the
easy goal. (All significance tests were
verified
according
to the
Newman-Keuls procedure.)
Next,
we
calculated
an
intercorrelation matrix
to
examine
the
relationship between goals, valence, instrumentality,
and
the
four
instrumentality factors.
As
seen
in
Table
4,
assigned
goal level
was
positively
and
significantly
related
to
instrumen-
tality
and to
each
of the
four
instrumentality factors. Factor
1
(pride)
and
Factor
3
(school
benefits)
were most strongly related
to
goal level. There
was a
strong negative relationship between
achievement valence
and
both assigned goals
and the
instru-
mentality
factors.
Study
8
This experiment
was
conducted
by
Howard Klein indepen-
dently
and
without knowledge
of the
other seven studies. Never-
theless, this study made several important additions
to our

GOAL LEVEL,
VALENCE,
INSTRUMENTALITY
401
7.0-
6.5-
6.0-
5.5-
5.0-
4.5-
4.0-
Achievement
Valence
School
Outcomes
(F-3)
Pride,
Competence
(F-1)
Job
Benefits
(F-2)
Career
&
Life
Benefits
(F-4)
4.0
-3.5
-3.0
-2.5
-1.5
\
I
C B
-1.0
Grade Goal
or
Outcome
Figure
2.
Achievement valence
and
instrumentality functions
for
grade goals
or
outcomes
in
Study
7.
knowledge.
First,
because
a
between-subjects design
was
used,
we
were able
to
test
an
implication
of our
hypothesis
that
goals
and
achievement valence
are two
aspects
of the
same phenome-
non.
If
this hypothesis
is
true, then goal
and
valence measures
can
be
used interchangeably
and
equally
successfully
to
predict
performance.
Second, Klein tried
a
number
of
additional
ways
of
scoring
the
valence ratings.
In our
studies,
we
used
the
over-
all
(summated) valence measure
in all
cases.
In
this
study,
four
additional valence measures were tried. Third, Klein used
a
bipolar
rather than
a
unipolar valence measure. Fourth, Klein
had
subjects rate
outcome
attractiveness rather than anticipated
satisfaction.
Method
Subjects
Participants
were
356
undergraduate students enrolled
in an
intro-
ductory organizational behavior course.
Task
The
task
was
exam performance,
specifically,
performance
on the
second
of
three exams required
for
that course.
The
exam
was
worth
100
points
and
represented
25% of the
grade
in the
course.
All
subjects
were
volunteers
who
participated
as
part
of an
in-class exercise. Partici-
pants
filled out a
questionnaire
one
week before
the
second exam.
Measures
Valence.
Participants rated
the
attractiveness
of
attaining
six
possi-
ble
performance
levels
ranging
from
50 to 100
points
in
10-point
incre-
ments. Participants indicated
how
good they would
feel
about
receiv-
ing
each
of
those possible performance levels
on
their
second
exam,
using
a
7-point Likert scale ranging
from
-3
(very
unattractive)
to 3
(very
attractive).
From these
six
ratings,
five
different
valence measures were
derived.
Achievement
valence
was
computed
by
summing
the six
attractive-
ness ratings each participant provided
for the
different
performance
levels.
This
was
essentially
the
same measure used
in the
previous
seven
studies. Goal
valence
was
computed
by
taking
the
attractiveness
rating that corresponded
to the
performance level subsequently chosen

402
A.
MENTO,
E.
LOCKE,
AND H.
KLEIN
Table
4
Means,
Standard
Deviations,
and
Intercorrelations
Between
Goals,
Valence,
Instrumentality,
and
Instrumentality
Factor
Scores
(Study
7)
Variable
1
.
Assigned goal
2.
Achievement valence
3.
Instrumentality
total
4.
Factor
1
(pride,
respect,
and
confidence)
5.
Factor
2
(job benefits)
6.
Factor
3
(school benefits)
7.
Factor
4
(career
and
life
benefits)
M
2.06
5.21
52.04
26.65
9.56
8.32
4.66
SD
0.81
1.08
17.17
9.61
3.74
3.99
2.01
1234
—
-.76
.79 .80
—
-.59 -.59
—
.96
—
5
.55
-.39
.81
.67
—
6
.78
-.59
.85
.77
.63
—
7
.27
-.23
.56
.48
.47
.31
—
Note.
N =
251.
In
this study, assigned goals
and
personal goals were
the
same.
All
correlations
are
significant
at p <
.01.
by
the
participant
as his or her
goal.
