37
DEVELOPMENT AND EVALUATION OF A SCALE TO
MEASURE PERCEIVED SELF-EFFICACY IN PEOPLE
WITH ARTHRITIS
KATE LORIG, ROBERT L. CHASTAIN, ELAINE UNG, STANFORD SHOOR,
and HALSTED R. HOLMAN
There is evidence that the psychological attribute
of perceived self-efficacyplays a role in mediating health
outcomes for persons with chronic arthritis who take the
Arthritis Self-Management Course. An instrument to
measure perceived self-efficacy was developed through
consultation with patients and physicians and through
study of 4 groups of patients. Tests of construct and
concurrent validity and of reliability showed that the
instrument met appropriate standards. Health outcomes
and self-efficacy scores improved during the Arthritis
Self-Management Course, and the improvements were
correlated.
We describe the development and testing of an
instrument to measure patients’ perceived self-efficacy
(SE) to cope with the consequences of chronic arthri-
tis. The need for such an instrument arose in the
course of studying the effects of the Arthritis Self-
Management Course (ASMC). When evaluated in ran-
domized studies, subjects who took the ASMC were
found to have less pain and to be more active than
controls (1). These results persisted, albeit attenuated,
for 20 months after the course, without reinforcement.
However, when the data were examined for antici-
pated associations between changes in behavior (ex-
From the Department of Medicine, Stanford University,
Stanford, California.
Supported by NIH Multipurpose Arthritis Center grant
20610-05 and by the Arthritis Foundation.
Kate Long, DrPH; Robert L. Chastain, MS; Elaine Ung,
BS; Stanford Shoor, MD; Halsted R. Holman, MD.
Address reprint requests to Kate Lorig, DrPH, Senior
Research Associate, Department of Medicine, Stanford University,
HRP Building, Room 6 , Stanford, CA 94305.
Submitted for publication May 29, 1987; accepted in revised
form July 8, 1988.
Arthritis and Rheumatism, Vol. 32, No. 1 (January 1989)
ercise, relaxation, and walking) and changes in health
status (pain, disability, and depression), the expected
associations were weak or were absent (2).
These unexpected findings precipitated an in-
terview with the participants and evaluation of their
experiences in the course. Fifty-four participants were
asked why they found the course helpful or not help-
ful. For half the people interviewed, pain and/or
disability had decreased; for the other half, pain and/or
disability had not changed or had increased. The
former group attributed their benefits to an increased
sense of influence over the consequences of arthritis;
the latter group believed that they could do little to
improve their situation (3).
The findings from these interviews indicated
that a sense of one’s personal ability to affect the
consequences of disease was strong in some subjects
and relatively weak in others, and interacted with the
course to create the health outcomes. This sense of
ability to effect change (akin to confidence) is similar
to the psychological concept of perceived SE (43). In
a preliminary study using early instruments to measure
perceived SE to cope with the consequences of arthri-
tis, we found statistically significant correlations be-
tween perceived SE and health status (ref. 6 and
unpublished observations). We therefore sought to
develop a reliable and valid instrument for measuring
perceived SE.
Perceived SE, as postulated by Bandura (4), is
one’s belief that one can perform a specific behavior or
task in the future. It refers to personal judgments of
performance capabilities in a given domain of activity.
Although it is related to other psychological concepts,
such as locus of control, learned helplessness, and
self-esteem, it differs in that it is behavior-specific.
38
Thus, SE is measured as a specific state, not a gener-
alized trait. For example, one may have high SE for
using medications correctly but low S E for utilizing
relaxation to relieve pain. Perceived SE is the belief
that one can achieve a behavior or a state of mind; it is
not an actual measure of accomplishment.
Self-efficacy theory states that 1) perceived SE
for behaviors that affect health status will predict
future health status, given that subjects believe that
the outcome of the behavior will be improved health
status and that they value improved health status, 2)
SE is not a static trait; it can be altered, and 3)
