0022-3018/93/1814-0246$03.

00/0
Tm-:Jo1IRNm.oI-‘ NERVOUS AND MENTAL DISEASE Vol. 181, No.4
Copyright © 1993 by Williams & Wilkins Printed in USA.

The Defense Style Questionnaire

GAVIN ANDREWS, M.D.,‘ MICHELLE SINGH, B.SC. (HONS),‘ AND MICHAEL BOND, M.D.2

The Defense Style Questionnaire has proven of interest as the first questionnaire to reliably
describe defense styles. The 72-item DSM-III-R-labeled Defense Style Questionnaire was ad-
ministered to 388 controls and 324 patients. Eight statistical and two a priori criteria were
used in choosing two items to represent each of the 20 defenses. A new 40—item Defense
Style Questionnaire is published together with normative and reliability data on a normal
population, patients with anxiety disorders, and child-abusing parents. The scores are unaf-
fected by the sex of the respondent, but the endorsement of immature defense styles de-
creases with age.
-——JNe7'v Ment Dis 181:246—256, 1993

Anna Freud (1966, p. 5) held that ego defense mecha-
nisms are “the Ways and means by which the ego wards
off unpleasure and anxiety, and exercises control over
impulsive behavior, affects and instinctive urges.” The
DSM-III-R (American Psychiatric Association, 1987, p.
393) offers a more user-friendly definition of defense
mechanisms as “patterns of feelings, thoughts, or be-
haviors that are relatively involuntary and arise in re-
sponse to perceptions of psychic danger. They are de-
signed to hide or to alleviate the conflicts or stressors
that give rise to anxiety.” Although the concept of de-
fense mechanisms is a useful heuristic in both clinical
and research contexts, evidence has been dependent
on clinical anecdote, which is not surprising given the
lack of agreed operational definitions. Vai1lant’s (1971,
1976) work was a significant exception. Utilizing a hier-
archy of defenses derived from psychodynamic theory,
he was able to measure defenses from vignettes with
considerable reliability, and reported that mature de-
fenses were positively correlated and immature de-
fenses negatively correlated with an objective measure
of life success.
Bond et al. (1983, p. 334) developed a self-report
measure to assess possible conscious derivatives of
defense mechanisms, the intention being “to elicit man-
ifestations of a subject’s characteristic style of dealing
with conflict, either conscious or unconscious, based
on the assumption that persons can accurately com-
ment on their behavior from a distance.” Scores from
that 67-item Defense Style Questionnaire (DSQ) were
found to discriminate between psychiatric patients and
normal subjects, with the patients endorsing more im-
mature defense styles. While the conceptualization of
‘School of Psychiatry, University of New South Wales at St Vin-
cent's Hospital, 299 Forbes Street, Darlinghurst, Sydney, NSW 2010,
Australia. Send reprint requests to Dr. Andrews.
2Department of Psychiatry, Institute of Community and Family
Psychiatry, Sir Mortimer B Davis-Jewish General Hospital, Montreal,
Quebec, Canada.
246
defense mechanisms as unconscious processes would
seem to preclude any possibility of self-report, two
points can be made in support of this methodology.
First, we often are aware of the results of the operation
of unconscious processes, asking ourselves in hindsight
why we acted in a certain way and inferring from our
behavior that some unconscious process was responsi-
ble. Second, we believe that the habitual use of any
particular defense will leave tracks in an individual’s
belief or attitude system and that endorsement of cer-
tain attitudes or beliefs can be taken as an indicator
of the habitual use of that defense. More substantive
support for the self-report methodology derives from
subsequent findings in which scores on the DSQ (Bond
et al., 1983) were found to be correlated with clinical
assessments, providing some evidence of the construct
validity of the instrument (Vaillant et al., 1986).
The conceptual relationship between defense style
and diagnosis is unclear. Whether, as Freud (1959) pos-
tulated, there is an intimate connection between special
forms of defense and particular illnesses is a question
that has yet to be answered categorically. Bond and
Sagala Vaillant (1986) thought that if this were the case
then information pertaining to an individual’s preferred
defense style would bring little additional information
to the therapy situation. If the two were independent,
however, then explicit categorization of defense style
may aid the planning and execution of any therapeutic
intervention (Strayhorn, 1983). Bond and Sagala Vail-
lant (1986) investigated the relationship between de-
fense style, as measured by the revised 88-item DSQ
(Bond, 1986b), and four diagnostic groups: psychoses,
affective disorders, anxiety disorders, and other disor-
ders. Patients with affective disorders had defense
styles comparable to normal subjects, whereas the
other three groups endorsed fewer mature and more
immature defenses, but in a manner that rendered them
indistinguishable from each other. Furthermore, Bond
 THE DEFENSE STYLE QUESTIONNAIRE
(1990) found that the 88-item DSQ did not discriminate
one diagnostic group from another. However, a recent
study (Bond etal., 1993) comparing the defense styles
of 78 female borderline patients and 72 female patients
with other types of personality disorders indicated that
the borderline subjects reported using the maladaptive
and image-distorting defense styles significantly more
often and the adaptive defense styles significantly less
often than the other subjects. A larger sample and more
rigorous diagnostic criteria yielded correlations that
were obscured in the previous work.
Andrews et al. (1989) modified the 88-item DSQ to
make it largely consistent with the DSM-III—R draft glos-
sary of defense mechanisms (Advisory Committee on
Defense Mechanisms, 1986). Exceptions involved ex-
cluding the defense of repression which, although cen-
tral to the whole notion of defense mechanisms, is, by
definition, unable to be tapped; excluding the defense
of intellectualization, as no consensus could be reached
regarding appropriate items; and including anticipation
as a mature defense (Vaillant, 1971), as to do so ap-
peared germane to such a modification. The resulting
72-item DSQ was subsequently administered to 413 sub-
jects. The 20 defenses loaded on three factors, labeled
as mature, neurotic, and immature. The factor structure
was comparable to that obtained from the Bond-labeled
88-item version. Normal subjects and anxiety patients
could now be distinguished, with anxiety patients using
more immature and neurotic and less mature defense
styles (Andrews et al., 1989). Pollock and Andrews
(1989), reanalyzing the same data, showed that the 72-
item DSQ could distinguish between individual anxiety
disorders, and Brennan et al., (1990) showed that child-
abusing parents were more likely than normal subjects
or anxiety patients to endorse the immature defenses
of projection, denial, and splitting.
The above results served to illustrate the apparent
construct validity of the DSQ as a measure of defense
style. However, closer scrutiny showed that the DSQ
suffered from two major shortcomings, reflecting the
incomplete nature of the scale (Bond et a1., 1983). The
first was that two of the items may have been measur-
