The Effects of Learning Goal Difficulty Level and Cognitive Ability
Gary P. Latham
University of Toronto
Gerard Seijts and Dan Crim
University of Western Ontario
The relationship between the difficulty level of a learning goal and a person’s (N⫽146) performance
on a task that required the acquisition of knowledge to perform effectively was examined. Multiple
hierarchical regression analysis revealed that the higher the learning goal, the higher the person’s
performance. Cognitive ability and goal commitment also positively affected performance. The results
showed that the person’s cognitive ability moderated the learning goal–performance relationship.
Contrary to previous research findings on performance goals for tasks that are straightforward for people,
the performance of individuals lower in cognitive ability was more positively affected by the setting of
a difficult learning goal than was the case for people higher in cognitive ability.
Keywords: learning goal, cognitive ability, performance
A core premise of Locke and Latham’s (1990, 2002) goal-
setting theory is that, given goal commitment, there is a linear
relationship between the difficulty level of a goal and a person’s
subsequent task performance. Commitment is the sine qua non for
goal setting. As Locke and Latham (1990) stressed, a person who
is not committed to a goal by definition does not have one.
More than 1,000 studies have shown that when people have the
ability to perform a task, and the task only requires the choice to
exert effort and to persist until it is accomplished, the setting of and
commitment to a specific high goal leads to higher performance
than a vague goal such as an exhortation to “do your best”
(Mitchell & Daniels, 2003). On a task that is objectively complex
in terms of number of components and the dynamic relationship
amongst them (Wood, 1986; Wood, Mento, & Locke, 1987), a
fourth moderator of the goal–performance relationship, in addition
to choice, effort, and persistence, is strategy development (Locke,
Shaw, Saari, & Latham, 1981). People draw on their “repertoire of
knowledge and experience from which to develop a suitable plan”
(Locke & Latham, 1990, p. 96). For example, drivers developed
ways to assess the weight of their respective logging trucks so as
to attain a specific high goal for their performance (Latham &
Baldes, 1975). They also used their two-way radios to coordinate
with one another so that there was always a truck available when
the timber was ready to be loaded (Latham & Saari, 1982). People
with a specific high goal for their test score started to write notes
in their margins (Terborg & Miller, 1978). People who were trying
to lose weight chose low calorie foods and refused second helpings
in order to attain a specific difficult goal for losing weight (Ban-
dura & Simon, 1977). In each of these studies, the participants
already possessed the requisite knowledge for developing an ef-
fective strategy or plan for attaining their respective goals. Goal
attainment only required choice, effort, persistence, and reliance
on knowledge the person already possessed. This was not the case
for participants in Kanfer and Ackerman’s (1989) and Mone and
Shalley’s (1995) experiments.
When people are in the declarative stage of learning, Kanfer and
Ackerman (1989) found that urging them to do their best resulted
in higher performance than assigning them a specific high goal to
attain regarding their performance. In their review of this study,
Locke and Latham (1990) made three observations. First, the
“subjects had to learn the best strategy to use” (p. 105). Second,
because the Air Force trainees had no prior experience or training
at the air traffic control task, they had no proven strategies or
problem solving processes to fall back on. Third, those participants
with a specific high goal for their performance likely felt pressure
to perform effectively immediately. They were not informed that
they should take the time to learn the task before focussing on
performing well. Thus, people with a specific high goal, relative to
those who were urged to do their best, “may have had tunnel
vision, focussing more on the desire to get immediate results rather
than learning the best way of performing the task” (Locke &
Latham, 1990, p. 105).
Mone and Shalley (1995) replicated Kanfer and Ackerman’s find-
ings. Using a human resources staffing task that required the acqui-
sition of knowledge before it could be performed effectively, they too
found that a specific high goal had a detrimental effect on perfor-
mance relative to urging people to do their best. Contrary to their
expectations, Mone and Shalley also found that multiple perfor-
mance trials over a three-day period did not lead to the acquisition
of the knowledge necessary to perform the task when people were
committed to attaining a specific high goal for improving their
performance. In fact, the deleterious effect on performance in-
creased over the three-day period as a result of having a specific
high goal. The very opposite was true of the three day performance
of those who were simply urged to do their best. The mediating
variable that explained these findings was strategy. Those with a
Gary P. Latham, Joseph L. Rotman School of Management, University
of Toronto, Toronto, Canada; Gerard Seijts and Dan Crim, Richard Ivey
School of Business, University of Western Ontario, London, Canada.
Correspondence concerning this article should be addressed to Gerard
Seijts, Richard Ivey School of Business, University of Western Ontario,
1151 Richmond Street North, London, Ontario, N6A 3K7, Canada. E-mail:
Canadian Journal of Behavioural Science Copyright 2008 by the Canadian Psychological Association
2008, Vol. 40, No. 4, 220–229 0008-400X/08/$12.00 DOI: 10.1037/a0013114
specific high goal to attain regarding their performance appeared
to mindlessly switch strategies relative to their counterparts who
had been urged to do their best. Those in the latter condition
appeared to search systematically for one or more effective strat-
egies. Mone and Shalley concluded that the setting of a specific
high goal for performance interfered with learning the strategies
necessary for the person to perform the task effectively. A limita-
tion of goal-setting theory appeared to have been found.
Kanfer and Ackerman’s (1989) results, Locke and Latham’s
(1990) critique of them, as well as Mone and Shalley’s (1995)
findings regarding strategy search support the concept of a learn-
ing goal. Whereas a performance goal refers to the desired aim or
end of an action in terms of level of performance to be attained on
a task (Locke & Latham, 1990), a learning goal refers to learning
the requisite strategies, processes, or procedures for performing
effectively as opposed to relying on the knowledge and skill one
already possesses. A specific high-performance goal, Locke and
Latham (1990) argued, only has a beneficial effect when a person
already possesses the knowledge for developing suitable plans or
strategies for attaining it; hence, the importance of studying the
concept of learning goals when this is not the case. Latham and his
colleagues speculated that a specific high-learning goal only has a
beneficial effect when the requisite knowledge for performing
effectively is not known to the individual.
Following Locke and Latham’s (1990) explanation of Kanfer
and Ackerman’s (1989) results, Winters and Latham (1996) asked
two questions. The first was based on goal-setting theory: Does a
learning goal, as does a performance goal, have a positive effect on
subsequent task performance? The second question was based on
a moderator variable explicated in the theory, namely, ability.