enhanced SE will be associated with improved health
status in the areas affected by those specific behaviors.
As a psychological trait, it is likely that there are many
relevant antecedents and components of perceived
SE, such as inherent motor or mental skill, experience
with a particular behavior, general sense of ability or
self-worth, and motivation for accomplishment. Be-
cause it is, by definition, behavior-specific, specific
questions can be designed to describe it (4). Also,
particular learning and practicing activities or tech-
niques can be devised to facilitate and improve the
conduct of the relevant behavior and, thereby, pre-
sumably increase an individual’s perceived SE to
accomplish the behavior.
The behavior-specific nature of SE made it
necessary to develop an instrument specific for its
measurement in people who have arthritis. The stages
in the development process are illustrated in Table 1.
They include design of the initial scales from data
obtained from one population, refinement (replication)
of the scales using a second population, and confirma-
tory testing of factors that represent the refined scales
using both the original population sample and the
replication sample. The results from each stage of
development are described below, and preliminary
tests of hypotheses relating perceived SE to health
status follow.
PATIENTS AND METHODS
All tests of this instrument were conducted with
subjects who had been recruited for an ASMC by means of
public service announcements and referrals from health
professionals. After completing an initial application, the
subjects were randomized to a control group (no interm-
tion) or an experimental group (participation in the ASMC).
Data were collected by self-administered questionnairesthat
were mailed to the subjects at baseline and again 4 months
later, as previously described (1). We studied such variables
as pain (measured on a 10-cm visual analog scale) (7,8),
LORIG ET AL
Table 1. Steps in the development of the self-efficacy scale
I. Item generation
1. 23 items generated by a rheumatologist
2. 20 items refined in patient focus group
11. Initial instrument development study (n = 97)
1. Baseline and 4-month values
2. Factor analysis
3. Function self-efficacy
4. Other symptom self-efficacy
5 . Alpha coefficient (measure of internal reliability)
6 . Tests of construct validity
111. Concurrent validity test (n = 43)
IV. Replication study (n = 144)
1. Baseline and 4-month values
2. Factor analysis
3. Pain self-efficacy
4. Function self-efficacy
5 . Other symptom self-efficacy
6 . Alpha coefficient (measure of internal reliability)
V. Confirmatory analysis (n = 144 and n = 97)*
1. Factor analysis
2. Alpha coefficient (measure of internal reliability)
VI. Test-retest reliability study (n = 91)
* Scales generated on replication sample, then applied to original
development sample.
disability (measured by the Health Assessment Question-
naire) (9), and depression (measured by the short version of
the Beck Depression Scale) (10). All instruments have been
tested for reliability and, when appropriate, for validity.
Subjects composing the initial group entered the
ASMC study in the fall of 1984. Subjects for the replication
group entered the study in the spring of 1985. Subjects in the
concurrent validation and reliability group were past ASMC
participants who had not previously completed the efficacy
scale questionnaire. The correlations reported in Tables 5
and 6 are for combined data from experimental and control
subjects.
RESULTS
Generation of items. Based on the suggestions
from the 1981 Conference on Outcome Measures (1 l),
the specific behaviors of importance to persons with
arthritis were identified as controlling pain and disabil-
ity. A rheumatologist composed 23 questions related
to these behaviors (12). These questions were tested
with 3 focus groups, each of which consisted of 5-10
patients. Each focus group began by answering the
questions. Afterwards, participants discussed the
questions and described problems they had had in
understanding Or interpreting them. Next, the concept
of self-efficacy was explained, and participants were
asked to write down additional items related to control
of their arthritis symptoms. The focus groups helped
reshape many of the Original items and added 2o items
that dealt with arthritis symptoms. Thus, there were 43
items in the original item pool.
SELF-EFFICACY SCALE 39

