The Brief Symptom Inventory: An Introductory Report: Psychological Medicine, 1983, 13, 595-605 Printed in Great Britain

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Psychological Medicine, 1983, 13, 595-605

Printed in Great Britain

The Brief Symptom Inventory:


an introductory report
LEONARD R. DEROGATIS1 AND NICK MELISARATOS
From the Department of Psychiatry, Johns Hopkins University School of Medicine,
Baltimore, Maryland, USA

SYNOPSIS This is an introductory report for the Brief Symptom Inventory (BSI), a brief
psychological self-report symptom scale. The BSI was developed from its longer parent instrument,
the SCL-90-R, and psychometric evaluation reveals it to be an acceptable short alternative to the
complete scale. Both test-retest and internal consistency reliabilities are shown to be very good for
the primary symptom dimensions of the BSI, and its correlations with the comparable dimensions of
the SCL-90-R are quite high. In terms of validation, high convergence between BSI scales and like
dimensions of the MMPI provide good evidence of convergent validity, and factor analytic studies
of the internal structure of the scale contribute evidence of construct validity. Several criterion-
oriented validity studies have also been completed with this instrument.

Data rising from these measures may also be


PSYCHOLOGICAL ASSESSMENT BY directly incorporated into the clinical decision
SELF-REPORT systems (Fowler, 1969; Glueck & Stroebel, 1969;
Assessment of the psychological status of Lanyon, 1972, 1974), and self-report inventories
individuals by self-reports dates back to the First have been shown to be sensitive to a wide variety
World War, and the development of the of therapeutic interventions (Kellner, 1971;
Personal Data Sheet by Woodworth (1918). Lyerly & Abbott, 1964).
Woodworth's scale provided a means for each It should also be appreciated, however, that
man to 'interview himself and created a the self-report method tacitly assumes the validity
historical benchmark for a new modality of of the 'inventory premise', i.e. the assumption
psychological measurement. that the individual being assessed can and will
The self-report mode of assessment possesses accurately describe his current symptoms and
a number of unique advantages and characteris- behaviour (Wilde, 1972), a premise which cannot
tics to recommend it, as well as several inherent always be supported. Social desirability
weaknesses. The first advantage, particularly (Edwards, 1957) and extreme response styles
concerning psychopathology, is the accessibility (e.g. aquiescence) can contribute to systematic
of exclusive information that is ordinarily distortions, although careful work has produced
unavailable through other methods of evaluation. some serious doubts about the size and
Self-report measures reflect information derived pervasiveness of these effects (Block, 1965;
directly from the person experiencing the Fiske, 1971; Rorer, 1965).
phenomena, namely the patient himself. Sec- In spite of concerns about its deficiencies, the
ondly, there is the economy of professional time self-report method remains a useful form of
involved with the use of self-reports. The clinical measurement and, as Nunnally (1978,
technique can be integrated into institutional p. 141) has stated, 'even though self-inventories
routines with relative ease and is amenable to definitely have their problems as approaches to
actuarial methods of scoring and interpretation. the measurement of personality characteristics,
attitudes, values, and a variety of other
1
Address for correspondence: Dr L. R. Derogatis, Division of non-cognitive traits, they represent by far the
Medical Psychology, Department of Psychiatry, Johns Hopkins best approach available'.
University School of Medicine, Adolf Meyer Building, Room 200,
600 N. Wolfe Street, Baltimore, MD 21205, USA.