Valence
slope
was
measured
by
subtracting
the
attractiveness rating
of the
performance level immedi-
ately
below
the
participant's
personal goal from
the
attractiveness rat-
ing
of the
performance level chosen
by the
participant
as his or her
goal. Minimum goal
was
computed
as the
level
of
performance
at
which
the
participant's response pattern became positive.
Maximum
goal
was
computed
as the
performance level
at
which
the
participant's
response pattern reached
its
highest point (asymptote).
Self-efficacy.
Participants were also asked
to
rate their perceived
probability
of
attaining those same
six
possible performance levels.
Self-efficacy
strength
was
computed
by
summing
the
probability
rat-
ings
participants provided
for the six
different
performance levels.
Personal
goals, ability,
and
performance.
Participants were asked
to
indicate
the
performance level that they intended
to try for on the
second
exam.
Participants
were
also
asked
to
report
their cumulative
grade point average,
and
this
was
used
as a
surrogate measure
of
task
ability.
The
scores
participants actually received
on the
second exam
were
obtained
from
the
course
instructor
and
served
as the
measure
of
performance.
The
split-half reliability coefficients calculated
for
achievement
valence
and
self-efficacy
were
.70 and
.85, respectively.
After
correction
for the
reduced scale length with
the
Spearman-
Brown
formula,
the
reliabilities were
.83 and .92 for
achievement
va-
lence
and
self-efficacy,
respectively.
Results
Means,
standard deviations,
and
intercorrelations among
the
variables
are
presented
in
Table
5.
Valence
Table
5
indicates that valence, regardless
of
how
it was
opera-
tionalized,
was
significantly
correlated with ability
and
perfor-
mance.
With
one
exception
(valence
slope), valence also corre-
lated
significantly
with
personal goal
level
and
self-efficacy.
The
significant
relationships were
all
negative
for the
achieve-
ment
valence operationalization
and all
positive
for
the
remain-
ing
four
valence measures,
which
is
logical
given
the
different
operationalizations.
With
ability
and
performance
as
depen-
dent variables, goal valence, valence slope,
and
maximum goal
yielded
correlations that were
significantly
smaller than those
with
achievement valence
and
minimum goal
and
that were
not
significantly
different
from
each other.
This pattern also
held
true
with
personal goal
as the
depen-
dent variable,
with
two
exceptions.
The
correlation with mini-
mum
goal
was not
significantly
higher than that with goal
va-
lence,
and the
correlation
with
outcome valence
was
signifi-
cantly
higher than that
with
valence slope.
Achievement
Valence
as a
Measure
of
Goals
Multiple
regression analyses were employed
to
assess
whether
achievement valence
was
completely redundant
with
measures
of
personal goals.
Ability,
entered
in the first
hierar-
chical
step, accounted
for 22% of the
variance
in
performance.
Table
5
Means,
Standard
Deviations,
and
Intercorrelations
(Study
8)
1.
2.
3.
4.
5.
6.
7
8.
9.
Variable
Ability
Personal goal
Achievement
valence
Goal
valence
Valence slope
Minimum
goal
Maximum
goal
Self-efficacy
Performance
M
2.8
87.86
2.12
1.51
2.41
78.06
79.24
447.12
79.23
SD
0.49
5.25
2.80
1.06
1.18
5.67
5.35
73.75
9.24
1
.51*
-.56*
.15*
.17*
.53*
.28*
.44*
.44*
2
—
-.61**
.44**
-.01
.56**
.37**
.61**
.42**
3
—
.08
-.27**
-.89**
-.56**
-.45**
-.43**
4
—
.24**
-.11*
-.04
.28**
.18**
5
—
.29**
-.04
-.14
.18**
6
—
.41**
.37**
.38**
789
—
.19**
—
.21**
.30**
—
Note.
N
ranges from
232 to
356.
*p<.OS.
**p<.Ql.

GOAL LEVEL,
VALENCE,
INSTRUMENTALITY
403
Self-efficacy,
entered
in the
second
step,
accounted
for a
signifi-
cant increment (2%)
of the
remaining performance variance.
Personal goals
and
achievement valence were entered next
in
two
additional hierarchical
steps.
In the first
analysis, achieve-
ment valence
was
entered
in the
third
step
and
personal
goals
in
the
fourth.
In the
second analysis,
the
order
of
entry
was re-
versed.
The
results
of
these analyses
are
presented
in
Table
6.