Table 2. Characteristics of the 3 study samples used to develop and evaluate the perceived
self-efficacy scale
Initial development Replication Reliability
sample (n = 97) sample (n = 144) sample (n = 91)
~ ~

% female 86 80 78
Age (mean ? SD) 64.7 k 12.9 63.7 2 12.9 62.8 ? 13.2
Education (mean ? SD years) 14.1 ? 3.2 14.7 ? 2.8 14.0 ? 2.8
9% with osteoarthritis 56 58 43
% with rheumatoid arthritis 15 22 57
9% with other arthritis 29 20 0

Initial instrument development study. The 43 which represent the variance of the factors within each
items were transformed into questions. At baseline scale, were 4.83 and 4.76 for FSE and OSE, respec-
and 4 months later, 97 people who were enrolled in an tively. Alpha coefficients, which test for the internal
ASMC answered this self-administered questionnaire. reliability of items within a subscale, were estimated at
The characteristics of this sample are described in 0.93 for FSE and 0.90 for OSE.
Table 2. Using their answers, we reduced the original Self-efficacy theory predicts that present SE
set of 43 items to 25 items that were related to SE by will be related both to present health status and, more
1) removing items that were not related to the total importantly, to future health status. Using baseline
score, and 2) analyzing principal component factors and 4-month data from the above sample, correlations
with varimax (orthogonal) rotation. A factor is a were performed to determine whether the subscales
weighted linear composite of items, and it can be acted as the theory predicts. As illustrated in Table 3,
viewed as a discrete component of a large concept that construct validity was upheld by the finding of signif-
is addressed by a subset of questions that are separa- icant correlations between baseline SE and baseline
ble from those associated with other components. health status, between baseline SE and health status 4
Utilizing the baseline data, an exploratory max- months later, and between 4-month SE and 4-month
imum likelihood factor analysis identified 2 factors health status. FSE was most highly related to function
(1 3). We labeled these as self-efficacy for physical (disability); OSE was most highly related to depres-
function (FSE) and self-efficacy for controlling other sion.
arthritis symptoms (OSE). Item loadings (relative Concurrent validity test. In addition to construct
weights of the individual items to the whole factor validity, the FSE items were tested for concurrent
score) on the FSE ranged from 0.41 to 0.78, and on the validity. A new sample of 43 people with arthritis
OSE, item loadings ranged from 0.44 to 0.82. These completed the FSE scale by mail. A trained observer
loadings are all higher than the 0.40 criterion com-
who was blinded to their responses then visited in their
monly used in questionnaire design (14). Eigenvalues,
homes. Each subject was asked to perform the same
tasks on which they had previously rated their per-
Table 3. Associations between self-efficacy (SE) and health
status (HS) in initial development sample* ceived SE. The observer timed their performance of
the tasks and rated each according to a scale of 0 (no
Pain Disability Depression
difficulty) to 4 (great difficulty). Thus, actual perfor-
Baseline S E with baseline HS mance was compared with perceived ability, or effi-
FSE -0.35 -0.68 -0.45 cacy, to perform. Self-efficacy theory differentiates
OSE -0.45 -0.41 -0.45
Baseline S E with 4-month HS between perceived performance and actual perfor-
FSE -0.44 -0.68 -0.32 mance, but there is usually a moderately high correla-
OSE -0.40 -0.42 -0.39 tion between the two. For this sample, we found a
4-month SE with 4-month HS
FSE -0.48 -0.73 -0.48 correlation of r = 0.61 ( P < 0.01).
OSE -0.55 -0.54
~~
-0.60 Replication study. The same 43-item SE ques-
* Data were combined from a control group (n = 24) and an tionnaire was administered to 144 new subjects who
experimental group (n = 73) who took the Arthritis Self-Manage- had been recruited for an ASMC. Table 2 contains
ment Course (see text for details). Associations were analyzed using demographic information concerning this second
Pearson correlations. All correlations were significantly different
from zero (P < 0.01). FSE = function self-efficacy; OSE = other (“replication”) sample. Using the answers from this
symptom self-efficacy. second sample, the 43 items were reanalyzed in a
40 LORIG ET AL

Table 4. Arthritis Self-Efficacy Scale*

Self-efficacy pain subscale
In the following questions, we'd like to know how your arthritis pain affects you. For each of the following questions, please circle the
number which corresponds to your certainty that you can now perform the following tasks.
1. How certain are you that you can decrease your pain quite a bit?
2. How certain are you that you can continue most of your daily activities'?
3. How certain are you that you can keep arthritis pain from interfering with your sleep?
4. How certain are you that you can make a small-to-moderate reduction in your arthritis pain by using methods other than taking
extra medication?
5 . How certain are you that you can make a &gg reduction in your arthritis pain by using methods other than taking extra
medication'?

