mechanismsofaction inyouthpsychotherapy

Mechanisms of action in youth psychotherapy

V. Robin Weersing1 and John R. Weisz21University of Pittsburgh School of Medicine, USA; 2University of California, USA

Background: In this review, we address a basic, but unanswered, question about psychosocialinterventions for youth: How does psychotherapy work? Methods: We propose a framework for usingmediation analysis to answer this question, and we review the youth therapy outcome literature forevidence on mediating mechanisms. We focus our review on clinical trials of empirically supportedtreatments for youth anxiety, depression, and disruptive behavior (N � 67). Results: Contrary toprevious reports indicating that potential mediators are rarely assessed, 63% of the studies includedmeasures of potential mediating mechanisms in their designs. Across treatment domains, percentagesranged from 22% of the studies of learning-based interventions for anxiety (i.e., systematic desensi-tization, modeling, and reinforced practice) to 91% of parent training investigations. Despite the ratherextensive assessment of potential mediators, only six studies included any attempt to use the measuresin a formal mediation test. Thus, despite the positive effects of treatments and surprisingly ampleassessment of mediators, we still know remarkably little about how youth psychotherapies work.Conclusions: We note common problems that hampered mediation testing (e.g., the design of manytrials made it dif®cult to determine the temporal order of change in the mechanism and outcome), andwe offer recommendations for improving study design to better assess mechanisms of therapeuticaction. We also note the need to test mediation among referred youth treated in representative practicesettings to complement the laboratory-based evidence on therapy mechanisms that prevails to date.Keywords: Behavior therapy, cognitive therapy, mediation, methodology, psychotherapy. Abbrevia-tions: CBT: Cognitive behavioral therapy; RCT: Randomized controlled trials; ADHD: Attention de®cithyperactivity disorder; EST: Empirically supported treatments; CWD: Coping with depression; PT:Parent training; MST: Multisystemic therapy; MTFC: Multi-dimensional treatment foster care; CMT:Cognitive mediation training.

Fifty years ago, Eysenck's (1952) famous review ofthe adult psychotherapy literature appeared in print,with its pronouncement that therapy was no moreef®cacious than the passage of time. Reviews of thechild and adolescent treatment literature followedand drew equally grim conclusions about the ef®cacyof therapy for youth (Levitt, 1957, 1963). These pro-vocative papers spurred a ¯urry of research on theeffects of psychotherapy and led to numerous re-®nements in nosology, assessment, and clinical trialdesign and analysis (see Kazdin, 1978, for review).

A half-century later, after 1500 studies (Durlak,Wells, Cotton, & Johnson, 1995; Kazdin, 2000) andfour major meta-analyses (Casey & Berman, 1985;Kazdin, Bass, Ayers, & Rodgers, 1990; Weisz, Weiss,Alicke, & Klotz, 1987; Weisz, Weiss, Han, Granger, &Morton, 1995), there is little argument that psycho-therapy can have a bene®cial impact on the lives oftroubled children. In randomized controlled trials(RCTs), therapy for youth routinely outperformswaiting list and attention-placebo control conditions,and, for selected youth problems, evidence is accu-mulating that some forms of therapy work betterthan others (e.g., Weisz et al., 1987, 1995; Weiss &Weisz, 1995). As a ®eld, clinical child psychology hasmoved from asking the general question `Can ther-apy work?' to identifying empirically supportedtreatment protocols for speci®c youth disorders (seeLonigan & Elbert, 1998).

Despite this progress, several fundamental ques-tions about the effects of youth therapy remain un-answered. While we have ample evidence to suggestthat psychotherapy for youth can produce positiveeffects in RCTs (ef®cacy), we have very little infor-mation about whether psychotherapy works outsideof laboratory settings (effectiveness; Weisz, Weiss, &Donenberg, 1992). It is also unclear whether thestatistically signi®cant effects of therapy in clinicaltrials translate into clinically signi®cant changes inchildren's functioning and distress (e.g., Kendall &Grove, 1988). The real-world effectiveness of therapyand the clinical signi®cance of therapeutic changeare clearly matters of consequence, and we makereference to these issues throughout this review.However, the majority of our attention in this reviewis focused on yet another unanswered question inchild and adolescent therapy research. When youthpsychotherapy `works,' what are the mechanisms ofaction that produce therapeutic change?

Surprisingly little research exists to answer thisbasic question. At last review, it was reported thatless than 3% of published clinical trials of youthpsychotherapy included measures of the processesthought to underlie intervention effects (Kazdin,Bass, Ayers, & Rodgers, 1990). For those unfamiliarwith treatment outcome research, this may seem tobe a rather odd state of affairs: Treatments have beendeveloped, tested in clinical trials, re®ned, retested,

Journal of Child Psychology and Psychiatry 43:1 (2002), pp 3±29

Ó Association for Child Psychology and Psychiatry, 2002.Published by Blackwell Publishers, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA

and proposed for general adoption, all without a clearunderstanding of what makes therapy therapeutic.

As an example, consider the case of cognitive-behavioral therapy (CBT) for youth depression. CBTis the most widely tested psychosocial interventionfor depressed youth, employed in 13 of the 15published RCTs (Brent, Gaynor, & Weersing, inpress). The theory of psychopathology underlyingCBT posits that a major cause of depression iscognitive distortions ± faulty information processingthat leads youth to take unrealistically negativeviews of themselves, the world, their future, and thecauses of signi®cant events in their lives (Beck,Rush, Shaw, & Emery, 1979; cf. Abramson, Metal-sky, & Alloy, 1989). CBT techniques, such as cog-nitive restructuring, are designed to interrupt andremediate these distorted ways of thinking and,through this mechanism, produce changes in thebroader syndrome of depression. At ®rst blush, re-sults of RCTs would seem to support the CBT model.Meta-analyses of the effects of CBT with depressedadolescents have yielded impressively large effectsizes (1.27, Lewinsohn & Clarke, 1999; 1.06, Rei-necke, Ryan, & Dubois, 1998). In addition, prelim-inary evidence suggests that CBT outperformsfamily and supportive therapies for depressed youth± treatments that do not speci®cally target depres-sive cognitions (Brent et al., 1997). Despite this in-direct evidence in support of the CBT model, theonly investigation that has directly tested mecha-nisms of action in a CBT trial found that change incognitive distortions actually did not mediate theeffects of treatment on depression symptoms (Kolko,Brent, Baugher, Bridge, & Birmaher, 2000). These®ndings present a substantial theoretical challengeto the CBT model. CBT may be an ef®cacious in-tervention and may be the psychosocial treatment ofchoice for depressed youth. However, given thelimited evidence available, it is possible that thesebene®cial CBT effects may have little to do with thetheoretical model of psychopathology upon whichCBT interventions are based.

This brief example illustrates the limits of ourknowledge about the mechanisms of action in in-terventions for youth. The example also demon-strates the useful role that studies of therapymechanisms can play in testing theoretical modelsof psychopathology. Ethical and practical con-straints limit the extent to which investigators canconduct experimental research on the origins andmaintenance of psychological dysfunction. Evenwere it possible, no researcher would suggest ran-domly assigning children to abusive parents orsubjecting youth to a string of uncontrollable lossesand failures. Psychotherapy research offers aunique opportunity to manipulate the processesinvolved in psychopathology, in the service of alle-viating symptoms in already impaired youth. Thisview of treatment research as an explanatory andtheory-testing tool is not new (e.g., Judd & Kenny,

1981; Scott & Sechrest, 1989; Cicchetti & Toth,1992; Kazdin, 1999), and yet commentators con-tinue to decry the small number of clinical trialsthat adopt theory-testing as a goal (e.g., Kazdin,2000).

We agree that it is the rare clinical trial that ex-plicitly sets out to test theory while demonstratingef®cacy (for a laudatory exception, see Eddy &Chamberlain, 2000). However, in our reading of theyouth therapy outcome literature, we have beenstruck by the number of RCTs that could have in-vestigated treatment mediation, but did not do so.We have noticed that many studies of CBT havemeasured cognitive distortions; investigations of ex-posure therapy have included measures of arousaland habituation; and studies of parent training haveassessed changes in discipline practices. In each ofthese cases, there have been opportunities to in-vestigate mediators of treatment effects. In the main,investigators have seemed to treat these possiblemechanisms as simple outcome variables, on parwith changes in symptoms, diagnoses, and func-tioning. In short, we have seen a good deal of unre-alized potential in clinical trial research.

In this review, we sought to uncover and explorethis untapped information and, in doing so, dem-onstrate how studies of therapy effects might bettercapitalize on opportunities for theory testing. Webegin by describing procedures for testing medi-ation, and we illustrate how mediational analysisprovides a useful framework for testing psychologi-cal theories. We next use this mediational frameworkas a template for reviewing the youth psychotherapyoutcome literature. In line with previous reports(e.g., Kazdin et al., 1990), we expected that very fewstudies would explicitly investigate mechanisms ofaction in youth psychotherapy. However, we didsuspect that many RCTs would include measuresthat could have been used to investigate mediationalhypotheses and that many of these studies wouldreport results bearing, at least indirectly, on mech-anisms of change. To test this notion, we conducteda focused review of what are arguably the mostthoroughly tested treatment programs available foryouth, namely those identi®ed by the Task Force onEmpirically Supported Procedures for Youth.1 TheTask Force reviewed treatments for youth withattention-de®cit hyperactivity disorder (ADHD),anxiety, autism, depression, and disruptive behaviordisorders. As no empirically supported treatments(ESTs) were identi®ed for autism, we excludedautism from our review. Empirically supportedtreatments were identi®ed for ADHD; however, much

1 Readers are referred to Lonigan, Elbert, and Johnson (1998)

for a complete description of the EST criteria, study selection

procedures, and history and mission of the EST review

committees. Findings of each of the disorder-speci®c review

committees were published in a special issue of the Journal of

Clinical Child Psychology (Lonigan & Elbert, 1998).

4 V. Robin Weersing and John R. Weisz

of the treatment research on ADHD used single-subject designs, raising unique issues in testingand interpreting mediating relationships. We, thus,limited our review to ESTs for youth anxiety,depression, and disruptive behavior problems.

Testing mediation in clinical trials

Answering the question `How does youth psycho-therapy work?'2 requires the identi®cation and as-sessment of mechanisms by which interventionsaffect change in clinically relevant outcomes. Thisprocess is fundamentally a search for mediatedeffects. Thus, before embarking upon our review ofthe EST clinical trials, we (a) brie¯y review the pro-cess of testing mediated effects and (b) illustrate howeach step in mediational analysis maps on to a test ofa different theoretical proposition about the effects oftherapy.

Mediation is established by demonstrating fourlogical relationships among treatment, mediator,and outcome (see Judd & Kenny, 1981; Baron &Kenny, 1986). In psychotherapy research, each ofthese analytic steps asks a different question aboutthe effects of therapy (see Figure 1). First, is psy-chotherapy ef®cacious? Second, does interventionaffect speci®c mechanisms? Third, do therapymechanisms affect psychopathology? And fourth,can therapy effects be accounted for through thiscausal pathway? We review the logic of these steps ingreater detail below.

Step one: Ef®cacy test

The ®rst step in a mediational analysis is to test thelink between participation in a treatment and im-proved symptoms or functioning. This step assessesthe ef®cacy of an intervention relative to a compar-ison condition, most often a wait list or no treatmentcontrol group. If intervention is not more ef®caciousthan control, then the search for mechanisms un-derlying treatment effects logically ceases. Shouldtreatment prove ef®cacious, the next three steps inanalysis unpack how therapy accomplishes thisbene®cial change.

In most clinical trial designs, symptoms are as-sessed at pre-treatment, post-treatment, and, lessfrequently, long-term follow-up assessments (Weisz,Huey, & Weersing, 1998). This assessment strategyis suf®cient for establishing ef®cacy but may causeproblems in the analysis of mediated effects. Estab-lishing that change in the mediator causes changesin outcome requires that the mediator change beforeoutcome. As will be discussed later, this may requiremore frequent and more sensitive assessments than

those necessary to simply establish the ef®cacy of anintervention.

Step two: Intervention/Speci®city test

In the second step of analysis, the relationship be-tween treatment and change in the candidate medi-ator is assessed. This step tests whether treatmentaffects the mechanism of action hypothesized to pro-duce intervention effects (theory of intervention; Scott& Sechrest, 1989; Kazdin, 1999). In clinical trial de-signs where active treatments are being compared,this analysis is often referred to as a test of treatmentspeci®city (e.g., Kolko et al., 2000). If treatment af-fects the speci®c, hypothesizedmechanismof change,mediation analysis proceeds to the next step.

If no relationship is found between treatment andmediator, several factors may be at work. First, themodel of intervention may be missing an importantmediating pathway. Thismaybe particularly likely formulti-component interventions that target multiplemediating processes (e.g., cognitive distortions, socialskills, family con¯ict). Second, a null ®nding may bedue to problems in the timing ormethod ofmeasuringthe mediating process. Logically, assessment oftreatment mechanisms should occur early enough inintervention that the temporal sequence betweenchanges in the mediator and later changes in symp-toms can be established. However, if assessment isconducted too early, themeasure of themediatormaynot be suf®ciently sensitive to detect treatment ef-fects, producing an inaccurate ®nding of no relation-ship between treatment and mediator. As a thirdpossibility, the relationship between participation intreatment and change in themediatormay bemaskedby the presence of signi®cant moderators; for exam-ple, if treatment is delivered with low ®delity for someportion of the sample (Yeaton & Sechrest, 1981).

Figure 1 Mediational analysis as a framework for test-ing psychological theories

2 Of course, answering this question also requires de®ning

`psychotherapy' and `work.' By focusing our review on the EST

clinical trials, we side-step some of these complexities.

Mechanisms of action in youth psychotherapy 5

Step three: Psychopathology test

The third step in mediation analysis evaluates thesigni®cance of the path between the mediator andtreatment outcome. This analysis can be viewed asproviding a partial test of the theory of psychopa-thology upon which the intervention is based. Forexample, Dodge and colleagues have proposed amodel linking early child abuse, development ofbiased social information processing, and subse-quent violent behavior in youth (Dodge, Bates, &Pettit, 1990). Interventions for violent youth basedon the work of Dodge and colleagues target biasedsocial information processing in an effort to decreaseaggression (e.g., Lochman, Burch, Curry, & Lam-pron, 1984). Establishing a link between change insocial information processing and change in ag-gression would not only help to establish the causalchain of treatment effects, but it also would providesome validation for the Dodge model. As discussedearlier, the change in the mediating process logicallyshould precede changes in the terminal outcome.

A ®nding of no signi®cant relationship betweenmediator and outcomemay indicate that themediatoris not an important maintaining factor in psychopa-thology. Alternately, the link between mediator andoutcome may suffer from the measurement problemsdiscussed in step two (i.e., poor timing of assessment,poor sensitivity of measures). And, as in step two, therelationship between mediator and outcome may bemoderated by other variables, and these moderatorsmay mask signi®cant relationships between media-tors and outcome. Returning to our social informa-tion-processing example, it has been argued thatthere are several distinct subtypes of antisocial youth(see e.g., Mof®tt, 1993; Loeber, 1990; Loeber et al.,1993). Some of these subtypes may exhibit consis-tently biased information processing (e.g., `reactive'aggressors), while others may have more speci®c in-formation processing de®cits. Thus, subtype of dis-order could moderate the pathway between mediatorand outcome, and the signi®cance of the relationshipbetween mediator and outcome may vary dependingon sample composition.3

Step four: Mediation test

In the fourth and ®nal step of analysis, the originalrelationship between treatment and outcome, es-tablished in the ef®cacy test, is examined whilecontrolling for the relationships between treatmentand process and between process and outcome. If

change in hypothesized mechanism mediates therelationship between treatment and outcome, thepreviously signi®cant path between treatment andoutcome should be substantially or wholly elimin-ated when controlling for the other paths in themodel (see Baron & Kenny, 1986; Holmbeck, 1997;for analytic issues). Thus, by demonstrating medi-ation, it is possible, in principle, to show that psy-chotherapy `works' and that it works through themechanisms speci®ed in the treatment's theories ofintervention and psychopathology.