Does the type of task, where the person has/has not the requisite
ability to perform it, moderate the effect of a learning versus a
performance goal? They found that when people were informed of
the correct ways to schedule classes for a university, a specific
high-performance goal led to higher performance than either urg-
ing people to do their best, or assigning them a specific high-
learning goal. When people were not informed how to perform this
task, Winters and Latham replicated the results obtained by both
Kanfer and Ackerman (1989) and Mone and Shalley (1995).
Urging people to do their best led to higher performance than the
setting of a specific high-performance goal. However, Winters and
Latham also found that the assignment of a specific high-learning
goal for the acquisition of knowledge led to higher performance
than a specific high goal for performance, or the exhortation to do
one’s best. In short, a learning goal, as does a performance goal,
has a positive effect on performance. But, the former is true only
when people have yet to acquire the knowledge to perform the task
effectively; the latter is true only when people already have the
ability to do what is required of them. These findings are consistent
with goal-setting theory regarding ability as a moderator of the
goal–performance relationship. Thus, Kanfer and Ackerman’s
(1989) finding as well as Mone and Shalley’s (1995) do not point
to a limitation of goal-setting theory. Rather, they point to the
necessity of taking into account the differential effect of two
different types of goals on performance, namely, learning versus
performance. Subsequent research has been conducted to deter-
mine similarities and differences between these two types of goals.
Drach-Zahavey and Erez (2002) replicated the above findings
on a different task, namely, a simulation requiring predicting stock
values. People who were given a learning goal regarding strategies
to be discovered/learned had significantly higher performance than
those who either had a specific high-performance goal or who
were urged to do their best. Similarly, Kozlowski and Bell (2006)
found that on a radar-tracking task where participants had yet to
learn how to perform it, a learning goal led to higher performance
than a performance goal.
In a field experiment, Latham and Brown (2006) found that first
year MBA students who set a specific high-learning goal regarding
learning ways to make their education more meaningful to them
had a significantly higher grade point average at the end of the
academic year than those who set a specific high distal perfor-
mance goal or were urged by the dean to do their best to obtain a
meaningful education. Furthermore, their satisfaction with the
MBA programme was significantly higher than it was in those
other two conditions.
In summary, both labouratory and field experiments have shown
that setting a learning goal has a differential effect on performance
relative to setting a performance goal. Both types of goals are
mediated by task strategies but in different ways. Performance
goals are mediated by knowledge/skill the individual already pos-
sesses when initially performing a task. Learning goals are medi-
ated by knowledge/skill the person has yet to acquire when ini-
tially performing a task. Hence both are moderated by ability, but
in different ways. A specific high-performance goal should be set
only when a person already has the ability to attain it. A specific
high-learning goal should be set only when the person lacks the
requisite ability to perform the task.
Only seven experiments, to the authors’ knowledge, have been
conducted on the effect of a specific high-learning goal on task
performance (Kaplan, Erez, & Van Dijk, 2004; Kozlowski & Bell,
2006; Latham & Brown, 2006; Noel & Latham, 2006; Seijts &
Latham, 2001; Seijts, Latham, Tasa, & Latham, 2004; Winters &
Latham, 1996). In each of those seven experiments, the assumption
of a relationship between goal difficulty level and subsequent
performance was assumed. This assumption, however, has yet to
be empirically tested. The purpose of the present study was to
address two questions central to goal-setting theory. First, is level
of difficulty a property of a learning goal that affects a person’s
performance? Second, if the answer is yes, does cognitive ability
moderate this relationship? For example, it would seem logical that
individuals higher in cognitive ability might be better able to direct
their attention to task knowledge acquisition than people with
lower cognitive ability. However, as is discussed below, an oppo-
site argument can also be made; hence, the need for empirical
research to find the answer.
Goal setting theory asserts that there is a linear relationship
between the degree of goal difficulty and performance. The linear
relationship levels off only when individuals reach the limits of
their ability. This assertion has been supported in literally hundreds
of empirical labouratory and field studies of specific high-perfor-
mance goals (Latham, 2007; Latham & Locke, 2007; Locke &
Learning goals are typically set in terms of a specific number of
task-relevant strategies to be learned for successful completion of
a task. Thus, the first hypothesis of this experiment was as follows:
LEARNING GOALS AND COGNITIVE ABILITY
Hypothesis 1: There is a positive relationship between the
difficulty level of a learning goal and the person’s perfor-
mance on a task that requires the acquisition of knowledge.
The rationale for this hypothesis is that, consistent with goal-
setting theory, a more difficult learning goal leads to greater
cognitive effort to acquire task-relevant strategies than an easier
goal, given that the individual is committed to goal attainment.
Learning appropriate strategies, processes, or procedures increases
a person’s performance as opposed to sheer effort and persistence
alone when a person lacks the requisite knowledge to perform
effectively. This finding, although not tested in the seven previous
experiments on learning goals, can be inferred from them. Never-
theless, the counterargument can be made that a focus on learning
processes or procedures for performing effectively in those exper-
iments is not qualitatively the same as a focus on attaining a
specific performance level as is required in the vast majority of
goal-setting studies. Thus, it is possible that a learning goal oper-
ates differently from a performance goal, and hence it may not
have the same moderating variables.
In his review of the literature, Vroom (1964) concluded that
when a person’s motivation is high, the role that ability plays on
performance is also high. That is, people with high ability likely
show a proportionately greater improvement in their performance
from an increase in motivation than do those individuals with low
ability. Subsequent research on goal setting supported this conclu-
sion. Specifically, Locke (1965) found that goal setting has a
greater effect on the performance of people with high as opposed
to low ability. This is because the more challenging the goal, the
more free rein people have to perform in accordance with their
skills, and thus the higher the association between one’s ability and
performance (Locke, 1982). Locke’s findings, however, were
based on tasks where people already possessed the knowledge to
perform well. Ability was defined in terms of a median split on
level of performance on the initial practise trial. Cognitive ability
was not assessed.
On a task where a person lacks the requisite knowledge, per-
formance is primarily a function of learning task-relevant strate-
gies (Latham, 2007; Locke, 2000). As discussed earlier, the as-
signment of a challenging learning goal has been shown to have a
positive effect on the acquisition of task-relevant knowledge and
hence subsequent performance. A moderating variable is likely to
be an individual’s cognitive ability.