When
achievement valence
was
entered third,
it
accounted
for
a
significant
4%
increment
in
performance. When personal
goal
was
entered fourth, however,
an
additional
3% (p <
.01)
of
the
remaining variance
was
also explained. When
the
order
was
reversed
and
personal
goal
was
entered
third,
a
significant
6%
of
performance
variance
was
explained. When achievement
valence
was
then entered fourth,
it
accounted
for a
nonsignifi-
cant
1
%
increment.
Discussion
These
analyses suggest that
personal
goals
and
achievement
valence,
while
highly
related,
are not
completely redundant
variables.
These same regression analyses were conducted
on
another
data
set
collected
by
Howard
J.
Klein. Although Klein
(in
press)
did not
report these analyses,
he
also measured
ability,
self-efficacy,
personal goal,
and
achievement valence
in
that
study
and
obtained
a
similar pattern
of
results. Personal goal
accounted
for
11%
of the
performance variance
(p <
.01)
when
entered
after
ability
and
self-efficacy.
When achievement
va-
lence
was
entered
in the
fourth
step,
it
accounted
for an
addi-
tional
1
% of the
performance variance
(p <
.05). When achieve-
ment
valence
was
entered third,
followed
by
personal goal,
va-
lence
accounted
for 9% of the
performance variance
and
personal goal
for 3% (p <
.01
for
both).
The
similar
pattern
of
results
in
Study
8 and the
previous
studies
with
respect
to
goals
and
valence suggests that
a
bipolar
measures
of
attractiveness
(as
compared with unipolar mea-
sures)
and
measures
of
outcome attractiveness
(as
compared
with
measures
of
anticipated satisfaction)
are
getting
at the
same phenomenon. Furthermore,
the
results suggest that
no
valence
measure works better than
the
summated measure,
al-
Table
6
Results
of Two
Regressions
of
Performance
on
Ability,
Self-
Efficacy,
Achievement
Valence,
and
Personal
Goal
(Study
8)
Step Variable
I
2
3
4
Ability
Self-efficacy
Achievement
valence
Personal goal
.22
.02
.04
.03
.47
.17
-.26
.27
48.20**
5.50*
10.15**
8.29**
Total
.32
19.84*
1
2
3
4
Ability
Self-efficacy
Personal goal
Achievement
valence
.22
.02
.06
.01
.47
.17
.33
-.16
48.20**
5.50*
15.35**
3.32
Total
.32
19.84*
>p<.05.
**;><.01.
though
Klein's minimum goal measure
was
comparable. This
was
also
the
case
in
Study
4
and,
to a
lesser extent,
in
Study
5
(although
we
have
not
presented
these
data
in
this article).
Conclusion
Taken
as a
group, these eight studies more
fully
reconcile
goal
theory
and
expectancy theory
from
the
valence
and in-
strumentality
perspective.
At the
same time, they point
out a
nonobvious limitation
in the way
people
usually think about
the
relation
of
valence, instrumentality,
and
goals.
The
sign
of
the
relationship between goal
level
and
(summated) achieve-
ment
valence
was
opposite
to the
sign
of the
relationship
be-
tween
goal
level
and
outcome instrumentality. This
difference
must
be
taken into account
in any
attempt
to
develop expec-
tancy
theory equations
for
predicting
the
performance
of
peo-
ple
trying
to
reach
particular
goals.
In an
expectancy theory
prediction equation,
the
achievement valence score needs
to be
weighted
negatively,
and the
instrumentality rating needs
to be
weighted
positively.
It
must
be
stressed that
the
negative weighting
of
valence
does
not
undermine expectancy theory.
In
these studies,
the
negatively
weighted achievement valence
was
roughly equiva-
lent
to the
goal-level measures, suggesting that
these
measures
are two
sides
of a
single coin. Klein's data (Study
8)
allowed
us to
test this interpretation.
The
results indicated that goal
level
and
achievement
valence scores
are
partly,
though
not
wholly,
inter-
changeable.
Both appear
to
explain
some unique performance
variance.
The
reasons
for
this
are not
entirely clear
and
need
to
be
explored
in
subsequent research. Whereas
the
valence mea-
sure
may be
more reliable
because
it is
based
on
multiple com-
ponents,
the
goal measure
may be
more valid because
it
reflects
more accurately what
the
subject
is
trying
to do.
Future studies
might
use
multiple
measures
of
personal goal (Locke
&
Bryan,
1968)
as
well
as
multiple valence measures
in
order
to
compare
their interrelationships
and
efficacy
in
predicting unique vari-
ance
in
performance.
