+Model

RPSM-524; No. of Pages 10 ARTICLE IN PRESS

Revista de psiquiatría y salud mental (Barcelona) xxx (xxxx) xxx---xxx

www.elsevier.es/saludmental

ORIGINAL ARTICLE

Spanish validation of the Detachment and

Compartmentalization Inventory (DCI) in a community
and clinical sample. A new instrument for measuring
dissociation
Salvador Perona-Garcelán a,∗,1 , Gabriel Rodenas-Perea a,1 ,
Elena Velasco-Barbancho a,1 , Cristina Senín-Calderón b,2 , Juan F. Rodríguez-Testal c,3 ,
Rosa Moreno-Buzón a,1 , Miguel Ruiz-Veguilla a,4 , Benedicto Crespo-Facorro a,4

a
University Hospital Virgen del Rocío, Seville, Spain
b
Department of Psychology, University of Cádiz, Cádiz, Spain
c
Personality, Evaluation and Psychological Treatment Department, University of Seville, Seville, Spain

Received 1 August 2020; accepted 3 December 2020

KEYWORDS Abstract
Dissociation; Introduction: Dissociative symptoms are a type of phenomenon which is present in a wide vari-
Detachment; ety of psychopathological disorders. It is therefore necessary to develop scales that measure
Compartmentalization; this type of experience for therapy and research. Starting out from the bipartite model of dis-
Assessment; sociation, this study intended to adapt and validate the Detachment and Compartmentalization
Inventory Inventory (DCI) in Spanish.
Material and methods: For this, 308 participants (268 from the community population and 40
with psychiatric pathology) completed the DCI, the Dissociative Experiences Scale (DES-II), the
Somatoform Dissociation Questionnaire (SDQ20) and the Mindfulness Attention Awareness Scale
(MAAS).
Results: The results showed that the Spanish version has a two-factor structure similar to the
original version and was invariant across participants. The reliability of DCI scores was adequate
and acquired evidence of validity related to other instruments.

∗ Corresponding author.
E-mail address: [email protected] (S. Perona-Garcelán).
1 Virgen del Rocío Outpatient Mental Hospital, Andalusian Health-Care Service, Avda. Jerez, s/n, 41013 Seville, Spain.
2 Department of Psychology, University of Cádiz, Spain; Ave. República Árabe Saharaui S/N. 11510, Puerto Real, Cádiz, Spain.
3 Personality, Evaluation and Psychological Treatment Department, University of Seville, Camilo José Cela, SN, 41018 Seville, Spain.
4 Virgen del Rocío University Hospital/ University of Seville/ IBiS/ CIBERSAM, Avda. Manuel Siurot, s/n, 41013 Seville, Spain.

https://doi.org/10.1016/j.rpsm.2020.12.004
1888-9891/© 2020 SEP y SEPB. Published by Elsevier España, S.L.U. All rights reserved.

Please cite this article in press as: S. Perona-Garcelán, G. Rodenas-Perea, E. Velasco-Barbancho et al., Spanish validation
of the Detachment and Compartmentalization Inventory (DCI) in a community and clinical sample. A new instrument for
measuring dissociation, Revista de psiquiatría y salud mental (Barcelona), https://doi.org/10.1016/j.rpsm.2020.12.004
+Model
RPSM-524; No. of Pages 10 ARTICLE IN PRESS
S. Perona-Garcelán, G. Rodenas-Perea, E. Velasco-Barbancho et al.

Conclusions: It is concluded that the DCI is a valid scale for detecting detachment and com-
partmentalization dissociative experiences, both in the clinic and research.
© 2020 SEP y SEPB. Published by Elsevier España, S.L.U. All rights reserved.

PALABRAS CLAVE Validación española del Inventario de desapego y compartimentación (DCI) en una
Disociación; muestra comunitaria y clínica. Un nuevo instrumento para medir disociación
Desapego;
Resumen
Compartimentación;
Introducción: Los síntomas disociativos son un tipo de fenómeno que están presentes en una
Evaluación;
amplia variedad de trastornos psicopatológicos. Por lo tanto, es necesario desarrollar escalas
Inventario
que midan este tipo de experiencia para terapia e investigación. Partiendo del modelo bipartito
de disociación, este estudio tuvo como objetivo principal adaptar y validar el Inventario de
Desapego y Compartimentación (DCI) en español.
Material y métodos: Participaron 308 personas (268 de la población general y 40 con patología
psiquiátrica). Completaron el DCI, la Escala de Experiencias Disociativas (DES-II), el Cuestionario
de disociación somatoforme (SDQ20) y Mindful Attention Awareness Scale (MAAS).
Resultados: Los resultados mostraron que la versión en español tiene una estructura de dos
factores similar a la versión original y fue invariante a través de los participantes. La fiabilidad
de las puntuaciones del DCI fueron adecuadas y se hallaron evidencias de validez en relación
con otras escalas.
Conclusiones: Se concluye que el DCI es una escala válida para detectar experiencias disocia-
tivas de desapego y compartimentación, tanto en la clínica como en la investigación.
© 2020 SEP y SEPB. Publicado por Elsevier España, S.L.U. Todos los derechos reservados.