Self-efficacy function subscale

We would like Lo know how confident you are in performing certain daily activities. For each of the following questions, please circle
the number which corresponds to your certainty that you can perform the tasks as of now, without assistive devices or help from
another person. Please consider what you routinely can do, not what would require a single extraordinary effort.
AS OF NOW, HOW CERTAIN ARE YOUTHAT YOU CAN:
I . Walk 100 feet on flat ground in 20 seconds?
2. Walk 10 steps downstairs in 7 seconds?
3. Get out of an armless chair quickly, without using your hands for support'!
4. Button and unbutton 3 medium-size buttons in a row in 12 seconds'!
5 . Cut 2 bite-size pieces of meat with a knife and fork in 8 seconds?
6 . Turn an outdoor faucet all the way on and all the way off?
7. Scratch your upper back with both your right and left hands?
8. Get in and out of the passenger side of a car without assistance from another person and without physical aids?
9. Put on a long-sleeve front-opening shirt or blouse (without buttoning) in 8 seconds'!

Self-efficacy other symptoms subscale

In the following questions, we'd like to know how you feel about your ability to control your arthritis. For each of the following
questions, please circle the number which corresponds to the certainty that you can now perform the following activities or tasks.
1 . How certain are you that you can control your fatigue?
2. How certain are you that you can regulate your activity so as to be active without aggravating your arthritis?
3. How certain are you that you can do something to help yourself feel better if you are feeling blue'?
4. As compared with other people with arthritis like yours, how certain are you that you can manage arthritis pain during your
daily activities?
5. How certain are you that you can manage your arthritis symptoms so that you can do the things you enjoy doing?
6. How certain are you that you can deal with the frustration of arthritis?

Each subscale is scored separately, by taking the mean of the subscale items. If one-fourth or less of the data are missing, the score is a mean
of the completed data. If more than one-fourth of the data are missing, no score is calculated. (The authors invite others to use the scale and
would appreciate being informed of study results.)

second factor analysis of baseline data. This time, 3
SE subscales, with a total of 20 items, were identified:
an FSE scale of 9 items, an OSE scale of 6 items, and
a pain-management self-efficacy scale (PSE) of 5 items
(Table 4). Item loadings ranged from 0.55 to 0.84 on
the FSE factor, from 0.63 to 0.81 on the OSE factor,
and from 0.48 to 0.75 on the PSE factor. Eigenvalues
were 4.47, 3.61, and 2.11 for FSE, OSE, and PSE,
respectively. Coefficient alpha estimates of internal
reliability were 0.89 for FSE, 0.87 for OSE, and 0.76
for PSE. The 3 subscales acted as self-efficacy theory
predicts, with SE being correlated with present and
future health status (Table 5 ) .
Confirmatory subscales analysis. The 3 sub-
scales, as derived from the replication sample, were
then applied to data from the initial 97-person sample
used for development of the instrument. Confirmatory
factor analysis identified item loadings ranging from
0.59 to 0.75 for the FSE factor, from 0.55 to 0.90 for
the OSE factor, and from 0.45 to 0.82 for the PSE
factor (Table 6). Eigenvalues were 4.60 for FSE, 3.61
for OSE, and 2.11 for PSE. Coefficient alpha estimates
of internal reliability were 0.90 for FSE, 0.87 for OSE,
and 0.75 for PSE.
Test-retest reliability study. Because self-
efficacy is a changeable psychological state, rather
SELF-EFFICACY SCALE 41