595 20-2
596 L. R. Derogatis and N. Melisaratos

dimensions or constructs. A brief definition of


THE BRIEF SYMPTOM INVENTORY each construct is provided in the sections below.
(BSI)
The Brief Symptom Inventory (Derogatis, 1975) I. Somatization (SOM)
is a 53-item self-report symptom inventory This dimension reflects psychological distress
designed to assess the psychological symptom arising from perception of bodily dysfunction.
status of psychiatric and medical patients, as well Complaints typically focus on cardiovascular,
as individuals who are not patients. The BSI is gastrointestinal, respiratory and other systems
essentially the brief form of the SCL-90-R, a with strong autonomic mediation. Aches and
self-report inventory that has been developed pains, and discomfort localized in the gross
and used in a wide variety of settings and musculature are also frequent manifestations.
applications (Derogatis et al. 1976; Derogatis,
1977). The instrument comprises 53 items II. Obsessive-compulsive (O-C)
selected to reflect best the 9 primary symptom The focus of this dimension is on thoughts and
dimensions of the SCL-90-R in a brief measure- actions that are experienced as unremitting and
ment scale. In addition to the 9 primary irresistible by the patient but are of an ego-alien
symptom dimensions, there are three global or unwanted nature. Examples are: having to
indices of distress associated with the BSI: the check and double-check actions, difficulty in
General Severity Index (GSI), the Positive making decisions, and trouble concentrating.
Symptom Distress Index (PSDI), and the
Positive Symptom Total (PST). These same three III. Interpersonal sensitivity (IS)
summary measures are also utilized by the
SCL-90-R, each communicating psychological This dimension focuses on feelings of personal
distress in a somewhat different fashion (Dero- inadequacy and inferiority. Self-deprecation,
gatis et al. 1975). feelings of uneasiness, and marked discomfort
during interpersonal interactions are characteris-
Scale characteristics tics of persons with high levels of interpersonal
sensitivity.
Each item of the BSI is rated on a 5-point scale
of distress (0-4), ranging from 'not-at-all' to IV. Depression (DEP)
'extremely'. Test administration ordinarily takes
less than 10 minutes, 1-2 minutes being devoted Depression reflects a broad range of signs and
to the introduction of theinstrument. Instructions symptoms of the clinical depressive syndromes.
preceding the inventory are brief and worded in Symptoms of dysphoric affect and mood,
simple language, as are the 53 test items. A withdrawal of interest in life activities, and loss
primary consideration in item development was of vital energy are reflected by this dimension, as
the selection of phrasing and language that are feelings of hopelessness and futility.
would be understood by a wide segment of the
population and still retain the intended item V. Anxiety (ANX)
meaning. The Thorndike-Lorge Word Book The anxiety dimension subsumes a set of
(1944) was used to equate the vocabulary levels symptoms usually associated clinically with high
of the 9 primary dimensions and to keep the manifest anxiety. Restlessness, nervousness and
general vocabulary as basic as possible. In tension are all indicative of anxiety, as are
general, the BSI can be used with individuals who experiences reflecting free-floating anxiety and
have reached a reading knowledge equivalent to panic.
that of an American sixth grade education. In
terms of age limitations, adolescents as young as VI. Hostility (HOS)
13 years have been evaluated via the BSI without The hostility dimension is organized around
apparent distortions. three categories of hostile behaviour: thoughts,
feelings, and actions. Typical experiences cover
Dimensional structure of the BSI feelings of annoyance and irritability, urges to
The BSI, like its parent instrument the SCL-90-R, break things, frequent arguments and uncontrol-
is conceived as measuring 9 primary symptom lable outbursts of temper.
Brief Symptom Inventory 597

The PSDI is a pure intensity measure, in a sense


VII. Phobic anxiety (PHOB) 'corrected' for the numbers of symptoms. It
The symptoms that comprise this dimension functions very much as a measure of response
have been observed to occur frequently in style, communicating whether the patient is
conditions usually termed phobic anxiety states 'augmenting' or 'attenuating' distress in his/her
or agoraphobia. Phobic fears oriented towards manner of reporting. The PST is simply a count
travel, open spaces, crowds, public places or of the symptoms which the patient reports
conveyances are all represented by this experiencing to any degree. The PSDI and the
dimension. PST are used in conjunction with the GSI to gain
a more meaningful understanding of the clinical
VIII. Paranoid ideation (PAR) picture.
The present definition of paranoid ideation rests
on the assumption that paranoid behaviour is Normative development of the BSI
most accurately viewed as syndromal in nature. A fundamental procedure for the interpretation
Paranoid phenomena are conceived as a mode of of an individual's psychological test performance
thinking. The primary characteristics of paranoid is the comparison of the patient's scores or
thought are projection, hostility, suspiciousness, profile with some relevant group of individuals
centrality, and fear of loss of autonomy. who have also completed the test. Typically, the
larger the normative group of subjects, the more
IX. Psychoticism (PSY) representative the sample will be of the population
The present definition of psychoticism represents of interest.
it as a continuum, progressing from a mildly There are 3 published norms available for the
alien life style at one extreme to floridly BSI. These are based upon: (a) a sample of 1002
psychotic status at the other. Signs of a schizoid, heterogeneous psychiatric out-patients; (b) a
alienated style of life are represented by this sample of 719 non-patient normal subjects; and
dimension as are dramatic symptoms of (c) a sample of 313 psychiatric in-patients.
psychosis. In most non-psychiatric populations The psychiatric out-patient norm comprised
this dimension measures social alienation. 425 males and 577 females who represent
psychiatric out-patients presenting for initial
Additional items evaluation at four treatment facilities: (1) the
There are 4 items of the BSI which are not Out-patient Psychiatry Division of the Johns
subsumed under any of the primary symptom Hopkins Medical Institutions; (2) the Out-patient
dimensions; these symptoms 'load' on several Psychiatry Division of the University of Maryland
dimensions, but are not unique to any of them. School of Medicine; (3) the Out-patient Psychi-
They are retained in the test because they atry Clinic of the Hospital of the University of
represent important vegetative and other clinical Pennsylvania; and (4) the Out-patient Psychiatry
indicators. Division of the University of Wisconsin School
of Medicine.
Global indices of distress The non-patient norm is based upon the
There are three global indices of distress responses of 344 males and 341 females. It
associated with the BSI: the General Severity represents a stratified random sample from a
Index (GSI), the Positive Symptom Distress single county in one of the large eastern states.
Index (PSDI), and the Positive Symptom Total The psychiatric in-patient norm is based on the
(PST). The function of each of these global intake scores of 115 males and 198 females who
measures is to communicate in a single score the were admitted as in-patients to the Phipps Clinic,
level or depth of symptomatic distress currently Johns Hopkins Hospital. All of these patients
experienced by the individual. The GSI is the were evaluated during the first week following
single best indicator of current distress levels, admission.
and should be utilized in most instances where
a single summary measure is required. The GSI Interpretation of the BSI
combines information on the numbers of Both the BSI and the SCL-90-R are designed for
symptoms and the intensity of perceived distress. interpretation at three distinct but related levels.
598 L. R. Derogatis and N. Melisaratos