Introduction Maaranen et al.5 found that the prevalence of pathologi-

cal dissociation in the community population was 3.4%. In a
Dissociation is considered a breakdown or lack of conti- recent review by Brand et al.,6 the rate of dissociative dis-
nuity between psychological processes that are normally orders in hospitalized psychiatric patients varied from 1% to
integrated, such as consciousness, memory, identity, per- 20.7% and from 12% to 29% in the outpatient context. Ross
ception and motor control.1 Other authors, such as Putnam,2 et al.7 found a lifelong prevalence of dissociative disorders
have also defined dissociation as a process which produces in psychiatric patients of 44.5%.
an alteration in an individual’s thoughts, feelings or actions There are presently two main models which attempt to
such that during a period of time, certain information is explain its etiology. The first postulates that dissociation is
not associated or integrated with other information, caus- a product derived from traumatic experiences (traumato-
ing a series of phenomena such as alteration of memory or genic model, e.g., Bailey and Brand8 ). It is based on studies
identity. which have found that this type of adverse childhood experi-
Dissociative disorders are included in the major diag- ence, such as sexual or physical abuse, is associated with the
nostic classifications (DSM-5 and ICD-11), providing better appearance of dissociative symptoms in adulthood. The sec-
agreement with each other in their organization than ond model (sociocognitive or fantasy model) proposes that
in previous editions. For example, the ICD-113 includes dissociative disorders are the consequence of social learning
Depersonalization-derealization disorder in this chapter of and of expectations, more specifically, hypothesizing that
dissociative disorders. Dissociative fugue is also framed as factors, such as inadvertent signals or cuing from the ther-
a specification of Dissociative amnesia. These changes were apist, influence of communication media and sociocultural
not in the previous version of the ICD. However, some dif- expectations would explain the symptomatology associated
ferences between the two classifications should also be with dissociative identity disorder (DID9 ). However, recent
highlighted. The ICD-11 includes conversion disorder in this reviews in which the predictive validity of these two models
chapter (as Dissociative neurological symptom disorder) and is compared have found that there is less scientific evidence
not within the Somatic symptoms and related disorders, as for the sociocultural model,8 although this debate is not
does the DSM-5.4 Other differences refer to the classifica- closed.10
tion Trance disorder, Possession trance disorder, and Partial Dissociation has traditionally been conceptualized as a
dissociative identity disorder, which in the DSM-5 are framed phenomenon on a continuum.11 Thus, there would be a
in a residual category (Other specified dissociative disor- structural dimension from occasional and normal experi-
ders). All of these differences show the lack of consensus ences, for example, states of absorption and daydreaming,
in definition and classification of this type of phenomenon. to severe psychiatric disorders such as DID. However, this

2
+Model
RPSM-524; No. of Pages 10 ARTICLE IN PRESS
Revista de psiquiatría y salud mental (Barcelona) xxx (xxxx) xxx---xxx

continuum hypothesis has been questioned by some authors. Material and method
Cardeña12 identified three dissociation categories, and later,
authors such as Brown13 and Holmes et al.14 proposed a Participants
bipartite model of dissociative disorders. This theoretical
framework proposes two main categories for classify-
The study sample was made up of 394 participants of whom
ing dissociation, detachment and compartmentalization.
337 were from the community population and 57 were
Detachment refers to an altered state of consciousness in
from a clinical population. The community population was
which there is a sensation of separation of the self or
comprised of university students and non-university parti-
from surroundings. This category includes absorption, some
cipants recruited by snowball sampling. The clinical group
types of depersonalization, derealization, affective flat-
was made up of patients under treatment by the mental
tening and out-of-body experiences. On the other hand,
health services of the Virgen del Rocio Hospital in Seville
compartmentalization is related more to a deficit in control
(Spain) by clinical referral. The diagnoses were made follow-
of processes or actions that normally are under one’s con-
ing ICD-10 criteria.22 In both groups, the inclusion criterion
trol, including the inability to recall information that would
was that they must be over 18 years of age, and in the
normally be remembered. Dissociative amnesia, conver-
clinical group, they had to be under treatment in the pub-
sion and somatoform symptoms and DID would be placed
lic mental health services for a psychiatric disorder. Any
here.14
participant who did not meet the DCI scale validity cri-
To date, a wide variety of scales and instruments have
teria, that is have a score of 0 on Items 8 and 15 (86
been developed to measure dissociative experiences. All of
participants, 69 (20.4%) from the general population and
them focus on evaluating psychological or somatic aspects
17 (29.8%) from the clinical population) was excluded.
of this construct, such as, e.g., the Dissociative Experience
The final sample consisted of 308 participants: 268 were
Scale (DES15 ), the Somatoform Dissociation Questionnaire-
from the community population, with a mean age of 31.53
2016 and Fisher’s Dissociative Ability Scale17 or specific
(SD = 13.44), of whom 117 were men (43.7%) and 151 were
symptoms, such as the Cambridge Depersonalization Scale
women (56.3%), and the 40 participants in the clinical
(CDS18 ). However, until recently, no instrument has been
group had a mean age of 32.58 (SD = 12.76), of whom 13
developed that evaluates dissociation from the viewpoint
were men (32.5%) and 27 were women (67.5%). No differ-
of the bipartite model. Mazzoti et al.,19 based on a con-
ences in age (t(306) = .46, p = .65) or sex (2 (1) = 1.78, p = .18)
firmatory factor analysis, found a three-factor structure
were found between the general and clinical population
of the DES-II,20 two of which would correspond to the
groups.
two types of dissociative symptoms proposed by the bipar-
Participants excluded from the general population dif-
tite model. Recently, Butler et al.21 developed the DCI
fered significantly from those who were not in age
(Detachment and Compartmentalization Inventory), which
(t(335) = 4.12, p < .05), but not sex (2 (1) = 3.14, p > .05). In
is a specific scale for measuring dissociation based on the
particular, the participants in the group excluded were sig-
bipartite model. The English version of this scale shows
nificantly older than those who were not excluded from
good psychometric properties, that is, it has good inter-
the study (39.19 and 31.53 years old, respectively). No
nal consistency, and evidence of convergent, discriminant
significant age (t(54) = 1.68, p > 0.05) or sex (2 (1) = 2.28,
and concurrent validity. Furthermore, the exploratory fac-
p > 0.05) differences were found in the clinical group
tor analysis showed a two-factor solution, which clearly
between those excluded and those who were not. Table 1
corresponded to the detachment and compartmentalization
presents the demographic characteristics of the study sam-
subscales.
ple.
The distinction between the two types of dissociative
phenomena is very important to research, knowledge of the
etiology of the dissociative phenomena and its treatment. It
is therefore necessary to develop psychometric instruments Measures
that capture both types of dissociation. This is why this arti-
cle presents the Spanish adaptation of the DCI scale and the Detachment and Compartmentalization Inventory (DCI)21
study of its psychometric properties. The specific objectives This is a self-report inventory with 22 items, designed to
of the study were: specifically measure dissociative detachment experiences
(10 items, for example, ‘‘What I see looks ‘flat’ or ‘lifeless’,
as if I am looking at a picture’’) and compartmentalization
(10 items, for example, ‘‘I do not feel in control of what
• Validate the DCI in a clinical and nonclinical Spanish pop- my body does as if there is someone or something inside
ulation. me directing my actions’’) when not under the effects of
• Confirm the internal structure of the two factors in the alcohol or drugs. Furthermore, it also has two items for mea-
original version by Butler et al.21 suring the validity of the responses (for example, ‘‘I cross
• Analyze measurement invariance across participants the street where there is no pedestrian crossing or cross-
(clinical and nonclinical population). walk, i.e., jaywalk’’). Each of the items is answered on an
• Estimate the reliability of the DCI scores and find evidence eight-point Likert-type scale which measures the frequency
of a validity-based relationship with other variables. of these experiences (0: never, 7: daily). In the original study
• Find a cutoff point for the DCI subscales that discrimi- by Butler et al.,21 the Cronbach’s alpha for the total scale
nates between a population with and without psychiatric score was .97, for the detachment subscale .93, and for the
pathology, and the scale’s sensitivity and specificity. compartmentalization subscale it was .96.