As Seijts and Latham (2005) explained, the purpose of a learn-
ing goal is to stimulate one’s intellect, to engage in discovery, to
acquire knowledge and integrate it with prior information, and to
“think outside the box.” Commitment to a challenging perfor-
mance goal, on the other hand, results in the choice to exert effort
and to persist until a desired level of performance is attained. The
performance goal cues people to use the knowledge, skill and
ability one already possesses (Locke & Latham, 1990). Commit-
ment to a challenging learning goal leads to the choice to system-
atically learn new ideas. The resulting behaviour is to execute a
specific number of newly acquired ideas in order to test newly
formed hypotheses (Seijts & Latham, 2005). The resulting behav-
iour of a person who commits to a difficult performance goal is to
focus on ways to quickly implement knowledge and skills that
have already been acquired in order to perform effectively (Locke
& Latham, 1990).
Cognitive ability or intelligence has been shown consistently to
predict learning and performance on tasks where people lack the
knowledge to perform effectively (Ackerman, Kanfer, & Goff,
1995; Hunter, 1986; Ree & Earles, 1991). As noted earlier, on a
task where cognitive mastery had yet to take place, Kanfer and
Ackerman (1989) found that following the declarative stage of
learning, the beneficial effects of setting a specific, high-perfor-
mance goal “are most likely to accrue to low-ability rather than to
high-ability persons” (p. 687). This is because such tasks are more
demanding for people with low rather than high cognitive ability.
Once initial learning has occurred, attentional demands go down,
and the positive effects of goal setting should prove to be espe-
cially beneficial for low ability individuals. Higher cognitive abil-
ity people in the same learning situation, they said, may not get the
same “boost” from goal setting because goal setting from the
outset has already heightened effort allocation. People with rela-
tively low cognitive ability, who now exert a great deal of effort,
have more room to improve their performance than their high
ability counterparts who are already exerting a great deal of effort.
Thus, there can be a “ceiling effect” for individuals with higher
cognitive ability when a specific high-performance goal is set.
Kanfer and Ackerman’s finding has yet to be tested with regard to
setting a learning goal.
The present study is a constructive replication of Kanfer and
Ackerman’s finding regarding cognitive ability as a moderator of
the goal–performance relationship. As Barrick, Bradley, Kristof-
Brown, and Colbert (2007) stated, a constructive replication is a
study that assesses the same relationships amongst the same con-
structs investigated in an earlier study, but varies the operational-
ization of those constructs. In the present study, a learning rather
than a performance goal was assessed with regard to the moder-
ating effect of cognitive ability as opposed to physical skill. In
addition, the goal difficulty–performance relationship, which is a
core premise of goal-setting theory, was investigated with regard
to a learning rather than a performance goal. As Eden (2002) has
noted, replications that are different from previous studies are
required for advancing theory.
Individuals who commit to a high-learning goal must engage in
effortful cognitive processes during the declarative stage of learn-
ing where the acquisition of knowledge/strategy is needed, but has
yet to take place. Individuals with lower cognitive ability may
benefit more from the assignment of a more difficult than less
difficult learning goal than do those with higher cognitive ability.
This is because, as noted earlier, people with high cognitive ability
may not derive the same benefit from an emphasis on discovering
ways to master the task when a learning goal is given to them.
They likely do so automatically and hence do not need to be
instructed to do so. Thus, there should be an interaction effect
between a learning goal and cognitive ability, such that a higher
learning goal is more effective than a lower learning goal for
individuals with lower rather than higher cognitive ability. There-
fore, the second hypothesis tested in this experiment, contrary to
Vroom’s (1964) conclusion from his review of the literature and
Locke’s (1965) finding regarding setting a performance goal, was
222 LATHAM, SEIJTS, AND CRIM
Hypothesis 2: The difficulty level of an assigned learning goal
has a greater effect on the performance of individuals with lower
cognitive ability than it does on the performance of individuals
with higher cognitive ability.
Sample and Design
Eighty-three female and 63 male students, whose mean age was
18.66 years (SD ⫽2.21), participated in the study. The participants
were first or second year university students who were enrolled in
the prebusiness programme at a large Canadian university. All
participants received course credit for their participation. The 146
participants were randomly assigned to one of two goal conditions:
a low-learning goal or a high-learning goal. Each condition con-
tained 73 participants.
The participants were required to produce unique class sched-
ules comprised of five nonredundant university classes. The task
was divided into three 8-min trials. Previous research indicated
that three 8-min trials is a sufficient time-span for acquiring the
knowledge to complete class schedules correctly (Seijts & Latham,
2001; Winters & Latham, 1996).
The instructions provided six rules for producing class sched-
ules. The rules are as follows: (1) each schedule indicates the
course name, its code, meeting times, and section; (2) each sched-
ule must have five different classes scheduled on the same day; (3)
each schedule must be unique, that is, it cannot duplicate another
class schedule; (4) any course with a quiz section must have the
quiz section scheduled on the same day as the class; (5) no two
marketing courses can be scheduled within one hour of each other;
and (6) any speech communication lecture class must have a
labouratory class scheduled as well.
This task was used for three reasons. First, it meets the criterion
for complexity set forth by Wood (1986). That is, performance on
the task is not increased solely through effort or persistence.
Effective task performance requires learning effective task strate-
gies. Second, previous studies indicate that individuals perceive
this task as complex for them (Earley, 1985; Seijts & Latham,
2001; Winters & Latham, 1996). Third, scheduling is an organi-
sationally relevant task (e.g., railways, trucks, inventory schedul-
ing of supplies, etc.). Using organisationally relevant tasks in
labouratory studies enhances the generalizability of findings from
labouratory simulations to field settings (Latham, 2007; Latham &
Each participant received a package that included an explana-
tion of the task requirements, a class schedule list, blank schedules,
and a series of questionnaires (see Measures). Participants were
informed that the computer broke down, and that “the Office of the
Registrar has requested you to complete class schedules.”
Consistent with recommendations by Locke and Latham (1990),
participants were given a 4-min pretest prior to the manipulation of
the independent variables so that the person’s current ability could
be used as a control variable in the analyses of performance.
Participants were encouraged to “schedule as many classes as
possible within the 4-min period.”
In developing the task, Earley (1985) created four unique,
task-effective strategies for producing correct class schedules.