Table 1. Demographic characteristics (%) of psychiatric out-patient (N = 1002), psychiatric


in-patient (N = 310) and non-patient subject (N = 719) normative samples for the BSI
Variable Psychiatric out-patient Psychiatric in-patient Non-patient

Age (mean±s.D.) 31O2± 121 33-31 ±14-85 4600 ±14-7


Sex
Male 42-4 370 50-7
Female 57-6 630 49-3
Race
White 671 55-7 86-8
Black 32-6 43-6 11-6
Other — 0-8 —
Marital Status
Single 45-7 44-5 —
Married 32-4 26 1 —
Other 21-6 29-3
Single or married — — 84-6
Divorced — — 15-4
Religion
Catholic 32-7 24-8 —
Protestant 363 64-8 —
Jewish 7-3 5-3
Other 23-4 50 —
Social class
I 4-3 2-3
II 7-8 2-6
III 24-2 16-2
IV 30-8 44-5 —
V 32-4 34-3 —

Initially, interpretation should focus on the 1002 psychiatric out-patients and 310 psychiatric
global scores to gain an appreciation of the in-patients plotted against the non-patient norm.
degree of overall distress. The evaluation should Focusing on the global indices first, the GSI for
then be refocused to integrate information at the both groups of patients is two standard
level of the primary symptom dimensions. The deviations above the normative mean. This
BSI primary symptom dimensions are designed magnitude of distress places the mean patient at
to provide a 'broad brush' profile of the the 98th centile of the non-patient norm. The
respondent which can delineate and underline PSDI and PST are each elevated about 1-6
specific areas of psychopathology. A more standard deviations, placing these groups in the
specific focus is provided at the level of the 95th centile of the norm. The 9 symptom
discrete symptoms; these individual items com- dimensions of the patient profiles are all clearly
municate detailed symptomatic manifestations elevated, ranging from 1-3 to 2-3 standard
which often further enhance and amplify the deviations above the normative means.
clinical decision process.
The 9 primary symptom dimensions essentially Reliability of the BSI
provide a profile of the patient's psychological The reliability estimates for the 9 symptom
status in psychopathological terms. They com- dimensions and the 3 global indices of the BSI
municate information on the nature and intensity are essentially of two types: internal consistency
of the patient's distress, and provide data and test-retest reliability. The former serves to
concerning the pattern of the patient's sympto- measure the homogeneity or consistency of the
matology. Table 2 and Fig. 1 depict the mean raw items selected to represent each symptom
scores on the 9 primary symptom dimensions construct. Test-retest reliability is essentially an
and the 3 global indices in tabular and graphical indicator of the stability of the measurement
forms for three normative clinical groups. across time. Table 3 provides internal consistency
Fig. 1 depicts the BSI symptom profiles of the (alpha) coefficients for the 9 symptom dimensions
Brief Symptom Inventory 599

Table 2. Mean raw scores on the 9 primary symptom dimensions and the 3 global indices of the
BSI for psychiatric out-patients, psychiatric in-patients and non-patient normals
Psychiatric out- Psychiatric in-
patients patients Non-patients
(mean±S.D.) (mean±s.D.) (mean±s.D.)

Symptom dimension
I. SOM 0-83 ±0-80 101 ±0-91 0-29 ±0-40
II. O-C •57+1-00 1-51 ±1-07 0-43 ±0-48
III. I-S 58±1O5 1 -48 ±1-11 0-32 ±0-48
IV. DEP •80+108 1-77±1-21 0-28 ±0-46
V. ANX •70 ±1-00 l - 7 0 ± 1-15 0-35±0-45
VI. HOS •16±0 93 1 00 ±0-97 0-35±0-42
VII. PHOB ()-86±0-88 1 -07 ±1-11 0 1 7 ± 0 36
VIII. PAR •14±0-95 1-26 ±1-02 0-34±0-45
IX. PSY •19±0-87 l-26±0-98 0 15±0-31
Global indices
GSI l-32±0-72 l-36±0-86 O-3O±O-31
PSDI 215±0-61 2-16 ±0-73 l-29±0-41
PST 30-80 ± 11-63 30-60± 13-40 11 -45 ±9-29

See text for a definition of abbreviations.

80

60-- --84

§ 55-- --70
H
50

45-- --30

40-- --16

35-- --7

30 4- -4- 2
SOM O-C +
I-S DEP ANX HOS PHOB PAR PSY GSI PSDI PST
FIG. 1. BSI symptom profiles for 1002 psychiatric out-patients (A # ) and 310 psychiatric in-patients ( 0 — 9 ) plotted against
the non-patient norm. Abbreviations are given in the text.

and test-retest coefficients for the 9 dimensions Test-retest reliability reflects the stability or
and the 3 global measures. consistency of measurement across time. As
The internal consistency reliability was estab- longer periods of time elapse between measure-
lished on the sample of 1002 out-patients and ments, there is a greater opportunity for change to
was determined by utilizing Cronbach's co- be effected; stability coefficients are typically
efficient alpha (a). Alpha coefficients for all 9 related inversely to time elapsing between
dimensions ranged from a low of 0-71 on the administrations of the measure. The stability
psychoticism dimension to a high of 0-85 for coefficients listed in Table 3 were generated from
depression. BSI data on a sample of 60 non-patient subjects
600 L. R. Derogatis and N. Melisaratos

Table 3. Internal consistency and test-retest Table 4. Correlations between like symptom
reliability coefficients for the 9 primary symptom dimensions of the SCL-90-R and the BSI based
dimensions and the 3 global indices of the BSI upon 565 psychiatric out-patients
Internal SOM 0-96 HOS 0-99
No. of consistency (a) Test-retest O-C 0-96 PHOB 0-97
items ( # = 719) I-S 0-94 PAR 0-98
DEP 0-95 PSY 0-92
Symptom dimension ANX 0-95
I. SOM 0-80 0-68
II. O-C 0-83 0-85
III. I-S 0-74 0-85
IV. DEP 0-85 0-84
V. ANX 0-81 0-79 Validity of the BSI
VI. HOS 0-78 0-81
VII. PHOB 0-77 0-91 Two major issues that should be appreciated
VIII. PAR 0-77 0-79 concerning the validity of psychological tests
IX. PSY 0-71 0-78
have to do with (a) the specificity of predictive
Global indices
GS1 0-90
validity, and (b) the programmatic nature of
PSDI 0-87 construct validation. The former issue refers to
PST 0-80 the observation that for the question ' Is this test
valid?' to have any scientific meaning, the
who were tested at a 2-week interval. Values conditional statement' For what purpose?' must
ranged from a low of 0-68 for somatization to a be appended.
maximum of 0-91 for phobic anxiety. The The second issue focuses on the fact that
stability coefficient for the GSI was 0-90, psychometric authorities have increasingly
strongly indicating that the BSI is a reliable stressed construct validity as the central criterion
measure over time. for the validation of psychological tests and the
Indices of stability for psychopathological assignment of meaning to these measures
syndromes ordinarily fall between those for (Messick, 1975, 1981). Such assertions demand
stable personality characteristics such as' intelli- an extensive programme of related experiments
gence' and more labile attributes such as that are analogous to the steps necessary for the
'mood'. The 9 dimensions of the BSI reflect high proof of scientific theory. Data from predictive,
levels of stability, with dimensions regarded as content, convergent, discriminant and other
being more state-determined, revealing some- types of validation studies contribute to the
what lower coefficients than those mediated ultimate validation of the hypothetical construct
more by the characterological (trait) structure of that the test serves to operationalize. It is
the individual. important to realize that this process should be
Another traditional form of reliability for represented by an ongoing series of experiments
psychological tests is that of alternate forms. that constantly extend and redefine the limits of
Alternate forms reliability is represented by the generalizability of the test as a definition of the
construct.
correlation between score distributionsdeveloped
from two different forms of the test, usually
administered within approximately 1-2 weeks of Convergent and discriminant validity
each other. Although the BSI and the SCL-90-R As Campbell & Fiske (1959) have demonstrated,
are not strictly speaking' alternate forms' of the convergent and discriminant relationships be-
same test, they do represent two tests measuring tween operational measures of constructs (e.g.
the same symptom constructs. For this reason psychological tests) and other operational mea-
we felt that correlations between the two would sures are necessary to establish the network that
provide useful data, and have presented them in forms the basis of construct validation. In simple
Table 4, based upon the responses of 565 terms, the concept requires that scores from a test
psychiatric out-patients. Correlations between designed to measure a particular construct
the two are uniformly very high across all 9 should correlate highly with other measures of
dimensions, demonstrating that they measure that construct, and show relatively low correla-
essentially the same symptom constructs. tions with measures of dissimilar constructs. Such
Brief Symptom Inventory 601

