Psychological Measurements in Psychopharmacology. Mod. Probl.

Pharmacopsychiat., vol. 7, pp. 151-169, ed. P. Pichot, Paris (Karger, Basel 1974)

Assessment of Depression:
The Depression Inventory
AARON T. BECK and ALICE BEAMESDERFER

Introduction

The study of depression poses a particularly difficult problem because its

diagnosis is frequently missed. BARRACLOUGH et al. [1968] found that al-
though almost all suicides in a small city had seen their family physicians
within a few weeks prior to their deaths, the physicians had uniformly failed
to recognize depression, despite ample evidence of its presence. NIELSEN et al.
[1972] compared the routine records of 129 medical outpatients with inde-
pendent ratings of depression on those patients. Depression was mentioned
in only 3.9 % of the charts, while 12 % scored in the depressed range on the
depression scale.
Because depression may often masquerade as another illness, it is under-
standable why the diagnosis is likely to be missed. Anxiety, on the other
hand, is relatively simple to detect. The anxious patient appears apprehen-
sive, complains of nervousness, palpitations, and faintness, sweats profusely,
and manifests rapid breathing, tachycardia, and a labile blood pressure. In
contrast, the identification of depression often requires time-consuming in-
terviews.
Epidemiological studies reflect the fact that depression has a greater
prevalence and incidence than is generally expected. KLERMAN estimates that
one American out of every eight will suffer a depression sometime during his
life [quoted in BRAND, 1972]. LEHMANN [1971] estimates the prevalence of de-
pression at 3-4 % of the general popUlation. Of these depressed patients, he
suggests that only 1 in 5 is treated, 1 in 50 is hospitalized, and 1 in 200 com-
mits suicide.
137.132.123.69 - 4/10/2017 12:23:27 PM

As shown by NIELSEN et al. [1972] standardized schedules are valuable

tools in identifiying and assessing cases of undiagnosed depression. Such
National Univ. of Singapore
Downloaded by:
 BECK/BEAMESDERFER 152

schedules help to chart the fluctuations of their depressions and to determine

the incidence and prevalence of the illness. The purpose of this present chap-
ter is to describe the development and utility of an inventory designed specifi-
cally to measure the depth of depression in patients.

Reliability of Psychiatric Diagnoses

Before the correlates of depression can be determined, it is necessary to

establish a valid and reliable measure of the illness. Some psychiatric re-
searchers have regarded the traditional diagnostic system with skepticism,
and many question the utility of subgroups (manic-depressive, involutional
depression, etc.). Reports of low inter-rater reliability of diagnoses have ad-
ded to the confusion.
BECK [1962] reviewed four studies concerned with psychiatric diagnosis
and found that agreement on specific diagnoses ranged from 32 to 42 %.
However, the studies in question were poorly designed, and in many cases the
diagnoses were rendered by psychiatric residents who were relatively inexpe-
rienced and unreliable clinicians.
With these methodological defects in mind, we set out to conduct our
own reliability study with a more stringent experimental design. Using expe-
rienced psychiatrists and controlling for extraneous factors which might artifi-
cially raise or lower the rate of agreement, we obtained a concordance rate of
54 %, which is substantially higher than that of the more loosely designed
studies. Thus, the proportion of agreement can be improved by the refine-
ment of experimental design [BECK et a/., 1962].
In the same study, we also compared the degree of agreement when only
the major diagnostic categories (psychosis, neurosis, and personality disor-
der) were used to classify patients. In this analysis, we found an agreement
rate of 70 % between the two clinicians who classified each patient. When
both diagnosticians indicated certainty regarding their diagnoses, the agree-
ment rate was even higher (81 %). In addition, when the diagnosticians pre-
sented an alternative diagnosis with their preferred diagnosis, there was at
least one matching pair of diagnoses in 82 % of the cases.
Our next step was an attempt to determine the reasons for diagnostic
disagreement. We conducted a systematic study in which each patient was
seen by at least two psychiatrists [WARD et a/., 1962]. When there was a disa-
137.132.123.69 - 4/10/2017 12:23:27 PM

greement in diagnosis, the psychiatrists conferred and attempted to establish

the reasons for the disagreement. In 97 such cases, the disagreement was at-
National Univ. of Singapore
Downloaded by:
 The Depression Inventory 153

tributed to one ofthese causes: fluctuations in the clinical state of the patient,
5 %; inconsistencies by one of the psychiatrists, 37 %; inadequacies in the
nosological system, 58 %.

