EFFECTIVENESS OF COGNITIVE-BEHAVIOURAL TREATMENT IN SOCIAL PHOBIA: A META-ANALYTIC REVIEW

Copyright 2001 by the
Psychology in Spain, 2001, Vol. 5. No 1, 17-25
Colegio Oficial de Psicólogos. Spain
EFFECTIVENESS OF COGNITIVE-BEHAVIOURAL
TREATMENT IN SOCIAL PHOBIA:
A META-ANALYTIC REVIEW
Pedro J. Moreno Gil, F. Xavier Méndez Carrillo and Julio Sánchez Meca
University of Murcia
Cognitive-behavioural treatments for social phobia are reviewed quantitatively. Mean effect size (weighted by the inverse
of the variance) was 0.77 for the posttest and 0.95 for the follow-up. These results enable us to affirm, in global terms, that
CBT for social phobia is clearly effective. However, the homogeneity test did not reach statistical significance, so that we
can assume that exposure techniques, cognitive restructuring techniques and social skills training are homogeneous in their
effectiveness. This fact raises questions about the psychological principles underlying the effectiveness of CBT. An expla-
nation is proposed based on exposure to feared social stimuli.
Se revisan, de modo cuantitativo, los tratamientos cognitivos y conductuales para la fobia social, obteniéndose un tamaño
del efecto medio ponderado de 0,77 para el postest y 0,95 para el seguimiento. Estos valores permiten afirmar que, en tér-
minos globales, dichos tratamientos resultan bastante efectivos para la fobia social. Sin embargo, la prueba de homoge-
neidad no es estadísticamente significativa, por lo que podemos asumir que las técnicas de exposición, las técnicas de rees-
tructuración cognitiva y el entrenamiento en habilidades sociales —los tratamientos más contrastados— no difieren en la
eficacia demostrada. Este hecho plantea cuestiones de interés sobre los fundamentos de la eficacia de dichos procedi-
mientos. Se propone una explicación a dicho fenómeno basada en la exposición a los estímulos sociales temidos.
Social phobia is an anxiety disorder that remained fears can attain diverse degrees of generalisation. The
practically unidentified until the mid-1980s
DSM-IV distinguishes a generalised subtype of social
(Heimberg, 1989a; Liebowitz, Gorman, Fyer and Klein,
phobia that is applicable to those people who fear the
1985), despite attaining prevalence rates of between 3%
majority of social situations. However, Heimberg, Holt,
and 13% (Kessler, McGonagle, Zhao et al., 1994).
Schneier et al. (1993) distinguish two additional subty-
This phobic disorder is usually complicated by work
pes: circumscribed subtype, applicable to those who fear
absenteeism, drug and/or anxiolytics abuse, alcoholism
only one or two discrete situations, and non-generalised
and depression (Barlow, DiNardo, Vermilyea and
subtype, applicable to those that, demonstrating adaptive
Blanchard, 1986; Bowen, Cipywnyk, D’Arcy and
functioning in some social areas, feel anxiety in a mini-
Keegan, 1984; Chambless, Cherney, Caputo and
mal number of interactive situations. Social phobia pre-
Rheinstein, 1987; Higgins and Marlatt, 1975; Kushner,
sents a high level of comorbidity with other disorders of
Sher and Beitman, 1990; Schneier, Martin, Liebowitz et
axis I (Brewerton, Lydiard, Ballenger and Herzog, 1993;
al., 1989). In some cases these problems are the expres-
Bulik, Beidel, Duchmann and Weltzin, 1991; Disalver,
sion of an undiagnosed social phobia, so that the preva-
Qamar and Del Medico, 1992; Schneier, Johnson,
lence of this clinical condition may be greater than esti-
Horning et al., 1992; Schneier, Martin et al., 1989;
mated (Stravynski, Lamontagne and Lavallee, 1986).
Schwalberg, Barlow, Alger and Howard, 1992; van
The central characteristic of social phobia is excessive
Amerigen, Mancini, Styan and Donison, 1991) and axis
and persistent fear of social situations in which the
II (Herbert, Hope and Bellack, 1992; Holt, Heimberg and
patient is exposed to the observation or scrutiny of others
Hope, 1992; Turner et al., 1992).
(American Psychiatric Association 1994). These social
The most studied cognitive-behavioural treatments are
social skills training, exposure techniques, cognitive res-
The original Spanish version of this paper has been previously
tructuring techniques and the combination of exposure
published in Psicothema, 2000, No 12, 346-352
and cognitive restructuring. Relaxation techniques, sys-
...........
tematic desensitisation and training in coping with
Address for correspondence: Julio Sánchez Meca, Facultad de
Psicología, Campus de Espinardo, Apdo. 4021, 30100 Murcia.
anxiety have been studied to a lesser extent. Exposure
Spain. E-mail: jsmeca@um.es
techniques are considered as the preferred treatment for
VOLUME 5. NUMBER 1. 2001. PSYCHOLOGY IN SPAIN
17