3
+Model
RPSM-524; No. of Pages 10 ARTICLE IN PRESS
S. Perona-Garcelán, G. Rodenas-Perea, E. Velasco-Barbancho et al.

Table 1 Participant demographic characteristics.

Characteristic Community population Clinical population
N 268 40
Gender
Men 117 (43.7%) 13 (32.5%)
Women 151 (56.3%) 27 (67.5%)
Age
Mean (SD) 31.53 (13.44) 32.58 (12.76)
Education
University 122 (45.5%) 12 (30%)
Vocational training 52 (19.4%) 5 (12.5%)
High school 43 (16.1%) 6 (15%)
Middle school 38 (14.2%) 7 (17.5%)
Elementary school 11 (4.1%) 8 (20%)
No formal education 2 (0.7%) 2 (5%)
Marital status
Single 170 (63.4%) 29 (72.5%)
Married/stable couple 89 (33.2%) 8 (20%)
Divorced 6 (2.2%) 3 (7.5%)
Widowed 3 (1.1%) 0
Diagnoses
Psychotic disorder 0 13 (32.5%)
Eating disorder 0 11 (27.5%)
Mood disorder 0 7 (17.5%)
Dissociative disorder 0 4 (10%)
Anxiety disorder 0 3 (7.5%)
Personality disorder 0 2 (5%)

Dissociative Experience Scale-II (DES-II)20 Mindful Attention Awareness Scale (MAAS)26

This self-report scale is made up of 28 items designed to It is a self-report scale based on a single factor of full
measure dissociative experiences in clinical and nonclini- awareness, which evaluates the dispositional capacity for
cal populations when the subject is not under the effects awareness or attention to experience at the moment in daily
of alcohol or drugs. This study used the Spanish version life. The Spanish version by Soler et al.27 was used. It is a
by Icarán et al.23 The items are answered on a percent- 15-item self-report scale scored on a Likert scale of 1---6 (1:
age scale which measures the frequency the item described almost always, 6: almost never), for example, ‘‘I could be
occurs in the subject’s daily life (for example, ‘‘Some people experiencing some emotion and not be conscious of it until
have the experience of feeling that other people, objects, sometime later.’’ The Cronbach’s alpha in this study was.88.
and the world around them are not real’’), where 0% is
never and 100% is always. For our purposes, the total scale
score and the sum of the items that make up the DES- Procedure
Taxon (3, 5, 7, 8, 12, 13, 22, 27) were used to measure
pathological dissociation.24 The Cronbach’s alpha in this Permission was requested from the authors for adaptation
study was .94 for the total scale score and .84 for the DES- and translation of the scale. The items in the original version
Taxon. of the DCI were translated into Spanish following the recom-
mendations of Muñiz et al.,28 using back-translation, with
two translators, one familiar with the Spanish culture and
the other familiar with the Anglo-Saxon culture. The first
Somatoform Dissociation Questionnaire (SDQ-20)16 translated the questionnaire into Spanish and this version
This scale is a self-report inventory designed to measure was then translated back into English by the second trans-
somatoform dissociation in 20 items (for example, ‘‘I can- lator. These two versions were compared with the original
not see for a while, as if I am blind’’) rated on a five-point English version for greater accuracy.
Likert-type scale (1: this applies to me not at all; 5: this The scales were applied to a sample of participants
applies to me extremely). In this study, the Spanish version from the community population and a sample of partici-
by González-Vázquez et al.,25 which has good psychometric pants with psychiatric pathology who were users of the
properties, was used. The Cronbach’s alpha in this study was public mental health services in Seville (Spain). The com-
.86. munity population sample was sent an email with a link

4
+Model
RPSM-524; No. of Pages 10 ARTICLE IN PRESS
Revista de psiquiatría y salud mental (Barcelona) xxx (xxxx) xxx---xxx

created with Google forms in which they were explained and comparative fit index (CFI), which had to be > .90,29
the objectives of the study, an informed consent sheet, and were employed to check the overall fit of the models. In
a sheet on which they were asked for their sociodemographic addition to these indices, the root mean square error of
and basic clinical information, and finally, the scales, in approximation (RMSEA) and its confidence interval (CI) at
this order: DCI, SDQ 20, DES-II and MAAS. Each participant 90%, which for good fit must be < .06, and the standardized
was assigned an alphanumeric code which was known only root mean square residual (SRMR), which had to be < .0830
to the study researchers and no personal information was were also calculated.
requested (name and last name) to safeguard anonymity. The measurement invariance across groups (clinical and
Fifteen days after the answers to the questionnaires had general populations) was tested using multi-group CFA.
been received, they were sent a second email with another First, the fit indices of the DCI model were tested in each
link with the DCI scale for the retest. The participants in group separately (clinical population vs general population).
the clinical sample were selected by their clinical referrals Measurement invariance was evaluated on three levels:
(clinical psychologist and psychiatrist) and were given the configural, metric and scalar.31 The fit of nested models
tests individually under adequate lighting and noise condi- was evaluated by comparing the CFI (CFI) and RMSEA
tions. Before the scales were applied, all the participants (RMSEA), using the recommended cut-off value of .01 for
in the clinical sample were also explained the objectives of CFI and .015 for RMSEA.32
the study and filled in the informed consent sheet. The tests Evidence Based on Relations to Other Variables were stud-
were applied by two resident clinical psychologists trained ied with the Pearson correlation coefficient. Reliability DCI
in application of the scales. The order in which the tests scores was analyzed with the ordinal alpha to calculate
were implemented and the time to retest were similar to internal consistency and test---retest reliability at 15 days
the community population group. applying the test with the Pearson correlation coefficient.
Finally, the ROC curve was calculated to measure the sen-
sitivity and specificity of the two DCI subscales and the total
Data analysis score.

Data analyses were performed with the SPSS v26 and LIS-
REL v8.7 programs. A descriptive analysis was made of the Results
sociodemographic variables as well as an analysis of skew-
ness and Kurtosis of the items on the scale. A confirmatory Preliminary analysis
factor analysis (CFA) was performed to calculate construct
validity with robust maximum likelihood estimation (RML) to Table 2 shows the means, standard deviations, skewness
test the suitability of the internal test structure. The Satorra and kurtosis of each of the items on the DCI scale. The
Bentler Scaled Chi-Square, non-normalized fit index (NNFI) mean score found in detachment and compartmentalization

Table 2 Descriptive statistics of the items on the DCI scale.

Community population Clinical population

Items Mean SD Skewness Kurtosis Mean SD Skewness Kurtosis

1 2.77 2.14 .25 −1.11 4.75 2.09 −1.06 .30
2 .66 1.33 2.41 5.77 3.03 2.81 .18 −1.63
3 3.80 2.06 −.22 −.90 5.28 2.05 −1.43 1.27
4 1.78 2.04 .89 −.44 4.03 2.25 −.43 −1.08
5 1.01 1.65 1.67 2.05 3.88 2.49 −.42 −1.20
6 .19 .75 5.36 32.82 2.08 2.53 .81 −.93
7 .82 1.47 2.10 4.06 3.45 2.65 −.16 −1.61
8 4.65 1.70 −.35 −.82 4.45 1.87 −.39 −.92
9 .80 1.38 1.99 3.55 3.33 2.40 −.09 −1.27
10 .24 .98 5.30 29.90 2.00 2.42 .79 −.83
11 .86 1.44 1.96 3.42 2.68 2.65 .44 −1.39
12 1.29 1.70 1.29 .84 3.85 2.63 −.57 −1.38
13 .96 1.43 1.74 2.77 3.03 2.35 .08 −1.40
14 .27 .79 3.59 14.15 1.60 2.21 1.20 .02
15 3.01 1.60 .67 −.13 4.00 1.85 −.13 −.93
16 .38 1.12 3.93 17.13 2.83 2.61 .28 −1.41
17 .56 1.15 2.38 5.49 3.98 2.69 −.40 −1.43
18 .47 1.10 2.66 6.90 2.55 2.64 .39 −1.47
19 .54 1.13 2.66 7.97 3.05 2.63 .24 −1.44
20 .15 .53 4.06 16.67 2.90 2.63 .21 −1.60
21 .51 1.07 2.75 9.18 3.30 2.52 .06 −1.48
22 1.11 1.63 1.52 1.34 3.18 2.63 .13 −1.53

5
+Model
RPSM-524; No. of Pages 10 ARTICLE IN PRESS
S. Perona-Garcelán, G. Rodenas-Perea, E. Velasco-Barbancho et al.

in the clinical group was 3.76 (SD = 1.57) and 3.03 (SD = 1.69) the MAAS mindfulness scale. As may be seen in this table,
respectively, and this difference was significant (t(39) = 3.84, the effect size of these correlations may be considered high
p < .001), and in the community population it was 1.57 as all of the coefficients are in the range of −.56 to −.73.
(SD = 1.07) and .53 (SD = .70) respectively, and this differ- The Spanish version of the DCI discriminated between
ence was also significant (t(267) = 20.63, p < .001). the participants in the general and clinical populations. As
shown in Table 5, the mean scores for the total scale and
the Detachment and Compartmentalization subscales were
Evidence of validity related to the internal scale
significantly higher in the clinical group.
structure

The CFA showed satisfactory fit to the two-factor model

proposed by the original version of the DCI, with the fol- Evidence of construct validity
lowing goodness-of-fit indicators: Satorra Bentler Scaled
2 (169) = 318.04, p < .001; NNFI = .98, CFI = .99, RMSEA = .054 Cohen’s q test33 was applied to find out whether there
(90% IC = .044, .063), SRMR = .058. The results for the final were significant differences between the correlations of
model were 20 items of which 10 items (1, 2, 3, 4, 7, 11, the DCI Detachment and Compartmentalization subscales
12, 18, 19 and 22) saturated on the Detachment factor and and the other scales for measuring dissociation used in this
10 items (5, 6, 9, 10, 13, 14, 16, 17, 20, 21) on the Com- study (DES-II, DES Taxon & SDQ-20, see Table 3). No sig-
partmentalization factor. The two validation items (8 and nificant effect size (q = .01) was found for the difference
15) were excluded from the CFA. Fig. 1 shows the CFA with between the correlations in Detachment and Compartmen-
the standardized factor weights. talization with respect to the DES-II, but there was small,
but significant effect size with respect to the DES-Taxon and
Measurement invariance of DCI across groups the SDQ-20, in which the effect size was greater for the
(patients and general populations) Compartmentalization subscale (q = .17 and q = .13, respec-
tively), showing that Compartmentalization was associated
First, individual CFAs were carried out in each group to more strongly than Detachment with the DES Taxon, which
find the baseline model, and the goodness-of-fit indicators measures pathological dissociation, and with the SDQ-20,
were found to be adequate (see Table 3). Multi-group CFA which measures somatoform dissociation.
(MGCFA) was then performed on the DCI scale across groups.
The configural invariance results (M0) revealed model good
fit (CFI > .95 and RMSEA < .08), which shows that the fac-
Estimation of DCI score reliability
tor structure is equivalent across patients and nonpatients.
In continuation, an MGCFA was performed to test the met-
Internal consistency of the items on the DCI was studied
ric invariance across groups (M1). The increase in CFI and
using the ordinal alpha and its temporal stability with test-
RMSEA in the comparison between the M1 and M0 models did
retest reliability. The ordinal alpha coefficient for the total
not exceed the recommended values, confirming the metric
score on the Spanish version of the DCI was .94. For the
invariance. This shows that both patients and participants
Detachment subscale it was .87 and for Compartmental-
from the general population perceived and interpreted the
ization it was .90. The test---retest was performed with 79%
items on the DCI in a similar manner. Finally, scalar invari-
(243 participants) of the sample using the Pearson correla-
ance was tested (M2). When M2 was compared with M1,
tion coefficient. The test---retest between pre and posttest
no significant increase was observed in the CFI or RMSEA
total scores was .81 (p < .001), for the Detachment subscale
indices, indicating that the dimensionality of the constructs
it was .75 (p < .001) and for Compartmentalization it was .81
was maintained across groups. These results show that the
(p < .001).
DCI structure, the relationship between the indicators of
each variable and their respective latent factor, and the
relationships among the latent variables were equivalent
across groups.
Calculation of the ROC curve

Evidence of validity based on relationships to The receiver operating characteristic curve (ROC) was com-
other variables puted to analyze the sensitivity and specificity of the DCI
subscales and its total score. The area under the ROC curve
The evidence of validity based on relationships with other was statistically significant for the Detachment subscale
variables was calculated from the Pearson correlation (area = .87, p < .001, 95% CI [.82, .93]) showing a sensitiv-
coefficients. As observed in Table 4 concerning evidence ity of 82% and specificity of 75% for a cutoff point of 17.50
of convergent validity, a positive correlation was found points. On the Compartmentalization subscale, a significant
between the total DCI score and its two subscales with the area under curve was also found (area = .93, p < .001, 95%
DES-II, DES Taxon and SDQ 20, which measure all these dif- CI [.89, .97]) for a cutoff point of 9.50 with a sensitivity of
ferent aspects of the dissociation construct (psychoform or 82% and specificity of 81%. Finally, the area under the ROC
somatoform). curve for the total DCI score was also significant (area = .93,
With respect to divergent validity evidence, a negative p < .001, 95% CI [.89, .97]) with a sensitivity of 85% and a
correlation was found between the two DCI subscales and specificity of 84% for a cutoff point of 31.5 points.

6
+Model
RPSM-524; No. of Pages 10 ARTICLE IN PRESS
Revista de psiquiatría y salud mental (Barcelona) xxx (xxxx) xxx---xxx

Figure 1 Model resulting from Confirmatory Factor Analysis (F1: Detachment, F2: Compartmentalization). Standardized factor
loadings and measurement error from the 20 items DCI are shown.

Table 3 Model fit indices for measurement invariance across groups.

Models SB2 df CFI RMSEA 90% CI NNFI SRMR CFI RMSEA
Clinical populations 202.76 169 .956 .072 [.02, .11] .95 .11
General populations 260.09 169 .989 .045 [.03, .06] .99 .08
[M0] Configural 492.75 338 .982 .055 [.04, .06] .98 .08
[M1] Metric 594.47 356 .973 .066 [.09, .10] .97 .08 <.01 <.15
[M2] Scalar 644.93 374 .967 .071 [.06, .08] .97 .09 <.01 <.15
Note: SB2 : Satorra---Bentler chi square; df: degrees of freedom; CFI: Comparative Fit Index; RMSEA: Root Mean Square Error of Approx-
imation; NNFI: Non-Normed Fit Index; SRMR: Standardized Root Mean Square Residual.

Discussion authors, such as Holmes et al.34 and Brown.13 The results

of the final model were comprised of the same number
The purpose of this study was to adapt and validate of items as in the original study. The final model con-
the DCI scale21 in a Spanish population. A CFA was per- sisted of 10 items for each of the factors, plus two items
formed to study the structural validity of the scale. in the validation scale. The analysis of invariance showed
The results confirmed the factor structure found in that patients and participants from the general population
the original version corresponding to the ‘‘Detachment’’ had an equivalent interpretation of the items on the DCI
and ‘‘Compartmentalization’’ factors described by other scale.

7
+Model
RPSM-524; No. of Pages 10 ARTICLE IN PRESS
S. Perona-Garcelán, G. Rodenas-Perea, E. Velasco-Barbancho et al.

Table 4 Correlation matrix between DCI subscales and DES-II, DES-Taxon, SDQ and MAAS.
1 2 3 4 5 6 7
1. DCI total
2. DCI detachment .94*
3. DCI compartmentalization .94* .77*
4. DES-II .65* .61* .62*
5. DES Taxon .68* .59* .69* .89*
6. SDQ 20 .73* .65* .72* .59* .60*
7. MAAS −.64* −.64* −.56* −.62* −.50* −.51*
Note. N = 308; DCI = Detachment and Compartmentalization Inventory; DES = Dissociative Experiences Scale; SDQ = Somatoform Dissocia-
tion Questionnaire.
* p < 0.001.

Table 5 Difference in mean scores on the DCI scale between the clinical and nonclinical populations.
Clinical population Nonclinical Student’s t
(Mean and SD) population
(Mean and SD)
Total DCI 3.38 (1.52) 1.02 (0.80) t (42,26) = 9.61, p < .001
DCI Detachment 3.76 (1.57) 1.57 (1.07) t (44,58) = 8,55, p < .001
DCI Compartmentalization 3.03 (1.69) .53 (.70) t (41,04) = 9.24, p < .001

Note. N = 308; DCI = Detachment and Compartmentalization Inventory.

In the Spanish version of the DCI, we kept Item 8 and 15 Similarly, for evidence of construct validity, on one hand,
from validation of the original test in English. The criterion we found no differences in the strength of the associa-
of application consisted in eliminating those participants tion between the two DCI subscales and the total score
who scored 0 on those items. The result was that a large on the DES-II, but on the other, there were differences in
number of subjects (86) had to be excluded from the study the association with the DES-Taxon and the SDQ-20. That
since their answers were not considered to be sincere, and is, the effect size of the correlation between Compart-
therefore, could have affected the overall response of the mentalization, pathological and somatoform dissociation is
inventory. This study did not provide clear enough informa- greater than the effect size found between these scales and
tion to determine whether these validation items enabled the subscale measuring detachment. This result is coherent
discrimination between valid and invalid answers to the test, with the idea that compartmentalization measures a more
and it would therefore be of interest in future to study the severe type of dissociative experience qualitatively differ-
adequacy of these items. ent from detachment-type dissociation. Furthermore, it is
The results on evidence of convergent validity were also also coherent with the hypothesis of some authors which
similar to those for the original version. The two subscales affirms that somatoform symptoms are more associated with
of the Spanish version of the DCI showed high positive cor- compartmentalization.16
relations with other scales that measure both psychoform The results also showed that the scores on the Spanish
(DES-II and DES-Taxon) and somatoform (SDQ-20) dissocia- version of the DCI discriminated between the participants
tion. This means that a high score on this scale is associated in the clinical group and those without a psychiatric diagno-
with a high score on the scales that measure dissociative sis. Similar to the original study, the scores found with this
symptoms in which psychological processes are altered as scale enable discrimination between the two populations,
well as dissociative symptoms affecting the body.16 It also but in addition, the Spanish version showed adequate sensi-
shows that the DCI scale has ample coverage and includes a tivity and specificity, with cutoff points that make it possible
wide diversity of dissociative experiences. to estimate whether the degree of dissociation is patholog-
With regard to evidence of divergent validity, nega- ical or not. Finally, the DCI scale scores also show adequate
tive correlations were found between the scores on the internal consistency (with values from .87 to .90) and test-
DCI subscales and the MAAS mindfulness scale, which mea- retest reliability scores, so it may be inferred that it is a
sure opposite constructs. This result is coherent with reliable scale, and that it also enables temporal follow-up
other studies done both in general and psychiatric popu- with repeated measures over time.
lations (more specifically with other studies in psychosis) The DCI scale can be very useful in both the clinic and
in which a negative correlation was found between mind- research. Differentiation between detachment and com-
fulness and dissociation in general,35 mindfulness and partmentalization constructs can facilitate more exhaustive
depersonalization36 and mindfulness and depersonalization evaluation of clinical cases, and therefore, make thera-
and absorption.37 peutic formulations better adjusted to the pathology and