Clinical versus Psychometric Ratings of Depression

In studies of depression, the question arises as to the relative merits of

clinical ratings of depression as compared with a more 'objective' instrument
such as the Depression Inventory (DI).
Clinical judgments have several advantages. First, they are based on
careful appraisals by specialists trained to identify clinical phenomena. Sec-
ond, the clinical rating can measure nonverbal, as well as verbal behavior.
Third, the diagnostician can deal with distorting factors such as exaggeration
or denial and can make a global rating which will reflect the patient's overt
behavior as well as his subjective symptoms.
Despite these advantages, clinical ratings present certain problems. As
mentioned previously, diagnosticians are often inconsistent, and the prob-
lems inherent in the nosological system are difficult to overcome. In addition,
even if quantitative assessments are satisfactorily established in one study,
other research groups have difficulty replicating the experiment because of
rater inconsistencies. Diagnostic differences between various institutions
have been proven in many studies.
An instrument such as the DI provides a standardized consistent mea-
sure that does not rely on the theoretical orientation of the interviewer. It is
far more economical than the psychiatric interview, since it can be adminis-
tered by minimally trained personnel or can be self-administered. In addi-
tion, it yields a numerical score which can easily be compared to other data.
Finally, an instrument such as the DI can objectively measure improvement
resulting from treatment, such as antidepressant drug therapy.
Prior to the development of the DI, the self-rating instruments were not
suitable for the study of depression in patients. In 1930, JASPER developed the
Depression-Elation test using a sample of normal college students, but he did
not refer to psychiatric patients. The Depression Scale (D-Scale) of the Min-
nesota Multiphasic Personality Inventory (MMPI) [HATHAWAY and McKIN-
LEY, 1942] also had various disadvantages. First, factor analytic studies
[COMREY, 1957] showed that the D-Scale contains a number of heterogeneous
137.132.123.69 - 4/10/2017 12:23:27 PM

factors, only one of which is consistent with the clinical definition of depres-
sion. Another study [O'CONNOR et al., 1957] isolated five clusters. Again,
National Univ. of Singapore
Downloaded by:
 BECKjBEAMESDERFER 154

since only one of these clusters was related to the clinical concept of depres-
sion, the authors questioned the attribution of unitary significance to the
D-Scale. Other studies suggest that the MMPI is sensitive to response sets
such as the social desirability response set and the acquiescence response set
[MESSICK, 1960].
HAMILTON'S [1960] rating scale for depression required administration
by experienced diagnosticians. More recently, adjective checklists have been
developed to measure depression and other affects [CLYDE, 1961; ZUCKER-
MAN and LUBIN, 1965]. Nevertheless, subjective feelings are only one aspect
of depressive illness, and failure to account for other dimensions in depres-
sion impairs the usefulness of these tests. Depression is much more than an
unpleasant feeling state, but is a complex disorder involving cognitive, moti-
vational, behavioral, and affective components [BECK, 1972].

Development of the DI

In the course of our research on the psychological correlates of depres-

sion, we became aware of the necessity of developing an inventory for mea-
suring the depth of depression [BECK et al., 1961]. The objective in applying
the scale to a given population of psychiatric patients was to identify as many
of the depressed patients as possible and to exclude as many of the nonde-
pressed patients as possible.
The DI assumed its present form because of the following observations:
(1) the number of symptoms increases with severity of depression, and the
frequency of depressive symptoms progresses in a step-like manner from
nondepressed, to mildly depressed, to moderately depressed, to severely de-
pressed patients; (2) the more depressed an individual is, the more intense a
particular symptom is likely to be [BECK, 1972].
For these reasons, the inventory was designed to include all symptoms in-
tegral to the depressive constellation, and at the same time to provide for grad-
ing the intensity of each. The scoring takes into account the number of symp-
toms reported by the patient, in addition to the intensity of each symptom.

Description of the Inventory

137.132.123.69 - 4/10/2017 12:23:27 PM

The items in the DI were primarily clinically derived. Systematic obser-

vations and records were made regarding the characteristic attitudes and
National Univ. of Singapore
Downloaded by:
 The Depression Inventory 155

symptoms of depressed patients. Those attitudes and symptoms which ap-

peared to be specific for depression and those which were consistent with de-
scriptions in the psychiatric literature were selected. On the basis of this selec-
tion process, we constructed an inventory of 21 categories of symptoms and
attitudes. Each category describes a particular manifestation of depression
and consists of a series of four self-evaluative statements which are assigned
values from 0 to 3 to indicate the degree of severity. The items and the alter-
natives within each item were chosen on the basis of their relationship to the
overt manifestations of depression and were not intended to reflect any theo-
ry regarding the etiology of depression. 1

The symptom-attitude categories were:

(1) mood; (2) pessimism; (3) sense of failure; (4) lack of satisfaction; (5)
guilty feeling; (6) sense of punishment; (7) self-dislike; (8) self accusations;
(9) suicidal wishes; (10) crying spells; (11) irritability; (12) social withdrawal;
(13) indecisiveness; (14) distortion of body image; (15) work inhibition; (16)
sleep disturbance; (17) fatigability; (18) loss of appetite; (19) weight loss;
(20) somatic preoccupation; and (21) loss of libido.

The inventory was administered by a trained interviewer who read aloud

each statement in the category and asked the patient to select the statement
that seemed to describe him best at present. In order that the inventory reflect
the current status of the patient, the items were presented so as to elicit the
patient's attitude at the time of the interview. The patient also had a copy of
the scale so that he could read each statement to himself as the interviewer
read it aloud. On the basis of the patient's response, the interviewer circled
the number next to the appropriate statement. The total score was obtained
by adding the scores of the individual symptom categories.
The patients for the original study were drawn from routine admissions
to the psychiatric outpatient department and psychiatric inpatient service of
a large metropolitan hospital and to the psychiatric outpatient department of
a university hospital. There was a predominance of white patients over Ne-
gro patients, an age concentration between 15 and 44, and a high frequency
of patients from the lower socioeconomic groups (IV and V on Hollings-
head's Two-Factor Index of Social Position).
Patients with organic brain damage or mental deficiency were automati-
137.132.123.69 - 4/10/2017 12:23:27 PM