Table 5 . Associations between self-efficacy (SE) and health ous participants in the ASMC. Characteristics of this
status (HS) in replication sample* third sample are also presented in Table 2, under the
Pain Disability Depression heading of “reliability sample.”
Baseline S E with baseline HS All subjects completed the 3 SE subscales twice
FSE -0.29 -0.76 -0.16 (mean time between completions 9.4 days, range 2-29
OSE -0.27 -0.25 -0.44
PSE -0.29 -0.21 -0.33
days). Utilizing Pearson correlations, item reliabilities
Baseline S E with 4-month HS ranged from 0.71 to 0.85. Subscale reliabilities were
FSE -0.28 -0.71 -0.21 0.85 for FSE, 0.90 for OSE, and 0.87 for PSE.
OSE -0.36 -0.21 -0.41
PSE -0.26 -0.24 -0.31 Tests of hypotheses using the SE scale. The
4-month S E with 4-month HS original reason for designing an arthritis self-efficacy
FSE -0.30 -0.7 I -0.30 instrument was to test 3 hypotheses: first, that there is
OSE -0.47 -0.21 -0.59
PSE -0.39 -0.21 -0.45 an association between perceived SE and both present
and future health status related to arthritis; second,
* Data were combined from a control group (n = 49) and an
experimental group (n = 95) who took the Arthritis Self- that SE can be changed by educational interventions;
Management Course (see text for details). Associations analyzed and third, that growth in SE is associated with im-
using Pearson correlations. All correlations were significantly dif- provement in health status. Our observations (Tables 3
ferent from zero (P < 0.01). FSE = function self-efficacy; OSE =
other symptom self-efficacy; PSE = pain self-efficacy. and 5) are consistent with the first hypothesis.
Outcome data about the experimental and con-
trol groups from the replication sample are given in
than a more permanent personality trait, one would Table 7. Levels of pain and depression declined sig-
expect some variance in the responses over time. To nificantly from baseline values, whereas perceived SE
test the reliability of the subscales, a third (new) for pain and for other symptoms rose significantly
sample of 91 subjects was selected from among previ- from baseline levels for the experimental group but not

Table 6 . Item factor loadings and itemitotal correlations

Two-factor scale Three-factor scale It e mitotal
correlation,
Initial Replication Initial Replication initial
Item no. sample sample sample sample sample
Function 1 0.74* 0.55* 0.75* 0.55* 0.56
Function 2 0.67 0.63 0.69 0.63 0.64
Function 3 0.61 0.66 0.62 0.66 0.69
Function 4 0.75 0.76 0.75 0.76 0.69
Function 5 0.76 0.84 0.75 0.84 0.76
Function 6 0.76 0.69 0.75 0.69 0.63
Function 7 0.59 0.68 0.59 0.68 0.59
Function 8 0.69 0.59 0.69 0.59 0.58
Function 9 0.71 0.72 0.71 0.72 0.65

Other 1 0.571‘ 0.66’F 0.57t 0.63t 0.60

Other 2 0.70 0.78 0.70 0.78 0.64
Other 3 0.54 0.70 0.55 0.7 I 0.61
Other 4 0.89 0.70 0.90 0.74 0.79
Other 5 0.86 0.79 0.87 0.81 0.79
Other 6 0.63 0.76 0.64 0.80 0.55

Pain 1 0.58 0.57 0.598 0.618 0.34

Pain 2 0.79 0.48 0.82 0.48 0.58
Pain 3 0.44 0.56 0.45 0.71 0.56
Pain 4 0.48 0.62 0.49 0.75 033
Pain 5 0.61 0.58 0.62 0.61 0.60
~ ~ ~

* These sections describe the self-efficacy function subscale.

t The following 11 items describe the self-efficacy painiself-efficacyother symptoms subscale (combined).
f These sections describe the self-efficacy other symptoms subscale (separate).
I These sections describe the self-efficacy pain subscale (separate).
42 LORIG ET AL

Table 7. Arthritis Self-Management Course outcomes in the replication sample: baseline values and 4-inonth changes“
Baseline scores Four-month change scores
Experimental Control group Experimental Control group
group (n = 95) (n = 49) group (n = 95) (n = 49) t
-
Health status
Pain (0-10 scale) 5.21 2 2.36 5.61 2 2.0 -1.31 -+ 2.42t -0.33 t 1.75 2.51$
Disability (0-3 scale) 0.72 ? 0.63 0.91 t 0.65 -0.02 -t 0.38 -0.05 t 0.35 -0.68
Depression (0-39 scale) 4.32 2 3.91 4.84 ? 3.72 -0.76 2 2.45s 0.47 2 3.31 2.52$
Self-efficacy
Function (10-100 scale) 73.27 2 20.22 67.94 2 22.45 0.76 2 14.62 0.00 -c 12.81 0.32
Other symptoms ( l & l O O 55.62 2 21.64 49.15 ? 20.56 5.85 i 19.219 2.87 2 13.29 0.98
scale)
Pain (10-100 scale) 52.04 ? 21.14 48.20 ? 17.88 7.10 -t 19.975 2.52 2 15.97 1.41

