The Psychological Record, 1988, 38, 533-541

GLOBAL VERSUS TASK-SPECIFIC MEASURES OF SELF-EFFICACY

ALVIN Y. WANG
University of Central Florida

R. STEPHEN RICHARDE
The Virginia Military Institute

Two studies were conducted in order to reconcile contextualist

accounts of self-efficacy with recent evidence supporting its
global nature. A correlational analysis (Study 1) indicated that
Tipton and Worthington's (1984) Generalized Self-Efficacy (GSE)
Scale was inversely related to Rotter's Internal-External Locus of
Control Scale and Beck's Hopelessness Scale. In contrast, no
discernible pattern of intercorrelations was obtained ac ross six
task-specific tests of self-efficacy. A single bipolar factor
emerged, however, when the GSE Scale was factor analyzed
with task-specific ratings of self-efficacy. Study 2 demonstrated
that task-specific, but not GSE scores, were sensitive to actual
performance on cognitive tasks. These results were interpreted
as evidence for the hypothesis that global and task-specific
measures assess relatively distinct aspects of the construct of
self-efficacy.

Since its introduction one decade aga the construct of self-efficacy

has gained increasing importance as a valid "person variable"
influencing both cognitive and affective behavior (see Bandura, 1981,
1982 for reviews). The preeminent definition of this construct was
developed by Bandura (1977) who characterized it as self-perceptions of
one's capability to organize and implement actions in specific situations
that may contain novel features.
Central to this notion of self-efficacy is the assertion that it is astate
rather than a trait variable. Accordingly, it is assumed that self-efficacy
can only be "indexed in terms of percepts ... that vary ac ross activities
and situational circumstances rather than as a global disposition
assayed by an omnibus test" (Bandura, 1982, p. 124). Since self-efficacy
does not present a crystallized target for stable measures, only
"microanalytic" tests directly relating percepts of efficacy to particular
behavior and their outcomes are considered valid (Bandura, 1977).
Support for the context-dependent nature of self-efficacy has been

Portions of the data were presented at the 1986 Annual Meeting of the Eastern
Educational Research Association, Miami Beach, FL. We appreciate the support received
from the Virginia Military Institute Research Laboratories.
Reprints should be requested from Alvin Y. Wang, Department of Psychology,
University of Central Florida, Orlando, FL 32816.
534 WANG AND RICHARDE

consistently demonstrated in studies showing that past experience with

a given task is critical for the formation of valid and predictive measures
of self-efficacy (Bandura & Schunk, 1981; Schunk, 1982; Wang &
RiCharde, 1987).
Despite claims and evidence documenting its situation-specificity, the
issue of whether changes in self-efficacy can nevertheless generalize
across various contexts and behavioral domains should be explored.
Evidence may be cited from three research areas suggesting that at
least a limited degree of generalization can be expected across similar
tasks and situations. Cognitive modeling, for example, enhances the
self-efficacy of snake phobics towards similar as weil as dissimilar
threats (Bandura, Adams, Hardy, & Howells, 1980). The same study also
reported that the coping behavior of agoraphobics can generalize across
threatening situations following self-directed "mastery" treatment
(Bandura et al., 1980). Second, a study on cognitive development found
that children's positive self-efficacy for learning word pairs can transfer to
a similar cognitive task (vocabulary learning) after generalized
instructions to monitor their acquisition performance (Wang & RiCharde,
1987). The third and perhaps most explicit challenge to context-specific
interpretations of self-efficacy is the Generalized Self-Efficacy (GSE)
Scale which was developed to measure a "relatively enduring set of
beliefs that one can co pe effectively in a broad range of situations"
(Tipton & Worthington, 1984, p. 547). The designers of the GSE Scale
report that this global measure of self-efficacy reliably predicted the
extent to which individuals persevere on a psychomotor strength task.
The GSE Scale was also correlated with a goal attainment scale
measuring perseverance for changing a problem behavior (e.g., smoking
or excessive eating).
The two studies reported here attempt to integrate contextualist
accounts of self-efficacy with evidence suggesting a more global nature.
Study 1 was designed to further establish the construct validity of the
GSE Scale with other global and trait measures of personality. The other
measures include Rotter's Internal-External Locus of Contral Scale
(1966), a Generalized Expectancy for Success Scale (Fibel & Haie,
1978) and Beck's Hopelessness Scale (Beck, Weissman, Lester, &
Trexler, 1974). These particular global measures were studied because
of their experimental and theoretical link to Bandura's construct of self-
efficacy as weil as to each other as mediators of self-referent thought
(Fibel & Haie, 1978; Fogg, Kohaut, & Gayton, 1977; Lefcourt, 1976;
Maddux, Sherer, & Rogers, 1982; Newman, 1977; Schunk, 1984; Tipton
& Worthington, 1984).
Study 1 also included several task-specific ratings of self-efficacy
(e.g., serial recall of numbers) so that the presumed distinctness of
generalized and specific measures of self-efficacy could be assessed.
To the extent that the GSE Scale possesses construct validity, one
would expect GSE scores to be correlated with other similar trait
measures of personality. However, if Bandura's (1982) contextualist
 SELF-EFFICACY 535

account of self-efficacy is correct, then little, if any direct correspondence

between GSE scores and task-specific ratings should be found.
The second study also evaluated the distinctness of global versus
task-specific ratings of self-efficacy by relating both measures to actual
performance on a variety of cognitive tasks. If the GSE Scale evaluates
stable and enduring percepts of efficacy while task-specific ratings of
self-efficacy focus on particular abilities, then both of these measures
should remain statistically distinct from one another. We therefore
hypothesized that task performance is more likely to correlate with
specific rather than global measures of this construct.

Study 1

Method

Design and Materials

A multimethod factor analysis was applied to task-specific measures
of self-efficacy and the measure of Generalized Self-Efficacy (GSE)
devised by Tipton and Worthington (1984). Three related trait measures
were also included in the analysis: Rotter's Internal-External (I-E) Locus
of Control Success Scale (1966), Fibel and Hale's (1978) Generalized
Expectancy for Success Scale (GESS), and Beck's Hopelessness Scale
(Beck et al. , 1974).
Task-specific measures of self-efficacy were obtained using the
technique devised by Bandura and Schunk (1981) but modified for
college-age individuals. Subjects were asked to refer to a 10-point Likert
scale (1 = not sure, 10 = very sure) when assessing the degree to which
they were certain of being able to perform six diverse tasks.
Specific measures of self-efficacy (SE) were obtained for the following
tasks:
(SE 1) Se rial recall of 10 names after 1 minute of study
(SE2) Identification of a painting by Rembrandt based on style alone
(SE3) Throwing a frisbee 50 yards
(SE4) Se rial recall of a 20-digit number after 1 minute of study
(SES) Drawing a 3-dimensional representation of a medieval castle
(SE6) Completion of a local newspaper crossword puzzle.
These particular tasks were chosen because they seemed to
represent relatively different skill domains. Inasmuch as different classes
of activities are tested, no generalization across these tasks should
occur according to contextualist accounts of self-efficacy.
The order whereby these measures were administered to all subjects
was as folIows: the I-E Scale, the GSE Scale, the GESS, Beck's
Hopelessness Scale, and finally the six task-specific ratings of self-
efficacy.

Subjects
Forty-eight male and 45 female Introductory Psychology students
536 WANG AND RICHARDE

from two sm all colleges in Southwestern Virginia participated as

subjects. None of the subjects had participated in prior psychological
studies. The questionnaires were administered to classes of 20-25
students and took approximately 40 min for completion.

Results

Table 1 shows the correlation matrix for all measures included in

Study 1. The GSE Scale was inversely related to both I-E scores (r =
-.24, p< .05) and Hopelessness scores (r = -.23, p< .05). This pattern
of results suggests that individuals with higher levels of generalized self-
efficacy are more likely to have an internal locus of control and a lesser
degree of hopelessness compared to others having an external locus of
control. Not surprisingly, a positive correlation was found between the
I-E and Hopelessness Scales (r = .31, P < .05), closely replicating the
findings of Fogg et al. (1977). This suggests that an internal locus of
control is associated with lower degrees of hopelessness. No discernible
pattern of intercorrelations was obtained for any of the task-specific tests
of self-efficacy. A principle components analysis carried out on all
measures yielded two factors: The first was a "global" factor with heavy
loadings for the I-E, GESS, and Hopelessness Scales. Interestingly, the
second was abipolar factor for GSE scores and the task-specific self-
efficacy measures, the two types of measures having heavy factor
loadings in opposite directions. The bipolarity of this self-efficacy factor
can be interpreted as meaning that global and task-specific ratings of
self-efficacy measured opposing aspects of the same construct
(Kerlinger, 1973).
Table 1

Correlation Coefficients for Task-Specific and Global Measures (Study 1)

IE GSE H GESS SE1 SE2 SE3 SE4 SES SE6

IE
GSE -.24*
H .31* -.23*
GESS .16 .10 .05
SE1 -.09 -.06 -.07 .03
SE2 -.03 .38* .05 -.03 .02
SE3 .07 .04 .22 .02 -.11 .05
SE4 .06 .13 .25* .02 .09 .13 .20
SES .10 .05 .00 .05 .51* .02 .21 .27*
SE6 .13 -.02 -.09 .00 .24* .15 .06 -.05 .33*

*Denotes significance at the p < .05 level.

The following measures are included in the correlation matrix: IE (Rotter's Internal-External
Scale), H (Beck et al.'s Hopelessness Scale), GSE (Tipton & Worthington's Generalized
Self-efficacy Scale), GESS (Fibel & Hale's Generalized Expectancy of Success Scale).
The SE1 through SE6 measures are the task-specific tests of self-efficacy presented in the
sequence detailed in the materials section.
 SELF-EFFICACY 537

Study 2

Method

Design and Materials

A pretest-posttest control group design was used to determine the
relationship between task-specific measures of self-efficacy and task
performance. In order to evaluate agiobai measure of self-efficacy
against task-specific measures and performance, the GSE Scale was
administered at the end of each experimental session.
Subjects were first pretested on task-specific self-efficacy for two
tasks: long division problems (three-digit numbers into five-digit
numbers) and 20-digit serial recall. They were shown examples of the
mathematics problems and the 20-digit number for approximately 2 sec
and asked to rate their degree of certainty for correctly performing these
tasks using a 10-point Likert scale. Schunk (1982) determined that 2 sec
was long enough for subjects to judge what the task involves without
having time to actively perform it. Subjects in the experimental group
were then asked to complete the two tasks in a timed test: They were
given 2 min to solve two of the division problems and 1 min to study a
20-digit number before attempting serial recall (3 min total). Control
subjects were not required to perform any of these tasks; instead they
were occupied with administrative "paper work" during a 3-min interval of
interpolated activity. In all other respects the control and experimental
groups were treated the same way. Following the 3-min period all
subjects were posttested for task-specific self-efficacy and the GSE
Scale.

Subjects
Forty-eight college males comprised the subjects in the second study.
One half (n = 24) were assigned either to the control or experimental
group on a random basis. None of the subjects had participated in any
prior psychological study.

Results

Analyses of variance (ANOVAs) indicated that the control and

experimental groups were statistically equivalent on measures of GSE
and pretest measures of task-specific self-efficacy (all Fs< 1). A multiple
regression analysis indicated that pretest ratings of task-specific self-
efficacy were not correlated with actual task performace for either task.
Posttest task-specific ratings for the experimental group, however, were
highly correlated with their performance measures (r = .52 and .49 for
division problems and digit span, respectively). No other correlations
were significant. Evidently, the GSE Scale did not relate statistically to
the task-specific ratings of self-efficacy or the performance measures in
either group.
538 WANG AND RICHARDE

Table 2
Mean Pretest and Posttest Ratings of Task-Specific Self-Efficacy (Study 2)

Task
Group Digit Span Division

Exp. (n = 24)
Pretest 5.8 8.1
Posttest 5.3 5.8
Control (n = 24)
Pretest 5.7 8.2
Posttest 5.8 8.3

Based on a 1O-point Likert Scale (1 = not sure, 10 = very sure).

Two separate 2 (Group) x 2 (Pretest, Posttest) MANOVAs were

performed on the task-specific self-efficacy ratings for digit-span and
division problems. None of the Fs for the analysis of digit-span ratings
were significant (all Fs < 1). In contrast, when self-efficacy for division
problems was considered, significant main effects were found for Group,
F(1, 46) = 9.75, and pretest-posttest comparisons, F(1, 46) = 23.97.
Moreover, the interaction between these two factors was also significant,
F(1, 46) = 11.93. Inspection of Table 2 suggests that the results of this
MANOVA were due solely to the difference between the pretest-posttest
ratings obtained in the experimental group. A posthoc comparison using
Tukey's HSD proved this to be the case, HSD (4, 42) = 1.28.
Table 3

Study 2 - MANOVA Summary (division problems)

SV SS df MS F P
A (Group) 59.23 1 59.23 9.75 .01
S/A 279.23 146 6.07

B (Pre-post) 42.31 1 42.31 23.97 .001

AB 21.05 1 21.05 11.93 .005
S/AIB 74.14 46 1.77

The findings reported herein suggest that experimental subjects

encountered more difficulty with the division problems (mean correct
was only .71) than they initially expected and their lower percepts of
efficacy following these tasks reflected this unanticipated difficulty.
Although the overall pretest-posttest ratings for digit span did not
change, experimental subjects individually became more realistic in
assessing their digit span after task performance (mean recall = 5.83).
This was indicated by the correlation between posttest (but not pretest)
ratings and digit-span scores. As expected, the pretest to posttest
 SELF-EFFICACY 539

ratings of control subjects remained unchanged since they did not have
the opportunity to perform either digit-span or division tasks.
Although others (Bandura, 1981; Schunk, 1982) have reported that
sense of efficacy may predict task performance, we were not able to
replicate this finding in Study 2. Neither the measure of GSE nor pretest
self-efficacy scores correlated with performance outcomes. Perhaps the
specialized nature of the tasks tested or the brief 2-sec exposure interval
prevented subjects from developing valid, initial estimates of self-
efficacy.

General Discussion

Two studies were designed to reconcile the contextualist account of

self-efficacy with recent attempts to derive global and trait measures of
this construct. The findings support our hypothesis that task-specific
ratings and the Generalized Self-Efficacy (GSE) Scale assess distinctly
different aspects of one's percept of efficacy. The following results
support this view:
(1) The GSE Scale which relies on broadly-stated self-report was
inversely related to two other global measures of personality: Rotter's I-E
Scale and Beck's Hopelessness Scale. Furthermore, a direct relation
was obtained between the I-E and Hopelessness Scales. In contrast, no
meaningful pattern of correlations involving any of the task-specific tests
of self-efficacy was obtained.
(2) When ratings of task-specific self-efficacy and the GSE Scale
were factor analyzed, they emerged as a single bipolar factor which
appeared to measure extreme ends of the same construct. Statistically,
there was no evidence for a direct, positive relationship between global
and context-dependent measures of self-efficacy.
(3) The GSE Scale and task-specific ratings were also evaluated
following actual performance on two cognitive tasks. Scores on the GSE
Scale were not sensitive to performance outcomes for division problems
or digit-span memory, nor were these scores correlated with task-
specific measures of self-efficacy. In contrast, task-specific posttest
ratings of self-efficacy were highly correlated with performance for both
tasks.
Is there some way in which the distinctness of global and task-specific
measures of self-efficacy can be approached theoretically? One
attractive notion is that measures of task-specific self-efficacy are more
valid when the task is clearly defined and somewhat familiar to
individuals. In contrast, a test such as the GSE Scale would be more
successful predicting performance in situations that persons find less
familiar and more ambiguous (Tipton & Worthington, 1984, p. 548).
In a study concerning the construct validity of their GSE Scale, Tipton
and Worthington (1984) found that it correlated with a self-determination
task (Iength of time one could hold an outstretched arm parallel to the
ground) and behavioral self-control tasks (e.g., reduced smoking, weight
540 WANG AND RICHARDE