1 An abridged version of the 01 and instructions for administering it are presented in

the Appendix.
National Univ. of Singapore
Downloaded by:
 BECK/BEAMESDERFER 156

cally excluded from the study. The percentages among major diagnostic cate-
gories were: psychotic disorder, 41 %; psychoneurotic disorder, 43 %; per-
sonality disorder, 16 %. The three largest subgroups were: schizophrenic re-
action, 28.2 %; psychoneurotic depressive reaction, 25.3 %; anxiety reaction,
15.5 %.
The psychiatrists in our study had several preliminary meetings during
which they reached an agreement regarding the criteria for each of the noso-
logical categories and focused special attention on the various types of de-
pression. The American Psychiatric Association's Diagnostic and statistical
manual of mental disorders [1952] was used, but considerable amplification of
the diagnostic descriptions was necessary.
The psychiatrists also established specific indices for use in making a
clinical evaluation of the depth of depression. For each specified sign and
symptom the psychiatrists made a rating on a four-point scale of none,
mild, moderate, and severe. These indices were used to increase uniform-
ity among the psychiatrists. However, in rating the depth of depression,
they made a global judgment and were not confined by the ratings in each
index.
The psychiatrists also rated the patient on the degree of overt anxiety
and agitation, and filled out a checklist to indicate the presence of other spe-
cific psychosomatic and psychiatric symptoms and disturbances in concen-
tration, recall, memory, reality-testing, and judgment. They also rated the
severity of the present illness on a four-point scale.
Each patient was seen by two psychiatrists who made independent judg-
ments of the depth of depression and the diagnosis. After the second inter-
view, the psychiatrists met and discussed the case to ascertain the reasons for
any disagreement. The DI was administered independently by a trained tech-
nician.

Reliability of the DI

Two methods of evaluating the internal consistency of the inventory

were used. First, the protocols of 200 consecutive cases were analyzed, and
the scores for each of the 21 items were compared with the total score on the
DI for each patient. Using the Kruskal-Wallis Non-Parametric Analysis of
137.132.123.69 - 4/10/2017 12:23:27 PM

Variance by Ranks, we found that all categories showed a significant rela-

tionship to the total score for the inventory. A subsequent item analysis of
National Univ. of Singapore
Downloaded by:
 The Depression Inventory 157

606 cases showed that each item had a significant positive correlation with
the total DI score [BECK, 1972].
The second method of evaluating internal consistency was the determi-
nation of the split-half reliability. 97 cases were selected for this analysis. The
Pearson Or' between the odd and even categories was computed and yielded a
reliability coefficient of 0.86; a Spearman-Brown correction for attenuation
raised the coefficient to 0.93.
Traditional methods of assessing the stability and consistency of inven-
tories were not appropriate for the evaluation of the DI. Test-retest did not
seem proper because of the possible influence of memory on the scores. If a
long interval were provided, the score would be influenced by fluctuations in
the intensity of depression. The inter-rater reliability method was not used
for the same reasons, i.e. two successive technicians would have to administer
the test.
Because of these considerations, we used two indirect methods of eval-
uating the stability of the inventory. The first was a variation of the test-retest
method. The inventory was administered to a group of 38 patients by a tech-
nician at two different times, with a mean interval of 4 weeks between the two
tests. Each time, a clinical rating of the depth of depression was made by a
psychiatrist. We found that changes in the DI scores paralleled changes in the
clinical ratings of the depth of depression.
An indirect measure of the inter-rater reliability was achieved by com-
paring the scores obtained by each of the three participating technicians with
the clinical ratings. The mean scores, respectively, obtained at each level of
depression were virtually identical among the interviewers. When the DI
scores were plotted against the depth of depression, the curves were notably
similar, indicating a high level of agreement among those who administered
the inventory.

Validity of the DI

In trying to determine how well a given test measures a specific personal-

ity variable, the researcher is often hampered by the absence of any estab-
lished criterion that defines the variable in question. The American Psycho-
logical Association's manual, Technical recommendations for psychological
137.132.123.69 - 4/10/2017 12:23:27 PM

tests and diagnostic techniques [1954] recommends the use of concurrent va-
lidity and construct validity criteria in evaluating personality tests.
National Univ. of Singapore
Downloaded by:
 BECKjBEAMESDERFER 158

Concurrent Validity

We evaluated the concurrent validity by determining how well the test

scores correlated with other measures of depression, for example, clinical
evaluation and scores on other psychometric tests of depression.
The concurrent validity of the DI was supported by a number of studies
comparing the test scores with clinicians' global ratings of depth of depres-
sion. In our own validation study [BECK et ai., 1961], the inventory was
found to correlate 0.65 with the clinicians' ratings; in a drug study by Nuss-
BAUM et al. [1963] the correlation was 0.66; in a British study by METCALFE
and GOLDMAN [1965] the correlation was 0.616.
MAY et ai. [1969] compared a modified version of the DI to ratings of
doctors and nurses. These authors obtained a correlation of 0.65 between the
clinicians' ratings and the modified inventory. In a study of the DI in general
practice, SALKIND [1969] found the correlation coefficient between the DI
and depth of depression ratings to be 0.73. Similar correlations were ob-
tained in comparisons between the DI scores and clinicians' ratings in Cze-
choslovakia [VINAR and GROF, 1969], Finland [STENBACK et ai., 1967],
France [DELAY et ai., 1963], and Switzerland [BLASER et ai., 1968].
Concurrent validity has also been demonstrated through comparisons
with other standardized measures of depression. NUSSBAUM et al. [1963]
found initial and final correlations between the MMPI D-Scale and the DI to
be 0.75 and 0.69, respectively. In a comparison of the DI and Hamilton's
Rating Scale for depression [HAMILTON, 1960], SCHWAB et ai. [1967b]
obtained a correlation coefficient of 0.75 between the two scales. NIELSEN
et al. [1972] also compared the DI to Hamilton's scale. The authors assumed
that their correlation of 0.54 was deflated due to the fact that low DI and
Hamilton scores were excluded from their analysis.
SPITZER et al. [1967] computed the correlations between the Mental Sta-
tus Schedule (MSS) and a number of other measures, including the DI. The
'feelings-concerns' scale ofthe MSS correlated 0.58 with the DI; the 'depres-
sion-anxiety' scale correlated 0.55 with the DI. BLOOM and BRADY [1968]
compared the depression scale of the Multiple Affect Adjective Check List
(MAACL) to the DI. These authors found a correlation of 0.66 between
the DI and the MAACL.