* Values are the mean k SD. t values calculated by Student’s 2-tailed, 2-sample f-test comparing change scores of experimental and control
groups. The 2-tailed, paired t-test was used to compare baseline with 4-month change scores for experimental and control groups.
i P < 0.01 versus baseline.
$ P < 0.05, experimental versus control change scores.
0 P < 0.05 versus baseline.

for the control group. The change scores for pain and
depression were significantly different between the 2
groups, but those for S E were not. The correlations of
change scores for FSE, OSE, and PSE with disability,
depression, and pain were, respectively, -0.32,
-0.29, and -0.41 in the experimental group and
-0.20, 0.07, and 0.08 in the control group. Thus, the
health outcome improvement values were similar to
those found in prior studies (1,2), SE increased during
the ASMC, although not to a statistically significant
level, and the improvements in health outcomes and
SE were correlated in the experimental group. These
observations are consistent with the second and third
hypotheses.
This consistency serves as evidence of con-
struct validity of the instrument. However, the data
were obtained from the sample used to develop the
instrument, and participants from courses that did not
emphasize enhancement of SE were included. There-
fore, inferences concerning the validity of the hypoth-
eses must remain tentative until independent samples
are tested and appropriate interventions are used.
DISCUSSION
It is widely recognized that many of the at-
tributes of patients and their circumstances (e.g.,
biological, psychological, sociological characteristics)
influence the outcome of illness, especially chronic
illness. However, the mechanisms and relative contri-
butions of each attribute are poorly understood. One
obstacle to understanding has been the absence of
instruments for measuring many of the attributes.
Experience with the Arthritis Self-Management
Course prompted our formal effort to create an instru-
ment with which to measure an individual’s perceived
self-efficacy to cope with and/or ameliorate the conse-
quences of chronic arthritis. We used standard meth-
ods for designing and testing this questionnaire, based
on specific issues that are important to people who
have arthritis. The initial instrument was derived from
one sample, refined with another sample, and then the
refined instrument was retested with the original sam-
ple. This process yielded an instrument with 3 sub-
scales and a total of 20 questions, which met reason-
able standards for construct and concurrent validity
and test-retest reliability.
When used to evaluate the 2 groups of the
replication sample (those who took the ASMC and
those who did not), the instrument revealed that there
was significant growth in self-efficacy among subjects
in the ASMC group, and a high association between
the increase in perceived SE and the decline in pain.
Thus, the instrument performed well during its devel-
opment and in a preliminary test, discriminating pa-
tients who received educational intervention from
patients in the control group.
The design and performance of these scales
raise a number of issues. First, do the scales actually
measure perceived self-efficacy for behavior, or do
they measure outcome, or some combination of the
two? The interrelationships between perceived SE for
behavior and expectations of outcomes from the be-
havior are a highly disputed issue in psychology. The
controversy can be illustrated in this case by the
question “How certain are you that you can decrease
SELF-EFFICACY SCALE
your pain quite a bit?” In this case, the behavior is
represented by the word “decrease,” and the outcome
is represented by a pain level. It would be difficult to
distinguish absolutely a person’s perceived SE to
decrease pain (by whatever means) from some expec-
tation about ultimate levels of pain, and although the
distinction is clearly important in psychological the-
ory, we do not believe it is central to our purposes.
Our goal is to reduce the patient’s level of pain
in order to improve the patient’s level of well-being
and activity potential. We therefore seek to identify
those elements of health education that contribute to
this goal. At the outset of these studies, we anticipated
a straightforward sequential relationship, progressing
from intervention, through behavior change (e.g., spe-
cific exercise, use of relaxation, and general activity),
to reduced pain and depression, accompanied by in-
creasing physical capability. The associations actually