ing symptoms of specific anxiety disorders, "and one
must always be aware of the possibility of current state
confounding a trait measure. Item 78 (“I have habits or
rituals which I feel compelled to do or else something
terrible will happen”), an item contributing to the de-
fense of undoing, is also a symptom of obsessive-com-
pulsive disorder (OCD), and was thus deleted. Item 41
(“I’m very shy about approaching people”), originally
labeled by Bond as detecting inhibition (Bond, 1986a),
a non-DSM—III-R defense, but used as an indicator of
devaluation in the relabeled DSQ, was found to be func-
tioning to detect symptoms of social phobia and was,
therefore, also deleted. Thus, we reanalyzed the data
247
from the Pollock and Andrews (1989) study after re-
moving these two items. although all other findings
remained, the DSQ score no longer significantly dis-
criminated among the three anxiety disorders.
A further problem with the 72-item DSQ was that
the 20 defenses are unequally represented, with some
defenses being assessed by as many as 10-items, and
others by only one. The defenses with more items
tended to be more internally consistent, and so the
internal consistency of the corresponding factor score
was inflated. Homogenous scales may increase the con-
struct validity of an instrument and allow for ease of
interpretation of scores, but the central question is
“whether the criterion that the test is trying to predict
is itself relatively homogenous or heterogenous . a
. .
single homogenous test is obviously not an adequate
predictor of a highly heterogenous criterion” (Anastasi,
1982, p. 115). The individual defenses in the DSQ are
essentially first-order factors and should comprise ho-
mogenous items, since each defense represents a ho-
mogenous construct. Conversely, the factor scores, es-
sentially second-order factors, should be sampling
equally from this domain, which by definition is likely
to be heterogenous.
Ideally, one would generate a large number of items
for each defense or first-order factor and then select
five to 10 items on the basis of a maximized coefficient-
alpha to represent this factor. Our experience with the
72-item version of the DSQ made it clear that patient
fatigue reduced compliance to such an extent that the
minimum 200-item version consistent with the above
ideal would be impractical. As we had found the origi-
nal 36—item short version to be relatively acceptable,
we decided, in the current version, to have each defense
equally represented by the “best" two items, thereby
trading the sin of methodological expediency for the
virtue of practicality.
The aim of the present study was, therefore, to create
a psychometrically acceptable instrument in which the
heterogeneity of the defenses was preserved in the fac-
tor scores, while aiming for internal consistency at the
level of the individual defenses.
Methods
Subjects
The data from 712 subjects, including the 413 sub-
jects who participated in the relabeling study (Andrews
et al., 1989), were utilized in the reconstruction of the
DSQ. All subjects completed the 72-item DSM-lII-R-la-
beled version of the DSQ. Ages ranged from 12 to 78
years (X : SD, 35 : 13 years); a total of 457 (64%) of
the subjects were female.
The following are subsamples of the above. The nor-
mal control sample (N 388) included 204 persons
=
248 ANDREWS
from the general population; convenience normal sub-
jects constituted the remainder. Their ages ranged from
12 to 75 years (X = 34 years); there were 224 (58%)
female subjects. A subsample (N = 89) was tested on
two occasions, separated by an interval of 4 weeks, in
order to determine the test-retest reliability of the 72-
item DSQ. An independent sample from the general
population (N 56) was also tested 18 months apart
= were discarded and others with less than ideal
on the earlier 36—item DSQ (Andrews et al., 1989). A
general practice sample (N =67) and a psychiatric
outpatient sample (N = 225) were incorporated into
the larger sample. The latter included 45 patients with
panic disorder, 94 with agoraphobia, 44 with social pho-
bia, and 17 with OCD. The clinical diagnoses were made
according to DSM-III-R criteria, and confirmed against
the Diagnostic Interview Schedule. Ages ranged from
16 to 68 years (X = 36 years); there were 156 (69%)
females. A sample of 32 chi1d—abusing parents was also
included. Ages ranged from 17 to 39 years (X 24
= specific anxiety diagnoses.
years); twenty-seven (84%) subjects were female.
DSQ Development
We sought to represent each defense by two items.
Given the homogenous nature of the individual de-
fenses, and the heterogenous nature of the factors, sim-
ple item selection procedures such as those based
solely on the coefficient—alpha statistic or on factor
analytic methods were considered inadequate. Eight
selection criteria were used to assess the construct
validity, the criterion-related validity, and the reliability
of each item. The construct validity of each item was
evaluated in terms of the following:
a) the strength of the item-to-defense correlation,
given that this was indicative of the construct
being assessed.
b) the strength of the item-to—factor correlation to
indicate which items did not contribute to the
scale at all.
c) the extent of unique contribution to the variance
in the corresponding factor score made by each
item, as assessed by the order of entry of that
item in a hierarchical regression analysis with the
factor score as the dependent variable. In those
instances where the item contributed significantly
to the factor score variance but performed poorly
on the other criteria, the item was excluded.
(1) the relation to the corresponding factor and not
to the other two factors. This index Was com-
puted by calculating the absolute value of the
item’s standardized covariance with the factor to
which it supposedly related, and subtracting from
this the sum of the absolute values of the item’s
standardized covariance with the two factors to
which it was supposedly unrelated.
e) face validity as indicated by the degree of
et al.
agreement among five expert raters as to which
defense each item represented. An independent
sample of raters agreed to 74% of the original
relabeling (Weighted Kappa = .75; Andrews et al.,
1989). Even in this present revision, there are still
some items for which a perfect consensus was
not reached. Some items with good face validity
face validity were included. It must be remem-
bered, however, that the allocation of such items
to a particular defense was made in the context
of each item’s performance on other criteria as-
sessing construct and criterion-related validity.
The criterion-related validity of the item was as-
sessed by two further criteria:
f) the item’s capacity to discriminate between nor-
mal subjects and anxiety patients.
g) the item’s capacity to discriminate between the
These two criteria were assessed by the ETA square
test statistic, which indicates the proportion of variance
in item scores accounted for by group membership. A
high value was sought for the former criterion, consid-
ered a useful indicator of the construct validity of the
item, as it appeared likely that, for most defenses, pa-
tients would score quite differently than normal sub-
jects. Items with high values on the second criterion