Table 5. Correlations between BSI symptom dimensions and MMPI clinical, Wiggins (IV) and
Try on (T) scores*
I. Somatization Schizophrenia (MMPI) 0-48
Body symptoms (T) 0-38 Poor morale (W) 0-45
Hypochodriasis (MMPI) 0-37 Autism (T) 0-42
Organic symptoms (W) 0-36 Resentment and aggression (T) 0-41
Poor health (W) 0-33 Organic symptoms (W) 0-40
II. Obsessive-Compulsive Phobia (W) 0-40
Schizophrenia (MMPI) 0-44 VI. Hostility
Psychasthenia (MMPI) 0-44 Resentment and aggression (T) 0-56
Depression (W) 0-39 Manifest hostility (W) 0-48
Organic symptoms (W) 0-38 Depression (W) 0-42
Autism (T) 0-38 Suspicion and mistrust (T) 0-35
Resentment and aggression (T) 0-36 Family problems (W) 0-35
III. Interpersonal Sensitivity Anxiety (T) 0-31
Poor morale (W) 0-63 VII. Phobic Anxiety
Psychasthenia (MMPI) 0-55 Phobias (W) 0-45
Introversion (T) 0-52 Psychasthenia (MMPI) 0-35
Schizophrenia (MMPI) 0-49 Anxiety (T) 0-30
Depression (T) 0-48 Poor morale (W) 0-30
Depression (W) 0-47 Depression (W) 0-30
Social maladjustment (W) 0-47 VIII. Paranoid Ideation
Social introversion (MMPI) 0-44 Suspicion and mistrust (T) 0-47
IV. Depression Resentment and aggression (T) 0-42
Depression (W) 0-72 Manifest hostility (W) 0-41
Depression (T) 0-67 Family problems (W) 0-41
Poor morale (W) 0-57 Paranoia (MMPI) 0-35
Schizophrenia (MMPI) 0-52 Autism (T) 0-35
Resentment and aggression (T) 0-52 IX. Psychoticism
Psychasthenia (MMPI) 0-46 Schizophrenia (MMPI) 0-48
Anxiety (J) 0-45 Psychopathic deviate (MMPI) 0-40
Autism (T) 0-43 Poor morale (W) 0-39
V. Anxiety Psychoticism (W) 0-38
Anxiety (T) 0-57 Psychasthenia (MMPI) 0-38
Depression (W) 0-48 Autism (T) 0-37
Psychasthenia (MMPI) 0-48 Paranoia (MMPI) 0-32

• Correlations below 0-30 are omitted.

a pattern of relationships should hold if the test dimensions showed excellent convergence (e.g.
is to be considered a valid reflection of the Somatization with' Body symptoms', Depression
construct under consideration. with 'Depression' and Paranoid Ideation with
A previous study, based on a sample of 209 'Suspicion and mistrust'). Even the Obsessive-
symptomatic volunteers, demonstrated impres- Compulsive dimension, for which there is no
sive convergent validity for the SCL-90-R with the truly analogous construct among the MMPI
MMPI (Derogatis et al. 1976). Since the 53 items scales, revealed a pattern of maximal correlations
of the BSI are contained within the SCL-90-R, (i.e. Schizophrenia, Organic symptoms, Psychas-
the data set was re-analysed, scoring for the BSI thenia) which is highly consistent with the
instead of the SCL-90-R. Table 5 contains the definition of the clinical construct.
correlational results of this re-analysis, listing The current re-analysis, in terms of the BSI,
coefficients ^ 0-30 between the 9 dimensions of also revealed excellent convergence, although in
the BSI, and (1) the clinical scales of the MMPI the case of several dimensions the overall
(Dahlstrom, 1969), (2) the Wiggins Content magnitudes of correlations were somewhat
Scales of the MMPI (Wiggins, 1966), and (3) the reduced. Interpersonal Sensitivity, Depression,
Tryon Cluster Scores (Tryon, 1966). Anxiety, Hostility, Phobic Anxiety, Paranoid
In the original SCL-90-R study (Derogatis et Ideation and Psychoticism all demonstrated
al. 1976), 8 of the dimensions of the SCL-90-R maximum correlations with MMPI scales that
demonstrated directly convergent counterparts were clearly convergent (see Table 5). The
among the 30 MMPI scales evaluated, and all 8 magnitudes of maximal correlation coefficients
602 L. R. Derogatis and N. Melisaratos

for the former three were almost identical with The first factor generated was essentially the
those in the SCL-90-R study. In the cases of the Psychoticism factor, with 4 of the 5 items
latter four measures, the magnitudes of correla- comprising the hypothesized dimension found to
tions were reduced by about 010, although the load on the empirically derived factor. Two items
pattern of correlations remained very similar in assigned to the Interpersonal Sensitivity dimen-
the two studies. In the cases of the Somatization sion also correlated highly on this factor; so did
and Obsessive-Compulsive dimensions, although one item from the Depression dimension.
the patterns of correlations were retained, the The second factor to emerge from the analysis
magnitudes of coefficients decreased by approxi- was the Somatization factor, with all 7 hypothe-
mately 015. Apparently, items deleted from the sized items showing marked correlations with the
shorter form of the test were more important for dimension. Loadings were high, and the pattern
convergence with the MMPI in these dimensions was consistent with previous confirmatory factor
than in the other 7 measures. analysis of the SCL-90-R (Derogatis & Cleary,
The general finding of high convergence for 1977 a). Analogous findings were observed with
the dimensions of the BSI with MMPI scales, Hostility (Factor IV), Obsessive-Compulsive
although not unexpected, confirms the fact that (Factor VI), and Paranoid Ideation (Factor VII),
the reduction of the length of the SCL-90-R which also revealed well-saturated loadings for
dimensions has not had a significant effect upon all items derived from the a priori rational
their validity. Reductions in the magnitude of structure.
coefficients almost certainly reflect some loss of Among the anxiety measures, loadings on the
reliability associated with shortening the scales, dimension of Phobic Anxiety were substantial
but convergent patterns of relationship remain and the hypothesized pattern was intact, except
clearly in evidence. for the item' feeling nervous when you are alone'
which did not correlate in this sample. An
Internal structure and construct validity additional item from the Interpersonal Sensitivity
Particularly when dealing with multidimensional dimension, 'feeling very self-conscious with
tests, the issue of internal structure is also critical others', also loaded on the Phobic Anxiety
to the question of construct validation. If the test dimension. The general Anxiety dimension split
is designed to measure a certain number of into two more specific component dimensions
dimensions or constructs defined via specified that were labelled 'panic anxiety' (Factor VII)
series of item sets, then the dimensions defined in and 'nervous tension' (Factor IX), afindingthat
this way should emerge in the analysis of a was evident to a lesser degree in the confirmatory
representative empirical problem. The technique SCL-90-R study alluded to above (Derogatis &
usually employed to test the equivalence of Cleary, 1977 a).
hypothetical versus empirical test structures is The Depression dimension (Factor III) was
factor analysis, and we have utilized the method well defined in the present instance, with only one
here in a similar fashion. item, 'feelings of worthlessness', showing a high
To assess the reproducibility of the internal loading on another factor. In addition, the
structure of the BSI, the scores of the psychiatric fourth of the Interpersonal Sensitivity items was
out-patient sample (N = 1002) described earlier also observed to load on the Depression factor.
were subjected to a principal components There is no obvious explanation why the
analysis with 100s on the diagonal of the Interpersonal Sensitivity dimension was not
correlation matrix. The correlation matrix reproduced in the present study. This dimension
analysed was 49 x 49, omitting the 4 'additional' has been consistent and invariant in the context
items in the test since they are not hypothesized of both the SCL-90-R (Derogatis & Cleary,
to have univocal loadings on any of the 9 1977 a, b) and its predecessor, the Hopkins
primary BSI dimensions. Nine interpretable Symptom Checklist (Derogatis et al. 1972,1974).
factors were derived from a normal varimax It is possible that some peculiarity of the sample
rotation of the principal components (Harman, composition acted to 'unravel' the linear
1967), which accounted for 44% of the variance composition of this dimension; however, it
should also be appreciated that 4 items (the
in the matrix. A detailed representation of the
smallest factor in the BSI) may simply be too few
factor loadings ^ 0-35 is given in Table 6.
Brief Symptom Inventory 603

Table 6. Orthogonal varimax loadings for 9 factors determined from a principal components
analysis of 49 items of the BSI
Symptom Symptom Symptom
dimension Loading dimension Loading dimension Loading

I. Psycholicism II. Sotnatization III. Depression


3. The idea that someone else — 2. Faintness or dizziness 0-45 9. Thoughts of ending your life 0-42
can control your thoughts 7. Pains in heart or chest 0-50 16. Feeling lonely 0-65
14. Feeling lonely even when 0-54 23. Nausea or upset stomach 0 44 17. Feeling blue 0-64
you are with people 29. Trouble getting your breath 0-64 18. Feeling no interest in things 0-57
34. The idea that you should 0-42 30. Hot or cold spells 0-62 35. Feeling hopeless about the 0-36
be punished for your sins 33. Numbness or tingling in future
44. Never feeling close to 0-45 parts of your body 0-56 50. Feelings of worthlessness 0-35
another person 37. Feeling weak in parts of 060 20. Your feelings being easily 0-42
53. The idea that something 0-48 your body hurt
is wrong with your mind
22. Feeling inferior to 0-51
others
42. Feeling very self- 0-47
conscious with others
50. Feelings of worthlessness 0-63

IV. Hostility V. Phobic Anxiety VI. Obsessive-Compulsive


6. Feeling easily annoyed or 0-42 8. Feeling afraid in open 0-44 5. Trouble remembering things 0-62
irritated spaces 15. Feeling blocked in getting 0-37
13. Temper outbursts that 0-63 28. Feeling afraid to travel 0-56 things done
you could not control on buses, subways or 26. Having to check and
40. Having urges to beat, 0-56 trains double-check what you do 0-48
injure or harm someone 31. Having to avoid certain 0-44 27. Difficulty making decisions 0-43
41. Having urges to break or 0-57 things, places or 32. Your mind going blank 0-53
smash things activities because they 36. Trouble concentrating 0-53
46. Getting into frequent 0-64 frighten you
arguments 43. Feeling uneasy in crowds 0-66
47. Feeling nervous when you —
are alone
42. Feeling very self-conscious 044
with others

VII. Anxiety VIII. Paranoid Ideation IX. Anxiety


1. Nervousness or shakiness — 4. Feeling others are to blame 0-37 1. Nervousness or shakiness 0-45
inside for most of your troubles inside
12. Suddenly scared for no 0-54 10. Feeling that most people 0-59 12. Suddenly scared for no —
reason cannot be trusted reason
19. Feeling fearful 0-57 24. Feeling that you are 0-56 19. Feeling fearful —
38. Feeling tense or keyed — watched or talked about by 38. Feeling tense or keyed up 0-71
up others 45. Spells of terror
45. Spells of terror or 0-49 48. Others not giving you 0-35 or panic —
panic proper credit for your 49. Feeling so restless you 0-35
49. Feeling so restless you — achievements couldn's sit still
could not sit still 51. Feeling that people will 0-53
take advantage of you if
you let them
21. Feeling that people are 0-54
unfriendly or
dislike you

to sustain invariance across important population hypothesized symptom constructs were repro-
parameters (Gorsuch, 1966). duced with little or no disjuncture of items; an
Although there are certain minor differences eighth dimension simply split into two well-
between the empirical factor structure and the defined clinical component dimensions. The
dimensional structure rationally hypothesized, ninth dimension did not stay together as a linear
there is more agreement than disagreement combination, but the set of only 4 items that
between the two. Essentially, 7 of the 9 define this dimension may well be too small to
604 L. R. Derogatis and N. Melisaratos

ensure invariance. It may be necessary to add Campbell, D. & Fiske, D. W. (1959). Convergent and discriminant
two or three additional items from the dimension validation by the multitrait multimethod matrix. Psychological
Bulletin 56, 81-105.
as defined in the SCL-90-R. Results from the Dahlstrom, W. G. (1969). Recurrent issues in the development of the
structure-comparing factor analysis lend strong MMPI. In MM PI: Research Development and Clinical Applications
additional weight to the construct validation of (ed. J. N. Butcher), pp. 1-40. McGraw-Hill: New York.
Derogatis, L. R. (1975). Brief Symptom Inventory. Clinical Psycho-
the BSI. metric Research: Baltimore.
Derogatis, L. R. (1977). The SCL-90 Manual I: Scoring, Admini-
Predictive validity stration and Procedures for the SCL-90. Clinical Psychometric
Research: Baltimore.
In spite of the scientific value of construct Derogatis, L. R. & Cleary, P. A. (1977a). Confirmation of the
validation, the average clinician/investigator is dimensional structure of the SCL-90: A study in construct
validation. Journal of Clinical Psychology 33(4), 981-989.
usually interested in the more practical side of Derogatis, L. R. & Cleary, P. A. (1977 A). Factorial invariance across
test validity, i.e. predictive validity. Most test gender for the primary symptom dimensions of the SCL-90. British
Journal of Social and Clinical Psychology 16, 347-356.
users are interested in the practical issues of how Derogatis, L. R., Lipman, R. S., Covi, L. & Rickets, K. (1972).
well a test can register changes in psychological Factorial invariance of symptom dimensions in anxious and
status arising from mental disorders, treatment depressive neuroses. Archives of General Psychiatry 27, 659-665.
Derogatis, L. R., Lipman, R. S., Rickels, K., Uhlenhuth, E. H. &
interventions, stress-inducing life events, or a Covi, L. (1974). The Hopkins Symptom Checklist (HSCL): A
host of other factors. The sensitivity of a self-report symptom inventory. Behavioural Science 19(1), 1-15.
measuring instrument to factors effecting such Derogatis, L. R., Yevzeroff, H. & Wittelsberger, B. (1975). Social
class, psychological disorder and the nature of the psychopathologic
changes is reflected in its correlations with indicator. JournalofConsultingandClinicalPsychoIogy43,183-191.
various relevant criterion measures which, in Derogatis, L. R., Rickels, K. & Rock, A. F (1976). The SCL-90 and
the MMPI: A step in the validation of a new self-report scale.
turn, represent summary statements concerning British Journal of Psychiatry 128, 280-289.
its specific predictive validities. Edwards, A. L.(1957). The Social Desirability Variable in Personality
Although during the 1970s research with the Assessment and Research. Holt: New York.
Fiske, D. W. (1971). Measuring the Concepts of Personality. Aldine:
SCL-90-R tended to take precedence over work Chicago.
with the BSI in our laboratory, this situation has Fowler, R. D. (1969). Automation and the computer. In MMPI:
now changed substantially. Several predictive Research DevelopmentsandClinicalApplications(eA.i. N. Butcher),
pp. 105-126. McGraw-Hill: New York.
validation studies with the BSI will soon be Glueck, B. C. & Stroebel, C. F. (1969). The computer and the clinical
completed, and a number of other investigators decision process. American Journal of Psychiatry 125(7), (Suppl),
have also begun to use the brief form of the scale. 2-7.
Gorsuch, R. L. (1966). A comparison of Biquartimin, Maxplane,
Several have published reports showing high Promax and Varimax. Educational Research 36, 566-587.
sensitivity for the BSI. Marshal & Bougsty (1981) Harman, H. H. (1967). Modern Factor Analysis. University of
Chicago Press: Chicago.
have published an interesting epidemiological Kellner, R. (1971). Part 1. Improvement criteria in drug trials with
report on psychological disorders in one com- neurotic patients. Psychological Medicine 1, 416-425.
munity, and Amenson & Lewinsohn (1981) have Kremer, E. & Atkinson, H. J. (1981). Pain measurement: construct
validity of the affective dimension of the McGill Pain Questionnaire
completed a detailed investigation of sex differen- with chronic benign pain patients. Pain 11, 93-100.
ces in unipolar depression with the BSI. Kremer Lanyon, R. I. (1972). Technological approach to the improvement of
& Atkinson (1981) have shown high convergent decision making in mental health services. Journal of Consulting and
validity for the BSI with a number of other scales Clinical Psychology 39, 4 3 ^ 8 .
Lanyon, R. I. (1974). Technology of personality assessment: The
in predicting affective status among chronic pain psychological screening inventory. In Progress in Experimental
patients, and Peterson and his colleagues (1981) Personality Research, Vol. 7 (ed. B. A. Maher). Academic Press:
New York.
have reported the BSI to have substantial Lyerly, S. B. & Abbott, P. S. (1964). Handbook of Psychiatry Rating
predictive value in a counselling centre popula- Scales (1950-1964). NIMH: Bethesda, Maryland.
tion. The utilization of the BSI in a broad range Marshal, P. S. & Bougsty, T. (1981). Social/psychological problems
in an energy-impacted community. In Proceedings of the First
of substantive contexts indicates that a number International Symposium on the Human Side of Energy (ed.
of additional criterion-oriented validity studies J. A. Davenport and J. Davenport), pp. 55-66. University of
will soon be forthcoming. Wyoming: Laramie, Wyoming.
Messick, S. (1975). The standard problem: Meaning and values in
measurement and evaluation. American Psychologist 30, 955-966.
Messick, S. (1981). Constructs and their vicissitudes in educational
and psychological measurement. Psychological Bulletin 89(3),
REFERENCES 575-588.
Amenson, C. S. & Lewinsohn, P. M. (1981). An investigation into the Nunnally, J. C. (1978). An overview of psychological measurement.
observed sex difference in prevalence of unipolar depression. In Clinical Diagnosis of Mental Disorders: A Handbook (ed.
Journal of Abnormal Psychology 90(1), 1-13. B. B. Wolman), pp. 97-146. Plenum Press: New York.
Block, J. (1965). The Challenge of Response Sets. Appleton-Century- Peterson, R. L., Hodge, E. A., Kafer, L. G., Matthews, S. R., Pfeifle,
Crofts: New York. H. H. & Van Heuse, C. L. (1981). A group counselling emphasis
Brief Symptom Inventory 605

at a university counselling service. Psychotherapy Theory and Wiggins, J. S. (1966). Substantive dimensions of self-report in the
Practice 18, 525-536. MMPI item pool. Psychological Monographs 80 (22, whole no.630).
Rorer, L. G. (1965). The great response-style myth. Psychological Wilde, G. J. S. (1972). Trait description and measurement by
Bulletin 63, 129-156. personality questionnaires. In Handbook of Modern Personality
Thorndike, E. L. & Lorge, I. (1944). The Teacher's Work Book of Theory (ed. R. B. Cattell), pp. 69-103. Aldine: Chicago.
30000 Words. Bureau of Publications: New York. Woodworth, R. S. (1918). Personal Data Sheet. Stoelting: Chicago.
Tryon, R. C. (1966). Unrestricted cluster and factor analysis with
application to the MMPI, and Holzinger-Harman problems.
Multivariale Behavioral Research 1, 229-244.

You might also like