observed, however, did not sustain a strong sequential
relationship. Rather, the health outcomes correlated
much more strongly with a perceived ability “to do
things” (perceived self-efficacy) that would yield the
desired outcomes. Although the patients’ perceived
ability “to do things” grew during the course, the
“things” themselves were apparently unspecified in
the course or were not the principal behaviors that
were taught. So, to return to the issue of the inter-
relationship between a perceived ability to decrease
pain and the expected outcome of a lower level of
pain, the interrelationships can be pursued meaning-
fully only when the “things” that can be done to
diminish pain are known. We are currently attempting
to identify those “things.”
What are the “things” that the patients might
do to decrease their pain or depression level? The
principal behaviors that were taught in the ASMC
probably play a role, but the role appears to be small
even as an aggregate, given the weak associations with
outcomes and the low degree of outcome variance that
they explain. Other possible behaviors are seeking or
using social support in spite of handicap; reducing fear
about an activity causing physical harm; learning
useful techniques from discussions with others in the
class (e.g., cognitive distraction, acupressure, massage
and/or counter-stimulation); developing more realistic
expectations concerning the prognosis; and develop-
ing a greater capacity to tolerate discomfort. Pres-
ently, these ideas are only speculative. However, in a
small preliminary study, we have used conventional
efficacy-enhancing methods (e.g., individual goals.
specific instruction and practice, contracting, model-
43
ing, and reinterpretation of physiological symptoms)
over a short term to educate a group of patients with
significant rheumatoid arthritis, and we have observed
even greater gains in perceived SE and larger correla-
tions with health outcomes (6). It is thus the task of
further study to identify the operational components of
efficacy in this setting, their relative importance to a
favorable outcome, ways in which they can be altered,
and possibly, the relationship between perceptions of
efficacy for behavior and expectations for outcome.
The foregoing discussion bears on a second
issue concerning the interpretation of scale results.
Could the strength of the correlations between per-
ceived SE and health outcomes be the result of a
similarity between efficacy and outcome expectations?
That is, are we measuring very similar things and,
hence, finding them to be correlated? Again, until we
can identify the behaviors about which the individuals
perceive themselves to be efficacious, we cannot disen-
tangle their independent effects. Yet, without conscious
attention to this concept in the ASMC, perceived SE
was enhanced and outcomes were improved, and
changes in SE were correlated with changes in health
status. Perhaps SE grew because of the course’s
general emphasis on self-management or because of
similarities between some of the course content and
efficacy-enhancing techniques. Whatever the ultimate
explanation of the results and the correlations, it is
clear that desired events occurred and that under-
standing and strengthening those events is the appro-
priate focus for future studies. We do not know
whether the results will fit neatly into present concepts
of SE or will require others.
An interesting related issue concerns the rea-
sons SE and outcomes improved for pain and depres-
sion but not for specific measures of physical function.
The data do not indicate an answer, but we surmise
that physical capability was probably relatively fixed
over the time of the study. The ASMC primarily
influenced the ways people perceived their physical
disease and used their existing physical capabilities;
that is, the course improved patients’ adjustments to
their situations. The absence of change in both the
perceived SE for physical function and the measured
physical function suggests that the individuals knew
their physical limits and that their outcome expecta-
tions in this domain influenced their SE perceptions.
A third issue is whether perceived SE is a
mediator of the outcomes, and if it is a mediator, what
is the magnitude of its effect. At best, imperfect
answers can be inferred. Perceived SE correlated with
44
outcomes both before and after the course, and as
outcomes improved during the course, perceived SE
grew. Thus, the data reveal associations that are
compatible with SE causing outcomes, outcomes
causing SE, an interaction both ways, or a third factor