were closely examined to ensure that they were not
simply symptoms specific to a particular disorder.
The final criterion concerned the test-retest reliabil-
ity of the items:
h) given that defense style is presumed to be a per-
sonality trait and stable over time, items with low
test-retest correlations were considered to be ei-
ther unduly influenced by current mental state
or to be simply unreliable in the sense that the
wording of the item was confusing.
Two additional a priori criteria were considered and
used to further refine the questionnaire. Items that rep-
resented symptoms of specific diagnoses (items 41 and
78, see above), and those items that were applicable to
a particular subgroup only (for example, item 85 [“I
smoke when I am nervous”] and item 79 [“I take drugs,
medicine, or alcohol when I am tense”]) were deleted.
Items within the defenses were then rank ordered on
each of the above criteria, and, guided by the sum of
these ranks, two items were chosen for each defense.
Given that the original DSQ contained only one item
for autistic fantasy and for rationalization, and that the
face validity of the existing rationalization item was
very low, it was necessary to create three additional
items, one for the defense of autistic fantasy and two
for the defense of rationalization. In addition, three
items (16, 59, and 768) were reworded because it ap-
peared that lack of clarity had contributed to poor 4-
 THE DEFENSE STYLE QUESTIONNAIRE
week test-retest correlations. Fifty normal control sub-
jects from a representative census district were then
used to obtain normative data, and 50 medical students
in their final year were used to obtain 1-week test-retest
data on the new and reworded items.
The new 40-item DSQ is in the Appendix. The struc-
ture of the DSQ 40 is evident from Table 1. There are
two items for each of the 20 defenses. The item num-
bers associated with each defense are displayed against
each defense accordingly. Four defenses correspond to
the mature factor (sublimation, humor, anticipation,
and suppression), four to the neurotic factor (undoing,
pseudo-a1truism—reverting to Bond’s [1986a] original
label for this defense, idealization, and reaction forma-
tion), and 12 to the immature factor (projection, passive
aggression, acting out, isolation, devaluation, autistic
fantasy, denial, displacement, dissociation, splitting, ra-
tionalization, and somatization).
The DSQ 40 can yield both 20 individual defense
scores and the three higher—order factor scores (ma-
ture, neurotic, and immature). The scoring is uncompli-
cated. Individual defense scores are simply the average
of the two items for that defense (no items are reverse
scored) and factor scores are simply the average of the
defense scores contributing to that factor.
Results
The performance of the 40 items on each of the selec-
tion criteria are displayed in Table 1, together with the
general population mean and standard deviation for
that item. Only means, standard deviations, and test-
retest data are available for the new items. Does the
new scale measure the same constructs as the old
scale? The correlations among the mature, neurotic,
and immature factors derived from the two versions
were .97, .98, and .95, respectively, evidence of the con-
struct validity of the revised instrument.
Normative data on defenses and factors derived from
the 72- and 40-item versions are displayed in Table 2.
Normative data on the three new items were available
only within the sample of 50 population control sub-
jects, so mean values for the three new items were
imputed using a series of regression analyses (Lepkow—
ski, 1989). We examined the effects of age and sex on
the 72- and 40-item factor scores. There were no sex
differences in defense style in the normal population,
but there was a tendency to use immature defense
styles less with age (DSQ 40: F = 21.83, df = 1,340,
p < .001; Bonferroni-adjusted [Harris, 1985]). The mean
immature factor score fell .82 SD units from age 18 to
25 (3.91) to age 26 to 35 (3.42), to age 36 to 50 (3.46) to
age >50 (3.13). This finding is consistent with previous
research (Vaillant, 1976), and with evidence from a pro-
spective study of young people (Andrews et al., in
249
press) in which a controlled stressor accelerated per-
sonality maturation. In the anxiety patient group, males
scored higher than females on the immature factor on
both versions (DSQ 40: male)? 4.31, female X’ 3.73,
= =
F = 13.22, df= 1,173, 1) = .00), even after diagnosis
and age were controlled for, but the influence of such
psychopathology on being referred for treatment can-
not be discounted.
The reliability coefficients associated with the de-
fense and factor scores for the 72- and 40-item DSQ are
displayed in Table 3. The internal consistency of the
defenses and factors, as assessed by the coefficient-
alpha statistic (Anastasi, 1982), was comparable for
both forms. The coefficient-alphas for the defenses of
undoing and devaluation fell for the 40-item form, mark-
edly in the case of devaluation, as a direct result of the
removal of the symptom items. The internal consis-
tency of the 72-item DSQ defense of devaluation was
not high to begin with. This perhaps reflects a problem
in both the 72- and 40-item questionnaires, given that
although one of Bond’s (l986a) best two items for de-
valuation (item 24) was retained in the DSQ 40, the
other (item 18) could not be agreed upon and was thus
discarded during the DSM—III—R relabeling of the DSQ
(Andrews et al., 1989). Item 29, the other of the two
devaluation items used in the DSQ 40, was originally
labeled by Bond as detecting inhibition (Bond, 1986a),
a non—DSM-III-R defense. On the one hand, this further
highlights the need for an agreed taxonomy of defenses,
while on the other, it illustrates that items are not al-
ways read by patients in ways that psychiatrists would
presume. The addition of items to the defenses of autis-
tic fantasy and rationalization made it possible to calcu-
late coefficient-alphas for these two defenses; both of
these defenses performed well on this measure. The
coefficient-alphas of the mature, neurotic, and imma-
ture factors were moderate to high and comparable for
both versions of the scale. Test—retest coefficients were
uniformly high across the 72- and 40—item forms. In
addition, using the original short form of the DSQ (An-
drews et al., 1989), we found 18-month test-retest corre-
lations of .71 for the mature factor and .60 for the imma-
ture factor in a sample of 56 normal subjects.
Profile and discriminant function analyses (with
Bonferroni adjustment) were used to compare the pat-
tern of defenses of the normal control subjects, the
anxiety patients, and the child-abusing parents (Table
4). As with the 72-item DSQ, the new 40-item scale
significantly discriminated anxiety patients from nor-
mal controls, child-abusing parents from normal con-
trols, and child-abusing parents from anxiety patients.
In comparison to normal subjects, the anxiety-disor-
dered patients were less likely to endorse the mature
defenses of humor, suppression, and sublimation, and
more likely to endorse the neurotic defense of reaction
250 ANDREWS et al.
TABLE 1
Performance of the DSQ 40“
DSM-III-R Defense Item 1 2 3 4 5 6 7 8 Mean SD