---

phobic disorders (Echeburúa, 1990; Echeburúa and
d) Does not treat work with waiting list groups as inde-
Salaberría, 1991; Marks, 1991), or at least as a funda-
pendent studies, as is the case in Taylor (1996).
mental component of any effective therapeutic package
e) Includes calculation of effect sizes differentiated for
(Echeburúa, 1993; Heimberg, 1989b; Heimberg and
between-groups and within-group designs.
Juster, 1995), though their application to social phobia is
f) Includes all types of measure directly related to the
quite recent (Echeburúa and Salaberría, 1991). Also,
descriptions of clinical conditions in the nosological
there is a wide consensus on the fact that cognitive
systems employed (e.g., scales applied by intervie-
mediation is present in social phobia more than in the
wers, ‘Fear of Negative Evaluation Scale’ or ‘Social
rest of anxiety disorders (cf. Beck and Emery, 1985;
Avoidance and Distress Scale’).
Butler, 1989; Heimberg and Barlow, 1988), though the
g) Excludes measures not directly related to the des-
application of cognitive restructuring techniques has not
criptions of clinical conditions in the nosological
provided conclusive results (Echeburúa, 1993;
systems (e.g., irrational beliefs or depressive state),
Heimberg, 1989b; Heimberg and Juster, 1995;
since studies differ in the assessment of type and
Salaberría and Echeburúa, 1995). Feske and Chambless
number of additional constructs analysed. It is our
(1995) affirmed, after a meta-analysis of 21 studies, that
view that, in principle, the inclusion of constructs not
exposure techniques are equally effective applied alone
considered in the definition of social phobia may
or in combination with cognitive restructuring techni-
generate an additional heterogeneity that would lead
ques. Taylor (1996), whose meta-analysis included 42
to confusion as regards the magnitude of the diffe-
studies, found that all the treatments considered in his
rences between treatments.
work —placebo pills, exposure, cognitive restructuring
h) Employs data-analysis techniques based on weighted
(without exposure exercises), integrated combination of
least squared. Currently, classical statistical techni-
cognitive restructuring and exposure, social skills trai-
ques, which do not weight studies according to their
ning— gave effect sizes superior to those of the waiting
precision, are not recommended in meta-analyses
list condition, but that only the combination of cognitive
(Cooper and Hedges, 1994).
restructuring and exposure produced better results than
those of the group treated with placebo pills.
METHOD
The aim of the present meta-analysis is to estimate the
Distinction between “research report” and “study”
effectiveness of exposure techniques, cognitive restruc-
In the meta-analyses carried out we distinguish between
turing techniques and social skills training, and combi-
“research report” and “study”. By research report we
nations of them, in social phobia patients. Furthermore,
understand the framework employed to report the results
we shall identify the variables that moderate therapeutic
of one or more comparisons of groups in relation to the
effectiveness in this phobic disorder. The main differen-
effectiveness of psychological treatments for social pho-
ces between the present study and the two previously-
bia. By study we understand the comparison between a
published meta-analyses are that our work:
group receiving psychological treatment and a control
a) Includes all studies that examine exposure techni-
group. In those cases where a treated group was not
ques, social skills training, cognitive restructuring
compared with a non-treated group, but an assessment
techniques and their combinations in a group design
was made of the therapeutic gain comparing a posttest
with at least pretest-posttest data (this implies the
measure with a pretest measure for the treated group, we
inclusion of, for example, groups that receive only
also considered it as an independent study.
cognitive restructuring without exposure, or that
receive cognitive restructuring after exposure).
Inclusion/exclusion criteria for research reports
b) Excludes groups treated with placebo pills, exclusi-
With the aim of homogenising the empirical sample and
vely or in combination with the cognitive-behaviou-
applying the meta-analytic technique in an appropriate
ral treatments considered, given the confusion cau-
way, we established the following conceptual and met-
sed, for our purposes, on comparing a group with
hodological criteria for including or excluding a rese-
only psychological treatment with others that add the
arch report:
expectations of receiving pharmacological treat-
a) The research report should examine the effectiveness
ment.
of exposure techniques, cognitive restructuring tech-
c) Includes studies carried out with Spanish popula-
niques, social skills training or a combination of
tions.
these treatments.
VOLUME 5. NUMBER 1. 2001. PSYCHOLOGY IN SPAIN
18

