The Italian version of the Response Evaluation Measure–71
Antonio Prunasa,⁎, Fabio Madeddua, Stefano Pozzolib, Cristina Gattic,
Richard J. Shawd, Hans Steinerd
aDepartment of Psychology, University of Milan-Bicocca, Milan, Italy
bInstitute of Psychology, University of Milan, Milan, Italy
cDepartment of Psychology, Stanford University, Palo Alto, CA, USA
dDepartment of Psychiatry and Behavioral Sciences, Stanford University School of Medicine, Palo Alto, CA, USA
Objective: The aim of the present study was to analyze the psychometric properties of the Italian version of the Response Evaluation
Measure (REM-71), a 71-item self-report measure previously developed for the assessment of defenses in adults and adolescents. The authors
also examined the differences in the use of defenses based on sex and age (early adolescence, late adolescence, and early adulthood), and the
association between defenses, psychosocial health, and psychologic distress in a large community sample.
Method: The Italian version of REM-71 was obtained through back-translation and administered to 1648 (1020 female subjects, mean age =
19.5years, SD = ±5.77) community subjects, aged between 13 and 68 years, voluntarily recruited among high school and university studentsin
Milan, Italy, and the surrounding area. All subjects completed a self-report measure to assess demographic variables and satisfaction with life.
A subsample (n = 1197) completed the Italian version of the Symptom Checklist–90–Revised to assess symptoms of psychologic distress.
Results: Results were highly consistent with those obtained in the original English version of the REM-71 and included satisfactory internal
consistency of the measure. Factor analyses yielded 2 principal factors that showed overall stability across age and sex subgroups. Factor 1 and
Conclusions: Results provide further support for the structure and validity of the REM-71 as a useful instrument for the assessment of
defenses in adolescents and adults and suggest no major cross-cultural differences in the organization of these defenses.
© 2009 Elsevier Inc. All rights reserved.
The Diagnostic and Statistical Manual of Mental
Disorder, Fourth Edition, Text Revision ( p807) defines
defenses as “automatic psychological processes that protect
the individual against anxiety and from the awareness of
internal or external dangers or stressors.” Although interest
in the scientific study of these processes has waxed and
waned considerably over the years , the concept of
defenses, a form of normative, unconscious, and automatic
self-regulation , is arguably one of Sigmund Freud's most
important contributions to psychiatry . A large body of
evidence suggests that defenses are related to psychologic
health and well-being, psychiatric symptoms, and psycho-
logic distress [5-14]. Defenses have been shown prospec-
tively to predict health and illness in longitudinal studies
. More recently, research from the field of neuroscience
has provided some insights into how such forms of
automatic, reflexive self-regulation are conceptually possible
and what the supporting neuroarchitectures might be .
Much of the early research on defenses has relied upon
self-report measures such as the Defense Style Questionnaire
(DSQ) . Despite methodological challenges in studying
automatic, unconscious processes by self-report, evidence
supports the reliability and validity of the DSQ in both cross-
sectional and longitudinal research [7,17-21]. However, one
of the more problematic aspects of the DSQ is the relative
instability of its factor structure, with studies yielding results
only partially corresponding to the original categories of
defenses described by Bond et al . For example,
Australian research by Andrews et al  showed a 3-factor
solution, whereas Reister et al  in a German version of
the DSQ found 4 factors. By contrast, Rutherford et al  in
Available online at www.sciencedirect.com
Comprehensive Psychiatry 50 (2009) 369–377
⁎Corresponding author. Dipartimento di Psicologia, Università degli
Studi di Milano-Bicocca, Piazza dell’Ateneo Nuovo 1, 20126, Milan, Italy.
Tel.: +39 02 6448 3757; fax: +39 02 6448 3705.
E-mail address: email@example.com (A. Prunas).
0010-440X/$ – see front matter © 2009 Elsevier Inc. All rights reserved.
a sample of patients receiving methadone therapy were only
able to clearly delineate 1 immature dimension of defense.
Bonsack et al , using a French version of the DSQ, found
4 dimensions; but analysis of their factor structure showed
that their allocation of defenses was different from that
originally described by Bond et al. Using an Italian version
of the DSQ, San Martini et al  identified 3 factors but
concluded that only one had sufficient reliability and internal
consistency to warrant use in its present form.
The Response Evaluation Measure (REM-71) was devel-
oped as an alternative to the DSQ , extending the
assessment into juvenile cohorts [20,27]. This scale was
developed to improve several psychometric shortcomings of
the DSQ that became apparent in the attempt at modification
for younger age groups . Operating from a perspective of
developmental psychiatry, the scale defines defenses as
normative and ubiquitous self-regulatory processes, in line
with Vaillant's  suggestions to regard defenses as metrics
of positive psychologic health. The REM-71 has versions for
observer-rated (parent) report . Across these age groups,
an unrotated 2-factor solution was preferred on the ground of
methodological considerations [20,27,28]. Factor 1 (F1)
contains 14 defenses that distort reality in accordance with
expected outcomes, leading to less adaptive functioning.
Factor 2 (F2), by contrast, contains 7 defenses that attenuate
unwelcome reality, allowing more adaptive functioning.
The aim of the present study was to analyze the
psychometric properties of the Italian version of the REM-
71, including its factor structure and its stability in sex and
age subgroups and across different cultural contexts. We also
assessed differences in defenses based on sex and age, the
association between defenses and self-report of psychosocial
health, and symptoms of psychologic distress. Hypotheses
were that the original 2-factor solution would be preserved;
that there would be age effects in line with developmental
expectations (decrease in F1, increase in F2 with increasing
age); and that there would be correlations, in line with
theoretical expectations, between the REM-71 and the self-
report of well-being on the one hand and psychopathology
on the other.
2. Material and methods
This study was approved by the institutional review
board. Subjects were 1648 volunteers between 13 and 68
years of age (mean age = 19.5 years, SD = ±5.77). Of the
subjects, 38.1% (n = 628) were male and 61.9% (n = 1020)
were female. The sample included both high school and
university students; the former (n = 959, 58.2% of the total
sample) were recruited from private and public schools in
Milan, Italy, or the surrounding area (46.8% female, mean
age = 16.29 years, SD = ±1.45, range = 13-24); 72.5% of
mothers and 94.2% of fathers of subjects were employed and
had completed high school (49.9% and 44.7%, respectively)
or university education (21.5% and 24.7%, respectively).
Most subjects (97%) were white. University students in the
sample (41.8%, n = 689, 82.9% female, mean age = 23.83
years, SD = ±6.71, range = 19-68) were recruited from
Milano-Bicocca State University, Milan. In this group,
69.3% of mothers and 97.3% of fathers of subjects were
employed and had completed high school (47.85 and 42.1%,
respectively) or university education (12.2% and 17.1%,
respectively). No data on ethnicity were obtained for this
subsample to ensure confidentiality.
Sex and age distributions for the 2 subsamples are
summarized in Table 1. Generally speaking, the participants
may be considered a convenience sample of the population
living in the urban and suburban area of Milan, with a
prevalence of middle-class subjects, with intermediate to
high levels of education.
2.2.1. Response Evaluation Measure
The REM-71  is a 71-item self-report questionnaire
allowing the evaluation of 21 defenses assessed by 3 or 4
items each and rated on a 9-point scale from “strongly
disagree” (scored as 1) to “strongly agree” (scored as 9). The
scale has favorable psychometric properties, as described in
the original validation study , including satisfactory
internal coherence with α values ranging from 0.36 to 0.85
(mean Cronbach α value = 0.56) for single defenses and
reaching 0.84 and 0.68 for F1 and F2, respectively. The
scale's validity is supported in several studies involving
adults, adolescents, and school-aged children [20,27]. The
current Italian version was obtained through back-translation
and previously administered to 18 bilingual subjects to
assess score correlations between the 2 versions .
2.2.2. Symptom Checklist–90–Revised
The Symptom Checklist–90–Revised  is a widely
used 90-item self-report questionnaire intended to measure
self-report symptom severity on a number of different
subscales. Items are scored on a 5-point Likert scale
indicating the rate of occurrence of the symptoms during
the previous 7 days. The checklist consists of 9 subscales and
3 global indexes of distress including the Global Severity
Sex and age distribution in the 2 subsamples
AgeHigh school students
(n = 959)
(n = 689)
(n = 510)
(n = 449)
(n = 118)
(n = 571)
370A. Prunas et al. / Comprehensive Psychiatry 50 (2009) 369–377
Index (GSI). A subgroup of the total sample (n = 1197,
64.9% female, mean age = 19.04 ± 3.91 years, range = 13-45
years) completed the Italian version of this measure .
2.2.3. Measures of psychosocial functioning
To explore the hypotheses regarding psychosocial
functioning, 5 additional self-report items used in the
original validation of the REM-71 were included in the
survey . Using a 9-point scale from “very unsatisfied”
(scored as 1) to “very satisfied” (scored as 9), subjects were
asked to indicate their satisfaction in the domains of work/
school, family, friendships, and recreation and to respond to
a general statement of satisfaction.
2.3. Statistical analysis
Statistical analysis included principal components factor
analysis, parallel analysis, multivariate analyses of variance,
and linear regressions as appropriate.
3.1. Internal consistency
The internal consistency for each of the 21 defenses is
shown in Table 2. The mean Cronbach α value was 0.56.
Two defenses, sublimation and denial, had α values less than
0.40. One item used to assess denial (isolation of affect) was
excluded because its correlations with the other 2 were low
(ie, Pearson r b .15) or not statistically significant;
correlation between the 2 remaining items was 0.32 (P b
.001). After deletion of one of the items (item 27), α value
because it showed low or no statistically significant correla-
tions withthe other 2 items:α value after its deletionwas 0.79.
Although sublimation did not show an acceptable internal
cohesion, this defense was retained to allow comparison with
the original version of the scale.
Defenses included in the REM-71: internal consistency and α coefficients
Defenses No. of
Current studySteiner et al 
aThe α values reported here were computed after deletion of item 27
(denial, F2) and item 34 (suppression, F1).
bThe α values of the original version are not comparable with those of
the Italian version because factor composition varies.
cThevalues reportedinbracketsare obtained after deletionof oneof the
Factor loadings of 21 defenses in the REM-71 for 1648 community subjects
from the unrotated 2-factor solution and the rotated 3-factor solution
Factor 1Factor 2Factor 3
Unrotated 2-factor solution
Rotated 3-factor solution
371A. Prunas et al. / Comprehensive Psychiatry 50 (2009) 369–377
In general, α values of the Italian REM-71 are broadly
comparable with those of the English version (Table 2).
3.2. Factor analysis
A principal components analysis (PCA) was performed
for 21 defenses (Table 3). The Kaiser-Meyer-Olkin index of
sampling adequacy was 0.85, indicating that the variables
shared sufficient common variance to justify the analysis. To
determine how many factors to retain, 2 statistical guidelines
were adopted. First, a scree plot showed that it was
appropriate to extract 3 factors. Second, the empirically
derived eigenvalues obtained from the PCA were compared
with those resulting from a series of Monte Carlo simula-
tions. The averaged results of 1000 Monte Carlo runs
indicated that retention of a third factor was questionable
because the eigenvalue of the empirically obtained third
factor was not significantly greater than that expected by
chance. Initial PCA resulted in 2 interpretable solutions: an
unrotated 2-factor solution and a 3-factor solution using
Varimax rotation. For both solutions, loadings greater than
0.30 were considered significant if the difference in loadings
between factors was at least 0.20.
In the unrotated solution, F1 contained 14 defenses. One
defense (reaction formation) had loadings on the first factor
that were within 0.20 of their F2 loadings. Because the
highest loading was on F2, it was included in this factor.
Factor 1 accounted for 23.2% of the variance and had an α
correlation of 0.88. Factor 2 contained 7 defenses that
accounted for 10.0% of the variance and had an α correlation
of 0.73. The 2 factors accounted for a total of 33.2% of
variance. Compared with the original 2-factor solution
described by Steiner et al , only 2 differences emerge:
sublimation was found to load on F2 in this sample, although
originally loading on F1; and suppression showed the
To explore the effects of maximizing the variance
accounted for by each factor, we used Varimax rotation on
the 3-factor solution explaining a total of 40.2% of variance
accounted for, a modest increase over the 33.2% in the 2-
factor solution. Factor 1 contained 12 defenses accounting
for 21.9% of the total variance. Factor 2 contained 5 defenses
accounting for 10.1% of the variance. Factor 3 contained 4
defenses accounting for 8.2% of the variance. The α
correlations for the 3 factors were 0.88, 0.73, and 0.66,
respectively. No defense in the 3-factor solution showed
secondary loadings greater than 0.30.
In addition, we tested the possibility that defenses could
cluster according to the more clinically/adult-based theore-
tical model found in other studies with the DSQ [19,21,33].
We therefore attempted to force a 4-factor solution; in the
Varimax rotated 4-factor solution, the total variance
explained was 46.3%, providing a further 6% increase in
variance explained. The first factor (23.3% of variance
explained) is similar in composition to F1 in the 2-factor
solution, with the exception of suppression (loading on F2)
and somatization (loading on F3). Factor 2 (10.0% of
variance) includes 4 defenses (humor, denial, idealization,
suppression), F3 (7% of variance) includes 3 defenses
(altruism, sublimation, somatization), and F4 (6% of
Factor structure resulting from the unrotated 2-factor solution of PCA carried out separately by age groups and sex
Young adol 13-16 y
(n = 530)
Late adol 17-20 y
(n = 575)
Adults 21 y and
older (n = 543)
(n = 628)
(n = 1020)
Steiner et al 
(n = 1875)
Denial (isolation of affect)
aEqual loadings on both F1 and F2.
372 A. Prunas et al. / Comprehensive Psychiatry 50 (2009) 369–377
variance) includes 2 defenses (intellectualization, reaction
formation). As the increase in variance explained was
negligible and 2 factors (F3 and F4) were composed only
of 3 and 2 defenses, the 2-factor solution proves preferable.
To test the stability of the retained factors, stability
coefficients of the factor scores across sex and age groups
(adolescents vs adults) were computed . Coefficients
greater than 0.90 are usually considered as indicating a good
level of correspondence among factors or components.
When 2 factors were extracted, very high replicability
coefficients were found (ie, greater than 0.94). The extraction
of 3 or more factors led to unreplicable factors across both
age and sex subgroups.
3.3. Sex and age differences in factor structure
groups (Table 4). Both F1 and F2 showed overall stability
across sex and age. Suppression, sublimation, and denial
(isolation of affect) showed higher loadings on F1 in the
adolescentsamples than in the adult sample.Sublimation and
denial (isolation of affect) showed equivalent loadings on F1
and F2 in the male sample.
Finally, a PCA was performed for 9 defenses showing
Cronbach α greater than 0.60 (Table 5). To determine how
many factors to retain, the same statistical guidelines used for
previous analyses were adopted.
Both the scree test and parallel analysis suggested a 2-
factor solution. The first factor includes 6 defenses
(projection, acting out, fantasy, conversion, suppression,
and withdrawal) and explains 24.8% of total variance
(Cronbach α = 0.79). The second factor includes 3 defenses
(idealization, humor, and altruism) and explains 15.7% of
total variance (Cronbach α = 0.70). No defense showed
significant secondary loadings.
Factor structure resulting from the unrotated 2-factor solution of PCA
carried out on 9 defenses with α values greater than 0.60
Factor 1Factor 2
Age and sex differences in the use of defenses by 1648 community subjects
Defenses Sex effect Age effectb
(n = 628)
(n = 1020)
Young adol 13-16 y
(n = 530)
Late adol 17-20 y
(n = 575)
Adults 21 y and older
(n = 543)
Factor 1 Acting out
Factor 2 Denial
Reaction formation 4.34 ± 1.56 4.03 ± 1.59 10.919 (=.001) 4.31 ± 1.61
Idealization6.11 ± 1.85 6.31 ± 1.76 11.292 (=.001) 6.50 ± 1.76A
6.64 ± 1.46 7.37 ± 1.22 79.137 (b.001) 6.91 ± 1.47A
Sublimation4.87 ± 1.50 5.20 ± 1.56 10.370 (=.001) 4.94 ± 1.55
Factor 14.26 ± 0.97 4.06 ± 1.01
5.25 ± 0.91 5.19 ± 0.83
4.45 ± 1.92 4.15 ± 1.79
5.71 ± 1.57 5.75 ± 1.78
4.12 ± 1.89 3.72 ± 1.84
3.67 ± 1.79 3.81 ± 1.80 16.102 (b.001) 4.06 ± 1.75A
4.50 ± 2.08 4.03 ± 2.033.293 (NS)
5.20 ± 1.65 4.47 ± 1.58 28.971 (b.001) 5.27 ± 1.62A
9.397 (=.002) 6.07 ± 1.63A
4.67 ± 1.95A
4.50 ± 1.78A
5.90 ± 1.68A
4.15 ± 1.82A
4.06 ± 1.81A
4.41 ± 2.05A
4.99 ± 1.59A
3.63 ± 1.63B
5.24 ± 1.69B
3.06 ± 1.61B
3.14 ± 1.68B
3.44 ± 1.84B
3.99 ± 1.45
4.41 ± 1.88A
4.78 ± 2.05A
3.10 ± 1.64 2.74 ± 1.55
4.28 ± 1.71 3.82 ± 1.77 15.036 (b.001) 4.23 ± 1.71A
5.42 ± 1.48 4.80 ± 1.55 29.052 (b.001) 5.39 ± 1.55A
4.74 ± 1.69 4.39 ± 1.890.191 (NS)
1.59 ± 1.19 1.34 ± 0.91 10.707 (=.001) 1.54 ± 1.15A
3.07 ± 1.75 4.16 ± 1.92 95.310 (b.001) 3.43 ± 1.89
5.40 ± 2.09 5.69 ± 2.09 5.249 (NS)
4.35 ± 2.06 3.93 ± 2.16 14.192 (b.001) 4.17 ± 2.15
4.39 ± 1.81 4.02 ± 1.78 17.856 (b.001) 4.27 ± 1.89
5.18 ± 1.70 4.88 ± 1.70 22.548 (b.001) 4.91 ± 1.76A
5.23 ± 1.45 4.53 ± 1.35 77.335 (b.001) 5.07 ± 1.46
1.593 (NS)3.25 ± 1.69A
3.17 ± 1.61A
4.22 ± 1.72A
5.26 ± 1.50A
4.84 ± 1.65A
1.51 ± 1.10A
3.64 ± 1.86
5.61 ± 2.08
4.14 ± 2.12
4.03 ± 1.81
4.93 ± 1.66
4.72 ± 1.42
4.13 ± 1.57
6.20 ± 1.79A/B
7.02 ± 1.39A
5.09 ± 1.46
4.31 ± 0.95A
5.16 ± 0.85A
2.20 ± 1.22B
3.52 ± 1.75B
4.45 ± 1.44B
3.60 ± 1.81B
1.26 ± 0.78B
4.16 ± 1.98
5.58 ± 2.05
3.98 ± 2.13
4.19 ± 1.70
5.15 ± 1.69
4.63 ± 1.37
4.00 ± 1.56
6.00 ± 1.82B
7.35 ± 1.18B
5.19 ± 1.62
3.66 ± 0.92B
5.22 ± 0.85A
5.13 ± 1.64A
5.55 ± 2.15
4.42 ± 0.97
5.27 ± 0.88A
NS indicates not significant.
aSex-age interaction effect was significant: repression: F(2,1642)= 8.003, P b.001; altruism: F(2,1642)= 6.086, P b.002; Factor: F(2,1642)= 4.305, P b.05.
bDifferent subscripts indicate a significant difference in scores between age groups, according to Bonferroni post hoc contrasts.
373 A. Prunas et al. / Comprehensive Psychiatry 50 (2009) 369–377
3.4. Age and sex differences in defensive array
To test for age and sex differences in defenses, we
performed multivariate analyses of variance by sex and age,
first with all 21 defenses entered simultaneously, then using
F1 and F2 scores. Post hoc analyses, using Bonferroni post
hoc contrasts, were carried out when significant univariate F
values emerged. The α levels were corrected according to the
Bonferroni procedure for multiple contrasts. Results at the
multivariate level indicated a significant effect of age (Wilks
λ = 0.746, F(42,3244)= 12.214, P b .001), sex (Wilks λ =
0.782, F(21,1622)= 21.524, P b .001), and their interaction
(Wilks λ = 0.960, F(42,3244)= 1.593, P b .05) on REM-71
subscales mean scores; as for F1 and F2 scores, sex effect
was not significant (age: Wilks λ = 0.913, F(4,3282)= 38.380,
P b .001; sex: Wilks λ = 0.998, F(2,1641)= 1.492, P N .05;
interaction: Wilks λ = 0.992, F(4,3282)= 3.5417, P b .05).
We found significant age and sex differences for most
defenses in F1 and F2 (Table 6). For F1, female subjects
reported greater use of splitting, dissociation, and somatiza-
tion, whereas male subjects reported greater use of passive
aggression, omnipotence, repression, suppression, and con-
version. For F2, female subjects reported greater use of
altruism, idealization, and sublimation, whereas male
subjects reported greater use of denial, humor, intellectua-
lization, and reaction formation. No sex effect was found for
total scores on F1 or F2.
With respect to age, there was a significant effect for all
but 3 of the F1 defenses (withdrawal, somatization, and
suppression), with adolescents reporting higher mean scores
for 11 of 14 defenses and the F1 total score compared with
adults. There was a significant age effect for F2 defenses
including humor, altruism, and idealization. Altruism
showed the expected variation of higher scores for older
subjects, whereas idealization showed the opposite pattern.
No defense showed significant differences in mean scores
between early and late adolescents.
3.5. Satisfaction with life and psychopathology
To examine the relationship between defenses and indices
of well-being in life domains and psychologic distress, a
series of hierarchical multiple regression analyses with
stepped variable entry was used, with age and sex entered as
a dyad (step 1), F1 and F2 scores as predictors (step 2), and
each measure of satisfaction and the GSI as dependent
variables (Table 7).
In all domains examined, defenses accounted for a
significant proportion of the variance, with the total amount
of variance ranging between 5.4% (family) and 11.5%
(overall satisfaction) for measures of satisfaction, and 34.2%
higher proportion of variance than that of age or sex for all
criterion measures. For all measures of satisfaction, higher F2
and lower F1 scores were associated with higher levels of
and lower use of F2 defenses predicted higher GSI scores.
Results of hierarchical regression analyses predicting life satisfaction and symptom severity by sex and age and by F1 and F2 scores
Model F (df)
374 A. Prunas et al. / Comprehensive Psychiatry 50 (2009) 369–377
This study replicates in a large community sample of
Italian adolescents and adults findings from the original
American study yielding a 2-factor classification of the 21
core defenses assessed in the REM-71 . The orthogonal
nature of these 2 factors is preserved in the PCA, an
important reason for the selection of this factor solution.
Additional support for the 2-factor solution is provided by
parallel analysis, the overall stability of both factors in
analyses with subjects divided by sex and age, as well as the
age effects and consistent correlations with indicators of
psychologic well-being and symptoms of psychopathology.
Such a finding calls for a careful reconsideration of
nomenclature, as suggested by Cramer . We have avoided
labeling our 2 factors based on our belief that there is
incomplete current evidence to support a simple division of
defenses into mature vs immature. However, we also note
that the 2 factors seem to predict, in experimental samples,
the activation of negative (F1) and positive (F2) emotions
during standardized stressors . Iffurther confirmed, these
patterns would suggest a nomenclature more in line with
recent findings from cognitive neuroscience [3,16,35].
4.1. Difficulties associated with the REM-71 factor structure
Although the current results are encouraging in that they
provide support for the factor solution reported in the
American sample , several problematic items and
First of all, some of the items showed low correlations
with the other items in the same scale. It seems advisable to
improve the internal consistency of some of the scales by
replacing deleted items (ie, denial) or finding new suitable
items (ie, sublimation).
Furthermore, the defense of sublimation loaded on F2,
consistent with the DSQ , whereas in the original
validation, sublimation loaded on F1 . By contrast,
suppression loaded on F2 in the American study, also
consistent with the DSQ, but in the current study loaded on
F1. In looking more closely at suppression, we found that
suppression loaded on F1 in younger subjects (13-20 years)
but shifted to F2 in the adult sample (Table 4). It is possible
that the alignment of suppression as an immature defense
may be explained by the relatively high proportion of
adolescents in the sample. Suppression is defined as dealing
“with emotional conflicts or internal or external stressors by
intentionally avoiding thinking about disturbing problems,
wishes, feelings or experiences” ( p813). We propose that
it is possible that in younger subjects the intentional,
deliberate nature of this defense cannot be fully appreciated,
thus making it similar to repression (ie, “expelling
disturbing wishes, thoughts, or experiences from conscious
awareness”) ( p813), for which the highest loading is on
F1. The defense of denial (isolation of affect) loads on F2 as
a mature defense in both studies, a finding that is in conflict
with traditional views regarding this defense. However,
Steiner et al  have noted that this item refers not to
“psychotic denial,” but rather to denial of affective
responses (ie, isolation of affect) that has been associated
with positive outcomes . Future studies should address
the specific shortcomings and inconsistencies regarding
Nonetheless, despite these limitations, strong support for
the validity of the 2-factor structure of the REM-71 has now
been provided in 2 large cross-cultural community samples.
4.2. Sex and age differences
Although no significant sex differences were found in
the organization of defenses in the 2-factor solution, there
were important differences in the frequency of use of 15
of 21 specific defenses. The defenses endorsed more
strongly by female subjects reflect an internalizing
(somatization) and relational (altruism, sublimation)
defense style, with the predominance of defenses such as
dissociation, idealization, and splitting. By contrast,
defenses endorsed more strongly by male subjects include
the repression of unwanted ideas, ignoring emotional
states (repression, reaction formation, conversion, suppres-
sion, denial, humor, intellectualization), and the regulation
of self-esteem by establishing dominance and control over
the environment (passive aggression, omnipotence). Pre-
vious studies on sex differences in defense style have
provided conflicting findings. Several studies have shown
no differences between male and female subjects [10,14],
whereas others do find a sex difference. Watson and Sinha
, for example, found that male subjects are more likely
to use isolation and suppression, whereas female subjects
are more likely to use “pseudoaltruism,” consistent with
the results of our study. In general, the results are
consistent with the original findings of Steiner et al ,
although, in the American sample, female subjects showed
higher scores in undoing and reaction formation and there
was no sex effect for conversion and humor. Differences
in scores between the American and Italian samples for
some defenses (ie, humor) may be indicative of cross-
cultural factors. Watson and Sinha  reported differ-
ences in defenses between Australian and Canadian
subjects and suggested socialization patterns as a possible
mechanism. Cultural environments may interact with
normative, ubiquitous processes differentially, reinforcing
or prohibiting the use of certain defenses.
With respect to age, significant differences emerged when
comparing the 3 age groups. Results demonstrated a decline
in the use of F1 defenses, in line with theoretical expectations
and previous studies [10,38]. Three of the 7 defenses in F2
showed a significant age effect. Idealization decreased in
magnitude with age, perhaps reflecting the sense of growing
realism and loss of innocence with adulthood. Altruism, by
contrast, increased in frequency with age, consistent with
other studies . Comparisons between early and late
375A. Prunas et al. / Comprehensive Psychiatry 50 (2009) 369–377
adolescents showed no significant differences on the level of
endorsement of any of the 21 defenses as well as on F1 and
F2 scores. This result is only partially at odds with previous
findings showing that defensive array tends to remain
globally stable during adolescence and that major changes
seem to occur in the transition from adolescence into
adulthood. Feldman et al , for example, using a modified
version of the DSQ, found out that differences between early
and late adolescents were few in number and modest in
magnitude. Ruuttu et al  reported significant differences
in only 2 (ie, rationalization and splitting) of 20 defenses
using the 40-item version of the DSQ.
4.3. Satisfaction with life and psychologic distress
Finally, a significant association was found between
defenses and self-reported indexes of psychosocial adjust-
ment and psychologic distress. Increased use of F2 defenses
was associated with greater satisfaction in all domains of life,
whereas use of F1 defenses predicted lower rates of
satisfaction. These findings are consistent with those of
Steiner et al  and previous studies . Factor 1 defenses
were also associated with GSI scores of psychologic distress;
the total amount of variance explained by F1 and F2 scores
alone (34.2%) is comparable with that found in previous
studies using the DSQ in community samples . Numerous
authors have demonstrated a clear correlation between
severity of psychopathology and maladaptiveness of
defenses in different clinical populations , whereas
studies in nonclinical samples suggest that the link between
defenses and psychopathology exists even for subclinical
forms of psychopathologic distress. Holi et al  found that
3 DSQ defense styles could explain 32.4% of Symptom
Checklist–90–Revised GSI score, whereas for the 20
defenses the total amount of variance explained was
40.5%. More recently, Ruuttu et al  found a significant
positive association between DSQ-40 immature defenses
and GHQ-36 total score (r = 0.65) and a negative association
between mature defenses and GHQ-36 total score (r = −0.34)
in a nonclinical sample of Finnish adolescents. These results
suggest that immature defenses may be considered a
psychologic marker for faulty self-regulation and vulner-
ability to stress and psychopathology.
This study suffers from several methodological limita-
tions. Besides the shortcomings previously mentioned, it has
been pointed out  that the use of a self-report strategy for
the assessment of defenses is itself a crucial factor that affects
considerably the way in which the individual self-reports on
other outcome measures; therefore, the only way to
adequately demonstrate the relationship between defense
and psychologic well-being or psychopathology should be
an independent assessment of the outcome measures. Future
studies might address this issue by comparing the self-report
of defenses with clinical ratings of psychopathologic severity
or overall level of functioning (eg, Diagnostic and Statistical
Manual of Mental Disorder, Fourth Edition, Text Revision
Axis V). Moreover, the cross-sectional design of this study
does not allow clarification of the causal relationship
between defenses and psychopathology; also, what might
appear to be age effects in the endorsement of defenses may
actually be explained as a combination of age and cohort
effects. As Michael Bond ( p272) has pointed out, “the
ideal study would be prospective and longitudinal, would
measure defenses in childhood, in adolescence, and at
several points in adulthood, and would note whether there
were significant correlations between preexisting defenses
and specific illnesses. Such a study has yet to be under-
taken.” Finally, future research should focus on clinical
samples to gain further support to the stability of factor
structure in psychiatric patients and provide data on the
predictive validity of the REM-71.
Despite these limitations, our findings provide further
evidence for the reliability and validity of the REM-71 for
the assessment of defenses in adults and adolescents.
Although the full version of the REM-71 needs some
revisions and integrations to improve its psychometric
properties, we propose here a short version (9 defenses, 28
items) of the scale to be adopted as a useful screening
instrument for detecting both individual's resources (F2) and
vulnerability for psychologic distress (F1).
The authors are grateful to Irene Sarno, PhD, for support
in data collection and to Helena Kraemer, PhD, Professor of
Biostatistics and Design, Stanford University, for expert
consultation and review of the statistical analyses.
 American Psychiatric Association. Diagnostic and statistical manual of
mental disorders. 4th ed. text revision. Washington, DC: American
Psychiatric Association; 2000 (DSM-IV-TR).
 Cramer P. Defense mechanisms in psychology today. Am Psychol
 Gross JJ, Thompson RA. Emotion regulation: conceptual foundations.
In: Gross JJ, editor. Handbook of emotion regulation. New York: The
Guilford Press; 2007. p. 3-26.
 Steiner H. Freud against himself. Perspect Biol Med 1977;20(4):
 Holi MM, Sammallahti PR, Aalberg VA. Defense styles explain
psychiatric symptoms: an empirical study. J Nerv Ment Dis 1999;187
 MurisP, WinandsD,HorselenbergR. Defensestyles,personalitytraits,
and psychopathological symptoms in non-clinical adolescents. J Nerv
Ment Dis 2003;191:771-80.
 Bond M, Perry JC, Gautier M, Goldenberg H, Oppenheimer J,
Simand J. Validating the self-report of defense styles. J Personal
 Sammallahti P. Evaluating defenses with the Defense Style Ques-
tionnaire: a review. Psychiatria Fenica 1997;28:145-7.
 Sammallathi P, Aalberg V, Pentisaari JP. Does defense style vary with
severity of mental disorder? An empirical assessment. Acta Psychiatr
376A. Prunas et al. / Comprehensive Psychiatry 50 (2009) 369–377
 Andrews G, Singh M, Bond M. The Defense Style Questionnaire.
J Nerv Ment Dis 1993;181:246-56.
 BondMP, GardnerS, Christian J, SigalJJ. Empirical studyof self-rated
defense styles. Arch Gen Psychiatry 1983;40:333-8.
 Bond M, Vaillant JS. An empirical study of the relationship between
diagnosis and defense style. Arch Gen Psychiatry 1986;42:285-8.
 Erickson SJ, Feldman S, Steiner H. Defense reactions and coping
strategies in normal adolescents. Child Psychiatry Hum Dev 1997;28:
 Muris P, Merckelbach H. The short version of the Defense Style
Questionnaire: factor structure and psychopathological correlates. Pers
Individ Dif 1996;20(1):123-6.
 Soldz S, Vaillant GE. A 50-year longitudinal study of defense use
among inner city men: a validation of the DSM-IV defense axis. J Nerv
Ment Dis 1998;186(2):104-11.
 Ochsner KN, Gross JJ. The neural architecture of emotion regulation.
In: Gross JJ, editor. Handbook of emotion regulation. New York: The
Guilford Press; 2007. p. 87-109.
 Perry JC, Hoglend P. Convergent and discriminant validity of overall
defensive functioning. J Nerv Ment Dis 1998;186:529-35.
 Vaillant GE, Bond M, Vaillant CO. An empirically validated hierarchy
of defense mechanisms. Arch Gen Psychiatry 1986;43:786-94.
 Bond M. Empirical studies of defense style: relationships with
psychopathology and change. Harv Rev Psychiatry 2004;12(5):
 Yasnovsky J, Araujo K, King M, et al. Defenses in school age children:
children's versus parents’ report. Child Psychiatry Hum Dev 2003;33
 San Martini P, Roma P, Sarti S, Lingiardi V, Bond M. Italian version of
the Defense Style Questionnaire. Compr Psychiatry 2004;45(6):
 Andrews G, Pollock C, Stewart G. The determination of defense style
by questionnaire. Arch Gen Psychiatry 1989;46:455-60.
 Reister G, Fellhauer RF, Franz M, et al. Psychometric measurements of
ego defense mechanisms: correlation between questionnaire and expert
rating. Psychother Psychosom Med Psychol 1993;43:15-20.
 Rutherford MJ, McDermott P, Cacciola JS, Alterman AI, Mulvaney F.
A psychometric evaluation of the Defense Style Questionnaire in
methadone patients. J Personal Disorder 1998;12:119-25.
 Bonsack C, Despland JN, Spagnoli J. The French version of the
Defense Style Questionnaire. Psychoth Psychosomat 1998;67:24-30.
 Feldman S, Araujo K, Steiner H. Defense mechanisms in adolescents
as a function of age, gender, and mental health status. J Am Acad Child
Adolesc Psychiatry 1996;35:1344-54.
 Araujo KB, Medic S, Yasnovsky J, Steiner H. Assessing defense
structure in school-age children using the Response Evaluation
Measure-71–Youth Version (REM-Y-71). Child Psychiatry Hum Dev
 Steiner H, Araujo KB, Koopman C. The Response Evaluation Measure
(REM-71): a new instrument for the measurement of defenses in adult
and adolescent. Am J Psychiatry 2001;158:467-73.
 Vaillant GE. Mental health. Am J Psychiatry 2003;160(8):1373-84.
 Prunas A, Baio A, Pozzoli S, Cioffi G, Madeddu F. La versione italiana
della Response Evaluation Measure–71 (REM-71): risultati prelimi-
nari [The Italian version of the Response Evaluation Measure–71
(REM-71): preliminary results]. Ricerche di Psicologia 2006;4:75-92.
 Derogatis LR. SCL-90-R. Administration, scoring and procedures
manual. Baltimore, MD: Psychometric Research; 1983.
 Conti L. Repertorio delle scale di valutazione in psichiatria. Rome:
Società Editrice Universo; 2001.
 Ruuttu T, Pelkonen M, Holi M, et al. Psychometric properties of the
defense style questionnaire (DSQ-40) in adolescents. J Nerv Ment Dis
 Everett JE. Factor comparability as a means of determining the number
of factors and their rotation. Multivariate Behav Res 1983;18:197-218.
 Steiner H, Erickson SJ, MacLean P, Medic S, Plattner B, Koopman C.
Relationship between defenses, personality, and affect during a stress
task in normal adolescents. Child Psychiatry Hum Dev 2007;38(2):
 Bar-On D. Different types of denial (isolation of affect) account for
short- and long-term recovery of coronary patients. Isr J Psychiatry
Relat Sci 1985;22:155-72.
 Watson DC, Sinha BK. Gender, age, and cultural differences in the
Defense Style Questionnaire-40. J Clin Psychol 1998;54(1):67-75.
 Vaillant GE. Adaptation to life. Boston (Mass): Little, Brown; 1977.
 Feldman SS, Araujo K, Steiner H. Defense mechanisms in adolescents
as a function of age, sex, and mental health status. J Am Acad Child
Adolesc Psychiatry 1996;35(10):1344-55.
 Vaillant GE. Ego mechanisms of defense and personality psycho-
pathology. J Abn Psychol 1994;103:44-50.
 Cramer P. Coping and defense mechanism: what’s the difference?
J Pers 1998;66(6):921-46.
377A. Prunas et al. / Comprehensive Psychiatry 50 (2009) 369–377