A
second
type
of
valence, outcome valence,
was not
exam-
ined
in
these eight studies. Outcome valence pertains
to the
value
of the
outcomes
to
which
the
various goals lead,
but be-
cause these values
do not
vary across goal levels,
any
difference
in
the
Outcome
Valence
X
Instrumentality
product
across
goal
levels
is due
solely
to
instrumentality. Thus,
the
Outcome
Va-
lence
X
Instrumentality product
will,
like instrumentality,
be
positively
associated
with
goal
level
(when
beneficial
outcomes
are
considered).
Let us now
consider
the
more general implications
of the
finding
that goal level
was
negatively related
to
achievement
valence.
The
differences
between people with
high
standards
and
those
with
low
standards
is
that
the
former must
accom-
plish
more
to
satisfy
themselves than
the
latter.
If it is
assumed
that
self-satisfaction
is
a
meta-goal
desired
by
most people, then
it
follows
that those
who
demand more
of
themselves
will
do
more
(in
terms
of
quantity,
quality,
or
both) than those
who
demand less
of
themselves when they
act or
before they cease
action
in a
given
sphere.
One can
observe this phenomenon
in
operation
in
many
areas
of
life,
including school
and
work settings.
For
example,
there
are
students
who
become quite upset when they
get a

404
A.
MENTO,
E.
LOCKE,
AND H.
KLEIN
grade
of B or C on an
exam. When questioned, they
will
say,
in
effect,"
I am not the
type
of
person
who
gets
Cs or Bs. I am a B
or A
student." Others
are
quite happy with
low
grade outcomes.
Similarly,
there
are
employees
who do not
care
how low
their
output
is or how
shoddy their work
is,
whereas others
will
sim-
ply
not
settle
for
doing
a
second-rate job. People
who
demand
a
lot of
themselves
in
their careers
are
usually called ambitious.
Ambitious
people
are
those
who
want
to get
ahead. Howard
and
Bray
(1988)
found
that ambition
was the
single best motiva-
tional
predictor
of
career progress
in
their 20-year
follow-up of
AT&T
managers.
This
is not to
claim
that
the
same
level
of
standard applies
to all
areas
of a
person's
life.
It is
likely
that
there
are
both trait-like
and
situationally
specific
aspects
to
standard setting, depending
on a
person's values
and
abilities.
Furthermore,
it is
undoubtedly
the
case
that setting high
standards
is
dependent,
in
part,
on
external
influences.
For
example,
a
child's parents
and
teachers play
a
role
by
encourag-
ing
high, low,
or no
standards—which
may or may not be ac-
cepted
by the
child.
In the
working world,
the
norms
of the
workplace
and the
quality
of
leadership
can
play
a
role
in en-
couraging
or
discouraging high standards (Peters
&
Waterman,
1982).
Why,
if it is
easier
to get
satisfaction
from
low
standards than
from
high standards, does
not
everyone
opt for low
standards?
Two
answers
can be
offered.
First,
if
standards depend
on
one's
self-concept
or
self-image,
it may be
much more
satisfying
to
meet
the
standards than
to
change
the
self-image, because
the
latter
may be
tied
to
one's
total
identity
as a
person.
Second, meeting high standards typically leads
to
greater
practical rewards
in the
real world than
does
meeting
low
stan-
dards.
For
example, good grades
in
high school lead
to
school
honors,
the
possibility
of
going
to
college, scholarships,
and the
respect
of
(some)
peers
and
parents
to a
greater
degree
than
do
low
grades.
In the
world
of
work, doing
a lot of
good quality
work
generally leads
to
higher salaries, greater
job
security,
and
more promotions than doing little work
or
poor quality work.
Making
a
higher
salary,
in
turn, leads
to
other
generally desir-
able outcomes with respect
to
housing, cars, vacations,
finan-
cial
security,
and the
like.
In
these eight studies, high goals were rated
as
more instru-
mental
than
low
goals with
respect
to
both immediate, internal
benefits
(viz.,
getting
a
sense
of
achievement, developing skills,
and
proving oneself)
and
remote, external benefits (such
as
career
and
life
success).
To
reverse
the
previous question, given
the
obvious practical
benefits,
why
does
not
everyone
opt for
high goals?