These strategies are: (1) recording class names and times chrono-
logically; (2) repeatedly scheduling the same subject; (3) schedul-
ing night classes; and (4) repeatedly scheduling the same section.
To the authors’ knowledge, no additional task-effective strategies
exist. The learning goal was framed as the number of unique,
task-relevant strategies to be identified and implemented. Some
participants were thus assigned an easier learning goal (to discover
and implement 1 or 2 strategies) than others, whereas others were
by definition assigned a more challenging learning goal (to learn 3
or 4 strategies). Defining a learning goal in terms of number of
task-relevant strategies to be discovered or learned for successful
completion of the task is consistent with Locke and Latham’s
(1990) goal-setting theory.
The experimental instructions were as follows: “There will be
three trials of 8 minutes each. Research has shown that thinking
about specific strategies to help you more quickly produce class
schedules results in the production of a larger number of schedules.
A pilot study of individuals with abilities similar to your own
indicated that a goal of discovering and implementing 1 or 2 (vs.
3 or 4) strategies is attainable. Research has shown that setting a
specific, yet attainable goal maximises productivity. Hence, your
goal for the next 24 minutes is to learn and implement 1 or 2 (vs.
3 or 4) strategies to produce class schedules.”
Perceived complexity of the task. A task can be “objectively”
complex in terms of meeting the criteria put forth by Wood (1986) yet
relatively straightforward for people who already have the ability to
perform it effectively (e.g., playing chess, obtaining funding for a
grant proposal, or making money in the stock market). A learning
goal, as explained in the introduction, only has a beneficial effect on
performance when a task is “subjectively” complex for an individual.
Thus, perceived complexity of the task was measured. Failure to
perceive the task as complex would explain why the specific high-
learning goal in this study did not have a positive effect on perfor-
mance. Perceived or subjective complexity of the task was measured
by five 5-point Likert-type items (e.g., “Many times, I had to cheque
one thing before I scheduled something else”) after the participants
completed the third 8-min trial. These items were used by Winters and
Latham (1996). Scale scores could range from 1 (not at all)to5(very
much so). This measure was included in the analyses because research
has shown that the extent to which individuals judge a task to be
complex for them affects motivation and subsequent performance
(Cervone, Jiwani, & Wood, 1991; DeShon, Brown, & Greenis, 1996).
Goal commitment. As noted in the introduction, goal commit-
ment was measured because a person who is not committed to a
goal by definition does not have one. Failure to commit to the goal
would therefore provide another explanation as to why an increase
in performance was not found in this study. Hence, it too was
measured. Commitment to the learning goal was measured prior to
each 8-min trial using five 5-point Likert-type items (e.g., “I am
strongly committed to pursuing this goal”) taken from Klein,
Wesson, Hollenbeck, Wright, and DeShon (2001). Scale scores
could range from 1 (completely disagree)to5(completely agree).
LEARNING GOALS AND COGNITIVE ABILITY
The goal referred to the specific number of strategies that was as-
signed to the participant. Commitment to the learning goal was
measured in order to determine whether the participants were in fact
attempting to attain it (Locke & Latham, 1990). The three commit-
ment scores were added to obtain an overall commitment score.
Learning. Learning was operationalized as the number of
unique strategies learned. In developing this task, Earley (1985), as
reported earlier, created four task-effective strategies for producing
correct class schedules. The extent to which each of these strate-
gies was identified and used was assessed by examining each class
schedule. For example, to measure the strategy of recording class
names and times chronologically, each schedule was examined to
determine whether classes and times written on the schedule
started with early morning classes on the first line of the schedule,
and ended with late classes on the last line of the schedule. One
point was given each time a particular strategy was used. The
strategies of repeatedly scheduling the same subject across the
completed schedules (e.g., business writing, accounting, or con-
sumer behaviour), repeatedly scheduling the same section (e.g.,
using the same 10:00 –10:50 a.m. Friday finance class across the
completed schedules), and recording class names and times chro-
nologically could be used only once for each schedule. In contrast,
the strategy of scheduling night classes could be used only twice;
from 4:00 to 5:20 p.m. and from 6:00 to 8:30 p.m. One point was
given each time a night class was scheduled.
Performance. Performance was operationalized as the number
of correct class schedules produced at the end of the 24-min period.
Cognitive ability. The Wonderlic Aptitude Test was used to
measure cognitive ability. The Wonderlic User’s Manual reports
test–retest reliability coefficients ranging from .82 to .94. The
internal consistency as assessed by Kuder–Richardson KR-20 is
.88. The predictive validity of the Wonderlic Aptitude Test with
performance is high. For example, using meta-analytic techniques,
Hunter and Hunter (1984) reported a predictive validity coefficient
of .63. Participants in the present experiment completed the test at
the beginning of the session, prior to reading the instructions of the
class-scheduling task. Participants were given 12 minutes to re-
spond to the 50 questions on the test.
Table 1 shows the means and SDs, as well as the intercorrela-
tions, of variables measured, collapsed across trials. The results of
the regression analyses are shown in Table 2.
Perceived complexity of the task. The coefficient alpha for the
five-item scale was .81. The mean overall score was 3.62 (SD ⫽
0.80), indicating that the task was viewed by the participants as
moderately complex. The Pearson correlation coefficient between
perceived complexity of the task and overall performance was
⫺.32 ( p⬍.05). That is, the higher the person perceived the
complexity of the task, the lower the person’s performance.
Goal commitment. The coefficient alphas for the five-item
goal commitment scale were .77, .86, and .88 for Trials 1, 2, and
3, respectively. The Pearson correlation coefficient between goal
commitment and overall performance was .29 ( p⬍.05). Thus, the
higher the commitment to the learning goal, the higher the person’s
performance. The reliability of commitment across the three trials
Learning. The first strategy, recording the classes and time in
chronological order, was used by 99% of the participants. The
second strategy, repeatedly scheduling the same subject, was im-
plemented by 98% of the participants. The third strategy, sched-
uling night classes, was implemented by 73% of the participants.
Descriptive Statistics and Correlations Among Variables Measured
1. Cognitive ability 24.14 4.63
2. Goal commitment 03.40 0.66 ⫺.01
3. Goal level 00.50 0.50 ⫺.02 ⫺.06
4. Task complexity 03.62 0.80 ⫺.10 ⫺.12 .11
5. Practice trial 00.86 0.33 .23
6. Strategies learned 03.39 0.66 .15
7. Performance 08.87 2.65 .40
Note.Nranges from 141 to 146. Goal level ranges from 0 ⫽“low- learning goal”to1⫽“high-learning goal.”