Of course, models may be more complicated thanthe simple single mediator model of Figure 1. Treat-ment may affect outcome through multiple inde-pendent mediators, through a chain of related,sequential processes, or in a non-linear fashion.Models may be analyzed with techniques as simpleas regression or with elaborate structural equationmodels. As discussed earlier, the relationships link-ing treatment±mediator and mediator±outcome maybe moderated by other variables. Similarly, there arelikely optimal times to intervene in the course ofyouth disorder, and developmental and contextualvariables may moderate the overall ef®cacy of inter-vention attempts (Cicchetti & Toth, 1992; Weisz &Hawley, in press). We ®nd the basic four-step logic ofmediation a useful starting place to begin thinkingabout these more complicated issues.

We also ®nd mediational analysis a useful con-ceptual framework to organize what we know and donot know about how psychotherapy for youth works.We have a great deal of information about the ®rststep in mediation analysis ± the ef®cacy of therapiesfor youth. In the remainder of this article, we reviewthe EST treatments for youth anxiety, depression,and disruptive behavior problems and seek to un-cover information about the remaining steps inmediational analysis. In this review, it is not our in-tention to provide an exhaustive summary of theresults and methodological details of the EST studies(see Lonigan & Elbert, 1998). Rather, we selectivelyreview ®ndings that bear on the theoretical models ofintervention and psychopathology hypothesized tounderlie EST effects.

Treatments for anxious and phobic youth

While many children experience transient fears andworries over the course of normal development, asubstantial portion of youth suffer from enduringand impairing levels of anxiety. Indeed, anxiety dis-orders are the most common diagnoses in youth(Bernstein & Borchardt, 1991), with communityprevalence estimates ranging as high as 20% (Bell-Dolan, Last, & Strauss, 1990). Untreated, youthanxiety disorders interfere with academic achieve-ment and the development of age-appropriate socialrelationships (e.g., Strauss, Frame, & Forehand,1987; Beidel, Turner, & Morris, 1999). In addition,

3 The presence of a signi®cant moderator of the mediator±

outcome relationship may indicate that the mediation model is

misspeci®ed. Our example model may require the inclusion of

additional mediating mechanisms, such as association with

deviant peers, in order to ®t the population of disruptive youth.

Alternately, the model could be left as is but restricted to only

apply to a subgroup of disruptive youth.


anxiety in youth may persist through childhood andadolescence into the adult years (Keller et al., 1992;Last, Hersen, Kazdin, Francis, & Grubb, 1987).

Modern theories of the etiology and maintenanceof anxiety disorders focus on the interplay between(a) biological vulnerability to acute stress reactions(e.g., Biederman et al., 1993); (b) the experience ofuncontrollable stressful life events (e.g., Chorpita &Barlow, 1998); (c) learned, maladaptive behavioralresponses to threat (e.g., parental avoidance beha-vior; Dadds, Barrett, Rapee, & Ryan, 1996); and (d)inaccurate, overly-threatening, cognitive interpreta-tions of events (e.g., anxious apprehension; Barlow,1988). Speci®c intervention programs for anxiousyouth target different combinations of these patho-logical processes. We now review the EST clinicaltrials for youth anxiety and phobias and evaluate theextent to which they assessed these possible mech-anisms of action. We broadly group the EST treat-ments into learning-based and cognitive-behavioralinterventions.

Learning-based interventions

We classify three of the empirically supported treat-ments for phobic youth as learning-based interven-tions: (a) systematic desensitization, (b) modeling,and (c) reinforced practice. Learning theories ofanxiety view pathological fears and phobic behaviorsas acquired responses. Thus, in learning-basedinterventions, anxious youth engage in activitiesdesigned to promote the `unlearning' of old fearresponses and the learning of new associations andcoping behaviors. This is generally accomplishedthrough graded exposure to fear-inducing stimuli.

Readers are referred to Ollendick and King (1998)for a detailed review of studies and a thorough des-cription of the EST interventions. In short, system-atic desensitization involves exposure to a hierarchyof feared situations, while anxious youth engage incompeting non-fear responses, such as relaxation.According to the original theory of desensitization,this exposure procedure should result in `counterconditioning' ± in which fearful responses are in-hibited by the newly conditioned non-fear responses(Wolpe, 1954). In mediational terms, systematicdesensitization should reduce arousal to fearedstimuli, which in turn should cause changes in thebroader subjective experience of anxiety.

Modeling treatments also rely on exposure toanxiogenic stimuli, but they are presumed to reduceanxiety through observational learning rather thanPavlovian conditioning. In a modeling intervention,youth repeatedly observe others interacting withfeared stimuli and see that no ill-effects ensue (e.g.,Bandura, Blanchard, & Ritter, 1969). Participantmodeling, the most ef®cacious of the modelingtreatments, follows up observation with youth par-ticipation in the same, feared activities. Modelinghas been hypothesized to work by (a) discon®rming

anxious youths' catastrophic expectations aboutengaging in feared behaviors; (b) enhancing youths'perceptions of control and environmental contin-gency; (c) teaching youth new, adaptive responses tothreat; and (d) promoting physiological habituationover the course of exposure (cf. Foa & Kozak, 1986;Mineka & Thomas, 1999).

Reinforced practice and other operant proceduresreward youth for engaging in feared behaviors (e.g.,petting a scary dog; Obler & Terwilliger, 1970). Aswith the other two learning-based interventions,youth are gradually exposed to feared stimuli. How-ever, in reinforced practice, the focus is teachingapproach behavior rather than reducing anxiousarousal per se. Through the experience of engagingin feared behaviors, youths may gain behavioraldiscon®rmation of feared catastrophic outcomes,learn adaptive coping behaviors, and have an op-portunity for physiological habituation over thecourse of the treatment (Mineka & Thomas, 1999).These processes may well serve as mediators oftreatment effects.

Evidence from the EST clinical trials indicates thatthese three learning-based interventions are ef®ca-cious treatments for phobic youth, with participantmodeling and reinforced practice attaining well-established treatment status (see Ollendick & King,1998).4 Thus, all three interventions have estab-lished the ®rst link in a mediational analysis ±demonstrating that treatment affects outcome. Asdiscussed previously, these interventions may `work'by reducing arousal, teaching adaptive coping, pro-viding evidence against catastrophic cognitions,and/or establishing a sense of control. We examinedeach of the RCTs to determine if any of these possiblemechanisms were measured, and, if so, if mediationwas tested. In Table 1, we summarize the results ofthis review. As can be seen in the table, 14 of the 18studies did not measure any of the putative treat-ment mediators, and none of the investigations ofreinforced practice assessed possible mechanismsof action. Of the four studies that did assesspathological processes, all measured arousal or ha-bituation to anxious stimuli at the end of treatment.

4 The Task Force on Empirically Supported Procedures for

Youth identi®ed empirically supported treatments as either

`well-established' or `probably ef®cacious' (see Lonigan, Elbert,

& Johnson, 1998). To be designated well-established, a treat-

ment must have been tested by independent teams of inves-

tigators and shown to be (a) more ef®cacious than placebo or

alternate treatment; or (b) as ef®cacious as an already estab-

lished treatment. In contrast, probably ef®cacious treatments

have either (a) met all the criteria for well-established status,

except independent replication; or (b) been shown to be more

ef®cacious than a no-treatment control group by independent

research teams. Alternate criteria are applied for interventions

supported by a series of single-case designs. The distinction

between well-established and probably ef®cacious is not

critical for the purposes of this review.


Table

1Mechanismsofactionin

learn

ing-b

asedESTsforphobic

youth

Clinicaltrial

Treatm

entconditions

Candidate

mediators

Ef®cacytest

Interventiontest

Psychopath

ology/Mediationtests

Bandura

etal.(1967)

MOD

livepositive

MOD

liveneutral

Dogalone

Positivealone

Notassessed

MOD

livetreatm

ents

superiorto

dogalone

orpositivecontext

alone

Notassessed

Notassessed

Mann&

Rosenth

al

(1969)

MOD

live

SYSin

vitro

Notx

Notassessed

Activetreatm

ents

superiorto

notx

Notassessed

Notassessed

Bandura

&Menlove

(1968)

MOD

®lm

multi

MOD

®lm

one

Attention

Notassessed

Activetreatm

ents

superiorto

attention

Notassessed

Notassessed

Hilletal.(1968)

MOD

®lm

Notx

Notassessed

MOD

®lm

superior

Notassessed

Notassessed

Ritter(1968)

MODP

MOD

Notx

Notassessed

Activetreatm

ents

superiorto

notx.

MODPsuperiorto

MOD

alone

Notassessed

Notassessed

Bandura

etal.

(1969)

MODP

MOD

®lm

SYSin

vitro

Waitlist

Arousal

(selfra

ted)

Activetreatm

ents

superiorto

wait

list.

MODPsuperiorto

altern

ate

treatm

ents

Activetreatm

ents

hadbettereffectth

an

wait

list.

MODPhad

besteffectin

some

analyses.

Notassessed

Blanchard

(1970)

MODP

MOD

+info

MOD

Notx

Arousal

(selfra

ted)

Activetreatm

ents

superiorto

notx.

MODPsuperiorto

altern

ate

treatm

ents

onsomemeasures

Activetreatm

ents

hadbettereffectth

an

notx.MODPhad

bettereffectth

an

MOD

+info.

Mediationnot

assessed.Arousal

signi®cantlyrelatedto

avoidanceand

anxiousattitudes

Murp

hy&

Bootzin

(1973)

MODPactive

MODPpassive

Notx

Notassessed

Activetreatm

ents

superior

Notassessed

Notassessed

Lewis

(1974)

MODP

MOD

Participation

Attention

Notassessed

MODPsuperiorto

MOD,participation,

andattention

Notassessed

Notassessed

Obler&

Terw

illiger

(1970)

Opera

nt

Notx

Notassessed

Opera

ntsuperior

Notassessed

Notassessed

Leitenberg

&Callahan(1973)

Opera

nt

Notx

Notassessed

Opera

ntsuperior

Notassessed

Notassessed

Sheslow

etal.

(1983)

Opera

nt

Copingself-talk

Opera

nt+coping

Attention

Notassessed

Opera

nttreatm

ents

superiorto

coping

self-talk

andattention

Notassessed

Notassessed


Table

1Continued

Clinicaltrial

Treatm

entconditions

Candidate

mediators

Ef®cacytest

Interventiontest

Psychopath


Menzies&

Clark

(1993)

Opera

nt

MOD

live

Opera

nt+MOD

Notx

Notassessed

Opera

nttreatm

ents

superiorto

MOD

liveandnotx

Notassessed

Notassessed

Kondas(1967)

SYSin

vitro

Relaxationonly

Hiera

rchyonly

Notx

Arousal

SYSin

vitro

superior

toaltern

ate

treatm

ents

andnotx

SYShadbettereffect

thanaltern

ate

treatm

ents

andnotx

Notassessed

Mann&

Rosenth

al

(1969)

SYSin

vitro

MOD

live

Notx

Notassessed

Activetreatm

ents

superiorto

notx

Notassessed

Notassessed

Milleretal.(1972)

SYSin

vitro

Eclectictx

Wait

list

Notassessed

Activetreatm

ents

superiorto

wait

list

Notassessed

Notassessed

Bara

basz(1973)

SYSin

vitro

Notx

Arousal,butno

oth

ermeasure

ofoutcome

SYSin

vitro

superiorto

notx

SYShadbettereffect

onarousal

Notapplicable,

arousalonly

measure

ofoutcome

Kuroda(1969)

SYSin

vivo

Notx

Notassessed

SYSin

vivo

superiorto

notx

Notassessed

Notassessed

Ulteeetal.(1982)

SYSin

vivo

SYSin

vitro

Notx

Notassessed

SYSin

vivosuperior

toSYSin

vitro

and

notx

Notassessed

Notassessed

NOTE:Stu

diesare

dividedinto

blocksto

re¯ectth

eclassi®cationsystem

usedbyth

eTaskForceonEmpiricallySupportedProcedures.Shadedentrieshavebeenidenti®edas`well

established'(O

llendick&

King,1998).SYS,systematicdesensitization;MOD,modeling;MODP,participantmodeling


Mediating role of arousal and habituation. In thefour studies in which arousal was assessed, treat-ment signi®cantly lowered arousal to previously fear-inducing stimuli. However, only one study (Blan-chard, 1970) assessed whether changes in arousalwere related to changes in the broader syndrome ofanxiety. In this study, Blanchard (1970) found thatparticipant modeling was signi®cantly more ef®ca-cious than vicarious modeling and a no-treatmentcontrol condition in reducing fear of snakes. In ad-dition, participant modeling produced signi®cantlygreater reductions in arousal than the other condi-tions, and arousal was signi®cantly related tomeasures of anxious avoidance and attitudes. Theseanalyses provide limited support for the theory thatmodeling works partly through arousal reductionand that inappropriate arousal is an importantcomponent of anxious symptomatology. However,the design of the investigation leaves substantialquestions about whether arousal actually mediatedtreatment gains or whether changes in arousal sim-ply represented reductions in overall anxiety (i.e.,arousal was measured at treatment termination,simultaneously with the other outcome measures).In addition, statistical procedures for formally test-ing mediation did not receive wide exposure until 16years after the publication of this investigation(Baron & Kenny, 1986), and, unsurprisingly, Blan-chard did not conduct analyses to test for mediatedeffects.

Cognitive-behavioral interventions

Three CBT programs have been identi®ed as prob-ably ef®cacious treatments for youth anxiety andphobias: (a) verbal self-instructional training forphobic youth (Kanfer, Karoly, & Newman, 1975;Graziano & Mooney, 1980); (b) the Coping Cat pro-gram for youth with a range of anxiety diagnoses(Kendall, 1994; Kendall et al., 1997; Barrett, Dadds,& Rapee, 1996); and (c) an extension of the CopingCat program, including a family therapy component(Barrett et al., 1996). These CBT interventions haveas a primary focus modifying the inaccurate cogni-tions characteristic of anxious and phobic youth.Anxious individuals tend to erroneously interpretambiguous situations as threatening, over-estimatethe likelihood of dangerous events, view the world asunsafe and full of risk, and perceive themselves asunable to successfully cope with these threats (seee.g., Barlow, 1988). Changes in these anxious modesof thinking are hypothesized to mediate the effects ofCBT on anxiety symptoms.

In verbal self-instructional training, phobic youthare taught to use positive, coping self-statements tocounter their anxious automatic thoughts. TheCoping Cat intervention utilizes this technique, aswell as more formal cognitive restructuring methods(e.g., monitoring and identifying irrational thoughts,challenging these cognitions, and developing more

realistic, coping thoughts). Coping Cat also includeselements of the learning-basedESTs for phobic youth,such as reinforcement for `brave' behavior; gradedexposure to fear-inducing stimuli; and training inadaptive coping skills (e.g., relaxation and socialproblem-solving skills). In the family extension toCoping Cat (CBT + FAM), parents are taught toeffectively carry out the CBT protocol at home (e.g., byrewarding brave behavior) and to better manage theirown anxiety, in order to avoid modeling anxiousthoughtsandbehaviors to their children.Thesemulti-componentCBTpackagesmayachieve their effects bychanging anxious thinking or through behavioral andphysiological pathways (e.g., arousal reduction; seedescription of the learning-based interventions).

Although there are far fewer clinical trials of CBTthan of learning-based interventions, a higher pro-portion of the CBT studies included tests of therapymechanisms. While only 22% of learning-basedESTs measured possible mediators, over half of CBTinterventions included measures of cognitive pro-cesses targeted by the treatments. As can be seen inTable 2, the Coping Cat studies measured changesin children's anxious self-talk, and CBT + FAM as-sessed children's interpretations of ambiguous situ-ations, with and without parental input. We examinethese studies in greater detail below.