causing both. Clarification of the relationship depends
upon identifying clearly behaviors which are affected
by perceived SE and which are distinct from the
outcome variables. The results of the aforementioned
preliminary study using deliberate SE-enhancement
techniques are compatible with a causal sequence
from efficacy to outcome. If perceived SE, a synonym
for which might be “confidence,” were the mediator,
it could achieve both its magnitude and its effects as a
result of a combination of experience and new, specific
learning activities. Judging from our prior study ( 2 ) , if
the magnitude of the SE effect were causal with regard
to pain, it would explain about 30% of the outcome
variance in pain scores.
With regard to the performance of the instru-
ment, it is highly improbable that the 3 categories of
either efficacy or health outcomes are independent of
each other. One can readily imagine the ways in which
physical function, pain, and depression interact. This
interaction would help explain our observations of
significant correlations between each efficacy category
and all outcome variables. In this setting, the choice
between 2 or 3 instrument subscales was largely
arbitrary. Because pain is such a major concern of
arthritis patients, and because of the superior perfor-
mance of the “other” category in relation to depres-
sion, we prefer the instrument with 3 subscales.
Testing the performance of the instrument with
other groups of patients in other settings is clearly
essential, as is comparing the instrument results with
those that measure related psychological states, such
as general coping skills (15), learned helplessness (16),
locus of control (17), and sense of coherence (18).
However, the results thus far encourage the view that
an arthritis-specific instrument that reveals important
psychological determinants of health outcomes for
persons with chronic arthritis can be constructed. Use
of such an instrument might help to disentangle and
illuminate the multiple factors that influence those
outcomes.
ACKNOWLEDGMENTS
The authors acknowledge the extensive, expert sta-
tistical assistance and contributions of Byron W. Brown, Jr.,
PhD and Daniel A. Bloch, PhD.
LORIG ET AL
REFERENCES
1 . Lorig K, Lubeck D, Kraines RG, Seleznick M, Holman
HR: Outcomes of self-help education for patients with
arthritis. Arthritis Rheum 28:68&685, 1985
2. Lorig K, Seleznick M, Lubeck D, Ung E, Chastain RL.
Holman HR: The beneficial outcomes of the arthritis
self-management course are not adequately explained by
behavior change. Arthritis Rheum 32:91-95, 1989
3. Lenker SL, Lorig K , Gallagher D: Reasons for the lack
of association between changes in health behavior and
improved health status: an explanatory study. Patient
Educ Counsel 6:69-72, 1984
4. Bandura A: Self-efficacy: toward a unifying theory of
behavior change. Psychol Rev 84: 191-215, 1977
5. Bandura A: Social Foundations of Thought and Action.
Englewood Cliffs, NJ, Prentice-Hall, 1986
6. O’Leary A, Shoor S , Lorig K , Holman HR: A cognitive-
behavioral treatment for rheumatoid arthritis. Health
Psychol (in press)
7. Dixon JS, Bird HA: Reproducibility along a 10-cm
vertical visual analogue scale. Ann Rheum Dis 40237-89,
1981
8. Downie WW, Leatham PA, Rhind VM, Wright V,
Branco JA, Anderson JA: Studies with pain rating
scales. Ann Rheum Dis 37:378-381, 1978
9. Fries JF, Spitz P, Kraines RG, Holman HR: Measure-
ment of patient outcome in arthritis. Arthritis Rheum 23:
137-145, 1980
10. Beck AT, Ward CH, Mendelson M , Mock J. Erbaugh J:
An inventory for measuring depression. Arch Gen Psy-
chiatry 4561-571, 1961
11. Decker JL: Summary of the Conference on Outcome
Measures in Rheumatological Clinical Trials, Hamilton,
Ontario, Canada, December 7-8, 1981. J Rheumatol
9:802-806, 1982
12. Shoor SM, Holman HR: Development of an instrument
to explore psychological mediators of outcome in
chronic arthritis patients. Trans Assoc Am Physicians
97~32.5-331, 1985
13. Lawly DN, Maxwell AE: Factor Analysis as a Statistical
Method. New York, Macmillan, 1971
14. Nunnally JC: Psychometric Theory. Second edition.
New York, McGraw-Hill, 1978
15. Lazarus RS, Folkman S: Stress Appraisal and Coping.
New York, Springer, 1984
16. Seligman M: Helplessness: On Depression, Develop-
ment, and Death. San Francisco, WH Freeman, 1975
17. Wallston BS, Wallston KA, Kaplan GD, Maides SA:
Development and validation of the health locus of con-
trol (HLC) scale. J Consult Clin Psychol 44:58&585,
I976
18. Antonovsky A: Health Stress and Coping. San Fran-
cisco, Jossey-Bass, 1980