---

b) Subjects treated should receive a diagnosis of social
coding protocol. The quality of the coding process was
phobia.
studied through the selection of a random sample of
c) The concept of social phobia employed in the rese-
33% of the total reports found, which was coded by two
arch report should coincide explicitly with the defi-
independent coders.
nitions included in the ICD nosological system
The variables whose possible moderating effect was
(World Health Organization, 1978, 1979, 1992) or
examined were classified, following Lipsey (1994), in
that of the DSM (American Psychiatric Association,
three large sections (see Table 1):
1980, 1987, 1994).
a) Substantive variables: these are intrinsic to the scope
d) The date of publication (or date of carrying out, if the
of the research being meta-analysed. They include:
study is unpublished) should fall within the period
treatment, subject and context variables. For exam-
1980-1997 inclusive. The beginning of this period is
ple, type of psychological treatment applied.
determined by the date of publication of the third
b) Methodological variables: deriving from aspects
edition of the Diagnostic and Statistical Manual of
related to the research design and methodology. For
Mental Disorders.
example, type of design used.
e) The research should have a group design, with at
c) Extrinsic variables: related to neither the methodo-
least 5 experimental subjects.
logy nor the scope of the research. Although such
f) The research report should provide sufficient data to
variables should not affect the results, they may
allow estimation of effect sizes.
occasionally be relevant. For example, date of the
g) Research reports that do not present pretest data
study.
were excluded.
h) Also excluded were reports that employed single-
Calculation of effect size
case design methodology, since it is impossible to
The index of effect size (ES) used is the standardised
combine in a quantitative manner the results of
mean difference d (Hedges and Olkin, 1985), conside-
group designs with those of single-case designs.
ring the following definitions according to the nature of
the design used in the assessed study: (a) For within
Literature search
group designs (pretest-posttest), d is defined as the dif-
In order to minimise selection biases, we used various
ference between the mean of the pretest and the mean of
processes in the search for research reports:
the posttest divided by the overall within-group standard
a) Computerised search. Using the databases PsycLIT,
MEDLINE and PSICODOC. The descriptors emplo-
Table 1
yed were: “social phobia”, “social anxiety”, “treat-
Moderator variables
ment” and “therapy”.
Treatment variables: 1) Psychological treatment employed, 2) Gender of
b) Review of articles and monographs by relevant aut-
professionals, 3) Experience of professionals, 4) Home tasks of the treat-
ment, 5) Follow-up programme of the treatment, 6) Utilisation of agents
hors in the field. Specifically: Echeburúa and
external to the therapeutic group, 7) Therapeutic contract, 8) Training
Salaberría (1991), Feske and Chambless (1995),
modality, 9) Number of therapists, 10) Number of treatment sessions, 11)
Heimberg (1989b), Heimberg and Juster (1995),
Duration of treatment, 12) Intensity of treatment, 13) Absenteeism.
Salaberría, Borda, Báez and Echeburúa (1996), and
Subject variables: 1) Age, 2) Gender, 3) Socio-economic level, 4)
Taylor (1996).
Educational level, 5) Diagnostic system employed, 6) Social phobia
c) Review of bibliographical references of the research
subtype, 7) Mean duration of the social phobia, 8) Comorbidity, 9)
Presence of previous treatments, 10) Type of previous treatments (psy-
reports already located, as a source of previous pri-
chological, pharmacological, or both).
mary studies.
Context variables: 1) Continent, 2) Country.
d) Request for papers from experts, published and
unpublished.
Methodological variables: 1) Subject recruitment mode, 2) Measures to
The literature search carried out allowed us to locate 25
preserve integrity of the treatment, 3) Type of design, 4) Random assign-
ment to groups, 5) Type of control group, 6) Total sample size, 7) Sample
research reports that fulfilled the selection criteria, pro-
size of treatment group, 8) Sample size of control group, 9) Experimental
viding a total of 39 studies.
mortality, 10) Design quality, 11) Months of follow-up, 12) Number of
dependent variables.
Coding of studies
Extrinsic variables: 1) Technical qualifications of first researcher, 2)
In order to define operationally the variables to be coded
Form of presentation of the study, 3) Range of publication, 4) Date of pre-
we prepared a manual with specific guidelines and a
sentation.
VOLUME 5. NUMBER 1. 2001. PSYCHOLOGY IN SPAIN
19

---

deviation (or, failing that, the average of the standard
teeing coding reliability (a value of at least 0.80 for agre-
deviations of the pretest and posttest); (b) For between-
ement rate and for Pearson’s correlation coefficient, and
groups designs (both experimental and quasi-experi-
a value of at least 0.70 for Cohen’s Kappa and the intra-
mental), with pretest and posttest measures, d is defined
class correlation). The reliability of the calculations of
as d= dE - dC, with dE and dC being the standardised mean
effect size (intra-class correlation and Pearson’s correla-
differences between the pretest and posttest of the expe-
tion) was greater than the value of 0.98 for the posttest
rimental and control groups, respectively. A d value
and follow-up, indicating an excellent consensus in the
above zero indicates a beneficial effect for the subjects
selection of dependent variables to be included in the
of the treated group, while a d value below zero indica-
calculation of the effect size.
tes a detrimental effect.
With the aim of optimising the comparison between
Study of publication bias
meta-analysed treatments, we maximised the homoge-
All of the studies included in the present meta-analysis
neity of the ESs. We selected only the dependent varia-
are contained in research reports published in specialist
bles directly related to the definition of social phobia,
journals, so that a possible threat to the validity of the
that is, fear, avoidance and/or uneasiness in social situa-
results obtained derives from a potential publication
tions, which implies deterioration in social, work and/or
bias, that is, that the publishers may be uneven in their
academic adaptation, and excluded the rest (trait anxiety,
treatment of the studies presented for publication, as a
depressive mood, self-esteem, locus of control, irratio-
function of the statistical significance reported.
nal beliefs, etc.).
Following Orwin (1983), we calculated the “index of
For each study we calculated a maximum of two ESs,
tolerance of null results”. According to this index, there
one for the posttest assessment and another for the
would have to be more than 180 unpublished studies
assessment of the longer follow-up. For each point in
(and not considered by the meta-analyst) filed away with
time we averaged the d values derived from the depen-
the publishers of the journals for the results of our meta-
dent variables that coincided with the established crite-
analysis to be invalidated. We can therefore conclude
ria.
that it is highly improbable that our results are affected
by the publication bias.
Statistical analysis techniques
We calculated a confidence interval around the mean ES
Meta-analysis in the posttest and follow-up
in order to estimate the population effect size and whet-
Five studies provide data only from the posttest and 34
her that effect was significantly different from zero (null
from the posttest and follow-up. The median of the time
effectiveness). We also applied a X2 test of homogeneity
interval between posttest and follow-up was three
of all effect sizes around the mean ES (Hedges, 1994;
months.
Hedges and Olkin, 1985).
The mean effect size (weighted by the inverse of the
Given that the homogeneity test may be less powerful
variance) was 0.769 in the posttest and 0.953 in the
when applied to a small group of studies, as in our case
follow-up, the parametric values being far from the null
(cf. Sánchez-Meca and Marín-Martínez, 1997), we deci-
value. These results allow us to affirm, in global terms,
ded to check the influence of variables that may theore-
that the cognitive-behavioural treatments reviewed are
tically be moderating the results, even though the homo-
clearly effective for social phobia. Moreover, conside-
geneity test was not found to be statistically significant.
ring the orientative classification proposed by Cohen
Following the meta-analytic approach of Hedges and
(1988), the mean value obtained in our meta-analysis
Olkin (1985), for the qualitative variables we carried out
approaches a high magnitude in the posttest (d = 0.80),
analyses of variance weighted by the inverse of the
and a higher one still in the follow-up.
variance of each ES. For the quantitative variables we
The homogeneity test was not found to be statistically
applied simple regression analyses weighted by the
significant [posttest: QT(38) = 19.163, p > .05; follow-
inverse of the variance of each ES.
up: QT(33) = 25.318, p > .05], so that we can assume
homogeneity of the different studies among themselves,
RESULTS
despite being derived from studies that differ in a large
Reliability of the coding
number of characteristics, both substantive and metho-
Inter-coder reliability for the moderator variables attai-
dological; the parametric values were representative of
ned the values established by Orwin (1994) for guaran-
those found in the meta-analysed studies.
VOLUME 5. NUMBER 1. 2001. PSYCHOLOGY IN SPAIN
20