Mature Factor
Sublimation 5 .26 .47 5 .46 5 .039” .008 .63 5.11 2.50
84 .37 .52 7 .47 3 .018 .011 .56 5.82 2.25
Humor 8 .46 .54 6 .52 5 .070” .080 .48 6.54 2.16
61 .46 .59 1 .58 5 .021 .028 .60 6.34 2.10
Anticipation 68 .20 .56 3 .55 5 .018 .010 .81” 6.62 1.92
81 .20 .48 2 .40 5 .005 .074 .52 4.47 2.62
Suppression 3 .04 .45 9 .44 5 .068” .007 .47 4.80 2.40
59 .17 .55 4 .54 5 .011 .064 .52” 6.23 2.23
Neurotic Factor
Undoing 71 .26 .52 3 .42 4 .012 .044 .73 4.56 2.59
88 .32 54 6 .32 4 .004 .006 .50 3.97 2.42
Pseudo—altruism 1 .11 .48 2 .43 5 .012 .003 .59 5.37 2.33
86 11 45 4 .42 4 .006 .009 .60 4.94 2.61
Idealization 51 .31 .55 1 .41 4 .003 .015 .56 3.32 2.45
58 .27 51 5 .45 4 .007 .000 .67 3.94 2.74
Reaction formation 13 .32 .33 12 .31 5 .033” .002 .86 3.47 2.56
63 .39 42 11 .34 5 .012 .035 .55 4.86 2.41
Immature Factor
Projection 12 .63 .55 2 .42 5 .032” .019 .53 2.25 1.79
66 .58 53 40 .39 5 .104” .024 .74 2.42 1.98
Passive aggression 54 .24 .39 34 .39 5 .001 .005 .71 2.85 2.31
82 .39 .50 39 .42 3 .011 .003 .47 3.56 2.30
Acting out 27 .53 .38 24 .31 5 .000 .017 .66 4.89 2.46
46 .38 .41 8 .39 5 .001 .000 .61 4.51 2.61
Isolation 76 .41 .44 10 .41 4 .019 .003 .66 4.28 2.70
83 .40 41 22 .37 4 .001 .017 .64 3.88 2.57
Devaluation 24 .02 .42 16 .41 4 .000 .051 .67 2.60 2.24
29 .42 40 21 .28 3 .129” .058 .63 3.50 2.25
Autistic fantasy 31 — —— —— —— —— — — .7 8‘ 2. 83 2. 52
40 — .60 1 .50 5 .007 .014 .76 3.95 2.68
Denial 16 .01 .43 15 .32 4 .018 .015 .50‘ 3.11 2.07
42 .01 26 6 .13 4 .044” .026 .63 2.92 2.25
Displacement 69 .18 .50 4 .39 3 .143” .005 .72 2.61 2.18
73 O0 .27 17 .17 4 .001 .035 .76 4.36 2.83
Dissociation 23 .29 .42 12 .39 4 .004 .062 .72 2.34 1.89
37 14 .27 23 .08 4 .045” .008 .69 3.36 2.36
Splitting 43 .07 .41 14 .35 4 .000 .015 .67 4.10 2.65
53 .29 32 20 .29 4 .002 .003 .53 3.46 2.52
Rationalization 6 — — ——- — —— — ——- .63” 6.38 1.51
38 — — — — — — — .52” 5.11 2.27
Somatization 28 .40 .49 11 .34 4 .117” .005 .60 3.21 2.44
62 .40 .43 5 .35 4 .018 .003 .69 2.89 2.43
“The table shows the best 40 items to delineate each defense using the following criteria: 1) item-defense correlation; 2) item-factor
correlation; 3) item-factor regression step; 4) item unique variance in factor; 5) face validity (rater agreement); 6) ETA square (normal subjects
versus patients); 7) ETA square (panic disorder/agoraphobia versus social phobia versus obsessive-compulsive disorder); and 8) item test-
retest correlation (N 89). The mean and SD for a population sample (N 388) are also displayed. Item numbers are as per Bond (1986b),
= =

except for the new items (6, 31, and 38) which have been given item numbers previously assigned to lie items.
"ETA square is significant when Bonferroni-adjusted for 72 tests.
‘Correlation coefficients for new/reworded items based on independent sample (N 50) tested on two occasions.
=

formation and the immature defenses of devaluation,
displacement, projection, and somatization. The child-
abusing parents were more likely to endorse the neu-
rotic defense of reaction formation and the immature
defenses of denial, devaluation, projection, and split-
ting. In comparison to the anxiety patients, the child-
abusing parents were more likely to endorse the imma-
ture defenses of dissociation and splitting. No signifi-
cant differences Were observed between the panic/
agoraphobia group and the social phobia group, the
panic/agoraphobia group and the OCD group, and the
social phobia group and the OCD group. The 40-item
DSQ thus behaves in a fashion similar to that of the 72-
item DSQ once the symptom items have been removed.
If one compares the differences, expressed in effect
sizes, between the diagnostic groups and the control
 THE DEFENSE STYLE QUESTIONNAIRE 251

TABLE 2
Normative Data by Defense and Factor for the 72—Ite'm and 40-Item Versions of DSQ“
DSM-III—R Defense 72—Item 40-Item

Mature Factor 5.95 i 1.03 5.76 i 1.15

Sublimation 5.88 i 1.56 5.45 : 1.82
Humor 6.44 i 1.77 6.44 : 1.77
Anticipation 5.72 i 1.79 5.72 1‘ 1.79
Suppression 5.76 i 1.36 5.50 : 1.81
Neurotic Factor 4.25 : 1.11 4.32 t 1.28
Undoing 3.73 i 1.69 4.26 : 1.96
Pseudo-altruism 5.14 2': 1.86 5.14 i 1.86
Idealization 4.00 i 1.72 3.64 1‘ 2.13
Reaction formation 4.08 : 1.32 4.17 i 1.92
Immature factor 3.31 i .93 3.54 i .95
Projection 2.38 i 1.10 2.34 1 1.63
Passive aggression 3.74 : 1.21 3.20 t 1.80
Acting out 4.11 : 1.52 4.70 i 2.03
Isolation 3.69 : 1.54 4.08 i 2.18
Devaluation 3.55 : 1.57 3.06 : 1.57
Autistic fantasy 3.95 : 2.68 3.63 : 2.25
Denial 2.88 : 1.57 2.88 : 1.57
Displacement 3.22 S:1.62 3.48 i 1.88
Dissociation 2.85 : 1.50 2.85 i 1.74
Splitting 3.25 : 1.56 3.78 i 1.91
Rationalization 3.01 : 2.26 5.57 i 1.25
Somatization 3.05 i 2.02 3.05 t 2.02

“Data are expressed as mean : SD (N = 388; see text).

TABLE 3
Reliability Coefiicients on Defenseand Factor Scores for the DSQ
Coefficient alpha for Test—Retest Correlation at
Normal Subjects 4 Weeks for
and Patients Normal Subjects
(N = 712) (N = 89)
Defenses 72-Item 40-Item 72-Item 40-Item

Mature Factor .59 .68 .68 .75

Sublimation .53 .42 .61 .67
Humor .59 .59 .66 .66
Anticipation .32 .32 ' .63 .63
Suppression .07 .39 .27 .38
Neurotic Factor .72 .58 .77 .78
Undoing .50 .37 .71 .61
Pseudo-altruism .19 .19 .62 .62
Idealization .38 .52 .75 .72
Reaction formation .65 .32 .69 .75
Immature Factor .89 .80 .86 .85
Projection .82 .64 .81 .77
Passive aggression .56 .38 .72 .67
Acting out .68 .49 .75 .72
Isolation .51 .56 .73 .70
Devaluation .42 — .01 .60 .57
Autistic fantasy — .89 .76 .73
Denial .10 .10 .48 . .48
Displacement .09 .17 .85 .80
Dissociation .27 .44 .71 .75
Splitting .36 .19 .72 .73
Rationalization — .73 .76 .53
Somatization .56 .56 .72 .72
252 ANDREWS et al.
TABLE 4
Differences on the 40-Item DSQ
Difference from Control group"
Obsessive-
Panic and Social compulsive All anxiety Child
agoraphobia phobia disorder patients abusers
Defense (N = 139) (N = 44) (N = 17) (N 200)
= (N = 32)
Mature Factor —.88” -1.16” —.90” —.95” —.31
Anticipation —.43 — .53” — .07 - .43 — .41
Humor —.69” — .94” -1.26” — .80” — .31
Suppression —.61” — .67” — .72” — .63” — .04
Sublimation -.50” — .76” — .34 — .54” - .38
Neurotic Factor .34” .48” .66 .40” .64”
Pseudo-altruism .36 .17 .40 .32 .17
Idealization .23 .11 .36 .21 .42
Reaction formation .37” .60” .39 .42” .65”
Undoing —.02 .28 .62 .10 .42
Immature Factor .42” .40 .76 .44” 1.01”
Acting out —.01 —.03 .32 .02 —.08
Denial —.08 —.36 —.17 —.15 .59”
Devaluation .34” .61” .77” .43” .65”
Displacement .35 .61” .72 .45” .08
Dissociation —.07 -.52” —.44 —.21 .54
Autistic fantasy .02 .31 .27 .11 .20
Isolation .17 .40 .47 .25 .51
Passive aggression .21 .17 .17 .20 .55
Projection .60” .78” 1.33” .71” 1.47”
Rationalization .07 .28 .36 .15 .42
Somatization .53” ‘ .56” .59 .54” .17
Splitting .04 —.01 .22 .04 .92”
Psychopathology score 1.64 2.04 2.32 1.79 1.96
(sum of deviation of effect sizes)
“Data represent differences in effect size units (from the normal means displayed in Table 1), on the 40-item DSQ between 388 control
subjects and 200 patients with anxiety disorder diagnoses, and between 388 control subjects and 32 child abusers.
"Difference from control group is significant at 1) < .05/23 (i.e., includes Bonferroni correction).

group for the 72-item form (displayed in Table 1 of
Pollock and Andrews [1989]), and for the 40—item form,
some interesting differences do emerge. Although the
discrimination between groups is as powerful, the con-
tribution made by the neurotic and immature factors is
lessened by the removal of the symptom items, and the
contribution to the discrimination made by the mature
factor is increased.
Discussion
The notion that unconscious mental mechanisms are
utilized to mediate the effects of stress is manifest to
psychiatrists and, indeed, to the educated public. For
clinicians, an understanding of how defense mecha-
nisms may augment or diminish acute neurotic symp-
tomatology can facilitate treatment. Moreover, some
awareness of defensive styles commonly used in every-
day life is often an asset in coping with colleagues who
unexpectedly become angry, compliant, or forgetful.
Our interest in how people cope with stress, including
attempts at measuring their defense style (Andrews
et al., 1978), stems not from any doctrinaire position,
but reflects an endeavor to improve our ability to pre-
dict the relationship between stress and illness. The
difficulty has always been the lack of an appropriate
measure of defense style and of a standardized taxon-
omy of defenses, and the original questionnaire devised
by Bond et al. (1983), together with the DSM-III-R draft
glossary of defense mechanisms (Advisory Committee
on Defense Mechanisms, 1986), constituted consider-
able advances, even if the face validity of some of the
current items does leave something to be desired.
The second stage in the development of any tech-
nique is often dull, and the heady excitement of initial
findings gives Way to a thorough examination of possi-
ble shortcomings Which may lead to refinement. The
strengths of the new DSQ 40 are that a brief question-
naire in the public domain has been shown to reliably
identify defense styles which correspond to hypothe-
sized patterns of unconscious psychological mecha-
nisms, and that the instrument possesses reasonable
psychometric properties, including internal consis-
tency and temporal stability appropriate in a trait mea-
sure. The availability of normative values established
on a large population sample, coupled with an absence
of any bias due to the sex of the respondent and a
 THE DEFENSE STYLE QUESTIONNAIRE
simple scoring procedure, add to the desirability of the
scale.
Consistent with expectations, there is evidence of a
decline in the use of immature defenses after adoles-
cence, and evidence that defense style varies among
three criterion groups—normal subjects, patients with
anxiety disorders, and child-abusing parents—in a pre-
dicted fashion quite independently of symptoms or as-
sociated clinical features. Although very comparable to
the 72-item DSQ, the DSQ 40 has yet to be validated
against the ratings of skilled clinicians or against other
instruments, such as the Defense Mechanism Inventory
(Gleser and Ihilevich, 1969) or the Defense Mechanism
Rating Scale (Perry and Cooper, 1986), as has been
done with the original DSQ (Bond et al., 1989).
Defense style information appears to be of value in
our clinic. Patients with anxiety disorders whose de-
fense style profile deviates less than 1 SD from normal
seem, even given concurrent stressors and a high level
of trait anxiety, to respond well to minimal interven-
tion, whereas those with a low mature and high neu-
rotic and immature defense pattern, the “susceptible”
pattern, seem to require considerably more treatment.
This is an area currently being researched. An unusually
aberrant score on one factor has, on occasion, clarified
the mechanism of neurotic breakdown in a patient who
previously did not appear vulnerable to a neurotic ill-
ness. This type of information would normally only be
elicited after a number of psychotherapy sessions.
The 72-item DSM—III—R-labeled DSQ has also proven
of interest in research settings (Andrews et al., in press;
Bennett et al., 1991; Kesby et al., 1991; Procter and
Hafner, 1991). We have been exploring the salience of
three personality factors associated with neurosis: trait
anxiety or neuroticism, locus of control, and defense
style. On two previous occasions (Andrews and Craig,
1988; Andrews and Moran, 1988), we have shown that
neuroticism or locus of control can change in response
to successful treatment of the disorder and that the
extent of such change predicts stability of improvement
after the conclusion of treatment. We now have evi-
dence that a susceptible defense style can improve with
cognitive behavior therapy (Hunt, 1992), but whether
such change immunizes against relapse is not yet
known.
In a large prospective study of adult twins (Andrews,
1991), we have been able to examine the genetic contri-
bution to defense style. In our current modeling of
these data, 38% of the variance in DSQ score can be
attributed to genetic factors and 62% to the influence of
environmental factors. Half of the genetic component
loaded on a common factor shared with the other vul-
nerability measures, neuroticism and locus of control,
but the other half of the genetic variance seemed spe-
cific to defense style itself. The important question is
258
how people learn to favor the use of a particular pattern
of defenses. In an ongoing prospective study, Ungerer
et al. (1990) measured the defense style of pregnant
women and suggested that susceptible patterns of de-
fense style are predictive of impaired mother/baby in-
teractions and impaired development of the self—regula-
tion of emotion in the baby. It remains to be seen
whether this is a harbinger of a susceptible defense
style in the child or adolescent. In conclusion, the DSQ
40 emerges as a convenient and reliable measure for
use in clinical practice and as a very promising research
instrument.
References
Advisory Committee on Defense Mechanisms (Work Group to Revise
DSM-III) (1986) A draft glossary of defense mechanisms for DSM-
III-R. In GE Vaillant (Ed), Empirical studies of ego mechanisms
of defense. Washington, DC: American Psychiatric Press, Inc.
American Psychiatric Association (1987) Diagnostic and statistical
manual of mental disorders (3rd ed, rev) (DSM-III-R). Washington,
DC: Author.
Anastasi A (1982) Psychological testing. New York: Macmillan.
Andrews G (1991) Anxiety, personality, and anxiety disorders. Int
Rev Psychiatry 3:293—302.
Andrews G, Craig A (1988) Prediction of outcome after treatment for
stuttering. Br J Psychiatry 1532236-240.
Andrews G, Moran C (1988) Exposure treatment of agoraphobia with
panic attacks: Are drugs essential? In H-U Wittchen (Ed), Panic
and phobias II. Treatments and variables aflecting coarse and
outcome. Heidelberg: Springer-Verlag.
Andrews G, Page A, Neilson M (in press) Sending your teenagers
away: Controlled stress decreases neurotic vulnerability. Arch Gen
Psychiatry.
Andrews G, Pollock C, Stewart G (1989) The determination of defense
style by questionnaire. Arch Gen Psychiatry 46:455-460.
Andrews G, Tennant C, Hewson D, Vai1lantG (1978) Life event stress,
social support, coping style, and risk of psychological impairment.
J Nerv Ment Dis 166:307-316.
Bennett E, Beaurepaire J, Langeluddecke P, Kellow J, Tennant C
(1991) Life stress and non-ulcer dyspepsia: A case—control study. J
Psychosom Res 35:579-590.
Bond M (l986a) An empirical study of defense styles. In GE Vaillant
(Ed), Empirical studies of ego mechanisms of defense. Washing-
ton, DC: American Psychiatric Press, Inc.
Bond M (1986b) Defense Style Questionnaire. In GE Vaillant (Ed),
Empirical studies of ego mechanisms of defense. Washington, DC:
American Psychiatric Press, Inc.
Bond M (1990) Are “borderline defenses” specific for borderline
personality disorders? J Pers Dis 4:251—256.
Bond M, Gardner ST, Christian J, Sigal JJ (1983) Empirical study of
self-rated defense styles. Arch Gen Psychiatry 40:333-338.
Bond M, Paris J, Zweig-Frank H (1993) Defense styles and borderline
personality disorder. Submitted for publication.
Bond M, Perry J C, Gautier M, Goldenberg M, Oppenheimer J, Simand
J (1989) Validating the self-report of defense styles. J Pers Disord
3:l0l—l12.
Bond MP, Sagala Vaillant J (1986) An empirical study of the relation-
ship between diagnosis and defense style. Arch Gen Psychiatry
43:285-288.
Brennan J , Andrews G, Morris-Yates A, Pollock C (1990) An examina-
tion of defense style in parents who abuse children. J Nerv Ment
Dis 178:592—595.
Freud A (1966) The ego and the mechanisms of defence. New York:
International Universities Press.
Freud S (1959) Inhibition, symptoms, and anxiety. London: Ho-
garth.
Gleser GC, Ihilevich D (1969) An objective instrument for measuring
defense mechanisms. J Consult Clin Psychol 33:51-60.
254
Harris RJ (1985) A primer of multivariate statistics (2nd ed). Lon-
don: Academic Press.
Hunt C (1992, July 4-8) Vulnerability factors and long-term outcome
after cognitive-behavioural treatment of anxiety disorders. Paper
presented at the Fourth World Congress on Behaviour Therapy,
Gold Coast, Queensland Australia.
Kesby G, Parker G, Barrett E (1991) Personality and coping style as
influences on alcohol intake and cigarette smoking during preg-
nancy. Med J Aust 155:229—233.
Lepkowski JM (1989) Treatment of wave nonresponse in panel sur-
veys. In D Kasprzyk, GJ Duncan, G Kalton, MP Singh (Eds), Panel
surveys. New York: John Wiley & Sons.
Perry JC, Cooper SH (1986) A preliminary report on defenses and
conflicts associated with borderline personality disorder. J Am
Psychoanat Assoc 34:863-893.
Pollock C, Andrews G (1989) Defense styles associated with specific
anxiety disorders. Am J Psychiatry 146:1500—1502.
ANDREWS et al.
Procter N, Hafner J (1991) Student nurses’ attitudes to psychiatry:
The influence of training and personality. J Adv Nursing 16:845-
849.
Strayhorn G (1983) A diagnostic axis relevant to psychotherapy and
preventive mental health. Am J Orthopsychiatry 53:677-696.
Ungerer JA, Dolby R, Waters BG, Bannett B, Kelk N, Lewin V (1990)
The early development of empathy: Self-regulation and individual
differences in the first year. Motivation Emotion 14:93—106.
Vaillant GE (1971) Theoretical hierarchy of adaptive ego mechanism:
A 30-year follow-up of 30 men selected for psychological health.
Arch Gen Psychiatry 24:107-118.
Vaillant GE (1976) Natural history of male psychological health: V.
The relation of choice of ego mechanisms of defense to adult
adjustment. Arch Gen Psychiatry 33:535-545.
Vaillant GE, Bond M, Vaillant CO (1986) An empirically validated
hierarchy of defense mechanisms. Arch Gen Psychiatry 43:786-
794.

D.S.Q. 40
INSTRUCTIONS: This questionnaire consists of a number of statements about personal attitudes. There are no
right or wrong answers. Using the 9-point scale shown below, please indicate how much you agree or disagree
with each statement by circling one of the numbers on the scale beside the statement. For example, a score of 5
Would indicate that you neither agree nor disagree with the statement, a score of 3 that you moderately disagree,
a score of 9 that you strongly agree.

Strongly
R i ‘T’ i i iii? Strongly
disagree agree

1. I get satisfaction from helping others and if this

were taken away from me I would get depressed

3. I’m able to keep a problem out of my mind until

I have time to deal with it ................................... ..

5. I work out my anxiety through doing something

constructive and creative like painting or wood-
work ......................................................................... ..

6. I am able to find good reasons for everything I

do ............................................................................. ..

8. I’m able to laugh at myself pretty easily ........... ..

12. People tend to mistreat me ................................. ..

13. If someone mugged me and stole my money, I’d

rather he be helped than punished .................... ..

16. People say I tend to ignore unpleasant facts as

if they didn’t exist ................................................. ..

23. I ignore danger as if I was Superman ................ ..

24. I pride myself on my ability to out people down

to size ...................................................................... ..

27. I often act impulsively when something is both-

ering me .................................................................. ..

Please go to next page=>

 THE DEFENSE STYLE QUESTIONNAIRE 255

Strongly Disagree => 1 23 1; 56 '7 8 .9 <= Strongly Agree

28. I get physically ill when things aren’t going well
for me ...................................................................... ..

29. I’m a very inhibited person .................................. ..

31. I get more satisfaction from my fantasies than

from my real life .................................................... ..

37. I’ve special talents that allow me to go through

life with no problems ............................................ ..

38. There are always good reasons when things

don’t work out for me .......................................... ..

40. I work more things out in my daydreams than in

my real life .............................................................. ..

42. I fear nothing ......................................................... ..

43. Sometimes I think I’m an angel and other times

I think I’m a devil .................................................. ..

46. I get openly aggressive when I feel hurt ........... ..

51. I always feel that someone I know is like a guard-

ian angel .................................................................. ..

53. As far as I’m concerned, people are either good

or bad ...................................................................... ..

54. If my boss bugged me, I might make a mistake

in my work or Work more slowly so as to get
back at him ............................................................. ..

58. There is someone I know who can do anything

and who is absolutely fair and just .................... ..

59. I can keep the lid on my feelings if letting them

out would interfere with what I’m doing .......... ..

61. I’m usually able to see the funny side of an other-

wise painful predicament ..................................... ..

62. I get a headache when I have to do something I

don’t like ................................................................. ..

63. I often find myself being very nice to people who

by all rights I should be angry at ........................ ..

66. I am sure I get a raw deal from life ................... ..

68. When I have to face a difficult situation I try to

imagine what it will be like and plan ways to
cope with it ............................................................ ..

69. Doctors never really understand what is wrong

with me ................................................................... ..

71. After I fight for my rights, I tend to apologize for

my assertiveness .................................................... ..

73. When I'm depressed or anxious, eating makes

me feel better ......................................................... ..

76. I’m often told that I don’t show my feelings .... .. 7

Please go to next page=>

256 ANDREWS et al.

Strongly Disagree => 1 23 1; 5678 .9 <= Strongly Agree

81. If I can predict that I’m going to be sad ahead of
time, I can cope better ......................................... .. 1 2 3 4 5 6 7

82. No matter how much I complain, I never get a

satisfactory reponse .............................................. .. 1 2 3 4 5 6 7

83. Often I find that I don’t feel anything When the

situation would seem to warrant strong emo-
tions......................................................................... .. 1 2 3 4 5 6 7

84. Sticking to the task at hand keeps me from feel-

ing depressed or anxious ..................................... .. 1 2 3 4 5 6 7

86. If I were in a crisis, I would seek out another

person who had the same problem .................... .. 1 2 3 4 5 6 7

88. If I have an aggressive thought, I feel the need

to do something to compensate for it ............... .. 1 2 3 4 5 6 7

PLEASE CHECK TO SEE THAT YOU HAVE ANSWERED ALL THE QUESTIONS