Mediating role of self-talk. The Coping Cat inter-vention is the product of a program of research onyouth anxiety by Kendall and colleagues. In twoclinical trials by the Kendall team, the interventionhas proven more ef®cacious than wait list control atpost-treatment (Kendall, 1994; Kendall et al., 1997)and over three-year follow-up (Kendall & Southam-Gerow, 1996). The superiority of Coping Cat overwait list also has been documented by Barrett andcolleagues (Barrett et al., 1996) ± the research teaminvestigating the utility of adding family treatment tothe basic Coping Cat protocol.

Referencing our mediational framework, CopingCat has established the ef®cacy link between treat-ment participation and positive symptomatic out-come. Coping Cat also has been shown to impactpotentially important mediators of intervention ef-fects. Using data from both clinical trial samples,Treadwell and Kendall (1996) tested whether chan-ges in anxious and negative self-talk mediatedtreatment effects. Support was found for the medi-ation hypothesis. Youth who received a course ofCoping Cat endorsed signi®cantly fewer negativeself-statements on a paper-and-pencil measure ofself-talk. In turn, changes in number of negative self-statements and the balance of positive to negativeself-talk mediated the impact of treatment on out-come. Results for the content speci®city of changesin anxious versus depressed self-talk were mixed.

While the Coping Cat research provides the mostcomplete information available on therapy mechan-ism in the treatment of anxious youth, the work has


several limitations. As with the Blanchard (1970) in-vestigation of participant modeling, the putative me-diator in the Coping Cat clinical trials was measuredsimultaneously with outcome, making it dif®cult todetermine the temporal order of changes in self-talkand changes in anxious symptomatology. In addi-tion, the relationship between self-talk and anxietymay have been due, in part, to shared measurementvariance. Self-talk was assessed by youth self-report,and changes in self-talk only mediated treatmentgains for youth-report, dimensional measures ofanxiety. Self-talk did not mediate treatment effectsfor teacher report on symptom scales or for parentsymptom measures and diagnostic interviews.

Mediating role of threat interpretations andavoidance. Additional evidence in support of thecognitive mediator model of CBT is provided by thework of Barrett and colleagues. Their CBT + FAMintervention has been shown to be more ef®caciousin reducing anxiety than Coping Cat alone and to besuperior to wait list control conditions (Barrett et al.,1996). Additionally, the clinical trial included`experimental measures' that appear to be potentialmediators of treatment effects. Anxious youth wereasked to interpret and respond to a series of ambi-guously threatening situations. Youth responses

then were rated on how threatening they viewed thescenarios and whether they chose to cope with thesituations by avoidance. Following these individualresponses, youth and their families jointly evaluatedtwo additional scenarios ± one involving unexplainedphysical sensations (feeling `funny in the tummy')and the other focusing on unclear social feedback(peers laughing). From this procedure, Barrett et al.calculated mean threat scores, mean avoidancescores, and scores indexing the impact of anxiogenicfamily processes on threat interpretations andavoidant behaviors.

Overall, Barrett et al. (1996) found that bothCBT + FAM and Coping Cat produced signi®cant re-ductions in children's threat interpretations and useof avoidant coping responses, both before and afterfamily discussion. CBT + FAM appeared to be some-what more effective than Coping Cat in reducingavoidant responding (Coping Cat was not superior towait list). However, the two active interventions didnot differ in threat and avoidance scores taken afterfamily discussion, a somewhat surprising result gi-ven the extra emphasis on family processes in CBT +FAM. In this investigation, the relationships betweenchange in threat and avoidance and change in overallanxiety symptoms were not assessed, and formalmediational analyses were not conducted.

Table 2 Mechanisms of action in cognitive-behavioral ESTs for anxious and phobic youth

Clinical trialTreatmentconditions

Candidatemediators Ef®cacy test Intervention test

Psychopathology/Mediation tests

Kendall (1994) Coping CatWait list

Content ofself-talk

CBT superiorto wait list onmultiplemeasures

CBT had bettereffect than waitlist on self-talk incombined samplefrom 1994 and1997 clinical trials

Self-talk mediatedtreatment gains incombined samplefrom 1994 and 1997clinical trials(Treadwell &Kendall, 1996)

Kendall et al.(1997)

Coping CatWait list

Content ofself-talk

CBT superiorto wait list onmultiplemeasures

Intervention linkdemonstrated(see above)

Psychopathology andmediation linksdemonstrated(see above)

Barrett et al.(1996)

Coping CatCoping Cat FAMWait list

Threat interpretationsAvoidant solutionsFamily impact onthreat and avoidance

CBT + familysuperior to CBT.

Active treatmentssuperior towait list

Active treatmentshad better effecton candidatemediators thanwait list.

Slight superiorityof CBT + family.

Not assessed

Kanfer et al.(1975)

CBT competenceCBTPlacebo

Not assessed CBT withcompetencefocused self-instructionssuperior

Not assessed Not assessed

Graziano &Mooney (1980)

CBTWait list

Not assessed CBT superiorto wait list


NOTE: Studies are divided into blocks to re¯ect the classi®cation system used by the Task Force on Empirically SupportedProcedures. CBT, cognitive-behavioral therapy


Summary

In sum, the EST clinical trials for anxious and phobicyouth provide only limited information on possiblemechanisms of therapeutic action. Evidence is espe-cially thin for learning-based interventions. Despitethe many possible mediators of these treatments' ef-fects, the studies of systematic desensitization andmodeling only assessed arousal, and only one of thesestudies veri®ed that changes in arousal were relatedto changes in anxiety more generally (Blanchard,1970). None of the clinical trials of reinforced practiceassessed candidate mediators.

Evidence on therapy mechanism is somewhatstronger for cognitive-behavioral interventions. Morethan half of the EST studies of CBT included meas-ures of the cognitive processes hypothesized to causethe interventions' effects. In all of these studies, CBTdid change both the cognitive mediator and symp-tomatic outcome, and, for the Coping Cat interven-tion, change in self-talk from pre- to post-treatmentmediated treatment gains.

Across clinical trials, re®nements in study designand measurement technology may have allowed forstronger inferences about therapy mechanism. Forexample, all of the studies that assessed possiblemediators did so at treatment termination, muddy-ing the temporal relationship between change in themechanism and change in outcome. In addition, asmall range of measurement strategies was em-ployed, with a general reliance on self-report ratingscales for both mediator and outcome assessment.This substantial method overlap raises questionsabout the validity of observed mediational relation-ships. We also suspect that the small number oflearning-based clinical trials measuring mechanismmay be partially due to the dif®culty in devising self-report scales for many of the mediators of interest(e.g., changes in conditioned associations betweencognitive stimuli and affective responses). New workon performance-based methods of assessment mayallow for more precise measurement of these con-structs (for review, see Vasey & Lonigan, 2000).

Treatments for depressed youth

It is only in the past 30 years that depression hasbeen recognized as a signi®cant psychiatric problemin children and adolescents. Indeed, recent epide-miological work in youth depression has suggestedthat the disorder (a) has a high lifetime prevalence bythe onset of puberty (28%; Lewinsohn & Clarke,1999); (b) may produce lifelong impairments in so-cial and occupational functioning (see, e.g., Rohde,Lewinsohn, & Seeley, 1994), and (c) substantiallyincreases the risk of early mortality by suicide(Shaffer et al., 1996; Gould et al., 1998).

Two major psychological theories have been pro-posed to explain the etiology and maintenance of

depression. Cognitive theory postulates that thesyndrome is the result of inaccurate, overly negativeviews of the self, the world, and possibilities for thefuture (Beck et al., 1979). Under situations of stress,this negative cognitive triad is hypothesized to in-terfere with accurate information processing andlead individuals to become increasingly despondent.Learned helplessness theory places a heavier em-phasis on the environmental and behavioral ante-cedents of depression (Abramson et al., 1989;Abramson, Seligman, & Teasdale, 1978). Depressionis thought to arise from a history of environmentalnon-contingency that results in negative, global, andstable beliefs about the world and the individual'sability to control his or her fate. As with the Beckmodel, this pessimistic explanatory style is believedto interfere with adaptive coping and to lead to be-havioral avoidance, withdrawal, and subsequentdepression. Both cognitive and learned helplessnesstheories were developed to explain the phenomenonof adult depression; however, there is a growing bodyof evidence that depressed youth exhibit patterns ofdepressogenic information processing similar to de-pressed adults (for a review, see Gladstone & Kaslow,1995).

Given the relative recency of research on depres-sion in youth, it is perhaps not surprising that thereare few child and adolescent depression clinicaltrials. At last review, a total of 15 RCTs had beenpublished, 13 of which assessed the effects of cog-nitive-behavioral interventions (Brent et al., inpress). The EST Task Force identi®ed two of theseCBT treatments as probably ef®cacious (see Kaslow& Thompson, 1998), and we review these interven-tions below. For youth depression, we also expan-ded our review to include all published CBT clinicaltrials. This decision was based on the marked si-milarity between the two speci®c CBT treatmentprograms featured in the Kaslow and Thompsonreview and the CBT programs used in other pub-lished youth depression trials.5

Cognitive-behavioral interventions

Two CBT programs for youth depression have beenidenti®ed as probably ef®cacious treatments: (a) theCBT program for depressed children, developed byStark and colleagues (Stark, Reynolds, & Kaslow,1987; Stark, Rouse, & Livingston, 1991); and (b) theCoping with Depression course (CWD), originallydeveloped for depressed adults by Lewinsohn andcolleagues and later adapted for adolescents (Lew-

5 Several of the Task Force review teams adopted a generic

approach to identifying ESTs and rated the empirical support

for broad categories of treatment; for example, identifying `CBT'

as a probably ef®cacious treatment for phobias (Ollendick &

King, 1998). Kaslow and Thompson appear to have adopted

more of a brand-name approach and, thus, evaluated the

empirical support of speci®c manualized protocols.


insohn, Clarke, Hops, & Andrews, 1990; Lewinsohn,Clarke, Rohde, Hops, & Seeley, 1996). An additionalten published RCTs have tested the effects ofCBT for depressed children and adolescents (seeTable 3). Across studies, these CBT interventionsshare a focus on teaching youth: (a) cognitive tech-niques to identify and modify irrational and depress-ogenic thought patterns; (b) behavioral strategies toregulate mood (e.g., pleasant activity scheduling,relaxation techniques); and (c) problem-solvingskills to proactively cope with environmental de-mands (e.g., social skills training). Implicit or expli-cit in these treatment models are the notions thatreductions in depression may be mediated bychanges in cognitive distortions, improved moodmanagement skills, or improved skills in socialproblem solving.

Overall, CBT for youth depression appears to be anef®cacious intervention, although treatment effectsmay be stronger for depressed adolescents (e.g.,Brent et al., 1997) than depressed children (e.g.,Butler, Miezitis, Friedman, & Cole, 1980). Turning tothe question of mechanism, a majority of CBT de-pression studies did include measures of cognitiveand behavioral processes targeted by treatment andtheorized to be important causal factors in depres-sion recovery. As can be seen in Table 3, sevenstudies assessed youths' positive self-concept orsense of perceived control. Five studies measuredmore generalized cognitive distortions and feelings ofhopelessness. Four studies focused on behavioralprocesses, such as youth engagement in pleasantactivities and adaptive social skills. We next examinethe evidence in support of the mediating role of thesecognitive and behavioral processes.

Mediating role of cognitive processes. In seven ofthe nine studies that assessed depressive cognitions,CBT produced greater change on cognitive measuresthan did comparison conditions. Following a courseof CBT, depressed youth reported a more positiveview of themselves, jumped to fewer negative con-clusions, and had a less hopeless view of their fu-tures. The mediating role of these cognitive changeswas assessed by Kolko and colleagues (2000) in afollow-up to the Brent et al. (1997) clinical trial.

In the original Brent RCT, CBT for depressed adol-escents was compared to family and nondirectiveinterventions (Brent et al., 1997). On multiplemeasures of depression, CBT was found to be moreef®cacious than these alternate treatments. Notably,CBT produced signi®cantly better effects on indicesof clinically signi®cant change in depression. Kolkoand colleagues sought to uncover the mechanisms ofaction responsible for these positive effects andinvestigated the mediating role of several cognitiveand family process variables. As hypothesized, CBTdid produce signi®cant, speci®c changes on ameasure of cognitive distortions, but CBT was notsuperior to alternate interventions in changing feel-

ings of hopelessness. In addition, contrary to hypo-theses, CBT was as effective as family therapy inchanging several indices of adaptive family func-tioning. Thus, while CBT affected one theoreticallyspeci®c mechanism of cognitive change, it also pro-duced nonspeci®c changes in `mediators' belongingto another theoretical model of intervention. In their®nal analyses, Kolko and colleagues did not ®nd thatchange in cognitive distortion mediated the impact ofCBT on depression outcome. However, mediationalanalyses were cut short by the ®nding of no differ-ence in ef®cacy between the three interventions. Asdiscussed earlier, CBT was more ef®cacious thanalternate treatments in terms of clinical remission ofdepression diagnoses and speed of symptomatic re-covery. The three interventions did not differ on thedimensional outcome measures used in the medi-ational analysis (i.e., BDI; Beck Depression Inven-tory).

Some additional, indirect support for the cognitivemediator model of CBT is provided by follow-upreports to three of the CBT clinical trials (Clarkeet al., 1992; Brent et al., 1998; Jayson, Wood, Kroll,Fraser, & Harrington, 1998). In these studies, poorself-concept, pervasive cognitive distortions, andhigh levels of hopelessness at intake related to poortreatment response (see Table 3). While thesestudies establish a link between maladaptive cog-nitions and depression symptoms, they do not an-swer the question of whether change in thesepathological processes causes improvements indepression.

Mediating role of behavioral processes. Fourstudies assessed possible behavioral mediators ofCBT effects. Lewinsohn and colleagues (1990)measured the frequency and enjoyment of pleasantactivities before and after CBT. As targeted in theintervention, CBT bene®cially impacted pleasantactivities. The relationship between engagement inpleasant activities and symptom relief was not di-rectly assessed, but low levels of pleasant activitiesat intake did predict later depression symptoms.Mediation was not tested.

Three clinical trials examined the role of socialskills and social adaptation in depression recovery.Evidence on the effects of CBT are mixed, with onestudy reporting improved social skills and function-ing after CBT (Vostanis, Feehan, Grattan, & Bick-erton, 1996a), and three investigations indicatingthat CBT did not outperform comparison conditions(RosselloÂ & Bernal, 1999; Wood, Harrington, &Moore 1996; Liddle & Spence, 1990). None of theseRCTs assessed mediation.

Summary

Over three-quarters of the CBT studies of depressionincluded measures of the cognitive and behavioralprocesses theorized to underlie intervention effects.


Table


cognitive-b

ehaviora

linterventionsfordepressedyouth

Clinicaltrial

Treatm

entconditions

Candidate

mediators

Ef®cacytest

Interventiontest

Psychopath


Stark

etal.(1987)

Selfcontrol

Problem-solving

Wait

list

Self-concept

Activetreatm

ents

superiorto

wait

list

CBThadbettereffect

onself-conceptth

an

problem-solvingtx

Notassessed

Stark

etal.(1991)

Selfcontrol+

NST

Cognitivedistortions

CBTsuperiorto

NST

onsomemeasures

CBThadbettereffect

oncognitivedistortions

Notassessed

Lewinsohnetal.(1990)

CWD

CWD

+parents

Wait

list

Pleasantactivities


Activetreatm

ents

superiorto

wait

list

Activetreatm

ents

hadbettereffecton

candidate

mediators

Mediationnotassessed.