---

A more intuitive interpretation of the effect sizes found
patient’s anxiety disorder; (c) presence of the generali-
can be obtained through their transformation into a
sed subtype of social phobia; (d) the mean duration of
correlation coefficient (r=d/[d2+4]1/2) in order to cons-
the clinical condition; (e) the presence of previous treat-
truct the binomial presentation of the effect size, BESD
ments; and (f) the type of design.
(binomial effect size display) proposed by Rosenthal
Table 2 shows the mean ESs obtained (d+) for the tre-
(1991). A correlation coefficient of 0.36, for example, is
atments considered in our meta-analysis. Also included
equivalent to an improvement rate of 68% in the treat-
are the limits of the confidence interval at 95% (Li ;
ment groups, as against a rate of only 32% in the control
Ls), the number of studies from which the ES is deri-
groups, representing a differential rate of 36% between
ved (K) and the sample size accumulated for each set
the two groups.
of studies (N).
According to the model of meta-analysis applied in our
study, the fact of obtaining non-significant results of the
Differences between posttest and follow-up
homogeneity tests should have put a halt to the analysis,
In order to examine the changes in effectiveness that
precluding the search for possible moderator variables
occurred between the posttest and the follow-up, we
of the effect sizes found in the empirical studies. Put
selected the 34 studies that provided data in posttest and
another way, these results should have led us to the con-
follow-up. In general, effect sizes tended to decrease
clusion that the studies meta-analysed here present
with respect to the posttest in the follow-up assessments,
homogeneous effectiveness in the psychological treat-
a finding usually attributed to a fading of the therapeutic
ment of social phobia and, consequently, the mean effect
effect of the treatment studied. However, in our research
sizes obtained, and their confidence intervals, represent
we found, as did Taylor (1996), that effect sizes were
in a valid way the set of studies in the meta-analysis. In
greater in the follow-up (as against the posttest), and that
fact, these results concur, in general terms, with those
this difference was marginally significant [T(33) = -
obtained in the meta-analyses by Feske and Chambless
1.946, p = .06].
(1995) and by Taylor (1996).
Nevertheless, a possible threat to the internal validity
Nevertheless, this conclusion may be seen as simplis-
of the relationship between increase in therapeutic effec-
tic, since the non-significant result obtained in the
tiveness and the point at which assessment was made
homogeneity test can be interpreted in different ways
may be constituted by selective mortality. In order to
(cf., e.g., Hall and Rosenthal, 1991). In fact, the homo-
analyse the relationship between experimental mortality
geneity test may be non-significant due to a lack of sta-
and magnitude of therapeutic effect, we defined two
tistical power, especially with a fairly small number of
variables: (a) Increase in posttest-follow-up experimen-
studies, as in our research, K=39 (cf., e.g., Harwell,
tal mortality ( Mpost-flwup), as the difference between
1997; Sánchez-Meca and Marín-Martínez, 1997).
pretest-follow-up mortality and posttest-follow-up mor-
It is interesting to note that the differences found bet-
tality for each study; and (b) increase in effect size post-
ween the values of effectiveness for the different studies
test-follow-up ( TEpost-flwup), as the difference betwe-
fail to reach statistical significance as a function of: (a)
en ES in the follow-up and ES in the posttest for each
the cognitive and/or behavioural technique employed;
study.
(b) the diagnostic system used for diagnosing the
Through the construction of a simple regression model,
Table 2
Effectiveness of the psychological treatments
Follow-up
Posttest
Treatment*
N
K
d+
Li
Ls
N
K
d+
Li
Ls
Exposure
228
10
0.835
0.61
1.06
138
8
0.946
0.67
1.22
CR (Ellis)
79
4
0.762
0.38
1.15
48
4
1.099
0.70
1.50
SST
78
3
0.606
0.20
1.02
68
3
0.889
0.46
1.32
Exposure + CR (Ellis)
195
9
0.835
0.57
1.10
159
8
0.962
0.69
1.24
Exposure + SST
95
6
0.693
0.39
0.99
57
4
0.944
0.59
1.30
Exposure + CR (Beck)
132
5
0.556
0.26
0.85
125
5
0.754
0.40
1.11
* CR: Cognitive restructuring. SST: Social skills training.
VOLUME 5. NUMBER 1. 2001. PSYCHOLOGY IN SPAIN
21