Results of Hierarchical Regression Analysis for Variables
Predicting Performance (N⫽141)
Practice trial 1.14
Cognitive ability .80
.19 .34 (2, 138) ⫽34.74
Practice trial .92
Cognitive ability .79
Complex task ⫺.72
Goal level .79
.17 .14 (3, 135) ⫽11.48
Practice trial .96
Cognitive ability 1.14
Complex task ⫺.72
Goal level .78
Goal Level ⫻
.34 .02 (1, 134) ⫽3.93
224 LATHAM, SEIJTS, AND CRIM
And finally, the fourth strategy, repeatedly scheduling the same
class section was used by only 70% of the participants. Thus, not
everyone learned the latter two strategies. This is further evidence
that the task was relatively complex in that 27% and 30% of the
participants failed to discover the third and fourth strategies, re-
spectively. A two-tailed ttest revealed that participants in the
high-learning goal condition learned more unique strategies (M⫽
3.52, SD ⫽0.60) than those in the low-learning goal condition,
M⫽3.26, SD ⫽0.69; t(144) ⫽2.43, p⬍.05. A series of
paired-sample ttests also showed that the number of unique
strategies identified increased over time: from 2.45 (SD ⫽0.88) on
Trial 1 to 2.73 (SD ⫽0.88) on Trial 2 to 2.93 (SD ⫽0.72) on Trial
3; all p’s ⬍.01. The number of unique task strategies learned by
the participants during the 24-min task correlated positively with
performance (r⫽.36, p⬍.01). Once the unique strategies were
identified, they could be used repeatedly in completing the sched-
ules. The number of times the various task strategies were actually
used by the participants correlated highly with performance (r⫽
.88, p⬍.001). Therefore, a strong argument can be made that
number of task strategies used was another measure of perfor-
Performance. The reliability of performance across the three
trials was .85. A repeated measures analysis of covariance
(ANCOVA) with goal level as a between-groups factor and trials
as a within-group factor was conducted. The results showed a
significant effect for trials, F(2, 286) ⫽15.15, p⬍.05,
A post hoc analysis using the Bonferroni test showed that perfor-
mance on Trial 2 (M⫽2.96, SD ⫽0.98) was higher than
performance on Trial 1 (M⫽2.45, SD ⫽0.81); and performance
on Trial 3 (M⫽3.47, SD ⫽1.20) was higher than performance on
Trial 2 (all p’s ⬍.05). These findings suggest that learning
increased across trials, thus providing further evidence of the
complexity of the task for the participants. The results also showed
a main effect for goal level, F(1, 144) ⫽5.88, p⬍.05,
The mean performance scores in the low and high-learning goal
conditions were 8.35 (SD ⫽2.83) and 9.40 (SD ⫽2.36), respec-
Cognitive ability. The Pearson correlation coefficient between
cognitive ability, as measured by the Wonderlic Aptitude Test, and
overall performance was .40 ( p⬍.05). Cognitive ability also
correlated positively with the number of strategies identified (r⫽
.15 p⫽.06) and the actual number of strategies used (r⫽.33, p⬍
Tests of Hypotheses
Hypotheses 1 and 2 were tested using multiple hierarchical
regression analyses of the data, collapsed across trials. Perfor-
mance on the 4-min pretest and cognitive ability were entered as
the first step in the regression analysis. Goal difficulty level, goal
commitment, and perceived task complexity were entered next in
order to test for their respective main effects. The goal difficulty
level ⫻cognitive ability interaction was entered in the third and
final step. All continuous predictors were standardised to correct
for ill-conditioned multicollinearity (Aiken & West, 1991). Goal
difficulty level was a dichotomous variable (low vs. high). Inter-
actions were formed by multiplying together the predictor vari-
ables. The results are shown in Table 2.
Hypothesis 1 stated that the difficulty level of the learning goal
is positively related to performance. The results, shown in Tables
1 and 2, support this hypothesis. The biserial correlation between
people who were assigned a low versus a high-learning goal and
performance was significant (r
⫽.20, p⬍.05). The results for
the hierarchical regression analyses show that the difficulty level
of the learning goal is positively related to performance; B⫽.78,
p⬍.05. Cognitive ability (B⫽1.14, p⬍.05) and goal commit-
ment (B⫽.57, p⬍.05) were also positively related to task
Hypothesis 2 stated that there is a goal difficulty level ⫻
cognitive ability interaction on performance. The results, re-
ported in Table 2, show that the interaction is significant; B⫽
⫺.66, p⬍.05. Consistent with the hypothesis, the multiple
hierarchical regression analysis revealed that the goal difficulty
level–performance relationship was stronger for participants
with lower cognitive ability. That is, the individuals with lower
cognitive ability benefited more from the setting of a learning
goal that increased in difficulty than did individuals with higher
Figure 1 shows the nature of the goal difficulty level ⫻cogni-
tive ability interaction graphically. As would be expected, people
with higher cognitive ability performed at higher levels than those
with lower cognitive ability. Figure 1 also illustrates that the
performance of participants with lower cognitive ability was more
positively affected by the setting of increasingly difficult learning
goals than it was for participants with higher cognitive ability; the
latter group did not appear to benefit from the assignment of
learning goals. Figure 1 also suggests that participants in the low
cognitive ability ⫻low goal difficulty condition performed worse
than those in the other conditions. We conducted a median split on
the scores for cognitive ability and then performed a series of
two-tailed ttests to explore differences in performance across the
four conditions. The means and standard deviations are shown in
Table 3. The performance of the participants in the low goal ⫻low
cognitive ability condition was significantly lower than the per-
formance of the participants in the other three conditions; there
were no other significant differences.
We also explored the effects of learning goals over time. Indi-
viduals may need time to discover the strategies but once they are
discovered, there should be an improvement in performance. We
contrasted the results for Trial 1 versus those for Trials 2 and 3.
The main effect for goal difficulty level, and the interaction effect
between goal difficulty level and cognitive ability, was not signif-
icant on the initial trial. This is common in learning studies
because participants have yet to acquire the knowledge necessary
to perform the task effectively. The combined results for Trials 2
and 3 were similar to the data collapsed across trials. Had the task
been more complex, there might have been improvement in per-
formance in Trial 3 relative to Trial 2.