In a cross-cultural study of depressive symptoms, ZUNG [1969] found a

137.132.123.69 - 4/10/2017 12:23:27 PM

correlation of 0.76 between the DI and his Self-Rating Depression Scale

[ZUNG, 1965] in England. The correlation was 0.72 in Germany.
National Univ. of Singapore
Downloaded by:
 The Depression Inventory 159

WILLIAMS et al. [1972] compared behavioral measures of depression to

the DI and Hamilton's scale. These authors obtained the following results:
the correlation between the Beck and Hamilton scales was 0.82; the correla-
tion between the DI and behavioral measures was 0.67.
The correlation of the DI scores with clinicians' ratings of depression
and with other psychometric tests for depression constituted solid support
for the concurrent validity of the DI.

Construct Validity

As pointed out by CRONBACH and MEEHL [1955], the most relevant in-
formation dealing with personality variables is obtained from an assessment
of the construct validity of the test. In short, the construct validity of a mea-
sure is determined by setting up a number of hypotheses regarding the per-
sonality variable (depression in our case). If the hypothesis is confirmed in an
experiment which uses the test as a criterion measure, the validity of the in-
strument is supported.
The hypotheses that we tested in our own investigations of depression
were: (1) depressed patients are likely to have a certain kind of dream charac-
terized by 'masochistic' content; (2) they are likely to have a negative self-
concept; (3) they identify with the 'loser' on projective tests dealing with suc-
cess and failure; (4) they have a history of deprivation that sensitized them to
depression in later life; (5) they respond to experimentally induced failure
with a disproportionate drop in self-esteem and increase in hopelessness; (6)
following a success experience, depressed patients will show a significant
subjective and objective improvement, and (7) they show a high correlation
between intensity of depression and suicidal intent.
Using the DI as the criterion measure, we found that these predictions
were largely supported. BECK and WARD [1961] found a significant relation-
ship between depression and 'masochistic' dreams. BECK and STEIN [1960]
found that depressives score highly on a self-concept test, with high scores in-
dicating negative self-concept. BECK [1961] found that depressed patients
identify with the 'loser' when presented with a series of pictorial stimuli.
BECK et at. [1963] obtained a significant relationship between childhood be-
reavement and adult depression. LOEB et at. [1964] demonstrated that de-
pressed patients make excessively pessimistic predictions after inferior task
137.132.123.69 - 4/10/2017 12:23:27 PM

performance. We also found that following successful completion of a man-

ual task, depressed patients showed a significant improvement in optimism,
National Univ. of Singapore
Downloaded by:
 BECK/BEAMESDERFER 160

self-evaluation, and performance [LOEB et al., 19711. SILVER et al. [1971]

found a significant correlation between depression and suicidal intent after
nonfatal suicide attempts.
Other investigations have provided further evidence of the construct va-
lidity of the DI. GOTISCHALK et al. [1963] found a significant correlation
(0.47) between scores on a 'hostility-inward' scale and the DI, and a negative
correlation between a 'hostility-out' scale and the DI. NUSSBAUM and MI-
CHAUX [1963] found a significant negative relationship between scores on the
DI and scores on a sense of humor test.
The DI was also used as a criterion measure in a study by MENDELS and
HAWKINS [1968], which showed that depressed patients experience a sleep
disturbance that reverts to normal when the depressive episode is terminated.
BECK [1972] found that depressed patients are more pessimistic than are
nondepressed patients, but that they return to normal after recovery.
Since all of our hypotheses were confirmed, we had strong support for
the construct validity of the DI.

Influence of Extraneous Variables

We analyzed the records of 606 patients included in our research, and

computed correlations between the total DI scores and background variables.
For purposes of comparison, correlation coefficients between the variables
and depth of depression ratings were also computed.
Sex. We found a significant positive correlation between females and the
DI score (point biserial r = 0.189). This was not an artifact, since the clini-
cian's ratings of depth of depression also correlated positively with female
sex.
Race. The correlations between race and the DI score and depth of de-
pression ratings were negligible.
Age. Contrary to popular belief that older patients are more likely to be
depressed, our product moment correlations with age were negligible.
Educational level. Using educational level as an index of social class, we
found a significant negative correlation between educational level and DI
score (r = -0.163). Thus, patients with lower educational attainment tended
to score higher on the DI than did those with a greater level of education.
This relationship was not found in the comparison with depth of depression
137.132.123.69 - 4/10/2017 12:23:27 PM

ratings (r = -0.026). While the reason for this discrepancy is not immediate-
ly apparent, it is possibly the result of a response set in less-educated patients.
National Univ. of Singapore
Downloaded by:
 The Depression Inventory 161

Vocabulary score. We used the vocabulary score as an index of intelli-

gence and found only slight correlations with the DI score and depth of de-
pression ratings.