loss). A positive correlation between sense of efficacy and effort

expenditure on these tasks was thus found. This supports Bandura's
(1981) contention that efficacy expectation contains a motivational
component that determines when and for how long one will engage in
overt behaviors to produce desired outcomes. Nevertheless, although
these tasks may require some cognitively evaluated commitment, they
do not entail the kind of performance for which there is an explicitly
correct or incorrect answer. Consequently, it seems likely that agiobai
scale such as the GSE may provide a stable and valid measure of an
individual's degree of perseverance for tasks which rely on a large
motivational component.
In contrast, the type of tasks employed by other researchers (Schunk,
1981,1982,1983,1984; Wang & RiCharde, 1987) require rather
specialized cognitive skills; perseverance alone would not lead to
successful performance without these skills. Task-specific self-efficacy
ratings appear to be highly dependent upon an individual's familiarity
with a task and the knowledge of whether the requisite skills are in his or
her repertoire. Therefore, performance outcomes encountered in these
situations, as weil as in the present study, might be more amenable to
specific rather than global measures of self-efficacy.
One final implication of our findings is that task-specific measures of
self-efficacy may become more realistic following performance feedback.
This was particularly evident for the experimental subjects in Study 2.
Only their posttest ratings of self-efficacy were found to be correlated
with actual performance. This is in line with Bandura's (1981) assertion
that self-efficacy is dependent upon self-referent thought. The ability to
introspect and engage in cognitive-monitoring processes (Flavell, 1979;
Wang & RiCharde, 1987) seems particularly important for percepts of
efficacy related to tasks involving clearly defined performance outcomes.

References

BANDURA, A. (1977). Self-efficacy: Toward a unified theory of behavioral

change. Psychological Review, 84, 191-215.
BANDUARA, A. (1981). Self-referent thought: A developmental analysis of self-
efficacy. In J. H. Flavell & L. Ross (Eds.), Social cognitive development:
Frontiers and possible futures. Cambridge: Cambridge University Press.
BANDURA, A. (1982). Self-efficacy mechanism in human agency. American
Psychologist, 37,122-147.
BANDURA, A., ADAMS, N. E., HARDY, A. B., & HOWELLS, G. N. (1980). Tests
of the generality of self-efficacy theory. Cognitive Theory and Research, 4,
39-66.
BANDURA, A., & SCHUNK, D. (1981). Cultivating competence, self-efficacy, and
intrinsic interest through proximal self-motivation. Journal of Personality
and Social Psychology, 41 , 586-598.
 SELF-EFFICACY 541

BECK, A. T., WEISSMAN, A., LESTER, D., & TREXLER, L. (1974). The
measurement of pessimism: The hopelessness scale. Journal of
Consulting and Clinical Psychology, 42, 861-865.
FIBEL, B., & HALE, W. (1978). The Generalized Expectancy for Success Scale:
A new measure. Journal of Consulting and Clinical Psychology, 46, 924-
931.
FLAVELL, J. H. (1979). Metacognition and cognitive monitoring: A new area of
cognitive-developmental inquiry. American Psychologist, 34, 906-911.
FOGG, M. E., KOHAUT, S. M., & GAYTON, W. F. (1977). Hopelessness and
locus of control. Psychological Reports, 40, 1Q70.
KERLINGER, F. N. (1973). Foundations of behavioral research (2nd ed.). New
York: Holt, Rinehart & Winston.
LEFCOURT, H. M. (1976). Locus of contro/: Current trends in theory and
research. New Jersey: Erlbaum.
MADDUX, J., SHERER, M., & ROGERS, R. (1982). Self-efficacy expectations
and outcome expectations: Their relationship and their effects on
behavioral intentions. Cognitive Therapy and Research, 6, 207-211.
NEWMAN, J. M. (1977). Comparison of the I-E Scale and a specific locus of
control measure in predicting risk-taking behavior under novel task
conditions. Psychological Reports, 40, 1035-1040.
ROTTER, J. B. (1966). Generalized expectations of internal versus extern al
control of reinforcement. Psychological Monographs, 80, 609.
SCHUNK, D. H. (1981). Modeling and attributional effects on children's
achievement: A self-efficacy analysis. Journal of Educational Psychology,
73,93-105.
SCHUNK, D. H. (1982). Effects of effort attributional feedback on children's
perceived self-efficacy and achievement. Journal of Educational
Psychology, 74, 548-556.
SCHUNK, D. H. (1983). Developing children's self-efficacy and skills: The rales
of social comparative information and goal setting. Journal of
Comparative Social Psychology, 75, 511-518.
SCHUNK, D. H. (1984). Self-efficacy perspective on achievement behavior.
Educational Psychologist, 19, 48-58.
TIPTON, R., & WORTHINGTON, L. (1984). The measurement of generalized
self-efficacy: A study of construct validity. Journal of Personality
Assessment, 48, 545-548.
WANG, A. Y., & RICHARDE, R. S. (1987). Development of memory-monitoring
and self-efficacy in children. Psychological Reports, 60, 647-658.