Finally, we tested for a significant interaction between goal
difficulty level and goal commitment. The results showed that the
interaction was not significant. This is because the variance was
not sufficient to create a moderating effect. Goal commitment is
typically high in labouratory experiments; in fact, considerable
effort on the part of an experimenter is typically required to get
variance in goal commitment (Locke & Latham, 1990).
LEARNING GOALS AND COGNITIVE ABILITY
The findings of the present experiment have both theoretical
significance for Locke and Latham’s (1990, 2002) goal-setting
theory and practical significance for managers. A goal for learning
is similar to a goal for performance in terms of the goal difficulty–
performance level relationship. From a theoretical standpoint, this
study shows that, as is the case with a goal for a specific perfor-
mance outcome, a challenging high-learning goal leads to high
performance when the task requires the acquisition of knowledge
in order to perform it effectively.
The second major finding from this experiment is of practical as
well as theoretical significance in that it is contrary to what has
been found with tasks where people already have the ability to
perform them. As noted in the introduction, previous research on
performance goals show that goal setting typically has a beneficial
effect on performance for people with higher as opposed to those
with lower ability, as defined by skill (Locke, 1965). The findings
from this study revealed the opposite result. A high-learning goal
was more beneficial for people lower in ability, as defined by
cognitive intelligence, than it was for those who scored higher on
this variable. It appears that by assigning high-learning goals,
people with lower cognitive ability can raise their performance
levels relative to those of people with higher cognitive ability.
Moreover, the findings from the present study suggest that high-
learning goals have little impact on individuals who have high
cognitive ability. As the results in Table 3 and Figure 1 indicate, a
learning goal and cognitive ability appear to be able to compensate
for one another to some degree. The performance of participants
with lower cognitive ability who were assigned a high-learning
goal approached the performance of participants with higher cog-
nitive ability. Those participants with lower cognitive ability,
however, who were assigned a lower learning goal performed
poorly. This pattern of results suggests that for people to eventu-
ally perform well on a task where they lack the ability to do so,
they must have either high cognitive ability or a high-learning
goal. In the absence of further research, we can only speculate as
to the reasons underlying the beneficial effect of a relatively
high-learning goal for people who score relatively low on cogni-
A learning goal may prompt people with lower cognitive ability
to take the time to reflect on what went wrong regarding the
execution of one or more strategies. It may keep them from
becoming ensnared in decision-making blunders that smarter peo-
ple avoid. That is, it may keep them from rushing to judgment,
from acting too swiftly. A high-learning goal may prompt them to
conduct an adequate search for alternative solutions to a problem.
Hence, they do not settle on a single idea early in their decision
making process. In short, a high-learning goal likely leads to
increases in the option pool and thus increases a person’s prospects
for success. People with high cognitive ability may do all of this
intuitively or automatically. The search for alternatives that con-
tain innovative solutions is important for both the intelligent and
less intelligent; the benefit of a high-learning goal is that it appears
to cue the less intelligent person to do so. Thus, this study provides
additional evidence of ability as a moderator of goal setting. The
Figure 1. Graphic depiction of the Goal Difficulty Level ⫻Cognitive Ability Interaction on performance.
Means and SDs for Performance
Low M7.83 9.36
SD 2.98 2.45
High M8.79 9.46
SD 1.98 2.36
226 LATHAM, SEIJTS, AND CRIM
effect of ability on performance, however, differs when a learning
versus a performance goal is set.
Limitations and Future Research
It is unlikely that the relationship between setting a learning goal
and subsequent performance will always be linear. As noted ear-
lier, previous research has shown that the linear relationship levels
off when people who have a high-performance goal reach the
limits of their ability (Locke & Latham, 1990). Similarly, the
relationship between the difficulty level of a learning goal and
performance will likely be shown in future studies to level off
when the appropriate number of effective strategies has been
learned. Future research may even show that the relationship is
curvilinear if people persist in their search for additional ones.
Such behaviour is described colloquially as “paralysis by analy-
A related limitation of this study is that the range in cognitive
ability scores of the university students who participated in this
experiment ranged from average to high. The mean score of 24
was higher than the approximate average score of 21 for all
workers in the United States (Wonderlic Personnel Test and Scho-
lastic Level Exam User’s Manual, 2002). Thus, the present results
may be indicating that learning goals benefit people who are
moderately high in cognitive ability. Learning goals may have
little or no benefit for people who are quite low on cognitive ability
even if they are motivated to acquire the knowledge necessary to
perform a give task. Conversely, the results of the present study
suggest that those with high cognitive ability may not have much
room to improve their performance on some tasks despite the fact
a high-learning goal is set. It may be that individuals with mod-
erate levels of cognitive ability have the greatest opportunity to
improve when motivated by difficult goals. Research is needed on
a sample of participants where cognitive ability ranges from low to
high in order to test for linear and nonlinear effects of goal effects.
Related to these issues, the generalizability of the present find-
ings to tasks of high complexity needs to be investigated. The
present task was perceived by the participants, on average, to be
moderately complex. It was likely seen by participants high in
cognitive ability as relatively low in complexity. Had the task been
very complex and the strategies very difficult to discover, the
effect of a difficult learning goal might have been stronger for high
cognitive ability participants and have had little effect on low
cognitive ability participants.
Assigning a challenging learning goal for discovering a specific
number of task strategies may not be functional for all tasks that
require the acquisition of knowledge for a person to perform well.
This is because knowledge acquisition is likely to be only one type
of learning that may be necessary for performing effectively.
Studies are now needed on the effect of a learning goal on the
acquisition of motor skills and abilities.
The present study did not include the setting of a specific
high-performance goal. The rival hypothesis that such a perfor-
mance goal would have had a similar, or better, effect on perfor-
mance than a learning goal can be rejected on the basis of the
findings obtained by Winters and Latham (1996). As noted in the
introduction, they found that a “do your best” goal led to higher
performance than a specific high-performance goal, and a specific
high-learning goal led to higher performance than urging partici-
pants to do their best on the same task that was used in the present
Field experiments are now required to test the external validity
of a learning goal (e.g., discover 5 ways to significantly increase
revenue) with employees where the number of available strategies
is not known. On what basis should the number of strategies,
processes or systems be set in such situations? On very complex
tasks, the number of strategies that are necessary (e.g., ways to
increase trust, or ways to prevent a hostile merger and acquisition)
are not known a priori. On other tasks that require learning (e.g.,
how to obtain legislation favourable to a company), it may very
well be that learning/discovering “the” optimal task strategy is
required. In such instances, there may be a diminishing return for
continuing to seek and apply additional strategies. Research is
needed to address such issues.
When it is known a priori what constitutes appropriate behav-
iour or strategies, behavioural goals are more effective than learn-
ing goals on a person’s performance (Brown & Latham, 2002).
Once one or more optimal ways have been discovered to perform
a task effectively, it would appear likely that employees can be
trained to implement them, and a specific high performance, rather
than a learning goal can be set. The search for additional strategies
might detract from performance. In sum, the appropriate timing for
switching from the setting of learning to performance goals should
be investigated, particularly regarding individuals with relatively
low cognitive ability.
Finally, learning goals should be investigated in combination
with Frese’s (2005) error-management training. Keith and Frese
(2005) found that this training induces both emotion control and
metacognitive activity, and that these processes enhance perfor-
mance on tasks that require finding new solutions. Error manage-
ment training and learning goals appear to go hand in hand in that
both interventions induce discovery type activities (Latham, 2007).
Moreover, Keith and Frese found that error training masks the
effects of a goal orientation disposition. As noted earlier, Seijts et
al. (2004) found that a specific high-learning goal masks a goal
orientation disposition. Implicit in the extant research on error
training may be self-set learning goals.
The effect size of a specific high-performance goal has been
shown to be smaller on complex tasks than on ones that are
straightforward for people (Wood, 1987); some studies have even
shown that setting a difficult performance goal can have an ad-
verse effect on performance on tasks that people have yet to learn
how to perform effectively (Earley, Connolly, & Ekegren, 1989).
The positive effects of a specific high-performance goal on such
tasks are often delayed or do not occur at all (Locke & Latham,
1990; Mone & Shalley, 1995). As Locke and Latham (2005)
acknowledged, the passage of time does not guarantee that people
will learn how to perform a task effectively.
Seven studies have shown that a specific high-learning goal can
help people acquire the knowledge they lack to improve their
performance relative to immediately setting a specific high-per-
formance goal (Seijts et al., 2004; Winters & Latham, 1996). The
unique contribution of the present study to the literature is that it
is the first to show that, as is the case with a specific high-perfor-
mance goal, there is a positive relationship between the difficulty
LEARNING GOALS AND COGNITIVE ABILITY
level of a learning goal and subsequent performance. Second, and
arguably more important, the present study shows that individuals
with lower cognitive ability can raise their performance levels
relative to those of people with higher cognitive ability when a
high-learning goal is set. This is likely due to a learning goal
enhancing what Schmidt and Ford (2003) called metacognition.
That is, attention is placed on understanding the task and “how
things work” before shifting one’s focus to attaining a specific
L’e´tude re´alise´e portait sur le rapport entre le niveau de difficulte´
d’un but d’apprentissage et le rendement d’une personne (N⫽
146) dans l’accomplissement re´ussie d’une taˆche qui requiert
l’acquisition de connaissances. De multiples analyses de re´gres-
sion hie´rarchique re´ve`lent que plus le but d’apprentissage est
e´leve´, meilleur est le rendement de l’individu. La capacite´ cogni-
tive et l’engagement a` l’e´gard du but influent aussi positivement
sur le rendement. Les re´sultats re´ve`lent que la capacite´ cognitive
d’une personne a des re´percussions sur le rapport but
d’apprentissage-rendement. Contrairement aux re´sultats de recher-
ches ante´rieures sur les objectifs de rendement pour des taˆches qui
sont simples pour les sujets, le rendement d’individus pre´sentant
une capacite´ cognitive infe´rieure e´tait influence´ positivement dans
une plus grande mesure lorsque le but d’apprentissage e´tait diffi-
cile que dans le cas d’individus ayant une plus grande capacite´
Mots-cle´s : but d’apprentissage, capacite´ cognitive, rendement
Ackerman, P. L., Kanfer, R., & Goff, M. (1995). Cognitive and non-
cognitive determinants and consequences of complex skill acquisition.
Journal of Experimental Psychology: Applied, 4, 270 –304.
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and
interpreting interactions. Newbury Park, CA: Sage.
Bandura, A., & Simon, K. M. (1977). The role of proximal intentions in
self regulation of refractory behaviour. Cognitive Therapy and Research,
Barrick, M. R., Bradley, B. H., Kristof-Brown, A. L., & Colbert, A. E.
(2007). The moderating role of top management team interdependence:
Implications for real teams and working groups. Academy of Manage-
ment Journal, 50, 544 –557.
Brown, T. C., & Latham, G. P. (2002). The effects of behavioural outcome
goals, learning goals, and urging people to do their best on an individ-
ual’s teamwork behaviour in a group problem-solving task. Canadian
Journal of Behavioural Science, 34, 276 –285.
Cervone, D., Jiwani, N., & Wood, R. (1991). Goal setting and the differ-
ential influence of self-regulatory processes on complex decision-
making performance. Journal of Personality and Social Psychology, 61,
DeShon, R. P., Brown, K. G., & Greenis, J. L. (1996). Does self-regulation
require cognitive resources? Evaluation of resource allocation models of
goal setting. Journal of Applied Psychology, 81, 595– 608.
Drach-Zahavy, A., & Erez, M. (2002). Challenge versus threat effects on
the goal–performance relationship. Organizational Behavior and Hu-
man Decision Processes, 88, 667– 682.
Earley, P. C. (1985). Influence of information, choice, and task complexity
upon goal acceptance, performance, and personal goals. Journal of
Applied Psychology, 70, 481– 491.
Earley, P. C., Connolly, T., & Ekegren, G. (1989). Goals, strategy devel-
opment and task performance: Some limits on the efficacy of goal
setting. Journal of Applied Psychology, 74, 24 –33.
Eden, D. (2002). Replication, meta-analysis, scientific progress, and
AMJ’s publication policy. Academy of Management Journal, 45,
Frese, M. (2005). Grand theories and midrange theories. Cultural effects on
theorizing and the attempt to understand active approaches to work. In
K. G. Smith & M. Hitt (Eds.), The Oxford handbook of management
theory: The process of theory development. New York: Oxford Univer-
Hunter, J. E. (1986). Cognitive ability, cognitive aptitudes, job knowledge,
and job performance. Journal of Vocational Behaviour, 29, 340 –362.
Hunter, J. E., & Hunter, R. F. (1984). Validity and utility of alternative
predictors of performance. Psychological Bulletin, 96, 72–98.
Kanfer, R., & Ackerman, P. L. (1989). Motivation and cognitive abilities:
An integrative/aptitude-treatment interaction approach to skill acquisi-
tion. Journal of Applied Psychology, 74, 657– 690.
Kaplan, E., Erez, M., & Van-Dijk, D. (2004, April). Reconciling potential
differences between the goal setting and self regulation theories. In M.
Erez (Chair), Goal setting, goal orientation, and self regulation focus:
An integration. Symposium presented at the annual meeting of the
Society for Industrial and Organizational Psychology, Chicago.
Keith, N., & Frese, M. (2005). Self-regulation in error management train-
ing: Emotion control and meta cognition as mediators of performance
effects. Journal of Applied Psychology, 90, 677– 691.
Klein, H. J., Wesson, M. J., Hollenbeck, J. R., Wright, P. M., & DeShon,
R. P. (2001). The assessment of goal commitment: A measurement
model meta-analysis. Organizational Behaviour and Human Decision
Processes, 85, 32–55.
Kozlowski, S. W. J., & Bell, B. S. (2006). Disentangling achievement
orientation and goal setting: Effects on self-regulatory processes. Jour-
nal of Applied Psychology, 91, 900 –916.
Latham, G. P. (2007). Work motivation: History, theories, research, and
practice. Thousand Oakes, CA: Sage.
Latham, G. P., & Baldes, J. J. (1975). The “practical significance” of
Locke’s theory of goal setting. Journal of Applied Psychology, 60,
Latham, G. P., & Brown, T. C. (2006). The effect of learning versus
outcome goals on self efficacy and satisfaction in an MBA program.
Applied Psychology: An International Review, 55, 623.
Latham, G. P., & Lee, T. W. (1986). Goal setting. In E. A. Locke (Ed.),
Generalizing from laboratory to field settings (pp. 101–117). Lexington,
MA: Lexington Books.
Latham, G. P., & Locke, E. A. (2007). New developments and directions
for goal setting. European Psychologist, 12, 290 –300.
Latham, G. P., & Saari, L. M. (1982). The importance of union acceptance
for productivity improvement through goal setting. Personnel Psychol-
ogy, 35, 781–787.
Locke, E. A. (1965). Interaction of ability and motivation in performance.
Perceptual and Motor Skills, 21, 719 –725.
Locke, E. A. (1982). Relation of goal level to performance with a short
time period and multiple goal levels. Journal of Applied Psychology, 67,
Locke, E. A. (2000). Motivation, cognition, and action: An analysis of
studies of task goals and knowledge. Applied Psychology: An Interna-
tional Review, 49, 408 – 429.
Locke, E. A., & Latham, G. P. (1990). A theory of goal setting and task
performance. Englewood Cliffs, NJ: Prentice Hall.
Locke, E. A., & Latham, G. P. (2002). Building a practically useful theory
of goal setting and task motivation: A 35-year odyssey. American
Psychologist, 57, 705–717.
Locke, E. A., & Latham, G. P. (2005). Goal setting theory: Theory building
by induction. In M. Hitt & K. G. Smith (Eds.), Oxford handbook of
228 LATHAM, SEIJTS, AND CRIM
management theory: The process of theory development. New York:
Oxford University Press.
Locke, E. A., Shaw, K. N., Saari, L. M., & Latham, G. P. (1981). Goal
setting and task performance: 1969 –1980. Psychological Bulletin, 90,
Mitchell, T. R., & Daniels, D. (2003). Motivation. In W. C. Borman, D. R.
Ilgen, & R. J. Klimoski (Eds.), Handbook of psychology, Vol. 12:
Industrial organizational psychology (pp. 225–254). New York: Wiley
Mone, M. A., & Shalley, C. E. (1995). Effects of task complexity and goal
specificity on change in strategy and performance over time. Human
Performance, 8, 249 –262.
Noel, T. W., & Latham, G. P. (2006). The importance of learning goals
versus outcome goals for entrepreneurs. The International Journal of
Entrepreneurship and Innovation, 7, 213–220.
Ree, M. J., & Earles, J. A. (1991). Predicting training success: Not much
more than g. Personnel Psychology, 44, 321–332.
Schmidt, A. M., & Ford, J. K. (2003). Learning within a learner control
training environment: The interaction effects of goal orientation and
metacognitive instruction on learning outcomes. Personnel Psychology,
56, 405– 429.
Seijts, G. H., & Latham, G. P. (2001). The effect of distal learning,
outcome, and proximal goals on a moderately complex task. Journal of
Organizational Behavior, 22, 291–302.
Seijts, G. H., & Latham, G. P. (2005). Learning versus performance goals:
When should each be used? Academy of Management Executive, 19, 1,
Seijts, G. H., Latham, G. P., Tasa, K., & Latham, B. W. (2004). Goal
setting and goal orientation: An integration of two different yet related
literatures. Academy of Management Journal, 47, 227–239.
Terborg, J. R., & Miller, H. E. (1978). Motivation, behavior and perfor-
mance: A closer examination of goal setting and monetary incentives.
Journal of Applied Psychology, 63, 29 –39.
Vroom, V. H. (1964). Work and motivation. New York: Wiley.
Winters, D., & Latham, G. P. (1996). The effect of learning versus outcome
goals on a simple versus a complex task. Group and Organization
Management, 21, 235–250.
Wonderlic Personnel Test and Scholastic Level Exam User’s Manual.
(2002). Libertyville, IL: Wonderlic Inc.
Wood, R. E. (1986). Task complexity: Definition of the construct. Orga-
nizational Behavior and Human Decision Processes, 37, 60 – 82.
Wood, R. E., Mento, A. J., & Locke, E. A. (1987). Task complexity as
a moderator of goal effects. Journal of Applied Psychology, 72,
416 – 425.
Received May 28, 2008
Revision received May 28, 2008
Accepted June 19, 2008 䡲
LEARNING GOALS AND COGNITIVE ABILITY