Response Sets

To obtain some idea of the effect of the social-desirability response set,

we used an inventory whose form was similar to the DI, but whose content
did not deal with symptoms. We found that depressed patients selected those
alternatives that reflected unfavorably upon themselves, since depressives
view themselves as undesirable. Thus, if a response set is operating, it is
probably a social undesirability set which may in itself be diagnostic of de-
pression.
Is the validity of the DI reduced by the presence of a response set? This
question may be answered by the following: (1) Correlations between clini-
cians' ratings and DI scores have been consistently high. In fact, the correla-
tions approach the upper limit imposed by the reliability of clinicians' ratings
(0.70-0.80). If the validity of the inventory were substantially reduced by re-
sponse sets, we would expect a much greater difference between DI scores
and clinicians' ratings. (2) The DI was found to be highly effective in discri-
minating between depression and anxiety. Again, if response sets were de-
creasing the effectiveness of the DI, we would expect more blurring between
these two states.
With these considerations in mind, we concluded that response sets do
not materially detract from the validity of the DI.

Factor Analytic Studies of the DI

Attempts have been made to classify depressive phenomena by the use of

factor analysis. These attempts are largely the result of the development of
instruments for measuring depression, such as the DI. After scores on specif-
ic test items are intercorrelated and a correlation matrix is prepared, factors
are extracted and rotated. In most studies, the isolated factors are tested for
independence (orthogonality) and ranked for magnitude of factor loadings.
137.132.123.69 - 4/10/2017 12:23:27 PM

Several factor analyses of the DI have been reported. DELAY et al. [1963]
administered the DI to 79 depressed patients in France. Since each individual
National Univ. of Singapore
Downloaded by:
 BECK/BEAMESDERFER 162

item of the DI correlated positively with the total DI score, the authors con-
cluded that there was evidence of a 'general factor' of depression. In a subse-
quent study, PICHOT and LEMPERIERE [1964] added 56 cases of depression to
their initial sample, making a total of 135 patients tested with the DI. Four
factors were then extracted from the data through factor analysis; factor A
consisted of the physiological signs of depression ('vital depression'); factor
B consisted of items relevant to the sense of self-derogation (,self-debase-
ment'); factor C contained items related to hopelessness and suicide ('pessi-
mism-suicide'); and factor D revolved around two motivational symptoms
('indecision-inhibition').
PICHOT et al. [1966] performed a factor analysis utilizing a modified ver-
sion of the DI (20 of the original 21 items, plus 13 additional items). In addi-
tion to a general factor of depression, these authors extracted 10 factors
which were submitted to Varimax rotation and grouped into categories. They
found 3 of the 10 factors (lethargy, intrapunitive, and affect) to be highly reli-
able measures of subjective symptomatology. Four other factors (somatic,
loss of libido, sleeping trouble, and anxiety) were 'interesting but weak'
measures. The remaining three factors were uninterpretable.
CROPLEY and WECKOWICZ [1966] and WECKOWICZ et al. [1967] also per-
formed a factor analysis using the DI. These authors reported three signifi-
cant factors: The first, called 'guilty depression', was heavily loaded on guilt
feelings, sense of punishment, self-accusation, sense of failure, self-punitive
wishes, self-hate, depressed mood, indecisiveness, and pessimism. The second
was identified as 'retarded depression', with high loadings on work inhibi-
tion, fatigue, lack of satisfaction, depressed mood, somatic preoccupation,
and indecisiveness. The third factor, 'somatic disturbance', was defined by
high loadings on weight loss, loss of appetite, and sleep disturbance. One oth-
er factor, called 'tearful depression', approached the authors' criterion of sig-
nificance. This factor loaded highly on body image, crying spells, and loss of
libido.
A more recent factor analytic study by WECKOWICZ et al. [1971] at-
tempted to relate the factors obtained from clinical evaluations, symptoms,
and complaints to physiological measures. Subjects were administered the
DI, psychomotor tests, the Shagass sedation threshold test, salivation tests,
and automatic nervous system activity tests. The following six significant fac-
tors were obtained: (1) 'somatic factor of retarded depression'; (2) 'atypical'
schizoid and involutional depression; (3) 'typical guilty depression'; (4) 'bod-
137.132.123.69 - 4/10/2017 12:23:27 PM

ily fatigue and neurasthenic exhaustion'; (5) 'somatization; and (6) 'hypo-
chondriasis'.
National Univ. of Singapore
Downloaded by:
 The Depression Inventory 163

In our own factor analysis using Varimax rotation, we obtained three

powerful factors which we labelled as follows: (1) 'Negative view of self and
future'; which was heavily loaded on sense of failure, expectation of punish-
ment, self-dislike, self-accusation, and suicidal ideas; (2) 'physiological',
which was heavily loaded on insomnia, anorexia, weight loss, and loss oflibi-
do; (3) 'Physical withdrawal', which was heavily loaded on work inhibition,
fatigability, and somatic preoccupation.