Candidate

mediators

predict

recovery

(Clark

eetal.,1992)

Lewinsohnetal.(1996)

CWD

CWD

+boosters

Wait

list

Notassessed

Notassessed

Activetreatm

ent

superiorto

wait

list

Notassessed

Butleretal.(1980)

CBT

Role-p

lay

Attention

Wait

list


Locusofcontrol

Role

playgenera

lly

superior,

but

incomplete

analyses

Role

playappearedto

havebettereffect,

but

incomplete

analyses

Notassessed

Reynolds&

Coats

(1986)

CBT

Relaxation

Wait

list

Self-concept

Activetreatm

ents

superiorto

wait

list

CBThadbettereffect

onacademic

self-

conceptatfollow-u

p

Notassessed

Kahnetal.(1990)

CBT

Relaxation

Self-m

odeling

Wait

list

Self-concept

Activetreatm

ents

superiorto

wait

list

Activetreatm

ents

hadbettereffecton

self-conceptth

an

wait

list

Notassessed

Liddle

&Spence(1990)

CBT

Attention

Notx

Socialskills

Notreatm

enteffect

Notreatm

enteffect

onsocialskills

Notassessed

Vostanis

etal.

(1996a,b;1998)

CBT

Supportive

Self-concept

Socialadjustm

ent

Notreatm

entsuperior

CBThadbettereffect

onsocialadjustm

ent

Notassessed

Woodetal.(1996)

CBT

Relaxation


Self-concept

Socialadjustm

ent

CBTsuperiorto

relaxation

CBThadbettereffect

onself-concept


Cognitivedistortionsdid

not

predictrecovery

(Jaysonetal.,1998)

Weiszetal.(1997)

CBT

Notx

Notassessed

Activetreatm

ents

superiorto

notx

Notassessed

Notassessed

Brentetal.(1997)

CBT

Family

Supportive


Hopelessness

Familyprocesses

CBTsuperiorto

altern

ate

txson

multiple

measures

CBThadspeci®c

effects

oncognitive

mediators

Mediationassessedbutnot

found(K

olkoetal.,2000)

Clark

eetal.(1999)

CBT

CBT+parents

Wait

list

Notassessed

Activetreatm

ents

superiorto

wait

list

Notassessed

Notassessed

RosselloÂ&

Bern

al

(1999)

CBT

IPT

Wait

list

Self-concept

Expressedemotion

Socialadjustm

ent

Activetreatm

ents

superiorto

wait

list

IPThadbettereffect

onsocialadjustm

ent

andself-conceptth

an

wait

list

Notassessed

NOTE:Stu

diesare

dividedinto

blocksto

re¯ectth


usedbyth

eTaskForceonEmpiricallySupportedProcedures.Notreatm

ents

were

identi®edas`well-established'

(Kaslow

&Thompson,1998);shadedtreatm

ents

havebeenidenti®edas`probably

ef®cacious.'CBT,cognitive-b

ehaviora

lth

era

py;CWD,CopingwithDepressionforadolescents;IPT,

interp

ersonalpsychoth

era

pyfordepression


However, while a large number of studies measuredthese processes, only one investigation assessedmediation (Kolko et al., 2000). This investigationfound that changes in cognitive distortions werespeci®c to CBT, relative to alternate treatments, butthat cognitions did not mediate the effects of CBT ontreatment outcome.

As with CBT studies of anxiety, mediational workin youth depression suffers from signi®cant overlapbetween the hypothesized cognitive mediators andtreatment outcome. Youth self-report was employedto assess all of the candidate cognitive processes andmany of the measures of depression symptoms (e.g.,the BDI). In addition to this methodological overlap,there may be problems with conceptual overlap be-tween measures of depression symptoms writbroadly and the constructs of hopelessness and de-pressive cognitions. It may be useful to unpack theconstruct of depressogenic thinking into more spe-ci®c cognitive processes such as enhanced recall ofnegative memories (e.g., Lyubomirsky, Caldwell, &Nolen-Hoeksema, 1998), impaired ability to usepositive memories to regulate mood (e.g., Rusting &DeHart, 2000), or tendency to ruminate in situationsof stress (e.g., Nolen-Hoeksema & Morrow, 1993).

Treatments for disruptive youth

The umbrella of `disruptive behavior' covers a varietyof youth rule-violations and hostile acts, ranging inintensity from swearing to criminal assault. Disrup-tive and violent behaviors in youth cause a great dealof familial, school, and societal concern. As a result,while community prevalence rates for each of the twodisruptive behavior diagnoses hover around 10% (seeCostello, 1990), conduct problems are the most com-mon cause for youth referral tomental health services(Achenbach & Howell, 1993; Weisz & Weiss, 1991).

A number of risk factors have been implicated inthe development and maintenance of conduct prob-lems in youth, including: (a) poor parent behaviormanagement practices and coercive family interac-tions (e.g., Dishion & Patterson, 1999); (b) neurobio-logical impairments in youths' executive functioning,verbal reasoning, and reactivity to threat (e.g., Tay-lor, Iacono, & McGue, 2000); (c) errors in youths'processing of social information (e.g., Crick & Dodge,1994); and (d) contextual, neighborhood, and peergroup norms for antisocial behavior (e.g., Mof®tt,1993). Different combinations of these risk factorsmay characterize different subtypes of conductproblem youth (for review, Loeber, 1990; Loeberet al., 1993; Mof®tt, 1993). For example, it has beenproposed that early-onset of conduct problems (be-fore age 10) may spring from heritable, neurologicalimpairments that manifest as youth irritability andpoor behavioral inhibition (e.g., Taylor et al., 2000).These children may be more challenging to manage,especially by parents with similar tendencies, and

over time, parent±child interactions may becomeincreasingly coercive and child behavior worse(Mof®tt, 1993; cf. Dodge et al., 1990). In contrast,adolescent-onset of disruptive behavior may be moreclosely tied to association with deviant peers andpoor parental monitoring of behavior (Taylor et al.,2000). These different patterns of etiology andcourse may be quite consequential; for example, theearly-onset subtype is much more strongly associ-ated with later adult criminality (Mof®tt, 1993).

We next evaluate the EST clinical trials and assessthe extent to which intervention was used as anopportunity to test these theories of disruptivebehavior. The majority of child and adolescentpsychotherapy research has evaluated treatment ofconduct problems, and the EST Task Force was ableto identify two well-established and ten probably ef-®cacious treatments for disruptive behavior in youth(Brestan & Eyberg, 1998). The ESTs fall into twogeneral categories: (a) parent- and system-focusedinterventions, including several varieties of parenttraining and multisystemic therapy; and (b) youth-focused interventions, such as anger managementtreatments and social problem-solving interventions.Given the maturity of the disruptive behavior treat-ment literature, we spend somewhat less time in thissection describing the details of individual studiesand treatment protocols and instead draw summaryconclusions across programs of work.

Parent- and system-focused interventions

Parent- and system-focused interventions aim toreduce disruptive behavior in youth by changing thecontext in which such behavior occurs. Parenttraining (PT) programs may involve relatively littletherapist±youth contact, as the treatment focus isprimarily on teaching parents behavioral manage-ment strategies. Parents receive instruction in howto set clear behavioral goals, monitor and trackyouth behavior, provide positive reinforcement fordesired behavior, use extinction techniques for dis-ruptive behavior, apply mild punishment strategies,use clear commands, and develop a positive com-munication style. Changes in these parenting prac-tices are the hypothesized mechanisms of action bywhich PT interventions affect youth behavior. PTprograms may differ in the mix of skills taught and inthe additional emphasis given to building positiveparent±youth relationships (cf. Hamilton & Mac-Quiddy, 1984; Eyberg, Boggs, & Algina, 1995).Across these variations, PT has a long and well-established history of positive effects with behaviorsfrom tantruming to repeat criminal violations. Bothof the well-established ESTs for youth conductproblems are PT interventions (Brestan & Eyberg,1998) ± Patterson's Living with Children program(Patterson & Gullion, 1968) and the Webster-Strat-ton videotape modeling treatment.


While PT interventions have shown reliable, posit-ive effects, a signi®cant portion of youth continue tohave conduct problems at post-treatment and overfollow-up (Dishion & Patterson, 1992). To improve onthe effects of traditional PT, some investigators haveadded CBT components designed to teach parentsand/or youth social problem-solving and stressreduction skills (Spaccarelli, Cotler, &Penman, 1992;Webster-Stratton, 1994; Kazdin, Siegel, & Bass,1992; Kazdin, Esveldt-Dawson, French, & Unis,1987b; Vitaro & Tremblay, 1994; Tremblay, Pangani-Kurtz, Masse, Vitaro, & Phil, 1995). These interven-tions have not produced dramatically superior effectson youth behavior, although they have been shown toaffect the targeted mediating constructs (e.g., parentproblem-solving skills; Webster-Stratton, 1994).

Multisystemic therapy (MST) was developed, inpart, to serve these more intractable cases of con-duct problem youth. MST attempts to reduce dis-ruptive behavior by changing the family, peer,school, and community contexts in which youth re-side (Henggeler, Schoenwald, Borduin, Rowland, &Cunningham, 1998). In general, MST interventionsfocus on improving parental monitoring of youthbehavior, increasing family cohesion and reducingcoercion, and disengaging youths from deviant peergroups (Huey, Henggeler, Brondino, & Pickrel,2000); thus improvements in youth behavior mightbe expected to be mediated through increased par-ental monitoring of the youth, improved family rela-tionships, and reductions in deviant peer contact.The MST model places a great deal of emphasis oncrafting an individually tailored treatment plan foreach disruptive youth. Indeed, MST protocols des-cribed in clinical trials refer readers to a list oftreatment principles, rather than a treatment man-ual per se (Henggeler, Schoenwald, Borduin, Row-land, & Cunningham, 1998). AlthoughMST is a morerecent arrival than traditional PT, MST has shownpositive effects with samples of adjudicated andseriously disruptive youth (e.g., Henggeler, Melton, &Smith, 1992) and has been identi®ed as a probablyef®cacious EST (Brestan & Eyberg, 1998).

In Table 4, we summarize the results of each of the22 EST clinical trials of parent- and system-focusedinterventions. We identify pathological processesassessed in the studies and summarize the results ofanalyses bearing on therapy mechanism. As can beseen in the table, almost every clinical trial (91%)measured at least one possible mediator of treatmenteffects. As would be expected, given the theoriesunderlying these interventions, candidate mediatorswere most often measures of parent behavior man-agement skills, measures of general family func-tioning, or indices of youth association with deviantpeers. In the next section, we brie¯y review each ofthese treatment mechanisms.

Mediating role of parent behavior managementskills. Sixteen of the 22 clinical trials included

measures of one or more behavior managementskills taught by the intervention programs. Treat-ment effects on speci®c skills were not uniformlysigni®cant across studies (e.g., Bernal, Klinnert, &Schultz, 1980), nor across maternal and paternalbehavior (e.g., Webster-Stratton, 1992). However,the overall pattern of results strongly suggests thatinterventions that teach parenting skills do, indeed,impact parenting practices. Participation in skills-focused treatment signi®cantly increased parentalmonitoring of youths' behavior, parental use of clearcommands, and parental praise and use of rewards.Treatment participation also decreased several un-desirable parenting practices, such as criticism,spanking, and coercive discipline practices. Positivechanges in parenting skills were evident in parents'reports of their own behavior (e.g., Spaccarelli et al.,1992); observations of lab-based parent±child inter-action tasks (e.g., Alexander & Parsons, 1973); andin-home observations of unstructured family inter-actions (e.g., Webster-Stratton, 1984).

Given the reliability and consistency of these ef-fects, we were surprised that none of the EST clinicaltrials directly tested whether changes in parentingpractices mediated the effects of treatment on youthbehavior.6 We searched the treatment outcome lit-erature for follow-up reports to these original clinicaltrials and found three additional studies bearing onthe role of parenting skills in treatment effects. Twoattempted to unpack the effects of the Patterson PTprogram (Patterson & Forgatch, 1995; Eddy &Chamberlain, 2000), and one examined the mecha-nisms of action in MST (Huey et al., 2000).

In a 1995 clinical trial, Patterson and Forgatchattempted to compare the effects of their well-established PT program for seriously disruptiveyouth to the outcomes of usual community therapy.Partway through the study, the community thera-pists independently sought out training in PT tech-niques, and, by the end of the investigation, therewere no differences in treatment type between PTand usual care. Combining these treatment groups,the investigators tested whether changes in theparenting practices targeted by PT predicted long-term functional outcomes of the treated youth. Par-enting skills at termination (discipline, monitoring,problem-solving) were signi®cant predictors of youtharrests and of out-of-home placement, two yearsafter therapy. As there was no treatment comparisoncondition after contamination of usual care, theseresults were not evidence of mediation, but they do

6 As with Blanchard's (1970) investigation of participant mod-

eling, this failure to test for mediated effects may be due, in

part, to the age of the EST clinical trials. Nine of the clinical

trials were published in or before 1986, the year in which

Baron and Kenny's seminal work describing methods for

testing mediation was published. Alexander and Parsons

(1973) tested all of the relationships involved in mediation,

except for the ®nal step.


Table


parent-

andsystem-focusedESTsfordisru

ptiveyouth

Clinicaltrial

Treatm

entconditions

Candidate

mediators

Ef®cacytest

Interventiontest

Psychopath


Bern

aletal.(1980)

PattersonPT

NST

Wait

list

Parentattention

Parentcommands

PTsuperiorto

NST

andwait

list

Notreatm

enteffect

Notassessed(butsee

Patterson&

Forgatch,1995;

Eddy&

Chamberlain,2000)

Firestoneetal.(1980)

PattersonPT

Wait

list

Notassessed

PTsuperiorto

wait

list

Notassessed

Notassessed(butsee

Patterson&

Forgatch,1995;

Eddy&

Chamberlain,2000)

Wiltz

&Patterson(1984)

PattersonPT

Wait

list

Parentcommands

PTsuperiorto

wait

list

Notassessed

Notassessed(butsee

Patterson&

Forgatch,1995;

Eddy&

Chamberlain,2000)

Alexander&

Parsons

(1973)

PattersonPT

PSYfamilytx

NSTfamilytx

Familyprocesses

indexingadaptive

communication

PTsuperiorto

altern

ate

treatm

ents

onmultiple

measures

PThadbettereffect

oncommunication


Communicationpredicted

recidivism.

Webster-Stratton(1984)

VPTgroup

VPTindividual

Wait

list

Parentcommands

Pra

ise/criticism

Spanking

Activetreatm

ents

superiorto

wait

list

VPTtreatm

ents

had

bettereffecton

candidate

mediators

Notassessed

Webster-Stratton

etal.(1988)

VPTindividual

VPTgroup

PTgroup

Wait

list

Parentcommands

Pra

ise/criticism

Spanking

Activetreatm

ents

superiorto

wait

list.

Trendtoward

VPT

groupsuperiority.

Activetreatm

ents

had

bettereffecton

candidate

mediators

thanwait

list

Notassessed


VPTnoth

era

pist

VPTw/th

era

pist

Wait

list

Parentcommands

Pra

ise/criticism

Parentalwarm

thSpanking

Activetreatm

ents

superiorto

wait

list

VPTtreatm

ents

had

bettereffects

on

spanking,pra

ise,and

warm

thth

anwait

list

Notassessed

Spaccarellietal.(1992)

VPT+CBT

VPT+attn

Wait

list

Coercivepunishment

Parentalstress

Parentprob-solving

VPTtreatm

ents

superiorto

wait

list

Both

VPThadbetter

effectonpunishment

andproblem

solvingth

an

wait

list.

VPT+CBT

improvedstress.