First,
high
goals
in the
real world have higher
costs
in
terms
of
effort
and
short-term
pleasures foregone (e.g., many entrepreneurs totally
neglect
their families
to
succeed
in
business). Thus, many peo-
ple
choose
not to try for
high goals especially
if
such goals
are
not
already part
of
their self-image. Second,
all
people
do not
think
they
can
attain high goals. People
differ
markedly
in
their
self-efficacy
for any
given task (task-specific self-confidence;
Bandura,
1986);
self-efficacy
affects
performance both directly
and
indirectly
through
its
effects
on
goal
choice
and
commit-
ment
(Locke
et
al.,
1984; Locke
&
Latham, 1990). Locke
and
Latham
(1990)
concluded that goal choice
is an
integration
of
what
one
wants
and
what
one
believes
is
possible.
Another
relevant
question
is
whether pursuing
high
goals
or
being
urged
to do so by an
employer
affects
individuals' health
and
well-being.
The
effect
of
self-chosen
high standards should
be
beneficial
with
respect
to
pride
and
practical
benefits
so
long
as the
person
is not
overly compulsive.
A
type
of
compul-
sive
is the
Type
A
personality,
who
tries
to do
more
and
more
in
less
and
less time. These individuals
do
experience more stress
than
do
non-Type
As.
(Recent work, e.g., Wright,
1988,
suggests
that
Type
A
compulsions stem
from
underlying
hostility
and
that this
hostility,
rather than work involvement,
is
responsible
for the
negative health consequences
of
this syndrome.)
If
high
standards
are
organizationally
imposed,
high
stress
can
also
be the
result (Locke
&
Shaw,
1984).
Japanese organiza-
tions
are
especially prone
to
driving their employees
to
continu-
ous
improvement
(in
Japanese, this
is
called
kaizen;
Imai,
1986)
in
virtually
every aspect
of
performance (quantity,
quality,
cost
savings,
etc.). This
has
proved
to be a
very
successful
strategy
for
Japanese companies, especially because Japanese workers
are
willingto
tolerate
such pressures. Furthermore,
challenging
em-
ployees
by
raising standards
on an
incremental basis
may be
more acceptable
and
less
stressful
than making standards high
all
at
once.
Finally,
the
experience
of
stress
is
fundamentally
dependent
on
self-efficacy
(Bandura, 1986); when
people
feel
confident
of
meeting
a
challenge, they experience less stress
than when their
self-efficacy
is
low.
It
should
be
possible
for
organizations,
therefore,
to
mitigate
the
effects
of
high stan-
dards through
careful
selection
and
training.
The
question
of the
effects
of
standards
on
well-being
should
not be
confined
just
to
high
standards.
The
same question
can
be
asked with respect
to low
standards.
For
example, people
with
low
standards
may
achieve success
and
satisfaction more
easily
than those
with
high standards,
but
they
are
also
likely
to
have
lower pride
in
accomplishment
and
fewer
practical
re-
wards.
The
latter,
in
turn, could
cause
greater
life
stress
(e.g.,
as a
result
of
poverty).
There
are
other
possibilities
as
well,
for
example, people
who
had
high ambitions when they were young
and
gave them
up
under
the
demands
of
life's
pressures.
But all
these speculations
are
grist
for
later studies.
References
Bandura,
A.
(1986). Social
foundations
of
thought
and
action:
A
social-
cognitive
theory. Englewood
Cliffs,
NJ:
Prentice-Hall.
Bandura,
A.
(1988). Self-regulation
of
motivation
and
action through
goal
systems.
In V
Hamilton,
G.
Bower,
& N.
Frijda
(Eds.),
Cognitive
perspectives
on
emotion
and
motivation
(pp.
37-61).
Dordrecht,
the
Netherlands: Kluwer Academic.
Dachler,
H. P., &
Mobley,
W
H.
(1973). Construct validation
of an
instrumentality-expectancy-task-goal
model
of
work motivation:
Some theoretical boundary conditions [Monograph
].
Journal
of
Ap-
plied
Psychology,
58,
397-418.
Garland,
H.
(1982).
Goal
levels
and
task performance:
A
compelling
replication
of
some compelling results. Journal
of
Applied
Psychol-
ogy,
67,
245-248.
Garland,
H.
(1985).
A
cognitive mediation theory
of
task goals
and
human performance. Motivation
and
Emotion,
9,
345-367.
Howard,
A.,
&
Bray,
D. W
(1988).
Managerial lives
in
transition.
New
York:
Guilford
Press.
Imai,
M.
(1986).
Kaizen:
The key to
Japans
competitive
success.
New
York:
Random House.

GOAL
LEVEL,
VALENCE,
INSTRUMENTALITY
405
Keppel,
G., &
Zedeck,
S.
(1989). Data analysis
for
research
designs.
New
York:
Freeman.
Klein,
H. J. (in
press). Further evidence
for the
integration
of
goal
setting
and
expectancy theories. Organization Behavior
and
Human
Decision
Processes.
Locke,
E. A.
(1964).
The
relationship
of
intentions
to
motivation
and
affect.
Unpublished doctoral dissertation, Cornell University,
De-
partment
of
Psychology.
Locke,
E. A.
(1966).
The
relationship
of
intentions
to
level
of
perfor-
mance.
Journal
of
Applied
Psychology,
50,
60-66.
Locke,
E. A.
(1982).
Relationship
of
goal
level
to
performance with
a
short
work
period
and
multiple
goal
levels. Journal
of
Applied
Psy-
chology,
67,
512-514.
Locke,
E.
A.,
&
Bryan,
J. F.
(1968).
Grade goals
as
determinants
of
academic achievement. Journal
of
General
Psychology,
79,
217-228.
Locke,
E.
A.,
Frederick,
E.,
Lee,
C,
&
Bobko,
P.
(1984).
Effects
of
self-efficacy,
goals,
and
task strategies
on
performance. Journal
of
Applied
Psychology,
69,
241-251.
Locke,
E.
A.,
&
Latham,
G. P.
(1990).
A
theory
of
goal
setting
and
task
performance.
Englewood
Cliffs,
NJ:
Prentice-Hall.
Locke,
E.
A.,
&
Shaw,
K.
N.
(1984).
Atkinson's inverse-U curve
and the
missing
cognitive variables.
Psychological
Reports,
55,
403-412.
Matsui,
T,
Okada,
A.,
&
Mizuguchi,
R.
(1981).
Expectancy theory
prediction
of
goal theory postulate "the harder
the
goals,
the
higher
the
performance." Journal
of
Applied
Psychology,
66,
54-58.
Mento,
A.
J.,
Cartledge,
N. D., &
Locke,
E. A.
(1980). Maryland
vs
Michigan
vs
Minnesota: Another look
at the
relationship
of
expec-
tancy
and
goal
difficulty
to
task performance.
Organizational
Behav-
ior
and
Human
Performance,
25,
419-440.
Meyer,
J. P., &
Gellatly,
I. R.
(1988).
Perceived performance norm
as a
mediator
in the
effect
of
assigned goal
on
personal goal
and
task
performance. Journal
of
Applied
Psychology,
73,
410-420.
Peters,
T.
J.,
&
Waterman,
R. H.
(1982).
In
search
of
excellence.
New
York:
Harper
&
Row.
Wright,
L.
(1988).
The
Type
A
behavior pattern
and
coronary artery
disease. American Psychologist,
43,
2-14.
Received
July
3,1990
Revision
received December
11,1991
Accepted December
12,1991
•
Low
Publication
Prices
for APA
Members
and
Affiliates
Keeping
You
Up-to-Date:
All
APA
members (Fellows; Members; Associates,
and
Student
Affiliates)
receive—as
part
of
their annual
dues—subscriptions
to the
American
Psychologist
and
APA
Monitor.
High School Teacher
and
International
Affiliates
receive subscriptions
to the APA
Monitor,
and
they
can
subscribe
to the
American
Psychologist
at a
significantly reduced rate.
In
addition,
all
members
and
affiliates
are
eligible
for
savings
of up to
60%
(plus
a
journal
credit)
on all
other
APA
journals,
as
well
as
significant
discounts
on
subscriptions
from
coop-
erating
societies
and
publishers (e.g.,
the
American Association
for
Counseling
and
Develop-
ment,
Academic Press,
and
Human Sciences Press).
Essential
Resources:
APA
members
and
affiliates
receive special rates
for
purchases
of APA
books, including
the
Publication
Manual
of
the
APA,
the
Master
Lectures,
and
Journals
in
Psychol-
ogy:
A
Resource
Listing
for
Authors.
Other
Benefits
of
Membership: Membership
in APA
also provides eligibility
for
low-cost
insurance plans covering life, income protection,
office
overhead, accident
protection,
health
care, hospital indemnity, professional liability, research/academic professional liability, stu-
dent/school
liability,
and
student health.
For
more
information,
write
to
American
Psychological
Association,
Membership
Services,
750
First
Street,
NE,
Washington,
DC
20002-4242,
USA