Cut-Off Scores

There is no arbitrary score that can be used for all purposes as a cut-off
point in the DI. The specific cut-off point depends upon the characteristics
of the patients in the sample and on the purposes for which the inventory is
being used.
The crux of the problem is: How many false-positives and false-nega-
tives occur at a particular cutting score? For identifying a relatively pure
group of depressed patients for research purposes, the investigator wants to
minimize false positives (i.e. high scores who are not really depressed). He
may not be concerned about false negatives, who would be excluded from his
study. A high cutting score, therefore, should be used (a score of more than
21 on the original DI).
As a screening device to detect depression among psychiatric patients, we
found a cut-off point of 13 is appropriate. This score gives fewer false-nega-
tives, but more false-positives than the higher cut-off point. For screening de-
pression among medical patients, SCHWAB et al. [1967a] found that a cutting
score of 10 was appropriate.

Abridged, self-administered DI

A new, short form of the DI has recently been developed to aid general
practitioners, as well as researchers, in the rapid screening of depressed pa-
tients [BECK and BECK, 1972]. Because depression may be masked by physi-
cal symptoms, the diagnosis is likely to be missed by a general practitioner,
and the depression often goes untreated. SALKIND'S [1969] data indicated
that48 % of office practice patients manifested depression ranging from mild
137.132.123.69 - 4/10/2017 12:23:27 PM

to severe. Patients on the medical services of general hospitals include an

even higher proportion of depressive [SCHWAB et aI., 1967a].
National Univ. of Singapore
Downloaded by:
 BECKjBEAMESDERFER 164

The DI has been recommended for all general practitioners in the British
Health Service by RAWNSLEY [1968]. In order to facilitate its use by family
physicians, a shorter, simplified version of the scale was developed to help
identify depressed patients. The new form requires approximately 5 min to
complete, and is suitable for self-administration by the patient.
We set the following criteria for the abridged version of the DI: (1) max-
imum correlation with clinicians' ratings of depth of depression, and (2) a 10-
to I5-item scale that would correlate better than 0.90 with the long form. In
our previous validation and reliability study, each of the 21 items was corre-
lated with the total DI score and with clinicians' depth of depression
ratings. The items were ranked for each of the two correlations, and the
two ranks were consolidated into a final rank.
Initially, the item that had the best correlation with the total DI score
was selected for the abridged form; then the sum of the best two items, then
the best three, and so on until the cumulative correlations levelled off. While
the correlation with the total DI score reached its criterion after seven items,
that with the clinicians' ratings did so after 13 items. Thus, we obtained a 13-
item questionnaire correlating 0.96 with the total DI score and 0.61 with the
clinicians' ratings of depression.
After rescoring the DIs of our original 599 patients on the basis of these
13 items, we computed the standard deviations and means for groups catego-
rized by clinicians' ratings of depth of depression and established cut-off
points for each category. While the clinician or investigator may have to
probe deeper if he wants a more complete estimate of the severity of depres-
sion, our cut-off points can alert him to the probable degree of severity with-
out SUbjecting the patient to a lengthy psychiatric interview and mental status
examination. The range of scores for the abridged DI are: 0--4, none or
minimal; 4-7, mild; 8-15, moderate; 16+, severe.

References

American Psychiatric Association: Diagnostic and statistical manual: mental disorders

(Amer. Psychiat. Ass., Wasinghton 1952).
American Psychological Association: Technical recommendations for psychological
tests and diagnostic techniques. Psychol. Bull. 51: suppl., pp. 13-28 (1954).
BARRACLOUGH, B. M.; NELSON, B., and SAINSBURY, P.: The diagnostic classification
and psychiatric treatment of 25 suicides. Proc. 4th Int. Conf. Suicide Prevention
Center (Delmar, Los Angeles 1968).
137.132.123.69 - 4/10/2017 12:23:27 PM

BECK, A. T.: A systematic investigation of depression. Comprehens. Psychiat. 2: 162-170

(1961).
National Univ. of Singapore
Downloaded by:
 The Depression Inventory 165

BECK, A. T.: Reliability of psychiatric diagnoses. I. A critique of systematic studies.