Notassessed


VPTindividual

Wait

list

Parentcommands

Pra

ise/criticism

Parentalwarm

thSpanking

IndividualVPT

superiorto

wait

list

VPThadbettereffecton

pra

ise/criticism,patern

al

commands,matern

al

warm

th,andspanking


Modera

tors

identi®edand

includedastx

targets

inWebster-Stratton(1994)


VPT

VPT+CBT

Childprob-solving

Parentcommands

Pra

ise/criticism

Parentprob-solving

Parentalstress

Parentalwarm

thSpanking

Communication

Notreatm

entsuperior

VPT+CBThadbettereffect

onparentandchild

problem-solvingskills

and

maritalcommunication


Criticism,matern

alstress,

patern

alcommands,and

patern

alproblem-solving

predicteddeviantbehavior

Vitaro

&Tremblay(1994)

DelinquencyPP

Notx

Associationwith

deviantpeers

DPPsuperiorat

long-term

follow-u

pDPPhadbettereffecton

peerassociation

Notassessed


Table

4Continued

Clinicaltrial

Treatm

entconditions

Candidate

mediators

Ef®cacytest

Interventiontest

Psychopath


Tremblayetal.(1995)

DelinquencyPP

Attention

Parentalmonitoring

Parentalpunishment

DPPsuperiorto

attention

onsomemeasures

Notreatm

enteffect

Notassessed

Henggeleretal.(1986)

MST

Communitytx

Familyfunctioning

Associationwith

deviantpeers

MSTsuperiorto

communitytx

MSThadbettereffects

on

peerassociationandsome

familymeasures(w

arm

th,

equality)

Notassessed(butsee

Hueyetal.,2000)

Henggeleretal.(1992)

MST

Probation

Familyfunctioning

Peerrelations

MSTsuperior

toprobation

MSThadbettereffects

on

candidate

mediators


Candidate

mediators

did

not

predictdeviancy(alsosee

Hueyetal.,2000)

Bord

uin

etal.(1995)

MST

Communitytx

Familyfunctioning

Peerrelations

MSTsuperiorto

communitytx

MSThadbettereffects

on

familyfunctioning

Notassessed(butsee

Hueyetal.,2000)

Zangwill(1983)

PCIT

Wait

list

Parentreward

sParentalpunishment

PCIT

superiorto

wait

list

onmanymeasures

PCIT

hadbettereffect

oncandidate

mediators

Notassessed

McNeiletal.(1991)

PCIT

Notx

Notassessed

PCIT

superiorto

notx

onmanymeasures

Notassessed

Notassessed

Eyberg

etal.(1995)

PCIT

Wait

list

Pra

ise/criticism

PCIT

superiorin

incomplete

sample

PCIT

hadbettereffect

oncandidate

mediators

Notassessed

Peedetal.(1977)

PT

Wait

list

Parentcommands

Parentattention

Parentreward

s

PTsuperiorto

wait

list

onmanymeasures

PThadbettereffect

oncandidate

mediators

Notassessed

Wells&

Egan(1988)

PT

Familytx

Parentcommands

Parentattention

Parentreward

s

PTsuperiorto

systemic

familytx

PThadbettereffect

oncandidate

mediators

Notassessed

Hamilton&

MacQuiddy(1984)

Tim

e-out

Tim

e-out+seat

Wait

list

Parentpra

ise

Spanking

Tim

e-out+signalseat

superiorto

typical

time-outandto

wait

list

Notassessedin

wait

list.

Tim

e-out+signalseathad

bettereffectonspanking

Notassessed

Kazd

inetal.(1987a)

PSST+PT

Attention

Notassessed

PSST+PTsuperior

toattention

Notassessed

Notassessed

Kazd

inetal.(1992)

PSST

PT

PSST+PT

Familyfunctioning

Parentalstress

AlsoseeTable

5.

PSST+PTsuperior

incombinationto

eitheralone

PSST+PThadbettereffect

oncandidate

mediators

AlsoseeTable

5.


Changein

childmeasuresdid

notrelate

tochangein

parent

measures.AlsoseeTable

5.

NOTE:Stu

diesare

dividedinto

blocksto

re¯ectth


usedbyth

eTaskForceonEmpiricallySupportedProcedures.Shadedentrieshavebeenidenti®edas

`wellestablished'(B

restan&

Eyberg,1998).PT,parenttraining;PCIT,parent±childintera

ctionth

era

py;PSST,problem-solvingskills

training;MST,multisystemic

thera

py;

PSY,psychodynamic

thera

py;NST,non-d

irectivesupportiveth

era

py;VPT,videotapeparenttraining


suggest that parenting may have mediated the im-pact of treatment on long-term outcome had thedesign of the study remained intact.

Additional evidence for the mediational role ofparent skills in the Patterson PT intervention isprovided by clinical trials extending PT to foster careand prevention samples (for general review, seeDishion & Patterson, 1999). One of these investiga-tions is suf®ciently comparable to the parent train-ing EST clinical trials to warranted extendeddiscussion here. In a recent study, Eddy andChamberlain (2000) compared the effects of multi-dimensional treatment foster care (MTFC) to usualgroup home placement on the subsequent criminalbehavior of seriously disruptive teenage boys.Working with the juvenile justice system, youth wererandomly assigned to group home or MTFC place-ments. MTFC youths received a broad package ofservices including individual therapy focusingon prosocial skill building, behavioral family therapywith their family of origin, and a behavioral man-agement program at school. In addition, foster par-ents of MTFC youth had been extensively trained in avariant of Patterson's parent training program. Incontrast, youths placed in group homes were super-vised by adult caretakers, but the general milieufocused on promoting a positive peer environment,and youths generally were encouraged to participatein the governance of the residential facility.

Eddy and Chamberlain (2000) used structuralequation modeling to test the effects of MTFCplacement on later antisocial behavior and to assessif the effects of placement were mediated by care-takers' use of behavior management strategies andyouths' association with deviant peers. Caretakerswere either the MFTC foster parents or the youths'primary adult contact in the group home. The effectsof these two mediators ± behavior managementpractices and peer association ± were combined intoa single latent variable for analysis. Results of theinvestigation strongly supported the ef®cacy ofMFTC, with MFTC youth having much lower levels ofself-reported delinquent behavior and of®cial crim-inal referrals 12 to 24 months after placement. Thesepositive effects of MFTC on behavior were fully me-diated by the latent variable capturing the combinedeffect of behavior management and peer association.In total, 32% of later antisocial behavior was ex-plained by the causal path from treatment throughmediator to outcome. Importantly, and unlike manyother investigations, assessment of the mediatingprocesses occurred three months into youths'placement, substantially before the measurement ofdelinquent behaviors at 12, 18, and 24 months.Mediators and outcomes were also assessed withdifferent measurement technologies, increasingcon®dence in the observed effects.

Evidence on the mediating role of parent disciplinepractices also is provided by studies of MST. Hueyet al. (2000) tested the role of parental monitoring in

MST effects, using data from the 1997 clinical trial ofMST with violent and chronic juvenile offenders(Henggeler et al., 1997). Results were then replicatedin a separate sample of participants drawn from astudy extending MST to the treatment of substance-abusing delinquents (Henggeler, Pickrel, & Brondino,1999). Unlike previous reports, in the 1997 clinicaltrial, MST did not have signi®cantly better effectsthan probation in reducing youths' future criminalactivity or in the percent of youth incarcerated overfollow-up. Additionally, parents in MST did not im-prove their monitoring of youth behavior; indeed,over time, parental monitoring decreased for bothMST and probation youth. Henggeler et al. (1997)attributed these poor effects to problems with MSTadherence. Expert ratings of adherence were notavailable for MST therapists, but the research teamwas able to obtain therapist-, parent-, and youth-report of therapists' use of MST strategies. Usingthese ratings as a criterion, MST therapists withgood adherence did produce signi®cant effects in thetargeted mediators and terminal outcomes.

Huey et al. (2000) extended these analyses andtested whether the relationship between MST ad-herence and youth delinquency was mediated bythree processes ± parental monitoring, family func-tioning, and association with deviant peers. Youthwho did not receive MST were excluded from inves-tigation, and the `treatment' variable consisted ofquality of MST implementation as rated by thera-pists, parents, and youths. A variant of path analysiswas used to test the hypothesized relationships be-tween treatment, mediators, and outcome; readersare referred to the original report for details of theanalytic procedure and model ®t statistics for the1997 and 1999 samples of youth. Across samples,the in¯uence of parental monitoring on youth de-linquency was clear, both through a direct path andindirectly though the effect of monitoring on youthassociation with deviant peers. Relationships be-tween MST treatment adherence and monitoringwere signi®cant, but weaker than the links betweenmonitoring and outcome. Some support was foundfor mediation of MST effects on delinquency throughchanges in parental monitoring; however, mediationwas only found when parent ratings of MST adher-ence were used (therapists and youths also assessedadherence).

Mediating role of family functioning. In addition tomeasures of parent behavior management, six ofthe EST clinical trials assessed family functioning.In this family functioning category, we includedmeasures of families' affective tone, parentalwarmth, family cohesion, and hierarchy/dominance.Family functioning was assessed by parent- andyouth-report (e.g., Henggeler et al., 1992); codedfamily interactions in the laboratory (Henggeleret al., 1986); and home observation (e.g., Webster-Stratton, 1992).


Results for family functioning paralleled ®ndingsfor parent behavior management strategies. On thewhole, the parent- and system-focused EST producedpositive changes in family functioning, although re-sults were not completely uniform across studies(e.g., Henggeler et al., 1986) or for changes in mater-nal versus paternal warmth (e.g., Webster-Stratton,1992). TheHuey et al. (2000) investigation, discussedearlier, also assessed whether these positive changesin family functioning mediated the effects of MSTadherence on delinquency. Improved family func-tioning and greater cohesion were signi®cantly asso-ciated with reductions in delinquency for all modelstested. Improved functioning was also reliably relatedto less association with deviant peers. As with themodels of parental monitoring, some support wasfound for the mediating role of family functioning ontreatment outcome, but only for models based onparent-ratings of MST adherence.

Mediating role of association with deviantpeers. Five of the EST clinical trials assessed youthassociation with deviant peers, including all four ofthe clinical trials of MST. In all of the studies, saveHenggeler et al. (1997), the EST interventions suc-ceeded in encouraging youth to associate with a lessdeviant peer group, primarily through improvedparental monitoring of youth behavior (Huey et al.,2000). In the Henggeler et al. (1997) study, positivechanges in peer af®liation were found in cases withhigh MST adherence ratings. As discussed previ-ously, Huey et al. (2000) found deviant peer af®li-ation to be an important predictor of delinquencyand a mediator between MST adherence and futureconduct problems (for parent ratings of adherence).In addition, Eddy and Chamberlain (2000) foundthat a latent variable composed of both caregiverdiscipline practices and youth association withdeviant peers mediated the impact of their multi-dimensional foster treatment program on latercriminal behavior.

Youth-focused interventions

Five youth-focused treatment programs have beenidenti®ed as probably ef®cacious interventions forconduct problem youth: Anger Control Training withStress Inoculation; the Anger Coping Program; as-sertiveness training; Problem-Solving Skills Train-ing; and Rational Emotive Therapy (see Brestan &Eyberg, 1998). These interventions target the dis-turbed cognitive processes and behavioral de®citsthought to produce aggressive and disruptive be-haviors. The interventions draw heavily from work byDodge and colleagues on social information pro-cessing. A full discussion of the social information-processing model is beyond the scope of this review;in short, the model postulates that socially compet-ent behavior is dependent on (a) accurate encodingof social cues and interpretation of others' intent; (b)

generation and selection of appropriate responses;and (c) skillful enactment of the chosen course ofbehavior (see Crick & Dodge, 1994). Using tech-niques such as cognitive restructuring (e.g., Loch-man, 1984) and social skills training (e.g., Huey &Rank, 1984), the youth-focused ESTs attempt toremediate de®cits at each point in the processingmodel. Several of these programs also place a greatdeal of emphasis on teaching youths how to solveproblems rationally and respond non-aggressivelywhen youths are actually aroused and angry (e.g.,`hot' processing; Schlicter & Horan, 1981).

Just over half of the youth-focused ESTs inclu-ded measures of the cognitive processes targeted bytreatment. Below, we review this evidence ontreatment mechanism, organizing our comments tomap on to the stages of the social information-processing model. Table 5 provides a study-by-study summary of treatment effects and data ontherapy mechanism.

Mediating role of social information process-ing. None of the EST studies assessed social infor-mation processing de®cits in cue encoding orinterpretation (e.g., hostile attributional bias). Oneclinical trial (Schlicter & Horan, 1981) did include ameasure of general cognitive distortions. In this in-vestigation, the anger control intervention had asigni®cant impact on self-rated anger (the mainoutcome measure); however, cognitive distortionswere unaffected.

Six of the EST clinical trials assessed disruptiveyouths' ability to generate solutions to interpersonalproblems and select appropriate behaviors. Speci®cmeasures included: (a) the total number of solutionsyouth were able to generate to hypothetical inter-personal scenarios (Feindler, Marriott, & Iwata,1984;Lochman et al., 1984; Webster-Stratton, 1994); (b)the quality and type of these hypothetical solutions(Lochman et al., 1984; Kazdin et al., 1992; Webster-Stratton, 1994); and (c) the quality and type ofresponse chosen by youth in role-plays of simulatedprovocations (Schlicter &Horan, 1981; Huey&Rank,1984). Overall, participation in the social problem-solving treatments had positive effects on thesemeasures of response generation and selection ± in-creasing the total number of responses considered,reducing the aggressiveness of chosen responses,and increasing the number of skilled, assertiveresponses. None of the clinical trials assessed therelationship between these positive changes inlab-based tasks and reduction in aggressive anddisruptive behavior at home or in school.

We searched for follow-up studies to these ESTclinical trials to see if later reports re-analyzedclinical trial data to assess mediation. We did not®nd any studies that met this description. How-ever, we did ®nd one non-EST investigation withrelevant results. Guerra and Slaby (1990) investi-gated the effects of cognitive mediation training


(CMT), a social problem solving intervention forviolent, incarcerated youth. In this study, theyassessed a number of social information processingvariables (e.g., cue interpretation, solutions gener-ated to social problems) and youths' general beliefsabout the legitimacy of aggressive behavior. Thesecognitive processes were directly targeted by theCMT program and hypothesized to mediate theeffects of CMT on violent behavior. Compared toattention-placebo and no treatment control, CMTproduced signi®cant improvements on youths' ag-gressive, impulsive, and in¯exible behavior, ratedwhile youths were incarcerated. CMT did not have

a signi®cant effect on recidivism after release fromthe detention facility. CMT also improved therationality of youths' social information processingand weakened youths' beliefs supporting the legit-imacy of aggression. Formal mediational analyseswere not conducted; however, the investigators didexamine the relationship between the scores on thecognitive measures and behavior ratings at post-treatment assessment. The majority of social in-formation processing and belief measures werenot signi®cantly related to aggressive behavior,although signi®cant effects were found for cueinterpretation (i.e., the hostile attributional bias)

Table 5 Mechanisms of action in youth-focused ESTs for disruptive youth

Clinical trialTreatmentconditions

Candidatemediators Ef®cacy test

Interventiontest

Psychopathology/Mediation tests

Schlicter &Horan (1981)

Anger controlRelaxationNo tx

Cognitive distortionsResponse toprovocation(imaginal and live)

No treatment effect Active treatmentshad better effecton response toprovocation

Not assessed

Feindleret al. (1984)

Anger controlNo tx

Problem-solving Anger controlsuperior onsome measures

Anger control hadbetter effect oncandidate mediator

Not assessed

Lochmanet al. (1984)

Anger + goalsAnger controlGoal-settingNo tx

Problem-solving Anger treatmentssuperior atreducingaggressiveness

Anger treatmentshad better effect oncandidate mediatorthan goals or no tx

Not assessed

Lochmanet al. (1989)

Anger + probsolv

Anger controlNo tx

Not assessed Anger treatmentssuperior to no tx


Huey &Rank (1984)

AssertivenessNSTNo tx

Response toprovocation (live)

Assertivenesssuperior to NSTand to no tx

Assertiveness hadbetter effect oncandidate mediator

Not assessed

Kazdinet al. (1987a)

PSST + PTAttention

Not assessed PSST + PT superiorto attention


Kazdinet al. (1987b)

PSSTNSTAttention

Not assessed PSST superiorto NST andattention


Kazdinet al. (1992)

PSSTPTPSST + PT

Response toprovocation(imaginal)

Also see Table 4.

PSST + PT superiorin combinationto either alone

PSST interventionshad better effect onresponse toprovocation.

Also see Table 4.

Also see Table 4.

Block (1978) RETPSYWait list

Not assessed RET superior toPSY and wait list



VPTVPT + PCBT

Youth problem-solving

Also see Table 4.

No treatmentsuperior

VPT + CBT hadbetter effect onproblem-solving

Also see Table 4.

Not assessed.Also see Table 4.

NOTE: Studies are divided into blocks to re¯ect the classi®cation system used by the Task Force on Empirically SupportedProcedures (Brestan & Eyberg, 1998). PSST, problem solving skills training; PSY, psychodynamic therapy; PT, parent training; VPT,videotape parent training; NST, non-directive supportive therapy; RET, rational-emotive therapy; PCBT, parent cognitive-behavioraltherapy


and the overall belief that aggression is legitimate.While CMT did not produce signi®cant effects onrecidivism, Guerra and Slaby also examined therelationship between the hypothesized cognitivemediators and later criminal behavior. Recidivismwas related to youths' endorsement of the hostileattributional bias, selection of hostile responses,belief that victims deserve aggression, and beliefthat aggression is a legitimate social response.These results provide some support for the medi-ating role of social information processing inyouth-focused treatments for aggression. However,results tying cognitive to behavioral variables weremixed, formal mediational analyses were not con-ducted, and assessment of the hypothesized medi-ators and outcomes occurred simultaneously atpost-treatment.

Summary

Seventy-seven percent of the EST clinical trials fordisruptive youth included at least one measure ofpossible therapeutic mechanisms. Evidence ontherapy mechanism is most complete for studies ofparent- and system-focused interventions. Acrossclinical trials, teams of investigators, and methods ofassessment, these interventions reliably improvedthe parent behavior management skills targeted bythe treatment protocols. Parent- and system-focusedinterventions also produced good effects on youthassociation with deviant peers and on more generalmeasures of family functioning and warmth.

Evidence on therapy mechanism was less com-plete for studies of youth-focused interventions.Many of these clinical trials included measures ofsocial information processing, and, overall, partici-pation in treatment led to positive changes inyouths' ability to generate solutions to interpersonalproblems and reduced the aggressive content ofyouths' action plans. Despite these promising re-sults, none of the EST studies formally assessedmediation or related changes in social informationprocessing to other measures of disruptive behav-ior. We did ®nd one non-EST clinical trial that re-lated social information processing measures tobehavior outcome. While the problem-solving inter-vention did produce positive change in both be-havior and a broad range of social informationprocessing measures, less than half of the cognitivemeasures actually were related to aggressive be-havior (Guerra & Slaby, 1990). As with the CBTclinical trials of anxiety and depression, thesestudies also suffered from conceptual overlap be-tween mediator and outcome. Indeed, some of theclinical trials classi®ed social information process-ing tasks as measures of aggressiveness. To main-tain consistency across studies, we reviewed allmeasures of social information processing as poss-ible mediators, regardless of the investigator's par-ticular operational de®nition.

Conclusions

We opened this review by posing a basic question:When youth psychotherapy works, why does itwork? According to previous reviews and commen-taries, there is very little evidence available to answerthis query (e.g., Kazdin et al., 1990). Through ourfocused review of the EST clinical trials, we came to asomewhat different conclusion: Considerable evi-dence exists, but it has not been fully exploited.Consistent with previous reports, very few studies oftreatment ef®cacy explicitly tested mechanisms ofaction underlying therapy effects. Although we re-viewed 67 clinical trials, we found only six investi-gations that attempted to test mediation (Treadwell& Kendall, 1996; Kolko et al., 2000; Patterson &Forgatch, 1995; Eddy & Chamberlain, 2000; Hueyet al., 2000; Guerra & Slaby, 1990). While completetests of mediation were rare, we found that a largeproportion of studies included measures that couldhave been used to investigate treatment mecha-nisms. Some 30% of anxiety trials, 79% of depres-sion trials, and 77% of disruptive behavior trialsincluded measures of processes hypothesized tomediate treatment effects. Within the disruptive be-havior category, an impressive 91% of the investi-gations of parent training assessed possiblemediators, most often parents' use of appropriatebehavior management techniques. Across disordersand interventions, the EST treatments producedsigni®cant changes in many of these targetedmechanisms. Thus, at a general level, we found goodevidence for the ®rst two steps in establishing me-diation. By de®nition, the EST interventions wereef®cacious, and a substantial proportion of treat-ments signi®cantly impacted possible mediatingmechanisms.

Despite this promising beginning, it would be dif-®cult to conclude that the ESTs for youth workthrough the mechanisms speci®ed in their theoriesof intervention. The vast majority of studies treatedtheir `mediator' as another outcome variable and,importantly, measured the mediator at treatmenttermination. This was true even for many of the in-vestigations that explicitly sought to demonstratemediated effects. As an example of the problems thisengenders, it may seem quite logical to stipulate thattraining parents to monitor and supervise theirteens' whereabouts decreases youth association withdeviant peers, which in turn reduces delinquent andantisocial acts (Huey et al., 2000). However, thiscausal chain cannot be established purely on thebasis of statistical relationships. It may be possiblethat different youth behaviors pull for different par-enting practices, and the direction of effects may ¯owfrom youth behavior to apparent changes in par-enting (e.g., youths who stop associating with gangmembers may be easier for parents to keep track of ).Post-treatment assessment of mediators and out-comes may be a useful ®rst step for investigators


seeking to understand mechanisms of therapeuticaction.7 However, multiple assessments of mediatorsand outcomes over the course of therapy and tem-porally based analytic strategies are required toconclusively demonstrate the direction of mediatedeffects.

The EST results bearing on mechanism also werelimited by conceptual and methodological overlapbetween mediators and outcomes. This dif®culty wasmost apparent in studies that assessed cognitivemediators. Youths' cognitive distortions were typic-ally assessed with self-report rating scales, as werethe symptomatic outcomes of interest (e.g., Kolkoet al., 2000; Lochman et al., 1984). Additionally, thecognitive measures employed in the clinical trialsfocused on cognitive products, rather than cognitiveprocesses. It may be splitting conceptual and meth-odological hairs to attempt to differentiate theseproducts from symptoms, as in the overlap betweenthe content of anxious self-talk and youths' self-rated anxiety symptoms (Treadwell & Kendall, 1996).One possible solution may be to refocus cognitiveassessment from products to processes, such asselective attention, impaired recognition, and biasesin recall. Speci®c processing de®cits have beendocumented for anxious versus depressed youths(Gotlib & MacLeod, 1997), and there has been in-creasing interest in developing standardized, per-formance-based measures that do not rely onyouths' ability to introspect and verbally report onthese processes (for review, see Frick, 2000). Alter-nately, there also may be value in testing behavioralmediators of CBT effects, a relatively neglected en-deavor. Only one of the investigations of CBT fordepression assessed engagement in pleasant activ-ities (Lewinsohn et al., 1990), a major skill taught todepressed youth to help them improve their mood.

Much of our discussion has highlighted limitationsin the extant clinical trials research relative to thegoal of mediation testing; however, we believe thatthese limitations are quite understandable in thecontext of an evolving ®eld. Three points are partic-ularly important in this regard. First, many of thestudies reviewed here were conducted before thepublication of articles detailing procedures for me-diation analysis (e.g., Baron & Kenny, 1986). In theseinstances, authors deserve considerable credit forcollecting data relevant to proposed therapy mecha-nisms, even though the data were not used in waysthat we now regard as appropriate for full mediationtesting. Second, for those studies conducted after thepublication of mediation testing procedures, we needto realize that an ef®cient strategy for treatment de-velopment may not include extensive mediation

testing at the front end. It may be reasonable to beginby determining whether a particular treatment pro-duces desired outcomes before investing the re-sources needed to identify and assess measures(especially in the case of time-intensive mediators,such as in-home behavioral observation of familyinteractions). This strategic reasoning may accountfor the several instances in our tables in which the®rst published study of a particular treatment doesthe least to investigate mediation. As a third, andrelated, point, crafting a treatmentmediation study isa complicated task, much more involved than de-signing a traditional ef®cacy clinical trial. Investiga-tors must decide which mediators to focus on, how tomeasure these mechanisms, when and how fre-quently to assessmediators and outcome, and how toanalyze these multiple measurements in a way thatilluminates the processes under investigation. Formany disorders and types of therapy, there currentlymay be little theory, and even less data, to guide thesespeci®c design decisions. Thus, the purpose of ourcritique has not been so much to criticize as tocharacterize the current state of knowledge and toidentify ways to enrich our understanding of treat-ment mechanisms in future work.

Future directions: mechanismsof action in the real world

Overall, we see a great deal of promise in clinicaltrials research on youth psychotherapy. Althoughvery few of the EST clinical trials formally assessedmediation, a majority of studies contained some in-formation relevant to mechanism. Re®nements inthe design of studies and assessments of mediatorsmay increase the theoretical yield of future investi-gations and help us to better understand the criticalprocesses and ingredients responsible for therapyeffects. Additionally, in a number of cases, already-completed studies may be revisited with new analyticmethods to take fuller advantage of the assessmentsalready completed.

Despite our emphasis on methodological rigor andprecise assessment, we do not wish to leave the im-pression that research of this kind should exclu-sively, or even primarily, be conducted within thetraditional settings and samples of clinical trials.Historically, the participants (recruited children andfamilies, researcher-employed therapists) and con-texts (e.g., grade schools, university research clinics)of clinical trials research have tended to differ rathermarkedly from the participants and contexts inwhich most everyday youth treatment occurs (Weiszet al., 1998). While there is certainly genuine valuein traditional clinical trials, the effects of treatmentmodels tested under carefully constructed, optimi-zing conditions may not generalize well to the wideruniverse of youth and families in need of mentalhealth services and referred to most mental health

7 Such analyses may lead to surprising outcomes. For exam-

ple, in their meta-analysis of CBT effects, Durlak, Fuhrman,

and Lampman (1991) found that CBT for youth produced

change on cognitive and behavioral outcome measures, but

changes in the two domains were not correlated.


service settings (Weisz, 2001; Weisz & Weersing,1999).

In addition to the social policy imperative to de-velop and re®ne treatments for real-world childrenand families, we believe that testing mediationalmodels in community samples may end up beingbetter science. Working with community families, forexample, may reveal signi®cant limitations in ourmodels of intervention effects. One can imaginemodel changes of two different forms, some leadingtoward increased complexity, some toward increasedsimplicity. Arguing for increased complexity is thefact that a major limitation in model building, boththeoretically and statistically, is the exclusion of keyconstructs. This form of model misspeci®cation maylead to illusory relationships between unrelatedvariables, obscure indirect effects, in¯ate errorterms, and provide skewed estimates of regressioncoef®cients. The pool of typical clinical trial partici-pants may be `low' on a number of theoretically im-portant characteristics, and models of therapyeffects tested in this sample may lead to modelmisspeci®cation by underestimating the necessity ofintervention components and mediational pathways.For example, the families of depressed youth incommunity settings have been reported to haveserious parental psychopathology, chaotic life cir-cumstances, and harsh parenting practices (Ham-men, Rudolph, Weisz, Rao, & Burge, 1999). Theoriesof depression suggest that environmental stressors,particularly uncontrollable negative events, may beimportant factors in the development and mainten-ance of symptoms (e.g., Abramson et al., 1989).However, none of the ESTs for depression includedintervention components designed to address thenegative familial context of depressed youth. Im-portantly, even if the ESTs had targeted family con-text, change in the predictability and tone of familyinteractions may not have mediated treatment ef-fects in the samples in which the ESTs were tested(e.g., mildly to moderately depressed schoolchildren;Butler et al., 1980).

Use of real-world samples of youth also mayhighlight opportunities to simplify our models ofpsychopathology and intervention. Consider the factthat youth recruited for clinical trials are oftenscreened to limit comorbid psychiatric problems ordiagnoses (e.g., Lewinsohn et al., 1990). In epide-miological studies, however, comorbidity among di-agnostic categories and subsyndromal symptoms ishigh (see Angold, Costello, & Erkanli, 1999; Cerel &Fristad, 2001). This suggests that clinical trial re-search that focuses on `pure' or constrained diag-nostic samples may be creating and re®ningtreatments for relatively rare groups of children andadolescents. Such research may also risk overlook-ing commonalities between different disorders andcommon mechanisms of therapeutic action. For in-stance, depressed, anxious, and aggressive childrenall may overestimate the hostile intent of others,

although their behavioral responses to this attribu-tional error may differ markedly (e.g., withdrawalversus aggression; Quiggle, Garber, Panak, & Dodge,1992). An ef®cient intervention for multi-problemyouths may target the shared aspect of disorders ±that is, the misattribution of others' ill-intent ± withthe hypothesis that change in this mechanismshould mediate treatment effects on multiple symp-toms. This is a rather different approach to buildingand testing treatments than the disorder-speci®cmodel of traditional clinical trials.

In our view, theory-testing and research in the®eld are best viewed as overlapping rather thandistinct or competing enterprises. It is quite possiblethat the fairest and fullest tests of our theories aboutwhy treatments work will be those that are focusedon real-world clients treated in real-world clinicalcontexts. But whether researchers share this beliefor not, and whether they structure their research inthis way or not, one point now seems quite clear:With recent advances in our understanding of me-diation testing, virtually any test of whether youthpsychotherapy works can now be designed to tell uswhy it works as well.

Acknowledgements

Preparation of this manuscript was facilitated byNational Institute of Mental Health (NIMH) Institu-tional Research Training Grant MH18269 support-ing V. Robin Weersing and NIMH Research ScientistAward K05-MH01161 supporting John R. Weisz.

Correspondence to

V. Robin Weersing, Child Psychiatry, Western Psy-chiatric Institute and Clinic, University of PittsburghSchool of Medicine, 3811 O'Hara Street, Suite 112,Pittsburgh, Pennsylvania 15213; Tel: (412) 624-3673; Fax: (412) 624-4326; Email: [email protected]

References

Achenbach, T.M., & Howell, C.T. (1993). Are America'schildren's problems getting worse? A 13-year com-parison. Journal of the American Academy of Childand Adolescent Psychiatry, 32, 1145±1154.

Abramson, L.Y., Metalsky, G.I., & Alloy, L.B. (1989).Hopelessness depression: A theory-based subtype ofdepression. Psychological Review, 96, 358±372.

Abramson, L.Y., Seligman, M.E.P., & Teasdale, J.(1978). Learned helplessness in humans: Critiqueand reformulation. Journal of Abnormal Psychology,87, 49±74.

Alexander, J.F., & Parsons, B.V. (1973). Short-termbehavioral intervention with delinquent families:Impact on family process and recidivism. Journalof Abnormal Psychology, 81, 219±225.


Angold, A., Costello, E.J., & Erkanli, A. (1999). Comor-bidity. Journal of Child Psychology, Psychiatry, andAllied Disciplines, 40, 57±87.

Bandura, A., & Menlove, F.L. (1968). Factors determin-ing vicarious extinction of avoidance behaviorthrough symbolic modeling. Journal of Personalityand Social Psychology, 3, 99±108.

Bandura, A., Blanchard, E.B., & Ritter, B. (1969).Relative ef®cacy of desensitization and modelingapproaches for inducing behavioral, affective, andattitudinal changes. Journal of Personality and SocialPsychology, 13, 179±199.