8
+Model
RPSM-524; No. of Pages 10 ARTICLE IN PRESS
Revista de psiquiatría y salud mental (Barcelona) xxx (xxxx) xxx---xxx

patient needs. It is also an instrument which can contribute dissociation in the general population. Aust N Z J Psychiatry.
to studying the efficacy of current treatments designed for 2005;39:387---94.
dissociation, since it can enable evaluation better adjusted 6. Brand BL, McNary SW, Myrick AC, Classen CC, Lanius R, Loewen-
to the type of dissociative experience. It can further be stein RJ, et al. A longitudinal naturalistic study of patients with
used in studying the role of different types of dissociative dissociative disorders treated by community clinicians. Psychol
Trauma. 2013;5:301---8.
experiences in psychopathological processes, such as anxi-
7. Ross CA, Duffy CMM, Ellason JW. Prevalence reliability and valid-
ety, depression and obsessive disorders,38 or hallucinations ity of dissociative disorders in an inpatient setting. J Trauma
in patients with psychosis37 in which detachment-type dis- Dissoc. 2002;3:7---17.
sociation seems to have a relevant role. 8. Bailey TD, Brand BL. Traumatic dissociation: theory research,
Nevertheless, although the results point to confirmation and treatment. Clin Psychol. 2017;24:170---85.
of the DCI as a valid and reliable scale for studying disso- 9. Lynn SJ, Lilienfeld SO, Merckelbach H, Giesbrecht T, van
ciative phenomenology, some limitations should be borne in der Kloet D. Dissociation and dissociative disorders: chal-
mind. There are several limitations having to do with the lenging conventional wisdom. Curr Dir Psychol Sci. 2012;21:
sample and participant selection: they were not found by 48---53.
10. Lynn SJ, Maxwell R, Merckelbach H, Lilienfeld SO, Kloet
random, and so there could be a bias in their recruitment.
D, van H, et al. Dissociation and its disorders: competing
The sample was not very large, and a considerable number
models, future directions, and a way forward. Clin Psychol Rev.
of subjects were excluded from both groups because they 2019;73:101755.
scored zero on the validity items. Furthermore, although 11. Putnam FW. Dissociation in children and adolescents: a devel-
the participants in the psychiatric pathology group were opmental perspective. New York, NY: The Guilford Press; 1997.
referred by their therapists (clinical psychologists or psychi- 12. Cardeña E. The domain of dissociation. In: Lynn SJ, Rhue JW,
atrists), their diagnoses could not be precisely confirmed. editors. Dissociation: clinical and theoretical perspectives. New
Moreover, as the scales used were the self-report type, the York: The Guilford Press; 1994. p. 15---31.
answers could have been biased by social desirability, dis- 13. Brown RJ. Different types of ‘‘dissociation’’ have different psy-
torted memories or even difficulty in understanding some chological mechanisms. J Trauma Dissoc. 2006;7:7---28.
14. Holmes EA, Brown RJ, Mansell W, Fearon RP, Hunter ECM,
of the items due to the uniqueness of this type of experi-
Frasquilho F, et al. Are there two qualitatively distinct forms
ence. In addition, antecedents of mental disorders in family
of dissociation? A review and some clinical implications. Clin
members was not explored in the sample of the general Psychol Rev. 2005;25:1---23.
population. Finally, and as the authors correctly mention 15. Bernstein EM, Putnam FW. Development reliability, and validity
in the original version of the DCI,21 its validity for distin- of a dissociation scale. J Nerv Ment Dis. 1986;174:727---35.
guishing between and belonging to the detachment and 16. Nijenhuis ERS, Spinhoven P, Van Dyck R, Der Hart OV, Vander-
compartmentalization constructs might be questioned, due linden J. The Development and Psychometric Characteristics
to the current debate on the dissociation concept itself of the Somatoform Dissociation Questionnaire (SDQ-20). J Nerv
(e.g., Dell39 ). In any case, we believe that this scale can Ment Dis. 1996;184:688---94.
contribute positively to this debate by providing informa- 17. Fisher W, Johnson A, Elkins G. Principal component analysis
of a measure of non-pathological dissociation: The Dissociative
tion than can be helpful in understanding dissociation and
Ability Scale. Contemp Hypn Integr Ther. 2013;30:32---41.
its underlying mechanisms.
18. Sierra M, Berrios GE. The Cambridge Depersonalisation Scale:
a new instrument for the measurement of depersonalisation.
Conflict of interests Psychiatry Res. 2000;93:153---64.
19. Mazzoti E, Farina B, Imperatori C, Mansutti F, Prunetti E, Sper-
The authors declare that they have no conflict of interest. anza AM, et al. Is the Dissociative Experiences Scale able to
identify detachment and compartmentalization symptoms? Fac-
tor structure of the Dissociative Experiences Scale in a large
Acknowledgments sample of psychiatric and nonpsychiatric subjects. Neuropsy-
chiatr Dis Treat. 2016:1295---302.
We would like to thank Dr. Martin Dorahy for his help and 20. Carlson EB, Putnam FW. An update on the Dissociative Experi-
support in performing this study. ence Scale. Dissociation. 1993;6:16---27.
21. Butler C, Dorahy MJ, Middleton W. The Detachment and Com-
partmentalization Inventory (DCI): an assessment tool for two
References potentially distinct forms of dissociation. J Trauma Dissoc.
2019;20:526---47.
1. Spiegel D, Loewenstein RJ, States U, Sar V. Dissociative disor- 22. World Health Organization (WHO). The ICD-10 classification
ders in DSM-5. Depress Anxiety. 2012;29:747. of mental and behavioural disorders: Clinical descriptions
2. Putnam FW. Dissociative phenomena. In: Spiegel D, editor. Dis- and diagnostic guidelines. Geneva: World Health Organization;
sociative disorders: a clinical review. Lutherville, MD: Sidran 1992.
Press; 1993. p. 1---16. 23. Icarán E, Colom R, Orengo-García F. Experiencias disociativas:
3. World Health Organization (WHO). The ICD-11 classifica- una escala de medida [Dissociative experiences: a scale of mea-
tion of mental and behavioural disorders. Ginebra: World sure]. Anuario de Psicología. 1996;70:69---84.
Health Organization; 2018. In https://icd.who.int/browse11/ 24. Waller N, Putnam FW, Carlson EB. Types of dissociation and
l-m/es#/http%3a%2f%2fid.who.int%2ficd%2fentity%2f108180424. dissociative types: a taxometric analysis of dissociative expe-
4. American Psychiatric Association (APA). Diagnostic and statis- riences. Psychol Methods. 1996;1:300---21.
tical manual of mental disorders: DSM-5. Washington, D.C.: 25. González-Vázquez AI, del Río-Casanova L, Seijo-Ameneiros N.
American Psychological Association; 2013. Validity and reliability of the Spanish version of the Somato-
5. Maaranen P, Tanskanen A, Honkalampi K, Haatainen K, Hin- form Dissociation Questionnaire (SDQ-20). Psicothema. 2017:
tikka J, Viinamäki H. Factors associated with pathological 275---80.

9
+Model
RPSM-524; No. of Pages 10 ARTICLE IN PRESS
S. Perona-Garcelán, G. Rodenas-Perea, E. Velasco-Barbancho et al.

26. Brown KW, Ryan RM. The benefits of being present: mindful- 33. Cohen J. Statistical power analysis for the behavioural science.
ness and its role in psychological well-being. J Pers Soc Psychol. 2nd ed; 1988.
2003;84:822---48. 34. Holmes E, Brown R, Mansell W, Fearon R, Hunter E, Frasquilho
27. Soler J, Tejedor R, Feliu-Soler A, Pascual JC, Cebolla A, Soriano F, et al. Are there two qualitatively distinct forms of dissocia-
J, et al. Psychometric proprieties of Spanish version of Mind- tion? A review and some clinical implications. Clin Psychol Rev.
ful Attention Awareness Scale (MAAS). Actas Esp Psiquiatr. 2005;25:1---23.
2012;40:19---26. 35. Baer RA, Smith GT, Allen KB. Assessment of mindfulness
28. Muñiz J, Elosua P, Hambleton RK. Directrices para la traduc- by self-report: the Kentucky inventory of mindfulness skills.
ción y adaptación de los tests: segunda edición [Indications for Assessment. 2004;11:191---206.
translating and adapting psychological tests: Second edition]. 36. Michal M, Beutel ME, Jordan J, Zimmermann M, Wolters S,
Psicothema. 2013:151---7. Heidenreich T. Depersonalization mindfulness, and childhood
29. Baumgartner H, Homburg C. Applications of structural equation trauma. J Nerv Ment Dis. 2007;195:693---6.
modeling in marketing and consumer research: a review. Int J 37. Escudero-Pérez S, León-Palacios MG, Úbeda-Gómez J, Barros-
Res Mark. 1996;13:139---61. Albarrán MD, López-Jiménez AM, Perona-Garcelán S. Dis-
30. Hu L, Bentler PM. Cutoff criteria for fit indexes in covari- sociation and mindfulness in patients with auditory verbal
ance structure analysis: conventional criteria versus new hallucinations. J Trauma Dissoc. 2016;17:294---306.
alternatives. Struct Eq Mod. 1999;6:1---55. 38. Soffer-Dudek N. Daily elevations in dissociative absorption
31. Byrne BM, van de Vijver F. Testing for measurement and and depersonalization in a nonclinical sample are related
structural equivalence in large-scale cross-cultural stud- to daily stress and psychopathological symptoms. Psychiatry.
ies: addressing the issue of nonequivalence. Int J Test. 2017;80:265---78.
2010;10:107---32. 39. Dell PF. An excellent definition of structural dissociation and
32. Chen FF. Sensitivity of goodness of fit indexes to lack of mea- a dogmatic rejection of all other models. J Trauma Dissoc.
surement invariance. Struct Equ Model. 2007;14:464---504. 2011;12:461---4.

10