Amer. J. Psychiat. 119: 210-216 (1962).
BECK, A. T.: Depression: causes and treatment (Univ. Pennsylvania Press, Philadelphia
1972).
BECK, A. T. and BECK, R. W.: A rapid technique for screening depressed patients in
general practice. Postgrad. Med. (in Press, 1972).
BECK, A. T.: SETID, B., and TUTHILL, R.: Childhood bereavement and adult depression.
Arch. gen. Psychiat. 9: 295-302 (1963).
BECK, A. T. and STEIN, D.: The self concept in depression; unpublished study (1960).
BECK, A. T. and WARD, C. H.: Dreams of depressed patients. Characteristic themes in
manifest content. Arch. gen. Psychiat. 5: 462-467 (1961).
BECK, A. T.; WARD, C. H.; MENDELSON, M.; MOCK, J. E., and ERBAUGH, J. K.: An
inventory for measuring depression. Arch. gen. Psychiat. 4: 561-571 (1961).
BECK, A. T.; WARD, C. H.; MENDELSON, M.; MOCK, J. E., and ERBAUGH, J.: Reliability
of psychiatric diagnoses. II. A study of consistency of clinical judgments and
ratings. Amer. J. Psychiat. 119: 351-357 (1962).
BLASER, R.; Low, D., and SCHAUB LIN, A.: Die Messung der Depressionstiefe mit einem
Fragebogen. Psychiat. clin. 1: 299-319 (1968).
BLOOM, P. M. and BRADY, J. P.: An ipsative validation of the Multiple Affect Adjective
Check List. J. clin. Psychol. 24: 45-46 (1968).
BRAND, D.: Beyond the blues. Wall Street J., p. 1 (April 7, 1972).
CLYDE, D. J.: Clyde Mood Scale (George Washington University, Washington 1961).
COMREY, A. L.: A factor analysis of items on the MMPI depression scale. Educ. pscyhol.
Measur. 17: 578-585 (1957).
CRONBACH, L. J. and MEEHL, P. E.: Construct validity in psychological tests. Psycho!.
Bull. 52: 281-302 (1955).
CROPLEY, A. J. and WECKOWICZ, T. E.: The dimensionality of clinical depression.
Austr. J. Psychol. 18: 18-25 (1966).
DELAY, J.; PICHOT, P.; LEMPERIERE, T. et MIROUZE, R.: La nosologie des etats depres-
sifs. Rapports entre l'etiologie et la semiologie. II. Resultats du questionnaire de
Beck. Encephale 52: 497-505 (1963).
GOTTSCHALK, L.; GLESER, G., and SPRINGER, K.: Three hostility scales applicable to
verbal samples. Arch. gen. Psychiat. 9: 254-269 (1963).
HAMILTON, M.: A rating scale for depression. J. Neurol. Neurosurg. Psychiat. 23: 56-61
(1960).
HATHAWAY, W.R. and McKINLEY, J.C.: A multiphasic personality schedule. III. The
measurement of symptomatic depression. J. Psychol. 10: 249-254 (1942).
JASPER, H. H.: A measurement of depression-elation and its relation to a measure of
extraversion-intraversion. J. abnorm. soc. Psychol. 25: 307-318 (1930).
LEHMANN, H. E.: Epidemiology of depressive disorders; in FIEVE Depression in the 70's
(Excerpta Medica, Amsterdam 1971).
LOEB, A.; BECK, A. T., and DIGGORY, J. C.: Differential effects of success and failure
on depressed and nondepressed patients. J. nerv. ment. Dis. 152: 106-114 (1971).
LOEB, A.; FESHBACH, S.; BECK, A. T., and WOLF, A.: Some effects of reward upon the
137.132.123.69 - 4/10/2017 12:23:27 PM

social perception and motivation of psychiatric patients varying in depression.

J. abnorm. soc. Psychol. 68: 609-616 (1964).
National Univ. of Singapore
Downloaded by:
 BECKjBEAMESDERFER 166

MAY, A.E.; URQUHART, A, and TARRAN, J.: Self-evaluation in various diagnostic

and therapeutic groups. Arch. gen. Psychiat. 21: 191-194 (1969).
MENDELS, J. and HAWKINS, D. R.: Sleep and depression. Arch. gen. Psychiat. 19: 445 to
452 (1968).
MESSICK, M.: Response style and content measures from personality inventories. Educ.
psychol. Measur. 22: 41-56 (1960).
METCALFE, M. and GOLDMAN, E.: Validation of an inventory for measuring depression.
Brit. J. Psychiat. 111: 240-242 (1965).
NIELSEN, A.; SECUNDA, S.; FRIEDMAN, R., and WILLIAMS, T.: Prevalence and recognition
of depression among ambulatory patients in a group medical practice. Proc. Meet.
Amer. Psychiat. Ass., Dallas 1972.
NUSSBAUM, K. and MICHAUX, W. W.: Response to humor in depression. A prediction
and evaluation of patient change? Psychiat. Quart. 37: 527-539 (1963).
NUSSBAUM, K.; WITTIG, B. A.; HANLON, T. E., and KURLAND, A. A.: Intravenous niala-
mide in the treatment of depressed female patients. Comprehens. Psychiat. 4:
105-116 (1963).
O'CONNOR, J.; STEFIC, E., and GRESOCK, c.: Some patterns of depression. J. clin.
Psychol. 13: 122-125 (1957).
PICHOT, P. et LEMPERIERE, T.: Analyse factorielle d'un questionnaire d'autoevaluation
des symptomes depressifs. Rev. Psychol. appl. 14: 15-29 (1964).
PICHOT, P.; PIRET, J. et CLYDE, D. J.: Aanlyse de la symptomatologie depressive sub-
jective. Rev. Psychol. appl. 16: 103-115 (1966).
RAWNSLEY, K.: The early diagnosis of depression. Office of Health Economics. Early
Diagnosis, Paper 4 (1968).
SALKIND, M. R.: Beck depression inventory in general practice. J. roy. ColI. Gen.
Practic. 18: 267 (1969).
SCHWAB, J. J.; BIALOW, M.; BROWN, J.M., and HOLZER, C.E.: Diagnosing depression
in medical inpatients. Ann. intern. Med. 67: 695-707 (1967a).
SCHWAB, J. J.; BIALOW, M., and HOLZER, C.E.: A comparison of two rating scales for
depression. J. clin. Psychol. 23: 94-96 (1967b).
SILVER, M.; BOHNERT, M.; BECK, A. T., and MARCUS, D.: Relation of depression to
attempted suicide and seriousness of intent. Arch. gen. Psychiat. 24: 495-500 (1971).
SPITZER, R. L.; FLEISS, J. L.; ENDICOTT, J., and COHEN, J.: Mental status schedule.
Properties of factor analytically derived scales. Arch. gen. Psychiat. 16: 479-493
(1967).
STENBACK, A; RIMON, R., and TURUNEN, M.: Validitet av Taylor manifest anxiety
scale. Nord. Psykiat. T. 21: 79-85 (1967).
VINAR, O. and GROF, P.: Die depressive Symptomatologie im Lichte des Beckschen
Fragebogens; in HIPPIUS Das depressive Syndrom (Publisher, Berlin 1968).
WARD, C.H.; BECK, AT.; MENDELSON, M.; MOCK, J.E., and ERBAUGH, J.K.: The
psychiatric nomenclature. Reasons for diagnostic disagreement. Arch. gen. Psy-
chiat. 7: 198-205 (1962).
WECKOWICZ, T. E.: MUIR, W., and CROPLEY, A J.: A factor analysis of the Beck
inventory of depression. J. cons. Psychol. 31: 270-278 (1967).
WECKOWICZ, T.E.; YONGE, K.A.; CROPLEY, A J., and MUIR, W.: Objective therapy
137.132.123.69 - 4/10/2017 12:23:27 PM