Bandura, A., Grusec, J.E., & Menlove, F.L. (1967).Vicarious extinction of avoidance behavior. Journal ofPersonality and Social Psychology, 5, 16±23.

Barabasz, A.F. (1973). Group desensitization of testanxiety in elementary school. The Journal of Psycho-logy, 83, 295±301.

Barlow, D.H. (1988). Anxiety and its disorders: Thenature and treatment of anxiety and panic. New York:Guilford.

Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable distinction in social psychologicalresearch: Conceptual, strategic, and statisticalconsiderations. Journal of Personality and SocialPsychology, 51, 1173±1182.

Barrett, P.M., Dadds, M.R., & Rapee, R.M. (1996).Family treatment of childhood anxiety: A controlledtrial. Journal of Consulting and Clinical Psychology,64, 333±342.

Beck, A.T., Rush, A.J., Shaw, B.F., & Emery, G. (1979).Cognitive Therapy of Depression. New York: TheGuilford Press.

Beidel, D.C., Turner, S.M., & Morris, T.L. (1999).Psychopathology of childhood social phobia. Journalof the American Academy of Child and AdolescentPsychiatry, 38, 643±650.

Bell-Dolan, D., Last, C.G., & Strauss, C.C. (1990).Symptoms of anxiety disorders in normal children.Journal of the American Academy of Child andAdolescent Psychiatry, 29, 759±765.

Bernal, M.E., Klinnert, M.D., & Schultz, L.A. (1980).Outcome evaluation of behavioral parent trainingand client-centered parent counseling for childrenwith conduct problems. Journal of Applied BehaviorAnalysis, 13, 677±691.

Bernstein, G.A., & Borchardt, C.M. (1991). Anxietydisorders of childhood and adolescence: A criticalreview. Journal of the American Academy of Child andAdolescent Psychiatry, 30, 519±532.

Biederman, J., Rosenbaum, J.F., Bolduc-Murphy, E.A.,Faraone, S.V., Chaloff, J., Hirshfeld, D.R., & Kagan,J. (1993). Behavioral inhibition as a tempermentalrisk factor for anxiety disorders. Child and AdolescentPsychiatric Clinics of North America, 2, 814±821.

Blanchard, E.B. (1970). Relative contributions of mod-eling, informational in¯uences, and physical contactin extinction of phobic behavior. Journal of AbnormalPsychology, 76, 55±61.

Block, J. (1978). Effects of a rational-emotive mentalhealth program on poorly achieving disruptive highschool students. Journal of Counseling Psychology,25, 61±65.

Borduin, C.M., Mann, B.J., Cone, L.T., Henggeler, S.W.,Fucci, B.R., Blaske, D.M., & Williams, R.A. (1995).

Multisystemic treatment of serious juvenile offenders:Long-term prevention of criminality and violence.Journal of Consulting and Clinical Psychology, 63,569±578.

Brent, D.A., Gaynor, S.T., & Weersing, V.R. (in press).Cognitive-behavioral approaches to the treatmentof depression and anxiety. Chapter to appear inM. Rutter & E. Taylor (Eds.), Child and adolescentpsychiatry: Modern approaches. London: BlackwellScienti®c Publications.

Brent, D.A., Holder, D., Kolko, D., Birmaher, B., Bau-gher, M., Roth, C., & Johnson, B. (1997). A clinicalpsychotherapy trial for adolescent depression com-paring cognitive, family, and supportive treatments.Archives of General Psychiatry, 54, 877±885.

Brent, D.A., Kolko, D., Birmaher, B., Baugher, M.,Bridge, J., Roth, C., & Holder, D. (1998). Predictors oftreatment ef®cacy in a clinical trial of three psycho-social treatments for adolescent depression. Journalof the American Academy of Child and AdolescentPsychiatry, 37, 906±914.

Brestan, E.V., & Eyberg, S.M. (1998). Effective psycho-social treatments of conduct-disordered children andadolescents: 29 years, 82 studies, and 5,272 kids.Journal of Clinical Child Psychology, 27, 180±189.

Butler, L., Miezitis, S., Friedman, R., & Cole, E. (1980).The effect of two school-based intervention programson depressive symptoms in preadolescents. AmericanEducational Research Journal, 17, 111±119.

Casey, R.J., & Berman, J.S. (1985). The outcome ofpsychotherapy with children. Psychological Bulletin,98, 388±400.

Cerel, J., & Fristad, M.A. (2001). Scaling structuredinterview data: A comparison of two methods. Journalof the American Academy of Child and AdolescentPsychiatry, 40, 341±346.

Chorpita, B.F., & Barlow, D.H. (1998). The developmentof anxiety: The role of control in the early environ-ment. Psychological Bulletin, 124, 3±21.

Cicchetti, D., & Toth, S.L. (1992). The role of develop-mental theory in prevention and intervention. Devel-opment and Psychopathology, 4, 489±493.

Clarke, G., Hops, H., Lewinsohn, P.M., Andrews, J.,Seeley, J.R., & Williams, J. (1992). Cognitive-beha-vioral group treatment of adolescent depression: Pre-diction of outcome. Behavior Therapy, 23, 341±354.

Clarke, G.N., Rohde, P., Lewinsohn, P.M., Hops, H., &Seeley, J.R. (1999). Cognitive-behavioral treatment ofadolescent depression: Ef®cacy of acute group treat-ment and booster sessions. Journal of the AmericanAcademy of Child and Adolescent Psychiatry, 38,272±279.

Costello, E.J. (1990). Child psychiatric epidemiology:Implications for clinical research and practice. In B.B.Lahey & A.E. Kazdin (Eds.), Advances in clinical childpsychology (vol. 13, pp. 53±90). New York: PlenumPress.

Crick, N.R., & Dodge, K.A. (1994). A review and refor-mulation of social information processing mecha-nisms in children's social adjustment. PsychologyBulletin, 115, 74±101.

Dadds, M.R., Barrett, P.M., Rapee, R.M., & Ryan, S.M.(1996). Family process and child anxiety and aggres-sion: An observational analysis. Journal of AbnormalChild Psychology, 24, 715±734.


Dishion, T.J., & Patterson, G.R. (1992). Age effectsin parent training outcome. Behavior Therapy, 23,719±729.

Dishion, T.J., & Patterson, G.R. (1999). Model buildingin developmental psychopathology: A pragmatic ap-proach to understanding and intervention. Journal ofClinical Child Psychology, 28, 502±512.

Dodge, K.A., Bates, J.E., & Pettit, G.S. (1990). Mecha-nisms in the cycle of violence. Science, 250, 1678±1683.

Durlak, J.A., Fuhrman, T., & Lampman, C. (1991).Effectiveness of cognitive-behavior therapy for mal-adapting children: A meta-analysis. PsychologicalBulletin, 110, 204±214.

Durlak, J.A., Wells, A.M., Cotton, J.K., & Johnson, S.(1995). Analysis of selected methodological issues inchild psychotherapy research. Journal of ClinicalChild Psychology, 24, 141±148.

Eddy, M.J., & Chamberlain, P. (2000). Family manage-ment and deviant peer association as mediators of theimpact of treatment condition on youth antisocialbehavior. Journal of Consulting and Clinical Psycho-logy, 68, 857±863.

Eyberg, S.M., Boggs, S., & Algina, J. (1995). Parent-child interaction therapy: A psychosocial model forthe treatment of young children with conduct prob-lem behavior and their families. PsychopharmacologyBulletin, 31, 83±91.

Eysenck, H.J. (1952). The effects of psychotherapy: Anevaluation. Journal of Consulting Psychology, 16,319±324.

Feindler, D.L., Marriott, S.A.A., & Iwata, M. (1984).Group anger control training for junior high schooldelinquents. Cognitive Therapy and Research, 8,299±311.

Firestone, P., Kelly, M.J., & Filke, S. (1980). Are fathersnecessary in parent training groups? Journal of ChildClinical Psychology, 9, 44±47.

Foa, E.B., & Kozak, M.J. (1986). Emotional processingof fear: Exposure to corrective information. Psychol-ogical Bulletin, 99, 20±35.

Frick, P.J. (2000). Laboratory and performance-basedmeasures of childhood disorders [Special section].Journal of Clinical Child Psychology, 29.

Gladstone, T.R.G., & Kaslow, N.J. (1995). Depressionand attributions in children and adolescents: A meta-analytic review. Journal of Abnormal Child Psychol-ogy, 23, 597±606.

Gotlib, I.H., & MacLeod, C. (1997). Information pro-cessing in anxiety and depression: A cognitive-devel-opmental perspective. In J.A. Burack & J.T. Enns(Eds.), Attention, development, and psychopatho-logy (pp. 350±378). New York, NY: The GuilfordPress.

Gould, M.S., King, R., Greenwald, S., Fisher, P.,Schwab-Stone, M., Kramer, R., Flisher, A.J., Good-man, S., Canino, G., & Shaffer, D. (1998). Psychopa-thology associated with suicidal ideation andattempts among children and adolescents. Journalof the American Academy of Child and AdolescentPsychiatry, 37, 915±923.

Graziano, A.M., & Mooney, K.C. (1980). Family self-control instruction for children's night-time fearreduction. Journal of Consulting and Clinical Psycho-logy, 48, 206±213.

Guerra, N.G., & Slaby, R.G. (1990). Cognitive mediatorsof aggression in adolescent offenders: II. Intervention.Developmental Psychology, 26, 269±277.

Hamilton, S.B., & MacQuiddy, S.L. (1984). Self-admin-istered behavioral parent training: Enhancementof treatment ef®cacy using a time-out signal seat.Journal of Child Clinical Psychology, 13, 61±69.

Hammen, C., Rudolph, K., Weisz, J., Rao, U., & Burge,D. (1999). The context of depression in clinic-referredyouth: Neglected areas in treatment. Journal of theAmerican Academy of Child and Adolescent Psychi-atry, 38, 64±71.

Henggeler, S.W., Melton, G.B., Brondino, M.J., Scherer,D.G., & Hanley, J.H. (1997). Multisystemic therapywith violent and chronic juvenile offenders and theirfamilies: The role of treatment ®delity in successfuldissemination. Journal of Consulting and ClinicalPsychology, 65, 821±833.

Henggeler, S.W., Melton, G.B., & Smith, L.A. (1992).Family preservation using Multisystemic Therapy: Aneffective alternative to incarcerating serious juvenileoffenders. Journal of Consulting and Clinical Psycho-logy, 60, 953±961.

Henggeler, S.W., Pickrel, S.G., & Brondino, M.J. (1999).Multisystemic treatment of substance abusing anddependent delinquents: Outcomes, treatment ®del-ity, and transportability. Mental Health ServicesResearch, 1, 171±184.

Henggeler, S.W., Rodick, J.D., Borduin, C.M., Hanson,C.L., Watson, S.M., & Urey, J.R. (1986). Multisys-temic treatment of juvenile offenders: Effects of adol-escent behavior and family interaction.Developmental Psychology, 22, 132±141.

Henggeler, S.W., Schoenwald, S.K., Borduin, C.M.,Rowland, M.D., & Cunningham, P.B. (1998). Multi-systemic treatment of antisocial behavior in childrenand adolescents. New York: Guilford.

Hill, J.H., Liebert, R.M., & Mott, J.E.M. (1968). Vicar-ious extinction of avoidance behavior through ®lms:An initial test. Psychological Reports, 22, 192.

Holmbeck, G.N. (1997). Toward terminological, concep-tual, and statistical clarity in the study of mediatorsand moderators: Examples from the child-clinicaland pediatric psychology literatures. Journal of Con-sulting and Clinical Psychology, 65, 599±610.

Huey, S.J., Henggeler, S.W., Brondino, M.J., & Pickrel,S.G. (2000). Mechanisms of change in multisystemictherapy: Reducing delinquent behavior through ther-apist adherence and improved family and peer func-tioning. Journal of Consulting and ClinicalPsychology, 68, 451±467.

Huey, W.C., & Rank, R.C. (1984). Effects of counselorand peer-led group assertiveness training on blackadolescent aggression. Journal of Counseling Psycho-logy, 31, 95±98.

Jayson, D., Wood, A., Kroll, L., Fraser, J., & Harrington,R. (1998). Which depressed patients respond tocognitive-behavioral treatment? Journal of the Ameri-can Academy of Child and Adolescent Psychiatry, 37,35±39.

Judd, C.M., & Kenny, D.A. (1981). Process analysis:Estimating mediation in treatment evaluations. Eval-uation Review, 5, 602±619.

Kahn, J.S., Kehle, T.J., Jenson, W.R., & Clark, E.(1990). Comparison of cognitive-behavioral, relaxa-


tion, and self-modeling interventions for depressionamong middle school students. School PsychologyReview, 19, 196±211.

Kanfer F.H., Karoly, P., & Newman, A. (1975). Reduc-tion of children's fear of the dark by competence-related and situational threat-related verbal cues.Journal of Consulting and Clinical Psychology, 43,251±258.

Kaslow, N.J., & Thompson, M.P. (1998). Applying thecriteria for empirically supported treatments tostudies of psychosocial interventions for child andadolescent depression. Journal of Clinical ChildPsychology, 27, 146±155.

Kazdin, A.E. (1978). History of behavior modi®cation:Experimental foundations of contemporary research.Baltimore, MD: University Park Press.

Kazdin, A.E. (1999). Current (lack of) status of theory inchild and adolescent therapy research. Journal ofClinical Child Psychology, 28, 533±543.

Kazdin, A.E. (2000). Developing a research agenda forchild and adolescent psychotherapy. Archives ofGeneral Psychiatry, 57, 829±836.

Kazdin, A.E., Bass, D., Ayers, W.A., & Rodgers, A.(1990). Empirical and clinical focus of child andadolescent psychotherapy research. Journal of Con-sulting and Clinical Psychology, 58, 729±740.

Kazdin, A.E., Esveldt-Dawson, K., French, N.H., &Unis, A.S. (1987a). Effects of parent managementtraining and problem-solving skills training combinedin the treatment of antisocial child behavior. Journalof the American Academy of Child and AdolescentPsychiatry, 26, 416±424.

Kazdin, A.E., Esveldt-Dawson, K., French, N.H., &Unis, A.S. (1987b). Problem-solving skills trainingand relationship therapy in the treatment of antiso-cial child behavior. Journal of Consulting and ClinicalPsychology, 55, 76±85.

Kazdin, A.E., Siegel, T.C., & Bass, D. (1992). Cognitiveproblem-solving skills training and parent manage-ment training in the treatment of antisocial behaviorin children. Journal of Consulting and Clinical Psy-chology, 60, 733±747.

Keller, M., Lavori, P., Wunder, J., Beardslee, W.,Schwartz, C., & Roth, J. (1992). Chronic course ofanxiety disorders in children and adolescents. Jour-nal of the American Academy of Child and AdolescentPsychiatry, 31, 595±599.

Kendall, P.C. (1994). Treating anxiety disorders inyouth: Results of a randomized clinical trial. Journalof Consulting and Clinical Psychology, 62, 100±110.

Kendall, P.C., & Grove, W.M. (1988). Normative com-parisons in therapy outcome. Behavioral Assessment,10, 147±158.

Kendall, P.C., & Southam-Gerow, M.A. (1996). Long-term follow up of a cognitive-behavioural therapy foranxiety-disordered youth. Journal of Consulting andClinical Psychology, 64, 724±730.

Kendall, P.C., Flannery-Schroeder, E.C., Panichelli-Mindel, S.P., Southam-Gerow, M.A., Henin, A., &Warman, M.J. (1997). Treating anxiety disorders inyouth: A second randomized clinical trial. Journal ofConsulting and Clinical Psychology, 65, 366±380.

Kolko, D., Brent, D., Baugher, M., Bridge, J., & Birma-her, B. (2000). Cognitive and family therapies foradolescent depression: Treatment speci®city, medi-

ation and moderation. Journal of Consulting andClinical Psychology, 68, 603±614.

Kondas, O. (1967). Reduction of examination anxietyand `stage fright' by group desensitization and relax-ation. Behaviour Research and Therapy, 5, 275±281.

Kuroda, J. (1969). Elimination of children's fears ofanimals by the method of experimental desensitiza-tion: An application of learning theory to childpsychology. Psychologia, 12, 161±165.

Last, C., Hersen, M., Kazdin, A., Francis, G., & Grubb,H. (1987). Psychiatric illness in the mothers ofanxious children. American Journal of Psychiatry,144, 1580±1583.

Leitenberg, H., & Callahan, E.J. (1973). Reinforcedpractice and reduction of different kinds of fears inadults and children. Behaviour Research and Ther-apy, 11, 19±30.

Levitt, E.E. (1957). The results of psychotherapy withchildren: An evaluation. Journal of Consulting andClinical Psychology, 21, 189±196.

Levitt, E.E. (1963). Psychotherapy with children: Afurther evaluation. Behaviour Research and Therapy,60, 326±329.

Lewinsohn, P.M., & Clarke, G.N. (1999). Psychosocialtreatments for adolescent depression. Clinical Psy-chology Review, 19, 329±342.

Lewinsohn, P.M., Clarke, G.N., Hops, H., & Andrews, J.(1990). Cognitive-behavioral treatment for depressedadolescents. Behavior Therapy, 21, 385±401.

Lewinsohn, P.M., Clarke, G.N., Rohde, P., Hops, H., &Seeley, J.R. (1996). A course in coping: A cognitive-behavioral approach to the treatment of adolescentdepression. In E.D. Hibbs & P.S. Jensen (Eds.),Psychosocial treatments for child and adolescentdisorders: Empirically based strategies for clinicalpractice (pp. 109±135). Washington DC: APA.

Lewis, S. (1974). A comparison on behavior therapytechniques in the reduction of fearful avoidancebehavior. Behavior Therapy, 5, 648±655.

Liddle, B., & Spence, S.H. (1990). Cognitive-behaviortherapy with depressed primary school children: Acautionary note. Behavioural Psychology, 18, 85±102.

Lochman, J.E., Burch, P.R., Curry, J.F., & Lampron,L.B. (1984). Treatment and generalization effects ofcognitive-behavioral and goal-setting interventionswith aggressive boys. Journal of Consulting andClinical Psychology, 52, 915±916.

Lochman, J.E., Lampron, L.B., Gemmer, T.C., & Harris,S.R. (1989). Teacher consultation and cognitive-behavioral interventions with aggressive boys. Psy-chology in the Schools, 26, 179±188.

Loeber, R. (1990). Development and risk factors ofjuvenile antisocial behavior and delinquency. ClinicalPsychology Review, 10, 1±41.

Loeber, R., Wung, P., Keenan, K., Giroux, B., South-amer-Loeber, M., Van Kammen, W.B., & Maughan, B.(1993). Developmental pathways in disruptive childbehavior. Development and Psychopathology, 5, 103±133.

Lonigan, C.J., & Elbert, J.C. (1998). Special issue onempirically supported psychosocial interventions forchildren [Special issue]. Journal of Clinical ChildPsychology, 27.

Lonigan, C.J., Elbert, J.C., & Johnson, S.B. (1998).Empirically supported psychosocial interventions for


children: An overview. Journal of Clinical Child Psy-chology, 27, 138±145.

Lyubomirsky, S., Caldwell, N.D., & Nolen-Hoeksema, S.(1998). Effects of ruminative and distracting respons-es to depressed mood on retrieval of autobiographicalmemories. Journal of Personality and Social Psychol-ogy, 75, 166±177.

Mann, J., & Rosenthal, T.L. (1969). Vicarious and directcounter-conditioning of test anxiety through individ-ual and group desensitization. Behaviour Researchand Therapy, 7, 359±367.

McNeil, C.B., Eyberg, S., Eisenstadt, T.H., Newcomb,K., & Funderburk, B. (1991). Parent±child interactiontherapy with behavior problem children: Generaliza-tion of treatment effects to the school setting. Journalof Clinical Child Psychology, 20, 140±151.

Menzies, R.G., & Clark, J.C. (1993). A comparison of invivo and vicarious exposure in the treatment ofchildhood water phobia. Behaviour Research andTherapy, 31, 9±15.

Miller, L.C., Barrett, C.L., Hampe, E., & Noble, H.(1972). Comparison of reciprocal inhibition, psycho-therapy, and waiting list control for phobic children.Journal of Abnormal Psychology, 79, 269±279.

Mineka, S., & Thomas, C. (1999). Mechanisms ofchange in exposure therapy for anxiety disorders. InT. Dalgleish & M. Power (Eds.), Handbook of cognitionand emotion (pp. 747±764). New York: Wiley.

Mof®tt, T.E. (1993). Life-course persistent and adoles-cent-limited antisocial behavior: A developmentaltaxonomy. Psychological Review, 100, 674±701.

Murphy, C.M., & Bootzin, R.R. (1973). Active andpassive participation in the contact desensitization ofsnake fear in children. Behavior Therapy, 4, 203±211.

Nolen-Hoeksema, S., & Morrow, J. (1993). Effects ofrumination and distraction on naturally occurringdepressed mood. Cognition and Emotion, 7, 561±570.

Obler, M., & Terwilliger, R.F. (1970). Pilot study on theeffectiveness of systematic desensitization with neu-rologically impaired children with phobic disorders.Journal of Consulting and Clinical Psychology, 34,314±318.

Ollendick, T.H., & King, N.J. (1998). Empirically sup-ported treatments for children with phobic and anxi-ety disorders. Journal of Clinical Child Psychology,27, 156±167.

Patterson, G.R., & Forgatch, M.S. (1995). Predictingfuture clinical adjustment from treatment outcomeand process variables. Psychological Assessment, 7,275±285.

Patterson, G.R., & Guillion, M.E. (1968). Living withchildren: New methods for parents and teachers.Champaign, IL: Research Press.

Peed, S., Roberts, M., & Forehand, R. (1977). Evalua-tion of the effectiveness of a standardized parenttraining program in altering the interaction of moth-ers and their noncompliant children. Behavior Modi-®cation, 1, 323±350.

Quiggle, N.L., Garber, J., Panak, W.F., & Dodge, K.A.(1992). Social information processing in aggressiveand depressed children. Child Development, 63,1305±1320.

Reinecke, M.A., Ryan, N.E., & Dubois, D.L. (1998).Cognitive-behavioral therapy of depression and de-

pressive symptoms during adolescence: A review andmeta-analysis. Journal of the American Academy ofChild and Adolescent Psychiatry, 37, 26±34.

Reynolds, W.M., & Coats, K.I. (1986). A comparisonof cognitive-behavioral therapy and relaxation train-ing for the treatment of depression in adolescents.Journal of Consulting and Clinical Psychology, 54,653±660.

Ritter, B. (1968). The group desensitization of children'ssnake phobias using vicarious and contact desensi-tization procedures. Behaviour Research and Ther-apy, 6, 1±6.

Rohde, P., Lewinsohn, P.M., & Seeley, J.R. (1994). Areadolescents changed by an episode of major depres-sion? Journal of the American Academy of Child andAdolescent Psychiatry, 33, 1289±1298.

RosselloÂ, J., & Bernal, G. (1999). The ef®cacy ofcognitive-behavioral and interpersonal treatmentsfor depression in Puerto Rican adolescents. Journalof Consulting and Clinical Psychology, 67, 734±745.

Rusting, C.L., & DeHart, T. (2000). Retrieving positivememories to regulate negative mood: Consequencesof mood-congruent memory. Journal of Personalityand Social Psychology, 78, 737±752.

Schlichter, K.J., & Horan, J.J. (1981). Effects of stressinoculation on the anger and aggression managementskills of institutionalized juvenile delinquents. Cogni-tive Therapy and Research, 5, 359±365.

Scott, A.G., & Sechrest, L. (1989). Strength of theoryand theory of strength. Evaluation and ProgramPlanning, 12, 329±336.

Shaffer, D., Gould, M.S., Fisher, P., Trautman, P.,Moreau, D., Kleinman, M., & Flory, M. (1996).Psychiatric diagnosis in child and adolescent suicide.Archives of General Psychiatry, 53, 339±348.

Sheslow, D.V., Bondy, A.S., & Nelson, R.O. (1983). Acomparison of graduated exposure, verbal copingskills, and their combination in the treatment ofchildren's fear of the dark. Child and Family BehaviorTherapy, 4, 33±45.

Spaccarelli, S., Cotler, S., & Penman, D. (1992). Prob-lem-solving skills training as a supplement to beha-vioral parent training. Cognitive Therapy andResearch, 16, 1±18.

Stark, K.D., Reynolds, W.M., & Kaslow, N.J. (1987). Acomparison of the relative ef®cacy of self-controltherapy and a behavioral problem-solving therapyfor depression in children. Journal of Abnormal ChildPsychology, 15, 91±113.

Stark, K., Rouse, L., & Livingston, R. (1991). Treatmentof depression during childhood and adolescence:Cognitive-behavioral approaches for the individualand family. In P. Kendall (Ed.), Child and adolescenttherapy (pp. 165±206). New York: Guilford.

Strauss, C.C., Frame, C.L., & Forehand, R.L. (1987).Psychosocial impairment associated with anxiety inchildren. Journal of Clinical Child Psychology, 16,235±239.

Taylor, J., Iacono, W.G., & McGue, M. (2000). Evidencefor a genetic etiology of early-onset delinquency.Journal of Abnormal Psychology, 109, 634±643.

Treadwell, K.R.H., & Kendall, P.C. (1996). Self-talk inyouth with anxiety disorders: Content speci®city andtreatment outcome. Journal of Consulting and ClinicalPsychology, 64, 941±950.


Tremblay, R.E., Pagani-Kurtz, L., Masse, L.C., Vitaro,F., & Phil, R. (1995). A bimodal preventative inter-vention for disruptive kindergarten boys: Its impactthrough mid-adolescence. Journal of Consulting andClinical Psychology, 63, 560±568.

Ultee, C.A., Grif®oen, D., & Schellekens, J. (1982). Thereduction of anxiety in children: A comparison ofthe effects of `systematic desensitization in vitro'and `systematic desensitization in vivo.' BehaviourResearch and Therapy, 20, 61±67.

Vasey, M.W., & Lonigan, C.J. (2000). Considering theclinical utility of performance-based measures ofchildhood anxiety. Journal of Clinical Child Psycho-logy, 29, 493±508.

Vitaro, F., & Tremblay, R.E. (1994). Impact of aprevention program on aggressive children's friend-ships and social adjustment. Journal of AbnormalChild Psychology, 22, 457±475.

Vostanis, P., Feehan, C., & Grattan, E. (1998). Two-yearoutcome of children treated for depression. EuropeanChild and Adolescent Psychiatry, 7, 12±18.

Vostanis, P., Feehan, C., Grattan, E., & Bickerton, W.(1996a). A randomised controlled out-patient trialof cognitive-behavioural treatment for children andadolescents with depression: 9-month follow-up.Journal of Affective Disorders, 40, 105±116.

Vostanis, P., Feehan, C., Grattan, E., & Bickerton, W.(1996b). Treatment for children and adolescents withdepression: Lessons from a controlled trial. ClinicalChild Psychology and Psychiatry, 1, 199±212.

Webster-Stratton, C. (1984). Randomized trial of twoparent-training programs for families with conduct-disordered children. Journal of Consulting and Clin-ical Psychology, 52, 666±678.

Webster-Stratton, C. (1990). Enhancing the effective-ness of self-administered videotape parent trainingfor families with conduct-problem children. Journal ofAbnormal Child Psychology, 18, 479±492.

Webster-Stratton, C. (1992). Individually administeredvideotape parent training: `Who Bene®ts?'. CognitiveTherapy and Research, 16, 31±52.

Webster-Stratton, C. (1994). Advancing videotape par-ent training: A comparison study. Journal of Consult-ing and Clinical Psychology, 62, 583±593.

Webster-Stratton, C., Kolpacoff, M., & Hollinsworth, T.(1988). Self-administered videotape therapy for famil-ies with conduct-problem children: Comparison of twocost-effective treatments and a control group. Journalof Consulting and Clinical Psychology, 56, 558±566.

Weiss, B.H., & Weisz, J.R. (1995). Relative effectivenessof behavioral and nonbehavioral child psychotherapy.Journal of Consulting and Clinical Psychology, 63,317±320.

Weisz, J.R. (2000). Agenda for child and adolescentpsychotherapy research: On the need to put scienceinto practice. Archives of General Psychiatry, 57,837±838.

Weisz, J.R., & Weersing, V.R. (1999). Psychotherapywith children and adolescents: Ef®cacy, effectiveness,and developmental concerns. In D. Cicchetti & S.L.

Toth (Eds.), Rochester symposium on developmentalpsychopathology: Vol. 9. Developmental approaches toprevention and intervention (pp. 341±386). Rochester,NY: University of Rochester Press.

Weisz, J.R., & Hawley, K.M. (in press). Developmentalfactors in the treatment of adolescents. Journal ofConsulting and Clinical Psychology.

Weisz, J.R., & Weiss, B.H. (1991). Studying the `refer-ability' of child clinical problems. Journal of Consult-ing and Clinical Psychology, 59, 266±273.

Weisz, J.R., Huey, S.M., & Weersing, V.R. (1998).Psychotherapy outcome research with children andadolescents: The state of the art. In T.H. Ollendick &R.J. Prinz (Eds.), Advances in clinical child psychology(vol. 20, pp. 49±92). New York: Plenum.

Weisz, J.R., Thurber, C.A., Sweeney, L., Prof®tt, V.D.,& LeGagnoux, G.L. (1997). Brief treatment of mild-to-moderate child depression using primary and secon-dary control enhancement training. Journal ofConsulting and Clinical Psychology, 65, 703±707.

Weisz, J.R., Weiss, B., & Donenberg, G.R. (1992). Thelab versus the clinic: Effects of child and adolescentpsychotherapy. American Psychologist, 47, 1578±1585.

Weisz, J.R., Weiss, B., Alicke, M.D., & Klotz, M.L.(1987). Effectiveness of psychotherapy with childrenand adolescents: A meta-analysis for clinicians. Jour-nal of Consulting and Clinical Psychology, 55, 542±549.

Weisz, J.R., Weiss, B., Han, S.S., Granger, D.A., &Morton, T. (1995). Effects of psychotherapy withchildren and adolescents revisited: A meta-analysisof treatment outcome studies. Psychological Bulletin,117, 450±468.

Wells, K.C., & Egan, J. (1988). Social learning andsystems family therapy for childhood oppositionaldisorder: Comparative treatment outcome. Compre-hensive Psychiatry, 29, 138±146.

Wiltz, N.A., & Patterson, G.R. (1974). An evaluation ofparent training procedures designed to alter inappro-priate aggressive behavior of boys. Behavior Therapy,5, 215±221.

Wolpe, J. (1954). Reciprocal inhibition as the mainbasis of psychotherapeutic effects. Archives of Neu-rology and Psychiatry, 72, 205±226.

Wood, A., Harrington, R., & Moore, A. (1996). Con-trolled trial of a brief cognitive-behavioural interven-tion in adolescent patients with depressive disorders.Journal of Child Psychology and Psychiatry, 37, 737±746.

Yeaton, W.H., & Sechrest, L. (1981). Critical dimen-sions in the choice and maintenance of successfultreatments: Strength, integrity, and effectiveness.Journal of Consulting and Clinical Psychology, 49,156±167.

Zangwill, W.M. (1983). An evaluation of a parenttraining program. Child and Family Behavior Ther-apy, 5, 1±6.

Manuscript accepted 14 August 2001


mechanismsofaction inyouthpsychotherapy

Documents