---

we found that DMpost-flwup was related to TEpost-flwup
lar to those found for exposure techniques applied exclu-
[F(1, 22) = 11.537, p = .003], explaining 31.4% of its
sively (ES for posttest and follow-up, respectively: 0.61
variance. However, this regression model appeared to be
and 1.28, in Emmelkamp, Mersch, Vissia and van der
affected by an outlier. The result in question belongs to
Helm, 1985; 0.98 and 1.77, in Mattick, Peters and
a study included in the report by Mersch et al. (1995),
Clarke, 1989; 0.59 and 1.05, in Mersch Emmelkamp,
and derives from the follow-up carried out with 3 of the
Bögels and van der Sleen, 1989). This could lead us to
7 patients that completed the posttest assessment, giving
the conclusion that the effectiveness of cognitive res-
an experimental mortality of 62.5%. After elimination of
tructuring techniques is not related to the inclusion of an
this outlier the relationship no longer showed statistical
explicit component of exposure to the phobic stimuli.
significance [F(1, 21) = 0.997, p = .329].
Given that we are considering a complex phobic con-
dition and a quantitative improvement criterion, an alter-
DISCUSSION
native hypothesis may reside in the existence of multiple
Considering in a global way the effectiveness of social
mechanisms of action through which both exposure
skills training, exposure techniques and cognitive res-
techniques and cognitive restructuring techniques may
tructuring techniques, we can conclude that they are
be effective, each type of technique acting on different
fairly effective. Smith, Glass and Miller (1980), despite
aspects of the social phobia (e.g., exposure techniques
the methodological and content differences between
on avoidance behaviour and high vegetative activation,
their study and our own, provide an alternative practical
and cognitive restructuring techniques on dysfunctional
illustration of the meaning of effect size obtained. They
cognitive content). In this case, though, the combination
point out that nine months of work to teach primary
of the two techniques should prove superior to each
school children to read translates into an effect size of
technique used alone (since the patient is treated in mul-
0.67, showing that the effectiveness of the treatments
tiple areas), and this is not the case. Nevertheless, we
analysed is clearly substantial for social phobia patients.
might ask ourselves whether response to treatment
The studies that examine the effectiveness of multi-
depends on the phobic profile of the patient. According
component therapeutic packages —made up of exposu-
to this hypothesis, patients with a predominantly cogni-
re techniques plus social skills training or cognitive res-
tive response would benefit more from cognitive res-
tructuring techniques— are not superior to the studies
tructuring procedures, whilst patients with a predomi-
that employ only exposure techniques, even though the
nantly physiological and motor response would benefit
number of subjects treated with exposure, alone or in
more from exposure to the phobic stimuli. The meta-
combination with other techniques, is greater than that
analysis carried out does not permit the testing of this
of subjects treated with cognitive restructuring or social
hypothesis, though the results of primary studies desig-
skills training.
ned ad hoc do not support the hypothesis of customised
This fact is interpreted by some authors as representing
treatment, at least for systematic desensitisation, social
a lack of support for the therapeutic principles underpin-
skills training, applied relaxation, Ellis’s rational emoti-
ning social skills training and/or cognitive restructuring
ve therapy and training in self-instructions (Jerremalm,
techniques in the treatment of social phobia (Feske and
Jansson and Öst, 1986; Mersch, Emmelkamp, Bögels
Chambless, 1995; Hope, Heimberg and Bruch, 1995;
and van der Sleen, 1989; Trower et al., 1978).
Mattick and Peters, 1988; Mersch, 1995; Scholing and
For Marks (1991), the equivalence of cognitive treat-
Emmelkamp, 1993; Stravynski, Marks and Yule, 1982).
ments and exposure techniques is due to the fact that
In this regard, the absence of differences between treat-
cognitive restructuring techniques implicitly include a
ments would derive from the fact that the different tech-
component of exposure to the feared social stimuli (cog-
niques possess common therapeutic elements that are
nitive exposure). Nevertheless, we believe it a little con-
effective in the treatment of phobic disorders. It is pro-
trived to consider that the Socratic dialogue (discussion
bable that exposure to feared social stimuli has been an
of beliefs such as “I should be perfect and not make mis-
element shared by the treatments applied to patients in
takes”, or “It would be terrible if no-one loved me”) wit-
the different studies. In fact, social skills training inclu-
hout explicit prescription of exposure (or self-exposure)
des exposure in vivo to the phobic stimuli.
tasks could constitute an element of exposure.
However, in the studies that examine cognitive restruc-
In our study we found no statistically significant diffe-
turing techniques and that explicitly controlled the
rences between the studies according to administration
exclusion of elements of exposure, we found ESs simi-
format of the treatment (individual vs. group), although
VOLUME 5. NUMBER 1. 2001. PSYCHOLOGY IN SPAIN
22

---

the tendencies in the ESs are similar to those obtained by
applied with tasks to carry out at home) may continue
Moreno (1999), who found that the most effective treat-
implementing self-exposure techniques by themselves
ments had been administered in group format. If we
in their own social environment, just as prescribed by
were to find a real difference in favour of group treat-
the therapist during the treatment phase. Self-exposure
ments, we could minimise the threat that the effective-
may be as effective as other variants of exposure techni-
ness of the treatments studied resides in what Frank
ques (Marks, 1991, p. 156), and may constitute the basis
(1988) called “common elements of psychotherapy”.
of the therapeutic gain that tends to be found in some
This conception encompasses, for example, the thera-
studies between posttest and follow-up. Nevertheless, in
pist-patient relationship, the explanation to the patient of
patients that received cognitive restructuring techniques
a conceptual scheme for understanding his/her problems
without (explicit) exposure tasks, an additional impro-
(a “myth”, in this author’s words, since its validity is
vement is likewise observed in the follow-up, and which
presupposed), or the provision of a therapeutic ritual to
is difficult to explain from the hypothesis of action
the patient (that is, a series of rules, techniques and exer-
mechanisms with a basis in exposure.
cises which, used correctly lead, a priori, to “cure”).
Some of these elements are more typical of clinical con-
REFERENCES
texts than contexts of research into therapeutic results.
American Psychiatric Association (1980). Diagnostic
Given the structure of the groups treated with these
and statistical manual of mental disorders (3ª ed.).
techniques, confirmation of the tendency in favour of
Washington, DC: Author. (Trad. Masson, 1981).
treatment application in group format would provide
American Psychiatric Association (1987). Diagnostic
empirical support for the principle of exposure to phobic
and statistical manual of mental disorders (3ª ed.,
stimuli as an explanation of the improvement achieved,
rev.). Washington, DC: Author. (Trad. Masson, 1987).
regardless of the specific treatment imparted. It is perti-
American Psychiatric Association (1994). Diagnostic
nent to point out that the fact of imparting a treatment
and statistical manual of mental disorders (4ª ed.).
through a series of group sessions of two hours’ dura-
Washington, DC: Author. (Trad. Masson, 1995).
tion could convert virtually any treatment (even that in
Barlow, D. H., DiNardo, P. A., Vermilyea, B. B.,
which the explicit prescription of exposure tasks is avoi-
Vermilyea, J. and Blanchard, E. B. (1986).
ded) into a treatment of exposure to ansiogenic stimuli
Comorbidity and depression among the anxiety disor-
with sufficient duration to produce habituation, accor-
ders: Issues in diagnosis and classification. Journal of
ding to the parameters generally recommended for pro-
Nervous and Mental Disease, 174, 63-72.
ducing therapeutic exposure sessions (Marks, 1991). In
Beck, A. T. and Emery, G. (1985). Anxiety disorders and
fact, cognitive treatments without explicitly implemen-
phobias. A cognitive perspective. New York: Basic
ted exposure techniques were carried out in group for-
Books.
mat with series of 6 to 8 sessions of 120 to 150 minutes
Bowen, R. C., Cipywnyk, D., D’Arcy, C. and Keegan,
each. It would remain to weight the therapeutic effect of
D. (1984). Alcoholism, anxiety disorders, and ago-
the inherent exposure to that situation (exposure not pro-
raphobia. Alcoholism: Clinical and Experimental
grammed or defined as such to the patient, in one group
Research , 8, 8-50.
of 4 to 7 persons with supposedly equivalent problems,
Brewerton, T. D., Lydiard, R. B., Ballenger, J. C. and
and over a period of 20 hours divided into 8 weekly ses-
Herzog, D. B. (1993). Eating disorders and social
sions).
phobia. Archives of General Psychiatry, 50, 70-79.
The observed increase in effectiveness of treatments
Bulik, C.M., Beidel, D.C., Duchmann, E. and Weltzin,
between posttest and follow-up is an unusual phenome-
T.E. (1991). An analysis of social anxiety in anorexic,
non in the scientific literature, even if it may be due, as
bulimic, social, phobic, and control women. Journal
Taylor (1996) points out, to the application of additional
of Psychopathology and Behavioral Assessment, 13,
treatment for those patients that require it. A possible
199-211.
alternative explanation, ruling out the phenomenon of
Butler, G. (1989). Issues in the application of cognitive
selective mortality, is that, even though researchers may
and behavioral strategies to the treatment of social
administer additional treatment to patients that suffer a
phobia. Clinical Psychology Review, 9, 91-106.
relapse, it is also sometimes the case that patients who
Chambless, D. L., Cherney, J., Caputo, G. C. and
manage to confront the feared social stimuli (the reader
Rheinstein, B. J. G. (1987). Anxiety disorders and
will recall that exposure techniques have mainly been
alcoholism: A study with inpatient alcoholics.
VOLUME 5. NUMBER 1. 2001. PSYCHOLOGY IN SPAIN
23

---

Journal of Anxiety Disorders, 1, 9-40.
in the diagnosis of social phobia. Journal of Anxiety
Cohen, J. (1988). Statistical power analysis for the
Disorders, 7, 249-269.
behavioral sciences (2ª ed.) Hillsdale, NJ: Erlbaum.
Heimberg, R. G. and Juster, H. R. (1995). Cognitive-
Cooper, H. M. and Hedges, L. V. (Eds.) (1994). The
behavioral treatments: Literature review. (pp. 261-
handbook of research synthesis. New York: Sage.
309) In R. G. Heimberg, M. R. Liebowitz, D. A. Hope
Dilsaver, S. C., Qamar, A. B. and Del Medico, V. J.
and F. R. Schneier (Eds.), Social phobia. Diagnosis,
(1992). Secondary social phobia in patients with
assessment, and treatment. New York: Guilford.
major depression. Psychiatry Research, 44, 33-40.
Herbert, J. D., Hope, D. A. and Bellack, A. S. (1992).
Echeburúa, E. (1990). La exposición y los enfoques cog-
Validity of the distinction between generalized social
nitivos en el tratamiento de los trastornos de ansiedad
phobia and avoidant personality disorders. Journal of
según los criterios diagnósticos del DSM-III y la CD-
Abnormal Psychology, 101, 332-339.
10.
Revista Española de Terapia del
Higgins, R. L. and Marlatt, G. A. (1975). Fear of inter-
Comportamiento, 8, 1-15.
personal evaluation as a determinant of alcohol con-
Echeburúa, E. (1993). Fobia social. Barcelona:
sumption in male social drinkers. Journal of
Martínez Roca.
Abnormal Psychology, 84, 644-651.
Echeburúa, E. and Salaberría, K. (1991). Tratamientos
Holt, C. S., Heimberg, R. G. and Hope, D. A. (1992).
psicológicos de la fobia social: Un análisis crítico.
Avoidant personality disorder and the generalized
Psicothema, 3, 1-23.
subtype of social phobia. Journal of Abnormal
Emmelkamp, P. M. G., Mersch, P. P. A., Vissia, E., and
Psychology, 101, 318-325.
van der Helm, M. (1985). Social phobia: A compara-
Hope, D. A., Heimberg, R. G. and Bruch, M. A. (1995).
tive evaluation of cognitive and behavioral interven-
Dismantling cognitive behavioral group therapy for
tions. Behaviour Research and Therapy, 23, 365-369.
social phobia. Behaviour Research and Therapy, 33,
Feske, U. and Chambless, D. L. (1995). Cognitive beha-
637-650.
vioral versus exposure only treatment for social pho-
Jerremalm, A., Jansson, L., and Öst, L. G. (1986).
bia: A meta-analysis. Behavior Therapy, 26, 695-720.
Cognitive and physiological reactivity and the effects
Frank, J. D. (1988). Elementos terapéuticos compartidos
of different behavioral methods in the treatment of
por todas las psicoterapias. In M. J. Mahoney and A.
social phobia. Behaviour Research and Therapy, 24,
Freeman (Eds.), Cognición y psicoterapia. Barcelona:
171 - 180.
Paidós (Orig. 1985).
Kessler, C. R., McGonagle, K., Zhao, S., Nelson, C.,
Hall, J.A. and Rosenthal, R. (1991). Testing for modera-
Hughes, M., Eschlemann, S., Wittchen, H. U. and
tor variables in meta-analysis: Issues and methods.
Kendler, K. S. (1994). Lifetime and 12-month preva-
Communication Monographs, 58, 437-448.
lence of DSM-III-R psychiatric disorders in the
Harwell, M. (1997). An empirical study of the Hedge’s
United States: Results from the National Comorbidity
homogeneity test. Psychological Methods, 2, 219-
Survey. Archives of General Psychiatry, 51, 8-19.
231.
Kushner, M. G., Sher, K. J. and Beitman, B. D. (1990).
Hedges, L. V. (1994). Fixed effects models. In H. M.
The relation between alcohol problems and the
Cooper and L. V. Hedges (Eds.), The handbook of
anxiety disorders. American Journal of Psychiatry,
research synthesis (pp. 285-299). New York: Sage.
147, 685-695.
Hedges, L. V. and Olkin, I. (1985). Statistical methods
Liebowitz, M. R., Gorman, J. M., Fyer, A. J. and Klein,
for meta-analysis. Orlando, FL: Academic Press.
D. F. (1985). Social phobia: Review of a neglected
Heimberg, R. G. (1989a). Social phobia: no longer
anxiety disorder. Archives of General Psychiatry, 42,
neglected. Clinical Psychology Review, 9, 1-2.
729-736.
Heimberg, R. G. (1989b). Cognitive and behavioral tre-
Lipsey, M. W. (1994). Identifying potentially interesting
atments for social phobia: A critical analysis. Clinical
variables and analysis opportunities. In H. M. Cooper
Psychology Review, 9, 107-128.
and L. V. Hedges (Eds.), The handbook of research
Heimberg, R. G. and Barlow, D. H. (1988). Psychosocial
synthesis (pp. 111-123). New York: Sage.
treatments for social phobia. Psychosomatics, 29, 27-
Marks, I. M. (1991). Miedos, fobias y rituales.
37.
Barcelona: Martínez Roca (Orig. 1987).
Heimberg, R. G., Holt, C. S., Schneier, F. R., Spitzer, R.
Mattick, R. P. and Peters, L. (1988). Treatment of seve-
L. and Liebowitz, M. R. (1993). The issue of subtypes
re social phobia: Effects of guided exposure with and
VOLUME 5. NUMBER 1. 2001. PSYCHOLOGY IN SPAIN
24

---

without cognitive restructuring. Journal of
Scholing, A. and Emmelkamp, P. M. G. (1993).
Consulting and Clinical Psychology, 56, 251-260.
Exposure with and without cognitive therapy for
Mattick, R. P., Peters, L. and Clarke, J. C. (1989).
generalized social phobia: Effects of individual and
Exposure and cognitive restructuring for social pho-
group treatment. Behaviour Research and Therapy,
bia: A controlled study. Behavior Therapy, 20, 3-23.
31, 667-681.
Mersch, P.P.A. (1995). The treatment of social phobia:
Schwalberg, M.D., Barlow, D.H., Alger, S.A. and
The differential effectiveness of exposure in vivo and
Howard, L.J. (1992). Comparison of bulimics, obese
an integration of exposure in vivo, rational emotive
binge eaters, social phobics, and individuals with
therapy and social skill training. Behaviour Research
panic disorder on comorbidity across DSM-III-R
and Therapy, 33, 259-269.
anxiety disorders. Journal of Abnormal Psychology,
Mersch, P. P. A., Emmelkamp, P. M. G., Bögels, S. M.
101, 675-681.
and van der Sleen, J. (1989). Social phobia:
Smith, M. L., Glass, G. V. and Miller, T. I. (1980). The
Individual response patterns and the effects of beha-
benefits of psychotherapy. Baltimore, MD: John
vioral and cognitive interventions. Behaviour
Hopkins University Press.
Research and Therapy, 27, 421-434.
Stravynski, A., Lamontagne, Y. and Lavallee, Y. J.
Mersch, P. P. A., Jansen, M. A., and Arntz, A. (1995).
(1986). Clinical phobias and avoidant personality
Social phobia and personality disorder: Severity of
disorder among alcoholics admitted to an alcoholism
complaint and treatment effectiveness. Journal of
rehabilitation setting. Canadian Journal of
Personality Disorders, 9, 143-159.
Psychiatry, 31, 714-719.
Moreno, P. (1999). Tratamientos psicológicos para la
Stravynski, A., Marks, I and Yule, W. (1982). Social
fobia social: un estudio meta-analítico. Unpublished
skills problems in neurotic outpatients: Social skills
doctoral thesis, University of Murcia.
training with and without cognitive modification.
Orwin, R. G. (1983). A fail-safe N for effect size in
Archives of General Psychiatry, 39, 1378-1385.
meta-analysis. Journal of Educational Statistics, 8,
Taylor, S. (1996). Meta-analysis of cognitive-behavioral
157-159.
treatments for social phobia. Journal of Behavioral
Orwin, R. G. (1994). Evaluating coding decisions. In H.
Therapy and Experimental Psychiatry, 27, 1-9.
M. Cooper and L. V. Hedges (Eds), The handbook of
Turner, S. M., Beidel, D. C. and Townsley, R. M. (1992).
research synthesis (pp. 134-172). New York: Sage.
Social phobia: A comparison of specific and genera-
Rosenthal, R. (1991). Meta-analytic procedures for
lized subtypes and avoidant personality disorder.
social research (ed. rev.). Newbury Park, CA: Sage.
Salaberría, K., Borda, M., Báez, C. and Echeburúa, E.
Journal of Abnormal Psychology, 101, 326-331.
(1996). Tratamiento de la fobia social: Un análisis
Trower, P., Yardley, K., Bryant, B. and Shaw, P. (1978).
bibliométrico (1974-1994). Psicología Conductual, 4,
The treatment of social failure: A comparison of
111-121.
anxiety-reduction and skills-acquisition procedures
Salaberría, K. and Echeburúa, E. (1995). Tratamiento
on two social problems. Behavior Modification, 2, 41-
psicológico de la fobia social: Un estudio experimen-
60.
tal. Análisis y Modificación de Conducta, 21, 151-
Van Ameringen, M., Mancini, C., Styan, G. and
179.
Donison, D. (1991). Relationship of social phobia
Sánchez-Meca, J. and Marín-Martínez, F. (1997).
with other psychiatric illness. Journal of Affective
Homogeneity tests in meta-analysis: A Monte Carlo
Disorders, 21, 93-99.
comparison of statistical power and Type I error.
World Health Organization (1978). Mental disorders:
Quality and Quantity, 31, 385-399.
Glossary and guide to their classification in accor-
Schneier, F. L., Martin, L. Y., Liebowitz, M. R.,
dance with the ninth revision of the international clas-
Gorman, J. M. and Fyer, A. J. (1989). Alcohol abuse
sification of diseases. Geneva: Author.
in social phobia. Journal of Anxiety Disorders, 3, 15-
World Health Organization (1979). The international
23.
classification of diseases, 9th revision, Clinical
Schneier, F.R., Johnson, J., Horning, C.D., Liebowitz,
Modification. Geneva: Author.
M.R. and Weissman, M.M. (1992). Social phobia:
World Health Organization (1992). The ICD-10 classifi-
Comorbidity and morbidity in an epidemiological
cation of mental and behavioral Disorders:
sample. Archives of General Psychiatry, 49, 282-288.
Diagnostic criteria for research. Geneva: Author.
VOLUME 5. NUMBER 1. 2001. PSYCHOLOGY IN SPAIN
25

---