predictors in depression: a multivariate approach. Clinical Psychology Publishing

Co., Monogr., suppl. 31.
National Univ. of Singapore
Downloaded by:
 The Depression Inventory 167

WILLIAMS, J. G.; BARLOW, D. H., and AGRAS, W. S.: Behavioral measurement of severe
depression (in press, 1972).
ZUCKERMAN, M. and LUBIN, B.: Manual for the multiple affect adjective check list
(Educational and Industrial Testing Service, San Diego 1965).
ZUNG, W. W. K.: A self-rating depression scale. Arch. gen. Psychiat. 12: 63-70 (1965).
ZUNG, W. W. K.: A cross-cultural study of symptoms in depression. Amer. J. Psychiat.
126: 154-159 (1969).

Author's address: AARON, T.BECK, MD, 429, Stauffer Building, Philadelphia

General Hospital, Philadelphia, PA 19104 (USA)

Appendix

Short Form of the Beck Depression Inventory

Instructions

This is a questionnaire. On the questionnaire are groups of statements. Please read

the entire group of statements in each category. Then pick out the one statement in that
group which best describes the way you feel today, that is, right now! Circle the number
beside the statement you have chosen. If several statements in the group seem to apply
equally well, circle each one.

Be sure to read all the statements in each group before making your choice.

A. (Sadness)
o I do not feel sad
1 I feel sad or blue
2 I am blue or sad all the time and I can't snap out of it
3 I am so sad or unhappy that I can't stand it
B. (Pessimism)
o I am not particularly pessimistic or discouraged about the future
1 I feel discouraged about the future
137.132.123.69 - 4/10/2017 12:23:27 PM

2 I feel I have nothing to look forward to

3 I feel that the future is hopeless and that things cannot improve
National Univ. of Singapore
Downloaded by:
 BECK/BEAMESDERFER 168

C. (Sense of Failure)
o I do not feel like a failure
1 I feel I have failed more than the average person
2 As I look back on my life, all I can see is a lot of failures
3 I feel I am a complete failure as a person (parent, husband, wife)
D. (Dissatisfaction)
o I am not particularly dissatisfied
1 I don't enjoy things the way I used to
2 I don't get satisfaction out of anything anymore
3 I am dissatisfied with everything
E. (Guilt)
o I don't feel particularly guilty
1 I feel bad or unworthy a good part of the time
2 I feel quite guilty
3 I feel as though I am very bad or worthless
P. (Self-Dislike)
o I don't feel disappointed in myself
1 I am disappointed in myself
2 I am disgusted with myself
3 I hate myself
G. (Self-Harm)
o I don't have any thoughts of harming myself
1 I feel I would be better off dead
2 I have definite plans about committing suicide
3 I would kill myself if I had the chance
H. (Social Withdrawal)
o I have not lost interest in other people
1 I am less interested in other people than I used to be
2 I have lost most of my interest in other people and have little feeling for them
1
3 I have lost all of my interest
1 in other people and don't care about them at all
1
I. (Indecisiveness) 1
o I make decisions about as1well 1 as ever
1 I try to put off making decisions
2 I have great difficulty in making decisioUil
3 I can't make any decisions at all any more
J. (Self-Image Change)
o I don't feel I look any worse than I used to
1 I am worried that I am looking old or unattractive
2 I feel that there are permanent changes in my appearance and they make me
look unattractive
3 I feel that I am ugly or repulsive looking
K. (Work Difficulty)
o I can work about as well as before
1 It takes extra effort to get started at doing something
137.132.123.69 - 4/10/2017 12:23:27 PM

2 I have to push myself very hard to do anything

3 I can't do any work at all
National Univ. of Singapore
Downloaded by:
 The Depression Inventory 169

L. (Fatigability)
o I don't get any more tired than usual
1 I get tired more easily than I used to
2 I get tired from doing anything
3 I get too tired to do anything
M. (Anorexia)
o My appetite is no worse than usual
1 My appetite is not as good as it used to be
2 My appetite is much worse now
3 I have no appetite at all any more

137.132.123.69 - 4/10/2017 12:23:27 PM

National Univ. of Singapore
Downloaded by: