Does Training Parents in Reinforcement Skills or Relationship Skills Enhance Individual Youths’ Cognitive Behavioral Therapy for Anxiety? Outcome,Specificity,and Mediation

Abstract

We conducted a dismantling design treatment study comparing individual cognitive behavioral therapy (CBT), CBT targeting parents’ reinforcement skills (CBT + Reinf), and CBT targeting parents’ relationship skills (CBT+ Relat) in 341 youths with primary anxiety diagnoses. At posttreatment, youths in CBT with parent involvement had lower anxiety than youths in CBT. At 12-month follow-up, youths in CBT + Relat maintained lower anxiety relative to CBT. At posttreatment, negative reinforcement was significantly lower in CBT + Reinf than CBT + Relat and CBT; negative reinforcement partially mediated anxiety reduction in youths. Reducing parental negative reinforcement in CBT + Reinf was associated with lower parental psychological control, which also partially mediated anxiety reduction in youths. Some of these mediational dynamics continued through follow-up. Targeting concrete behavioral parenting skills, especially negative reinforcement, produced treatment specificity and partial mediation relative to less concrete targeting and enhanced CBT. Findings highlight complexities in identifying mechanisms through which targeting of parenting skills produces anxiety reduction in youths and suggest avenues for future research.

Keywords

anxiety disorders clinical child intervention cognitive therapy CBT clinical trials evidence-based treatments

An abundance of randomized controlled efficacy trials have shown that despite a few encouraging findings (Barrett et al., 1996; Wood et al., 2006), cognitive behavioral therapy (CBT) with parental involvement generally fails to enhance anxiety outcome in youths compared with CBT without parental involvement (e.g., Kendall et al., 2008; Nauta et al., 2003; Silverman et al., 2009). A critical silver lining has emerged, nevertheless, from the cumulative body of research: Research has begun to shift from asking whether parental involvement in their child’s treatment makes a difference (e.g., Barrett et al., 1996; Kendall et al., 2008) to how parental involvement might make a difference (Settipani et al., 2013; Silverman et al., 2009, 2019).

Studies informed by this shift addressed the question of how by using treatment mechanism designs, which involve assessing and testing hypothesized mechanisms/mediators (La Greca et al., 2009; Nock et al., 2007; Weersing & Weisz, 2002). Although these studies represent important initial steps in addressing how, the treatments still targeted multiple parent components (e.g., Silverman et al., 2009). It therefore remains unknown which parent components are critical to understand how CBTs with parental involvement produce positive anxiety outcome in youths.

The current study represents a continuation and extension of past work to address how parental involvement might make a difference. Like past work, we used a treatment mechanism design (Settipani et al., 2013; Silverman et al., 2009, 2019). We further extended past work in a critical way: This trial is the first to use a dismantling design to address the question of how parental involvement might make a difference (e.g., Nock et al., 2007). Guided by theory and empirical findings about the role of parents in etiology and maintenance of anxiety in youths (e.g., Barber, 1996; Ginsburg et al., 1995; McLeod et al., 2007; van der Bruggen et al., 2008), we distilled key components from CBTs with parent involvement (e.g., Barmish & Kendall, 2005; Manassis et al., 2014). We isolated these components and targeted them specifically in two parent treatment arms. Families were randomly assigned to one of the following three arms: (a) CBT with parent involvement in which parents were trained in child behavior management skills—specifically, in using more positive reinforcement and less negative reinforcement (CBT + reinforcement skills [CBT + Reinf]); (b) CBT with parent involvement in which parents were trained in child-parent relationship skills—specifically, in using more acceptance and less psychological control (CBT + relationship skills [CBT + Relat]); and (c) individual CBT for youths only (CBT)—our baseline comparator control arm.

We addressed the following questions:

When delivering CBT + Reinf and CBT + Relat, do these two treatments produce superior effects on anxiety outcome in youths compared with CBT?

When delivering CBT + Reinf and CBT + Relat, do these two treatments produce specific effects on the targeted parent components?

When delivering CBT + Reinf and CBT + Relat, do the targeted parent components partially mediate the effect of the treatments on anxiety outcome in youths?

Do These Two CBTs With Parent Involvement Produce Superior Effects on Anxiety Outcome in Youths Compared With CBT?

Creswell et al. (2020) suggested that parent involvement has been ineffective in enhancing CBT outcomes because it may dilute delivery of key CBT procedures, especially when parent involvement is not directed toward facilitating exposure for youths. The authors called for a more “nuanced approach” to involving parents in their child’s CBT in future efforts to enhance outcomes (p. 6). This call echoes suggestions made earlier by Silverman and Kurtines (1996, Chapter 4) regarding the importance of using parent components that are “facilitative strategies” of exposure, such as contingency management (Silverman et al., 1999) and strategies that maintain exposure for youths through transfer of control parent-to-child mechanisms (e.g., reducing parental psychological control; e.g., Ginsburg et al., 1995).

Consistent with these suggestions, in one encouraging study, Wood and colleagues (2006) compared CBT and CBT with parental involvement (N = 40; ages 6–11; Building Confidence program). Building Confidence emphasized in vivo exposure procedures and rewards with parent training and changing parental communication patterns relating to intrusiveness and autonomy granting (i.e., psychological control). At posttreatment, Building Confidence produced greater improvement on independent evaluators’ ratings and parent ratings but not child ratings of child anxiety in comparison with CBT. Because the final assessment was conducted at posttreatment, it is unclear whether the greater improvement of Building Confidence was maintained over time.

Mirroring most individual studies, meta-analyses have failed to find enhanced outcomes in comparisons between CBT with and without parental involvement (Breinholst et al., 2012; In-Albon & Schneider, 2007; S. Reynolds et al., 2012; Silverman et al., 2008; Spielmans et al., 2007; Thulin et al., 2014). A recent meta-analysis similarly concluded that CBT with parent involvement “did not confer advantage over individual CBT” (Peris et al., 2021, p. 287). The meta-analysis also highlighted a research gap on measurement of parent/family variables noted more than a decade ago by Silverman and colleagues (2009). Specifically, only six of 11 trials of individual child CBT compared with CBT with parent involvement identified by Peris and colleagues (2021) included a parent/family measure to assess outcomes.

There are, however, two meta-analyses that adopted a more nuanced approach and showed encouraging preliminary results (e.g., Manassis et al., 2014; Sun et al., 2018). Manassis and colleagues (2014) compared (a) individual CBTs for youths with anxiety with limited parental involvement, (b) CBTs with active parental involvement with low emphasis on either contingency management or transfer of control, and (c) CBTs with active parental involvement with high emphasis on either contingency management or transfer of control. There were no significant pretreatment to posttreatment differences between any type of parental involvement and individual CBTs for youths (i.e., no significant differences between the first vs. the second vs. the third treatments). However, for youths with anxiety, CBTs with high emphasis on either contingency management or transfer of control showed better long-term maintenance at 12-month follow-up compared with CBTs with low emphasis on contingency management or transfer of control and individual CBT (third treatment > second and first treatments).

A similar pattern of findings was reported in Sun and colleagues’ (2018) meta-analysis of treatment features associated with internalizing outcomes in youths, although the findings for parental involvement were as encouraging at posttreatment as follow-up. Sun and colleagues thus echoed Manassis and colleagues (2014) that parental involvement in contingency management and transfer of control likely matters and that overall, “at least some involvement of parents (as opposed to none) will in general result in larger and more durable effects” (p. 11).

Given CBT + Reinf and CBT + Relat both have evidence for efficacy, at least when combined into a single intervention (e.g., Wood et al., 2006), and have similar medium effect sizes (e.g., Manassis et al., 2014), we had no reason to expect one parent treatment arm would be superior to another compared with CBT, the baseline comparator arm. However, because we designed each parent treatment arm to facilitate exposure for youths by isolating key parenting components (i.e., reinforcement skills, relationship skills) and delivered these components in a highly streamlined manner compared with past multicomponent parent CBTs, we expected that both CBTs with parent involvement would enhance outcomes relative to CBT at posttreatment and follow-up.

Do These Two Treatments That Involve Parents Produce Specific Effects on the Parent Components?

The emergent shift to how parental involvement might make a difference has led to refinements in randomized controlled efficacy trials’ measurement and design. Past trials rarely assessed parent components that were targeted, and when components were assessed, trials rarely tested whether changes on these parent components were specific to their targeting (treatment specificity). We are aware of four anxiety treatment studies of youths that examined whether treatment specificity occurs when parent components are targeted in CBTs with and without parent involvement (Jongerden & Bögels, 2014, N = 104; Kendall et al., 2008, N = 161; Silverman et al., 2009, N = 119; Silverman et al., 2019, N = 183). Parent components targeted in the CBTs with parent involvement were parenting skills and parent-youth relationships. Treatment specificity was not found in three of these studies: Parent components improved in the parent treatment arm (parent components targeted) and in the individual CBT arm (not targeted; Jongerden & Bögels, 2014; Kendall et al., 2008; Silverman et al., 2009). In contrast, a fourth study that compared CBT with parent involvement with group CBT for youths found that parental psychological control was significantly lower in CBT with parent involvement than group CBT, indicating specificity (Silverman et al., 2019).

Given that the literature is sparse and that most studies fail to find treatment-specific effects, we were more tentative regarding our expectations about specificity. Nevertheless, because we isolated the parent components and used them in two respective streamlined parent treatment arms, we were optimistic that the treatments would produce treatment specificity. We also had reason for optimism given that Wood and colleagues’ (2006) targeted these components and that the preliminary results of meta-analyses suggest prominent roles for contingency management and transfer of control (Manassis et al., 2014; Sun et al., 2018); both align with the two parent arms in the current study. Thus, we expected that parent reinforcement skills would improve more in CBT + Reinf than in CBT + Relat and CBT. We also expected that parent relationship skills would improve more in CBT + Relat than in CBT + Reinf and CBT.

Do the Targeted Parent Components (Partially) Mediate Anxiety Outcome in Youths?

Although it is widely assumed in the anxiety treatment literature on youths that targeting parenting components improves those parenting components, which, in turn, produces change in anxiety in youths (i.e., mediates), the lack of studies testing parenting as a mediator of treatment effects on anxiety in youths is a common lament (e.g., Creswell et al., 2020; Herres et al., 2015; Manassis et al., 2014; Sun et al., 2018; Thulin et al., 2014). We know of only two studies, but neither one was a direct comparison of CBT with parental involvement and individual CBT for youths with anxiety disorders. Casline and colleagues (2018) compared CBT with parent involvement to information management in a prevention study with anxious parents and their nonanxious children; in a randomized control efficacy study, Silverman and colleagues (2019) compared CBT with parent involvement to group CBT for youths with anxiety disorders. Findings revealed that parental negative reinforcement of avoidance behaviors in youths partially mediated treatment effects on anxiety outcome in youths at 12-month follow-up (Casline et al., 2018) and that parental psychological control mediated anxiety outcome in youths at posttreatment and at 12-month follow-up (Silverman et al., 2019). Findings varied across informants and highlighted complexities of deciphering mediational dynamics. This complexity is consistent with a larger pattern of diverse, intricate findings on the directionality of change in CBT with parent involvement, which makes it difficult to draw firm conclusions (Settipani et al., 2013; Silverman et al., 2009).

Given the above and again, because we compared two streamlined parent treatment arms that isolated key parent components to CBT, we expected that the parenting components we targeted would partially mediate treatment effects on anxiety outcome in youths. We viewed it premature to predict posttreatment mediational effects compared with 12-month follow-up mediational effects given the scarcity and inconsistency of the data (Manassis et al., 2014; Silverman et al., 2019; Sun et al., 2018; Wood et al., 2006). Likewise, we viewed it premature to predict how mediational dynamics would vary by informant. However, given that data reveal different findings for mediation in research on anxiety treatment in youths and research on treatment more generally (e.g., De Los Reyes & Kazdin, 2008), we expected our findings would vary by informants.

The Present Study

We first present anxiety treatment outcomes in youths in the two parent arms and in CBT. Because these are well-supported, evidenced-based approaches (e.g., Silverman et al., 2008), we hypothesized that all would produce statistically significant pretreatment-posttreatment changes on the study’s outcomes. As discussed above, we expected that both CBTs with parent treatments would enhance outcomes relative to CBT at posttreatment and at follow-up. As we also noted, we moved beyond efficacy outcome and examined treatment specificity and mediational dynamics of the two parent treatment arms. The four parenting components we isolated and studied are in line with those studied in past anxiety treatment studies focused on youths (but not isolated or dismantled) and are associated with etiology and maintenance of anxiety disorders in youths (e.g., Casline et al., 2018; Manassis et al., 2014; Silverman et al., 2019; Wood et al., 2006). In CBT + Reinf, the two components targeted were (a) increases in parental positive reinforcement of their child’s approach behaviors toward anxiety-provoking events and (b) decreases in parental negative reinforcement of their child’s avoidance behaviors, that is, decreases in permitting child avoidance of anxiety-provoking events, which allays distress in youths. In CBT + Relat, the two components targeted were (a) increases in parental expressions of acceptance toward their child and (b) decreases in psychological control of their child, that is, decreases in intruding on their child’s expressions of autonomy, which can improve the youth-parent relationship (e.g., Barber, 1996; Ginsburg et al., 1995; Wood et al., 2006).

With respect to treatment specificity effects, our first hypothesis was that CBT + Reinf compared with CBT + Relat and CBT would produce more parental positive reinforcement and less parental negative reinforcement because these were the components targeted. Our second hypothesis was that CBT + Relat compared with CBT + Reinf and CBT would produce more parental acceptance and less parental psychological control because these were the components targeted. With respect to treatment mediation effects, we hypothesized that changes produced on targeted parent components would partially mediate treatment outcome.

Method

Participants

Participants were 341 youths ages 7 to 16 (M = 9.52 years, SD = 2.48; 47.5% girls) and their parents (mostly mothers) who presented to an anxiety disorders research clinic for youths and began treatment in this randomized efficacy clinical trial (ClinicalTrials.gov identifier NCT00620958; for further details, see Fig. 1 and the Treatment Noncompletion section below). Major referral sources included pediatricians, school psychologists, school counselors, and other professionals. Two hundred seventy-nine youths were Hispanic/Latino (81.8%), 39 were European American (11.4%), five were Asian American (1.5%), four were African American (1.2%), eight reported “other” ethnicity (2.3%), and six did not report ethnicity (1.8%). Regarding marital status, 77.7% of parents were married, 10% were divorced, 3.8% were single, 3.5% were cohabiting with a partner, 2% were separated, 1.2% were widowed, and 1.8% did not report marital status. Regarding annual family income, 16.4% reported incomes less than $20,000, 19.1% reported incomes between $21,000 and $40,000, 15.2% reported incomes between $41,000 and $60,000, 14.1% reported incomes between $61,000 and $80,000, 9.7% reported incomes between $81,000 and $100,000, 13.8% reported incomes between $101,000 and $149,000, 5% reported incomes over $150,000, and 6.7% did not report income.

Fig. 1.

Consolidated Standards of Reporting Trials (CONSORT) diagram. Reinf = reinforcement; Relat = relationship; Tx = treatment.

To be eligible for inclusion, all youths were required to meet criteria for a primary anxiety disorder (fourth edition of the Diagnostic and Statistical Manual of Mental Disorders, American Psychiatric Association, 1994) using the Anxiety Disorders Interview Schedule for Children parent and child versions (ADIS-IV: P/C; Silverman & Albano, 1996). Primary diagnoses were separation anxiety disorder (28%), social phobia (23%), generalized anxiety disorder (23%), specific phobia (14%), and other anxiety disorders (12%). Developmental disabilities, psychosis, or current involvement in another psychosocial treatment were exclusion criteria. A small percentage of youths were on a stable dose of psychotropic medications, including selective serotonin reuptake inhibitors (SSRIs; 4.1% CBT + Reinf, 1.1% CBT + Relat, 5.6% CBT), stimulants (4.9% CBT + Reinf, 5.4% CBT + Relat, 1.6% CBT), other medications (e.g., anxiolytics, nonstimulant medication for attention-deficit/hyperactivity disorder; 0.8% CBT + Reinf, 2.2% CBT + Relat, 1.6% CBT), or a combination of medications (e.g., anxiolytic and SSRI; 1.6% CBT + Reinf, 0.0% CBT + Relat, 1.6% CBT).

Measures

Diagnostic instrument administered to parents and youths

Independent evaluators (IEs) masked to treatment arm and assessment wave administered the ADIS-IV: P/C to parents and youths, respectively. The disorder most interfering/impairing was considered as primary and was targeted in treatment. Before administering the ADIS-IV: P/C, IEs were trained in administration and scoring procedures and met 100% reliability criterion on five videotaped assessments of parents and youths. The ADIS-IV: P/C yields interrater reliability κ coefficients between .80 and 1.00 for diagnoses (Lyneham et al., 2007) and retest reliability κ coefficients between .80 and .92 for diagnoses (e.g., Silverman et al., 2001). In this study, interrater reliability κ coefficients were 1.0 for generalized anxiety disorder, 1.0 for social anxiety disorder, .86 for separation anxiety disorder, and .63 for specific phobia. ADIS-IV: P/C anxiety diagnoses have shown significant associations with anxiety ratings in youths (e.g., Wood et al., 2002).

Primary outcome measure completed by parents and youths

The Revised Children’s Manifest Anxiety Scale (RCMAS; C. R. Reynolds & Richmond, 1978) parent (RCMAS/P) and child versions are 37-item rating scales that assess anxiety symptom severity. All items are rated either yes (1) or no (0). Total anxiety scores range from 0 to 28, and higher scores indicate higher levels of severity. The RCMAS has extensive reliability and validity data (C. R. Reynolds & Richmond, 1978) and is sensitive to change in clinical trials (Silverman & Ollendick, 2005). The parent version (RCMAS/P) is reworded from “I . . . ” to “My child . . . ” with the identical stems, as in past studies (e.g., Kendall, 1994; Silverman et al., 1999). In this sample, McDonald’s (1999) total ω was .87 for both the RCMAS/P and the RCMAS.

Clinically significant change measures

The ADIS-IV: P/C (Silverman & Albano, 1996) was used to determine whether participants met diagnostic criteria (i.e., recovered, not recovered) for the primary targeted diagnosis at posttreatment and at 12-month follow-up. All IEs were masked to participants’ diagnoses at pretreatment and treatment arm.

The Children’s Global Assessment Scale (CGAS; Shaffer et al., 1983) is a clinician rating scale from 1 to 100 that assesses functional impairment. The CGAS was used to determine whether participants were in the clinical range at posttreatment and at 12-month follow-up. Scores less than 67 are in the clinical range (Shaffer et al., 1983). All case conference attendees (including study directors, therapists, and IEs) were masked to participants’ identity, status, and assessment point to reduce rater bias (Bird et al., 1993; Dyrborg et al., 2000). IEs’ ratings during case conference meetings were used to establish interrater reliability.

Parent measures completed by parents and youths

The Positive Reinforcement and Negative Reinforcement subscales of the Child Development Questionnaire (CDQ; Zabin & Melamed, 1980), parent (CDQ) and child versions (CDQ/C), include a list of 15 situations representative of common domains of fear and anxiety in youths (e.g., approaching a dog, separation from caregivers). The CDQ was designed to assess parental use of a range of learning mechanisms, including positive reinforcement and negative reinforcement, which were targeted specifically in CBT + Reinf. Similar to past research (e.g., Casline et al., 2018; van der Sluis et al., 2015), we used a modified version in which parents and their children are asked to rate how likely the parent would engage in positive reinforcement and negative reinforcement for each situation, using a rating scale from 1 (never) to 5 (always).

The Positive Reinforcement and Negative Reinforcement subscales each consist of 15 items, and scores range from 15 to 75. Higher scores on positive reinforcement indicate that parents are more likely to provide positive consequences to their child for facing fears. Higher scores on negative reinforcement indicate that parents are more likely to allow their child to avoid feared situations. Zabin and Melamed (1980) found a split-half reliability estimate of .67, and significant associations were found between positive reinforcement and negative reinforcement with anxiety in youths. In this sample, McDonald’s (1999) total ωs were .89 for parents’ ratings of positive reinforcement, .70 for parents’ ratings of negative reinforcement, .92 for youths’ ratings of positive reinforcement, and .81 for youths’ ratings of negative reinforcement.

The Parental Psychological Control and Acceptance Subscales of the Parenting Behavior Inventory (PRPBI/CRPBI; Schludermann & Schludermann, 1970) is a 30-item rating scale that assesses perceived parents’ behaviors toward their youth from the perspective of parents and children, respectively. The PRPBI/CRPBI was designed to assess a range of parental behaviors, including psychological control and acceptance, which were targeted specifically in one of the two parent-involvement arms. The two subscales corresponding to each of these mechanisms (i.e., psychological control, acceptance) were completed by both parents and youths. All items are rated as not like (0), somewhat like (1), or a lot like (2). Both Psychological Control and Acceptance subscales each consist of 10 items, and scores range from zero to 20 on each. Higher scores on psychological control indicate higher levels of perceived psychological control such that parents are more psychologically controlling of their child’s behaviors. Higher scores on acceptance indicate higher levels of perceived acceptance such that parents are warmer and more accepting of their child. Retest reliability estimates range from .79 to .89 (Schludermann & Schludermann, 1988). In this sample, McDonald’s (1999) total ωs were .74 for parents’ ratings of psychological control, .79 for parents’ ratings of acceptance, .79 for youths’ ratings of psychological control, and .87 for youths’ ratings of acceptance.

Procedure

The study was approved by the university’s institutional review board. All parents and youths provided informed consent/assent before participation. Study measures were completed at pretreatment, at posttreatment, and at 12-month follow-up. Using a random numbers table, we randomly assigned youths who met study inclusion to CBT + Reinf, CBT + Relat, and CBT in equal ratios. Families did not know to which treatment they would be assigned until the first session. Each treatment arm contained 12 to 14 weekly sessions of 60 min. In both CBT + Reinf and CBT + Relat, youths and parents were seen together in all sessions in a dyadic format, as in past studies (e.g., Silverman et al., 2009, 2019). CBT had three brief parent meetings, as described below. Treatment manuals were developed to standardize all treatment sessions. All arms targeted anxious symptoms in youths by using in- and out-of-session exposures and cognitive strategies.

CBT + Reinf

In addition to targeting anxious symptoms in youths, CBT + Reinf specifically targeted increasing parental positive reinforcement and decreasing parental negative reinforcement. To address increasing parental positive reinforcement, therapists assisted youths and parents in identifying tangible and intangible reinforcers (e.g., serving favorite meal) for parents to provide to child after completion of exposure tasks. To address decreasing parental negative reinforcement, therapists assisted youths and parents in identifying parental negative reinforcing behaviors (e.g., allowing child to skip a school field trip) for targeting (i.e., reducing) during completion of exposure tasks. Therapists modeled and conducted role plays to help parents identify and modify these parental behaviors. CBT + Reinf involved no parent training or instruction about the variables targeted in CBT + Relat.

CBT + Relat

In addition to targeting anxious symptoms in youths, CBT + Relat specifically targeted increasing parental acceptance and decreasing parental psychological control. To address increasing parental acceptance, therapists assisted youths and parents in identifying parental acceptance behaviors (e.g., helping child to feel better when upset). To address decreasing parental psychological control, therapists assisted youths and parents in identifying parental psychological control behaviors (e.g., guilt induction such as telling child all the things parent has done for them). Therapists modeled and conducted role-plays to help parents identify and modify these parental behaviors. CBT + Relat involved no parent training or instruction about the variables targeted in CBT + Reinf.

CBT

CBT was prototypical in that it emphasized targeting anxiety in youths by delivering psychoeducation and conducting in-session and out-of-session behavioral exposures and training children in adaptive/coping cognitive strategies (e.g., Silverman & Kurtines, 1996). CBT was a child-focused individual format; it contained three meetings only (beginning, middle, end of treatment), during which the therapist met with the parent along with the child for about 15 min, similar to our past studies (e.g., Lebowitz et al., 2019; Silverman et al., 2009, 2019). The purpose of these meetings was to provide information about the child’s CBT goals and procedures and to advise parents that on occasion their child might need their assistance to complete an exposure task (e.g., drop off child at a friend’s house). CBT involved no parent training or instructions, especially about the components targeted in the two parent treatment arms.

Therapists

Therapists were doctoral-level and master’s-level psychology graduate students. Therapists were randomly assigned to cases across arms because neither required differential skill level or background. W. K. Silverman provided didactic and clinical training and weekly supervision.

Treatment integrity

Treatment sessions were video recorded; therapists were unaware of the sessions that would be assessed. IEs, not involved in the study, rated a randomly selected 20% of videotapes on a checklist derived from past studies (Silverman et al., 1999, 2009). Ratings for the targeting of positive reinforcement and negative reinforcement variables yielded 100% present in CBT + Reinf and 0% present in CBT + Relat. Ratings for the targeting of acceptance and psychological control variables yielded 100% present in CBT + Relat and 0% present in CBT + Reinf.

Analytic approach

To evaluate models of treatment specificity and mediation, we used structural equation modeling (SEM) with variables representing the treatment arms, the posttreatment measures, and the follow-up measures. Because this is an efficacy trial, our focus was on a per protocol population (i.e., people who functionally complete treatment). Popular intent-to-treat (ITT) analyses are not appropriate for efficacy trials because they test the effects of random assignment to treatment conditions (i.e., they focus on effectiveness rather than efficacy). ITT analyses confound efficacy, implementation integrity, and patient adherence to treatment protocols and thus are problematic when applied to efficacy trials (Dallal, 2012; Feinman, 2009; Gross & Fogg, 2004). For details, see the Supplemental Material available online.

Figure 2 depicts the conceptual model that was tested using SEM. We defined dummy variables for the three treatment arms (CBT + Reinf, CBT + Relat, CBT) and entered two of them into the model for any given SEM analysis. In the model, the treatment arms were assumed to directly affect anxiety in youths at posttreatment (path a) and at 12-month follow-up (path b) and the mediators at posttreatment (paths c, d, e, f ) and at 12-month follow-up (paths g, h, i, j). Contemporaneous effects are reflected by the paths from the respective mediators to anxiety in youths at posttreatment (paths k, l, m, n) and at follow-up (paths o, p, q, r). The model also included autoregressive effects (paths s, t, u, v, w) and lagged effects (paths x, y, z, aa, ab, ac, ad, ae), per Figure 2, which we explicate in more depth below. We included the pretreatment measure of a given endogenous variable as a covariate in the linear equations predicting that endogenous variable. We used three additional covariates for each endogenous variable because of their past associations with anxiety: (a) biological sex, (b) age, and (c) ethnicity. The inclusion of all the covariates addresses sequential ignorability assumptions in mediation analysis (Imai et al., 2010; Ten Have & Joffe, 2012) and increases statistical power. Figure 2 excludes youths’ sex, age, and ethnicity and baseline exogenous variables to avoid clutter, but we included these variables in all model tests as covariates. Model estimation used Huber-White robust estimation (option MLR in Mplus [Version 7; L. K. Muthén & Muthén, 2012]). Missing data for the per protocol population was addressed using full information maximum likelihood (FIML; Enders, 2010), per Mplus defaults.

Fig. 2.

Conceptual model. PR = positive reinforcement; NR = negative reinforcement; PC = psychological control; ACPT = acceptance.

We used a variety of fit indices to evaluate global model fit. These included the overall χ² test of model fit, the comparative fit index (CFI), the root mean square error of approximation (RMSEA), the p value for close fit (pClose), and the standardized root mean residual (SRMR). In addition, we examined focused tests of ill fit, including modification indices and standardized residuals. For tests of null hypotheses of mediation, we used the joint significance test, which is generally preferable to product-coefficient methods for purposes of null hypothesis testing (see Mallinckrodt et al., 2006; Yzerbyt et al., 2018). For analyses using the product-coefficient approach, see the Supplemental Material. We did not use causal mediation frameworks because all our mediators and outcomes were continuous and we had multiple mediators with causal relationships among them, which greatly complicates causal mediation analysis (Daniel et al., 2015; Pearl, 2009).

Results

Preliminary analyses

Treatment nonstarters

Of the 414 youths randomly assigned, 341 began treatment (82.4%). There were no statistically significant differences between nonstarters and youths who began treatment on any of the primary outcome and mediator measures or sociodemographic variables (e.g., youths’ age, sex, ethnicity) as measured at baseline. Not starting treatment was due primarily to our inability to recontact the nonstarters to begin treatment. Families in the CBT-only arm were less likely to be nonstarters (8.8%) than the other two treatment arms combined (21.9%), but we could identify no basis for the difference other than naturally occurring sampling imbalance in random assignment. Given that nonstarter status was not significantly related to any of the variables studied, the imbalance should not affect results given our sample size (Strube, 1991, 2015).

Treatment noncompletion

Of the 341 youths who started treatment, 254 completed treatment (74.5%). These completion rates are similar to rates reported in past efficacy trials (e.g., Kendall, 1994; Silverman et al., 2019). There was no statistically significant difference in noncompletion rates as a function of study arm, and there were no significant correlates with baseline variables, with one exception. More completers (79.25%) than noncompleters (20.75%) were from families in which the mothers were in intact marriages. Note that there were no significant differences in marital status by treatment arm, and marital status was not significantly associated with any of the outcome measures at baseline or at posttreatment. Thus, there was no evidence that the marital status difference between treatment completers and noncompleters is relevant to our hypotheses. In sum, there was no evidence that random assignment was meaningfully compromised for the per protocol population of interest. This finding aligns with past treatment-noncompletion studies’ failure to demonstrate any consistent variable or set of variables that distinguishes treatment completers from noncompleters with respect to outcome differential (e.g., de Haan et al., 2013; Kendall & Sugarman, 1997; Pina, Silverman, Weems, et al., 2003).

Treatment sessions completed

Participants who began treatment completed an average of 11.25 treatment sessions (SD = 3.76) across treatment arms. There were no significant differences in sessions completed by treatment arm. There was not a statistically significant association between the number of treatment sessions completed and anxiety outcomes at posttreatment (controlling for baseline anxiety) either across arms or within arms.

Nonnormality and outliers

We conducted robust multivariate outlier analyses using the minimum generalized variance (MGV) method and the projection-type method (Wilcox, 2016). The MGV method revealed 18 outliers; the projection-type method revealed no outliers. Analyses were conducted with and without outliers, and results were comparable across analyses. Results from analyses are presented with the outliers included. Univariate indices of skewness and kurtosis were examined to determine whether the absolute value of any index was greater than 2.0. None of the skewness indices exceeded this standard, and only one variable, acceptance, yielded multiple kurtosis values greater than 3. This form of nonnormality is well accommodated by the robust algorithm (MLR) used in Mplus (B. Muthén et al., 2016), which was used in our primary analyses.

We evaluated missing data bias by creating dummy variables to indicate presence or absence of missing data on each variable. We examined associations between dummy variables and other study variables. No evidence of bias was found. We also examined baseline variables to determine whether there were sizable differences as a result of sampling imbalance during randomization. We found none.

Outcome analyses

Initial analyses compared outcome means to baseline means to determine whether anxiety reduction occurred in the treatment arms. The correlation between parents’ ratings of youths’ anxiety and youths’ self-ratings of anxiety was .25 at pretreatment and .30 at posttreatment. These modest correlations between parents’ and youths’ ratings are typical of past research (e.g., Silverman et al., 2009, 2019). We therefore treated each source’s ratings as separate primary outcome measures.

Primary outcomes

We analyzed primary outcomes in two ways. First, we used Huber-White robust likelihood ratio tests in Mplus (L. K. Muthén & Muthén, 2012), analogous to correlated groups t tests, to examine differences between pretreatment, posttreatment, and 12-month follow-up scores for youths’ anxiety, which we collapsed across treatment arms. Second, we used limited information Huber-White robust regression modeling (outside of the larger SEM model described below) to examine whether means at posttreatment and at 12-month follow-up varied by treatment arm. We included youths’ age, sex, ethnicity, and pretreatment scores on the outcome variable as covariates in the models. These analyses are analogous to analysis of covariance, which is preferable to other forms of randomized controlled trial modeling, such as mixed effects linear modeling (see Rausch et al., 2003). Our analytic strategy is preferable to analysis of covariance because it used robust estimation and FIML to accommodate missing data.

Table 1 presents means and standard deviations on anxiety variables by treatment arm. Collapsing across treatment arms, there was statistically significant pretreatment to posttreatment mean change on the RCMAS/P (pretreatment: M = 13.17, posttreatment: M = 8.34, z = 11.74, p < .001, d = 0.92). This was true as well for the RCMAS (pretreatment: M = 11.91, posttreatment: M = 7.15, z = 11.22, p < .001, d = 0.78). There was also a statistically significant mean change on the RCMAS/P from posttreatment to follow-up across treatment arms (posttreatment: M = 8.34, follow-up M = 7.45, z = 2.57, p = .01, d = 0.20), and on the RCMAS from posttreatment to follow-up (posttreatment: M = 7.15, follow-up: M = 5.43, z = 4.44, p < .001, d = 0.34).

Table 1.

Means (and Standard Deviations) for Anxiety Symptoms at Pretreatment, Posttreatment, and 12-Month Follow-Up

Measure	CBT + Reinf (n = 98)	CBT + Relat (n = 60)	CBT (n = 96)
RCMAS/P
Pretreatment	12.66 (5.58)	13.20 (5.28)	13.67 (6.00)
Posttreatment	7.77 (5.17)	8.04 (5.74)	9.14 (5.94)
Follow-up	6.80 (5.09)	5.70 (4.30)	8.99 (6.38)
RCMAS
Pretreatment	11.03 (6.04)	13.27 (6.47)	11.96 (6.44)
Posttreatment	6.17 (5.47)	6.58 (5.96)	8.48 (6.53)
Follow-up	5.27 (5.64)	4.77 (3.92)	5.88 (5.63)

Results of the robust regression analyses did not show differential treatment outcome means on the basis of parent ratings at posttreatment. Youths’ mean ratings at posttreatment showed differential treatment outcome: RCMAS mean scores were significantly lower for youths in the CBT + Relat arm than youths in the CBT arm (mean difference = −2.62, z = −2.92, p = .003, d = 0.37) and significantly lower for youths in the CBT + Reinf arm than youths in the CBT arm (mean difference = −1.80, z = −2.51, p = .012, d = 0.35). At follow-up, RCMAS/P scores were significantly lower for youths in the CBT + Relat arm than youths in the CBT arm (mean difference = −2.61, z = −2.98, p = .003, d = 0.37). There were no significant differences between any treatment arms at follow-up for youths’ anxiety ratings. Below, we consider these results relative to those in the broader SEM modeling.

Clinically significant change

Collapsing across treatment arms, the recovery rate for the primary targeted anxiety diagnosis was 80% at posttreatment and 73.1% at follow-up. These recovery rates are consistent with rates reported in past trials (e.g., Kendall, 1994; Silverman et al., 2009). For the CGAS, 66.1% of youths at posttreatment and 56.1% of youths at follow-up were no longer in the clinical range. Table 2 presents rates of clinically significant improvement by treatment arm. Modified linear probability models were used to examine probabilistic differences between treatment arms on rates of clinically significant improvement (Deke, 2014; Hellevik, 2009). Results showed no significant differences in these rates as a function of treatment arm at posttreatment or at follow-up.

Table 2.

Clinically Significant Change at Posttreatment and 12-Month Follow-Up: Percentage of Youths Not in Clinical Range

Measure	Result of assessment	Posttreatment			12-Month follow-up
Measure	Result of assessment	CBT + Reinf (n = 98)	CBT + Relat (n = 60)	CBT (n = 96)	CBT+ Reinf	CBT + Relat	CBT
ADIS: P/C	Recovered from primary targeted diagnosis	80.4%	85.7%	76.1%	71.4%	84.6%	67.7%
CGAS	Not in clinical range	68.0%	71.4%	60.9%	54.3%	64.1%	53.2%

Note: N = 254. CBT = cognitive behavioral therapy; Reinf = reinforcement skills; Relat = relationship skills; RCMAS = Revised Children’s Manifest Anxiety Scale–Child Version (C. R. Reynolds & Richmond, 1978); RCMAS/P = Revised Children’s Manifest Anxiety Scale–Parent Version; ADIS: P/C = Anxiety Disorders Interview Schedule, parent and child versions (Silverman & Albano, 1996); CGAS = Children’s Global Assessment Scale (Shaffer et al., 1983).

The SEM model

Initial results for the parental positive reinforcement and acceptance mediators yielded nonsignificant results for their relationship to outcomes, and contrary to our hypotheses, it was evident they had no role as active producers of treatment differences. For the sake of analytic and substantive parsimony, we therefore excluded them from the models and do not report on them further. Details surrounding this decision are in the Supplemental Material.

The initial model tested in Figure 2 showed ill fit as a result of a nontrivial modification index between negative reinforcement and psychological control reflecting substantial underprediction of the association between the two variables. The stress point could be addressed either by introducing a causal path between them or correlating their disturbances. We decided on the former by including a path from parent ratings of negative reinforcement to parent psychological control. This path made more theoretical, empirical, and clinical sense than correlating disturbances. In short, parent behaviors that facilitate child avoidance of anxiety-provoking situations (i.e., parental negative reinforcement) may also influence parents’ expressions of psychological control because messages convey to children that they are unable to handle such situations (e.g., Hudson & Rapee, 2004). Thus, the less parents view themselves as negatively reinforcing their child’s anxious avoidant behaviors, the less parents may rate themselves as psychologically controlling of their child. To determine whether the parent findings replicated with the data from the youths, we also included it in the youth model (see Table S1 in Supplemental Material) and at follow-up.

After making the adjustment, we found the two models provided good fit to the data. The global fit statistics for the parent model were χ²(4) = 2.72, p = .61, RMSEA = .00, pClose = .82, CFI = 1.00, SRMR = .005. For the youth model, they were χ²(4) = 5.93, p = .20, RMSEA = .04, pClose = .48, CFI = 1.00, SRMR = .01. The statistically significant estimated path coefficients and their associated margins of error (defined as the absolute half widths of 95% confidence intervals) appear in Table 3 and in Table S1 in the Supplemental Material. Table 3 focuses on the path coefficients from the model that included parent-rated anxiety in youths with parent-rated mediators (i.e., negative reinforcement, psychological control); Table S1 in the Supplemental Material focuses on the path coefficients from the model that included youth-rated anxiety with youth-rated mediators of parent variables (i.e., negative reinforcement, psychological control). The values of all key path coefficients, independent of statistical significance, are presented in the Supplemental Material. Although not predicted, we tested for possible differential estimated effects (path coefficients) of each parent mediator on the outcome as a function of treatment arm using interaction analysis with product terms. None of these effects were statistically significant and are not discussed further.

Table 3.

Statistically Significant Path Coefficients for Key Paths in Figure 2

Name of coefficient	Path coefficient	95% margin of error	p
D2 → NR-POST	−2.492	± 1.725	.005
D3 → NR-POST	−2.433	± 2.088	.022
D1 → ANX-FU	−2.644	± 1.686	.002
NR-POST → ANX-POST	0.092	± 0.067	.008
PC-POST → ANX-POST	0.248	± 0.21	.021
NR-POST → PC-POST	0.058	± 0.051	.026
NR-POST → NR-FU	0.634	± 0.132	< .001
ANX-POST → ANX-FU	0.558	± 0.177	< .001
PC-POST → PC-FU	0.557	± 0.158	< .001
NR-FU → ANX-FU	0.092	± 0.067	.008
PC-FU → ANX-FU	0.248	± 0.21	.021

Note: CBT = cognitive behavioral therapy; D1 = dummy variable for CBT + relationship skills – CBT; D2 = dummy variable for CBT + reinforcement skills – CBT; D3 = dummy variable for CBT + reinforcement skills – CBT + relationship skills; NR = negative reinforcement; PC = psychological control; ANX = anxiety; POST = immediate posttest; FU = 12-month follow-up.

We orient presentation of results using the parent data, after which we highlight results that replicated and differed for the youth data.

Parent data: treatment specificity and mediation

Treatment specificity effects in Figure 2 are reflected by the paths from the treatment arms to the respective mediators (negative reinforcement and psychological control, paths d and e, respectively). Results showed evidence of treatment specificity for negative reinforcement. Parent ratings of use of negative reinforcement at posttreatment were significantly lower in the CBT + Reinf arm than in the CBT arm (mean difference = −2.49, z = −2.83, p = .005, d = 0.36) and the CBT + Relat arm (mean difference = −2.43, z = −2.29, p = .022, d = 0.29; see Table 3). In addition, lower levels of parental use of negative reinforcement at posttreatment were significantly associated contemporaneously with lower levels of parent-rated anxiety in youths (path coefficient from negative reinforcement to anxiety [path l in Figure 2] = 0.09, z = 2.65, p = .008; see Table 3). These joint effects were consistent with mediation vis-à-vis the joint significance test (JST). The estimated effect on anxiety of CBT + Reinf relative to CBT only, considering both paths in the mediational chain, was −0.244. Lower levels of parental psychological control were also significantly associated contemporaneously with lower levels of parent-rated anxiety in youths (path coefficient from psychological control to anxiety = 0.25 [path m in Figure 2], z = 2.31, p = .021; see Table 3). However, there was no support for treatment specificity for psychological control.

As noted, the modification indices coupled with theory suggested a link between the two parent mediators at posttreatment. The path coefficient for this link was consistent with the proposition that the less parents perceived themselves as negatively reinforcing their child’s anxious avoidant behaviors, the less parents rated themselves as psychologically controlling of their child (path coefficient = 0.06, z = 2.22, p = .026; see Table 3). Because the CBT + Reinf treatment yielded lower levels of negative reinforcement than the CBT + Relat and CBT arms, this contributed affirmatively to parental psychological control, ultimately reducing somewhat the advantage of CBT + Relat in directly affecting parental psychological control relative to CBT + Reinf.

Additional mediational dynamics are of theoretical interest (see Fig. 2). Using JST logic, if all links in a given mediational chain are statistically significant, then (partial) mediation is implied vis-à-vis that chain. If any one of the links in the chain is not statistically significant, this represents a broken link in the chain, which disrupts mediation. For example, in Figure 2, the statistically significant path coefficient from the treatment arm variable (CBT + Reinf vs. CBT dummy variable) to negative reinforcement at posttreatment coupled with the statistically significant path from negative reinforcement to anxiety at posttreatment shows (a) that the treatment arm affects posttreatment anxiety and (b) that negative reinforcement partially mediates that association (see Table 3). As another example, the statistically significant coefficients in the causal chain from this dummy treatment arm variable to negative reinforcement, from negative reinforcement to psychological control and then from psychological control to posttreatment anxiety, show a mediational dynamic between these variables. Thus, path tracing through mediational chains underlies conclusions of mediation. Note that for the first example, the JST implies an effect of CBT + Reinf compared with CBT on posttreatment anxiety, but in the limited information analysis reported earlier, this contrast was statistically nonsignificant. The JST has more statistical power because it takes into account more information by introducing negative reinforcement into the analysis (for details, see Kenny, 2018; Shrout & Bolger, 2002). Yzerbyt et al. (2018) argued that effect size analysis of the individual links in the mediational chain is key to establishing the magnitude of the effects (see also Fiedler et al., 2011).

Although the direct paths from the posttreatment mediators to anxiety as measured at follow-up were not statistically significant, the mediators were still associated with follow-up anxiety by virtue of their estimated indirect effects through the autoregressive and contemporaneous effects in the model (see Fig. 2). For example, parental use of negative reinforcement at posttreatment was statistically significantly associated with negative reinforcement at follow-up (path coefficient = 0.63 [path t], z = 9.42, p = .000), which, in turn, was linked to anxiety at follow-up (path coefficient = 0.09 [path p], z = 2.65, p = .008; see Table 3). Likewise, parental psychological control at posttreatment was statistically significantly associated with psychological control at follow-up (path coefficient = 0.56 [path v], z = 6.91, p = .000), which, in turn, was linked to anxiety at follow-up (path coefficient = 0.25 [path q], z = 2.31, p = .021; see Table 3).

Youth data

We emphasize parent data in our presentation of results because we explicitly targeted parent components in the two parent-involvement CBT arms. Thus, parents may show reactivity to their targeted behaviors relative to their children (Wood et al., 2006). We therefore present the youth data in Figure S1 in the Supplemental Material. We comment here on the main findings. Like the parent data, lower levels of youths’ ratings of parental use of negative reinforcement at posttreatment were significantly associated contemporaneously with lower levels of anxiety in youths. However, the treatment difference for youths’ ratings of their parents’ use of negative reinforcement was not statistically significant. This means that the changes in negative reinforcement occurred for both parent-involvement treatments and for individual treatment. Like the parent data, lower levels of youths’ ratings of parental psychological control were also significantly associated contemporaneously with lower levels of youth-rated anxiety. There was no evidence of treatment specificity for either negative reinforcement or psychological control. In addition, although the direct paths from posttreatment negative reinforcement to anxiety as measured at follow-up were not statistically significant, negative reinforcement was still associated with follow-up anxiety by virtue of the estimated indirect effects through the contemporaneous and autoregressive effects. The same was true for parental psychological control.

Discussion

This is the first randomized controlled efficacy trial to use a dismantling design to compare CBT for individual youths and two CBTs with parental involvement, each of which was streamlined and targeted distinct parenting skills: reinforcement skills in CBT + Reinf and relationship skills in CBT + Relat. Our use of a dismantling design moves beyond past trials because most past trials lumped parenting components together into a single intervention and assessed outcome only (e.g., Barrett et al., 1996; Wood et al., 2006); only a few assessed and tested parent variables as mediators (Casline et al., 2018; Settipani et al., 2013; Silverman et al., 2009, 2019). This design allowed us to isolate/investigate those parenting components associated with etiology, maintenance, and reduction of anxiety disorders in youths (e.g., Barber, 1996; Manassis et al., 2014; McLeod et al., 2007; Wood et al., 2006) and to delve more deeply into mechanisms underlying parent-involvement CBTs for anxiety in youths. It is possible that the more concentrated approach that characterized the two parenting arms led to less dilution of key CBT procedures, especially those relevant to youths’ exposure (Creswell et al., 2020; Silverman et al., 2009). This may, in part, explain why both parent-involvement CBTs were superior to CBT at posttreatment on the basis of youths’ ratings. Not unexpectedly, informant discrepancies were found, which are discussed below.

The superior outcome, according to youths’ ratings in both parent-involvement CBTs, may be interpreted from the vantage point that CBT + Reinf is akin to contingency management and that CBT + Relat is akin to transfer of control (Manassis et al., 2014; Silverman & Kurtines, 1996). From this vantage point, exposure is viewed as a key change producing mechanism of reduction of anxiety in youths. Parental use of reinforcement, particularly decreasing negative reinforcement, a primary emphasis of CBT + Reinf training (i.e., not providing reinforcement of youths’ avoidance), is likely a highly potent and direct way to facilitate occurrence of youths’ exposure. This serves to enhance posttreatment outcome relative to CBT (Silverman & Kurtines, 1996). Likewise, parental expressions of autonomy granting couched within a transfer of control framework, a primary emphasis of CBT + Relat training, may also facilitate occurrence of youths’ exposures, including their continued occurrence after treatment ends (Silverman & Kurtines, 1996). This serves to enhance posttreatment and follow-up outcome relative to CBT.

Our findings relating to treatment specificity based on parent ratings are also partially supportive of our hypotheses. Specifically, the parent data showed that when we trained parents in the CBT + Reinf arm to reduce negative reinforcement of their child’s avoidant behavior, a decrease was found in parental use of negative reinforcement at posttreatment, which continued to be associated with negative reinforcement reduction at the 1-year follow-up relative to the CBT + Relat and CBT arms. These findings are the first demonstration of treatment specificity of a parent component in a comparative randomized controlled efficacy trial of individual CBT for youths and CBTs with parental involvement (Jongerden & Bögels, 2014; Kendall et al., 2008; Silverman et al., 2009). From a clinical perspective, this is an important contribution because it suggests that specificity can be produced when CBTs with parent involvement are highly concentrated on only key parenting components and when those components are prescriptive, concrete, and behavioral. Training parents, for example, in use of negative reinforcement in the context of their child’s avoidant behaviors, as in CBT + Reinf, may be more readily implementable and therefore specific relative to, for example, training parents in not using guilt induction, as in CBT + Relat, in which no treatment specificity was found vis-à-vis the parental psychological control mediator.

To be sure, CBT + Relat showed enhanced effects on anxiety relative to CBT for both the child (posttreatment) and parent (at follow-up) reports of anxiety (see the Supplemental Material). There is therefore something actively at work in CBT + Relat, but it appears too amorphous to detect treatment specificity of parental psychological control as well as its possible mediational role.

Our findings that parental use of negative reinforcement changed more when directly targeted in CBT + Reinf relative to CBT and CBT + Relat and that this difference was associated with lower levels anxiety in youths provide some insight into a mechanism by which CBT + Reinf reduces anxiety in youths. The finding provides partial support for our hypothesis that changes produced on targeted parenting components partially mediate treatment outcome—partial in that of the four parent components targeted, negative reinforcement was the one found to be consistent with mediation vis-à-vis the JST for parent ratings of child anxiety. The statistically significant effect of negative reinforcement on psychological control at posttreatment suggests another mechanism by which CBT + Reinf reduces anxiety in youths, which was not a facet of our hypotheses. That is, reducing parental use of negative reinforcement in CBT + Reinf may indirectly lead to decreases in anxiety in youths (as reported by the parent) at posttreatment via lower parental use of psychological control. This suggests that parents who did not allow their children to avoid feared situations were less likely to use shaming and guilt induction to control their children’s behaviors, which led to decreases in anxiety in youths. These findings coincide with findings from our other recent trial that highlight the intriguing possibility that causal relationships may exist among the mediators themselves (and once again, with the parent psychological control variable), thus complicating the theoretical tracing of relative treatment effects through mediators (Silverman et al., 2019). In the current study, these mediational dynamics continued through the follow-up period, as evidenced by statistically significant autoregressive paths from posttreatment to follow-up for both parent components (i.e., negative reinforcement and psychological control) and statistically significant relationships between both parent components and anxiety in youths at follow-up. However, given the many links in the mediational chain, it is not surprising that the strength of these effects is not strong.

Overall, our findings provide insight into two mechanisms of anxiety-reduction effects and, as noted, the complexity of mediational dynamics. Specifically, the data suggest that decreasing parental use of negative reinforcement reduced anxiety in youths, and the data were consistent with a model that this decrease in negative reinforcement reduces parental use of psychological control, which also likely reduces anxiety in youths. As noted, the statistical significance of paths and outcomes differed across informants. Such differential findings highlight the complexity of mechanistic research involving parents and children, a point we return to below. Nevertheless, these data provide evidence for directly targeting parental negative reinforcement in CBT for anxiety disorders in youths and suggest that such targeting partially explains how CBT with parental involvement reduces anxiety in youths. Identifying mechanisms of treatment outcome is the primary objective of dismantling designs (e.g., Nock et al., 2007). Thus, our trial was successful in moving the field closer to realizing the discovery of parenting-based mechanisms. Given the absence of past data on parent/family mechanisms/mediators and the frequent calls for such data (e.g., Creswell et al., 2020; Herres et al., 2015; Manassis et al., 2014; Peris et al., 2021; Sun et al., 2018; Thulin et al., 2014), our findings acquire even further significance. Altogether, the findings pave the way for additional mechanistic research, such as how parental reinforcement might be used to optimize outcomes further (e.g., dose effects, personalized treatment).

Our respective targeting of parental use of positive reinforcement in CBT + Reinf and of parental acceptance in CBT + Relat adds to the growing literature that negative reinforcement and psychological control are the components of parenting tied most consistently with anxiety in youths (e.g., Casline et al., 2018; McLeod et al., 2007; Wood et al., 2006). Furthermore, the findings suggest that continued targeting and researching of negative reinforcement specifically ought to be prioritized in future treatment studies and clinical application.

As found in past research on anxiety in youths (Settipani et al., 2013; Silverman et al., 2009, 2019), treatment outcome, specificity, and mediation dynamics varied by informants. Our findings appear to represent yet another common manifestation of informant discrepancy, which are manifested widely in clinical trials as well (De Los Reyes & Kazdin, 2008). In the context of CBT with parental involvement, it seems plausible that when concrete parent components such as using negative reinforcement are added to CBT, parents are actively engaged in applying these skills and therefore notice their changes on these behaviors. It is not surprising, however, that their children may not notice, especially when the children are focused themselves on learning and using CBT. For these reasons, we emphasize the parent data in our conclusions on parent behaviors. The findings further point to the need to improve our measurement of parental utility of the skills in which they are being trained to apply, which is now possible with ecological momentary assessment and mobile sensing methodologies. Overall, the findings likely represent meaningful and key differences in how parents and their children perceive their environments, especially about personally salient behaviors such as parenting (Silverman et al., 2019). Future research to advance understanding of different perceptions of therapeutic change in children and parents is needed.

Treatment outcome findings varied by measure such that significant differential outcome emerged on rating scales, not diagnostic recovery. Because rating scales are scored dimensionally, they are more sensitive to change and avoid many of the pitfalls of using dichotomized continuous constructs such as diagnostic recovery (MacCallum et al., 2002). Dichotomous outcomes also are much more sample size demanding relative to continuous outcomes in terms of statistical power (e.g., samples sizes of about 65 per group are needed to achieve power of .80 for medium effect sizes for continuous outcomes; sample sizes of about 400 per group are needed to detect percentage differences of 10%). It therefore is not necessarily surprising that although a consistent pattern of superior outcome in both parent-involvement CBTs was found using rating scales, statistical significance of superior outcomes was not attained using dichotomous outcomes (i.e., diagnostic recovery rates).

There are several study limitations. Some of the analyses were correlational in nature, and assumptions were necessitated about them in terms of the timing of dynamics between cause and effect and sequential ignorability. As in all studies that use complicated statistical modeling, equivalent/plausible model alternatives warrant consideration, including reciprocal causality or reverse causal dynamics. The use of more intensive and frequent measurement (Kraemer et al., 2002) in future research would not only provide higher resolution insight into timing issues but also allow for improved investigation of reciprocal causal mechanisms and alternative modeling strategies for autoregressive and lagged dynamics (Falkenström et al., 2020).

A second limitation is reliance on self-report data, although this was a reasonable place to begin in this underdeveloped line of inquiry and given the strengths/efficiency of self-report and that the informant data suggest intriguing possibilities (e.g., Etkin et al., 2021). Third, although it is a study strength that our sample contained a relatively high proportion of Hispanic/Latino participants and that previous randomized CBT efficacy trials focused on anxiety in youths with and without parent involvement showed that Hispanic and non-Hispanic samples responded similarly (Pina, Silverman, Fuentes, et al., 2003; Vaclavik et al., 2017), future research designed from the onset to understand the role of ethnicity-related variables as well as other variables (e.g., acculturation, treatment compliance) that might affect treatment use and outcomes is needed. Fourth, given that recent research supports the viability of a novel parent-based treatment that has no involvement of youths and targets the lowering of parental accommodation (Lebowitz et al., 2019), it is important to study which families might benefit from specific formats, including their combination (e.g., Raila et al., 2020) and the underlying mechanisms. Fifth, using a modification index, we included a path from parent ratings of negative reinforcement to parent ratings of psychological control. Our including this path made theoretical, empirical, and clinical sense compared with correlating the disturbances between these variables for the reasons noted earlier. Moreover, the latter would require specifying the substantive sources of the correlated disturbances, which is not straightforward because these sources would have to operate over and above the covariates for which we already controlled (e.g., relevant baseline measures). Nevertheless, the result is possibly open to different interpretations including correlated disturbances.

Finally, because our study was designed to address efficacy-based issues surrounding theoretical mechanisms of change, we cannot assume that the findings will generalize to effectiveness trials conducted in community settings. Such work requires an emphasis on understanding why participants do not complete treatment in community settings as well as issues of treatment fidelity. Further research on both efficacy and effectiveness processes (i.e., how is change produced; how can treatment adherence and completion be enhanced) is needed (Silverman et al., 2019).

Despite these limitations, this is only the second study that we are aware of to use a dismantling treatment design in anxiety treatment efficacy research for youths to focus on how parental involvement might make a difference. The first was conducted more than two decades ago, and the focus was not on whether or how parental involvement might make a difference (Silverman et al., 1999). In summary, our study addresses important efficacy issues that have received scant attention in research on anxiety treatment in youths. These include whether adding a parent component matters, treatment specificity and mediation, and intriguing causal relationships between parent components targeted in treatments. Taken altogether, our study illustrates the unique advantages of dismantling treatment designs to isolate/investigate putative mechanisms and provides important new insights for future research directions in clinical psychological science.

Supplemental Material

sj-pdf-1-cpx-10.1177_21677026211016402 – Supplemental material for Does Training Parents in Reinforcement Skills or Relationship Skills Enhance Individual Youths’ Cognitive Behavioral Therapy for Anxiety? Outcome, Specificity, and Mediation

Supplemental material, sj-pdf-1-cpx-10.1177_21677026211016402 for Does Training Parents in Reinforcement Skills or Relationship Skills Enhance Individual Youths’ Cognitive Behavioral Therapy for Anxiety? Outcome, Specificity, and Mediation by Wendy K. Silverman, Yasmin Rey, Carla E. Marin, James Jaccard and Jeremy W. Pettit in Clinical Psychological Science

Footnotes

Transparency

Action Editor: Christopher G. Beevers

Editor: Kenneth J. Sher

Author Contributions

W. K. Silverman developed the study concept. W. K. Silverman and J. Jaccard contributed to the study design. Testing and data collection were performed by Y. Rey and C. E. Marin. Y. Rey and C. E. Marin performed the data analysis and interpretation under the supervision of J. Jaccard. W. K. Silverman drafted the manuscript, and C. E. Marin, Y. Rey, J. Jaccard, and J. W. Pettit provided critical revisions. All of the authors approved the final manuscript for submission.

ORCID iD

Wendy K. Silverman

Supplemental Material

Additional supporting information can be found at

References

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.).

Barber

B. K.

(1996). Parental psychological control: Revisiting a neglected construct. Child Development, 67(6), 3296–3319. https://doi.org/10.2307/1131780

Barmish

A. J.

Kendall

P. C.

(2005). Should parents be co-clients in cognitive-behavioral therapy for anxious youth? Journal of Clinical Child and Adolescent Psychology, 34(3), 569–581. https://doi.org/10.1207/s15374424jccp3403_12

Barrett

P. M.

Dadds

M. R.

Rapee

R. M.

(1996). Family treatment of childhood anxiety: A controlled trial. Journal of Consulting and Clinical Psychology, 64(2), 333–342. https://doi.org/10.1037/0022-006X.64.2.333

Bird

H. R.

Shaffer

Fisher

Gould

M. S.

Staghezza

Chen

J. Y.

Hoven

(1993). The Columbia-Impairment-Scale (CIS): Pilot findings on a measure of global impairment for children and adolescents. International Journal of Methods in Psychiatric Research, 3(3), 167–176.

Breinholst

Esbjorn

B. H.

Reinholdt-Dunne

M. L.

Stallard

(2012). CBT for the treatment of child anxiety disorders: A review of why parental involvement has not enhanced outcomes. Journal of Anxiety Disorders, 26(3), 416–424. https://doi.org/10.1016/j.janxdis.2011.12.014

Casline

E. P.

Pella

Zheng

Harel

Drake

K. L.

Ginsburg

G. S.

(2018). Parental responses to children’s avoidance in fear-provoking situations: Relation to child anxiety and mediators of intervention response. Child and Youth Care Forum, 47(4), 443–462. https://doi.org/10.1007/s10566-018-9440-7

Creswell

Waite

Hudson

(2020). Practitioner review: Anxiety disorders in children and young people – Assessment and treatment. Journal of Child and Adolescent Psychiatry, 61(6), 628–643. https://doi.org/10.1111/jcpp.13186

Dallal

(2012). Intention to treat analysis. http://www.jerrydallal.com/lhsp/itt.htm

10.

Daniel

R. M.

De Stavola

B. L.

Cousens

S. N.

Vansteelandt

(2015). Causal mediation analysis with multiple mediators. Biometrics, 71(1), 1–14. https://doi.org/10.1111/biom.12248

11.

de Haan

A. M.

Boon

A. E.

de Jong

Hoeve

Vermeiren

(2013). A meta-analytic review on treatment dropout in child and adolescent outpatient mental health care. Clinical Psychology Review, 33(5), 698–711. https://doi.org/10.1016/j.cpr.2013.04.005

12.

Deke

(2014). Using the linear probability model to estimate impacts on binary outcomes in randomized controlled trials (Evaluation Technical Assistance Brief 6). U.S. Department of Health and Human Services, Administration on Children, Youth and Families, Office of Adolescent Health. https://opa.hhs.gov/sites/default/files/2020-07/lpm-tabrief.pdf

13.

De Los Reyes

Kazdin

A. E.

(2008). When the evidence says “yes, no, and maybe so”: Attending to and interpreting inconsistent findings among evidence-based interventions. Current Directions in Psychological Science, 7(1), 47–51. https://doi.org/10.1111/j.1467-8721.2008.00546.x

14.

Dyrborg

Larsen

F. W.

Nielsen

Byman

Nielsen

B. B.

Gautre-Delay

(2000). The Children’s Global Assessment Scale (CGAS) and Global Assessment of Psychosocial Disability (GAPD) in clinical practice: Substance and reliability as judged by intraclass correlations. European Child and Adolescent Psychiatry, 9(3), 195–201. https://doi.org/10.1007/s007870070043

15.

Enders

C. K.

(2010). Applied missing data analysis. The Guilford Press.

16.

Etkin

R. G.

Shimshoni

Lebowitz

E. R.

Silverman

W. K.

(2021). Using evaluative criteria to review youth anxiety measures, Part I: Self-report. Journal of Clinical Child & Adolescent Psychology, 50(1), 58–76. https://doi.org/10.1080/15374416.2020.1802736

17.

Falkenström

Solomonov

Rubel

(2020). Using time-lagged panel data analysis to study mechanisms of change in psychotherapy research: Methodological recommendations. Counseling and Psychotherapy Research, 20(3), 435–441. https://doi.org/10.1002/capr.12293

18.

Feinman

R. D.

(2009). Intention-to-treat. What is the question? Nutrition and Metabolism, 6(1). https://doi.org/10.1186/1743-7075-6-1

19.

Fiedler

Schott

Meiser

(2011). What mediation analysis can (not) do. Journal of Experimental Social Psychology, 47(6), 1231–1236. https://doi.org/10.1016/j.jesp.2011.05.007

20.

Ginsburg

G. S.

Silverman

W. K.

Kurtines

W. K.

(1995). Family involvement in treating children with phobic and anxiety disorders: A look ahead. Clinical Psychology Review, 15(5), 457–473. https://doi.org/10.1016/0272-7358(95)00026

21.

Gross

Fogg

(2004). A critical analysis of the intent-to-treat principle in prevention research. The Journal of Primary Prevention, 25(4), 475–489. https://doi.org/10.1023/B:JOPP.0000048113.77939.44

22.

Hellevik

(2009). Linear versus logistic regression when the dependent variable is a dichotomy. Quality and Quantity, 43(1), 59–74. https://doi.org/10.1007/s11135-007-9077-3

23.

Herres

Cummings

C. M.

Swan

Makover

Kendall

P. C.

(2015). Moderators and mediators of treatments for youth with anxiety. In Maric

Prins

P. J. M.

Ollendick

T. H.

(Eds.), Moderators and mediators of youth treatment outcomes (pp. 20–40). Oxford University Press.

24.

Hudson

J. L.

Rapee

R. M.

(2004). From anxious temperament to disorder: An etiological model. In Heimberg

R. G.

Turk

C. L.

Mennin

D. S.

(Eds.), Generalized anxiety disorder. Advances in research and practice (pp. 51–74). The Guilford Press.

25.

Imai

Keele

Yamamoto

(2010). Identification, inference and sensitivity analysis for causal mediation effects. Statistical Science, 25(1), 51–71. https://www-jstor-org-443.web.bisu.edu.cn/stable/41058997

26.

In-Albon

Schneider

(2007). Psychotherapy of childhood anxiety disorders: A meta-analysis. Psychotherapy and Psychosomatics, 76(1), 15–24. https://doi.org/10.1159/000096361

27.

Jongerden

Bögels

S. M.

(2014). Parenting, family functioning and anxiety-disordered children: Comparisons to controls, changes after family versus child CBT. Journal of Child and Family Studies, 24(7), 2046–2059. https://doi.org/10.1007/s10826-014-0005-6

28.

Kendall

P. C.

(1994). Treating anxiety disorders in children: Results of a randomized clinical trial. Journal of Consulting and Clinical Psychology, 62(1), 100–110. https://doi.org/10.1037/0022-006X.62.1.100

29.

Kendall

P. C.

Hudson

J. L.

Gosch

Flannery-Schroeder

Suveg

(2008). Cognitive-behavior therapy for anxiety disordered youth: A randomized clinical trial evaluating child and family modalities. Journal of Consulting and Clinical Psychology, 76(2), 282–297. https://doi.org/10.1037/0022-006x.76.2.282

30.

Kendall

P. C.

Sugarman

(1997). Attrition in the treatment of childhood anxiety disorders. Journal of Consulting and Clinical Psychology, 65(5), 883–888. https://doi.org/10.1037/0022-006X.65.5.883

31.

Kenny

(2018). Mediation. http://davidakenny.net/cm/mediate.htm#DI

32.

Kraemer

H. C.

Wilson

G. T.

Fairburn

C. G.

Agras

W. S.

(2002). Mediators and moderators of treatment effects in randomized clinical trials. Archives of General Psychiatry, 59(10), 877–883. https://doi.org/10.1001/archpsyc.59.10.877

33.

La Greca

A. M.

Silverman

W. K.

Lochman

J. E.

(2009). Moving beyond efficacy and effectiveness in child and adolescent intervention research. Journal of Consulting and Clinical Psychology, 77(3), 373–382. https://doi.org/10.1037/a0015954

34.

Lebowitz

E. R.

Marin

C. E.

Martino

Shimshoni

Silverman

W. K.

(2019). Parent-based treatment as efficacious as cognitive-behavioral therapy for childhood anxiety: A randomized noninferiority study of supportive parenting for anxious childhood emotions. Journal of the American Academy of Child and Adolescent Psychiatry, 59(3), 362–372. https://doi.org/10.1016/j.jaac.2019.014

35.

Lyneham

H. J.

Abbott

M. J.

Rapee

R. M.

(2007). Interrater reliability of the Anxiety Disorders Interview Schedule for DSM-IV: Child and parent version. Journal of the American Academy of Child and Adolescent Psychiatry, 46(6), 731–736. https://doi.org/10.1097/chi.0b013e3180465a09

36.

MacCallum

Zhang

Preacher

Rucker

(2002). On the practice of dichotomization of quantitative variables. Psychological Methods, 7(1), 19–40. https://doi.org/10.1037/1082-989X.7.1.19

37.

Mallinckrodt

Abraham

W. T.

Wei

Russell

D. W.

(2006). Advances in testing the statistical significance of mediation effects. Journal of Counseling Psychology, 53(3), 372–378. https://doi.org/10.1037/0022-0167.53.3.372

38.

Manassis

Lee

T. C.

Bennett

Zhao

X. Y.

Mendlowitz

Duda

Saini

Wilansky

Baer

Barrett

Bodden

Cobham

V. E.

Dadds

M. R.

Flannery-Schroeder

Ginsburg

Heyne

Hudson

J. L.

Kendall

P. C.

Liber

. . . Wood

J. J.

(2014). Types of parental involvement in CBT with anxious youth: A preliminary meta-analysis. Journal of Consulting and Clinical Psychology, 82(6), 1163–1172. https://doi.org/10.1037/a0036969

39.

McDonald

R. P.

(1999). Test theory: A unified treatment. Psychology Press. https://doi.org/10.4324/9781410601087

40.

McLeod

B. D.

Wood

J. J.

Weisz

J. R.

(2007). Examining the association between parenting and childhood anxiety: A meta-analysis. Clinical Psychology Review, 27(2), 155–172. https://doi.org/10.1016/j.cpr.2006.09.002

41.

Muthén

Asparouhov

(2016). Regression and mediation analysis using Mplus. Muthén and Muthén.

42.

Muthén

L. K.

Muthén

(2012). Mplus user’s guide (7th ed.).

43.

Nauta

M. H.

Scholing

Emmelkamp

P. M. G.

Minderaa

R. B.

(2003). Cognitive behavioral therapy for children with anxiety disorders in a clinical setting: No additional effect of a cognitive parent training. Journal of the American Academy of Child and Adolescent Psychiatry, 42(11), 1270–1278. https://doi.org/10.1097/01.chi.0000085752.71002.93

44.

Nock

M. K.

Janis

I. B.

Wedig

M. M.

(2007). Research designs. In Nezu

A. M.

Nezu

C. M.

(Eds.), Evidence-based outcome research: A practical guiding to conducting randomized controlled trials for psychosocial interventions (pp. 201–218). Oxford University Press.

45.

Pearl

(2009). Causality: Models, reasoning, and inference. Cambridge University Press.

46.

Peris

T. S.

Thamrin

Rozenman

M. S.

(2021). Family interventions for child and adolescent anxiety: A meta-analysis review of therapy targets, techniques, and outcomes. Journal of Affective Disorders, 286(1), 282–295. https://doi.org/10.1016/jad2021.02.053

47.

Pina

A. A.

Silverman

W. K.

Fuentes

R. M.

Kurtines

W. M.

Weems

C. F.

(2003). Exposure-based cognitive-behavior treatment for phobic and anxiety disorders: Treatment effects and maintenance for Hispanic/Latino relative to European-American youths. Journal of the American Academy of Child and Adolescent Psychiatry, 42(10), 1179–1187. https://doi.org/10.1097/01.chi.0000081804.49840.02

48.

Pina

A. A.

Silverman

W. K.

Weems

C. F.

Kurtines

W. M.

Goldman

M. L.

(2003). A comparison of completers and noncompleters of exposure-based cognitive and behavioral treatment for phobic and anxiety disorders in youth. Journal of Consulting and Clinical Psychology, 71(4), 701–705. https://doi.org/10.1037/0022-006X.71.4.701

49.

Raila

H. E.

Julian

Lebowitz

E. R.

Silverman

W. K.

(2020). Separating from the mothership: A coordinated individual and parent-based approach to severe agoraphobia in a young adult. Cognitive and Behavioral Practice. Advance online publication. https://doi.org/10.1016/j.cbpra.2020.06.004

50.

Rausch

J. R.

Maxwell

S. E.

Kelley

(2003). Analytic methods for questions pertaining to a randomized pretest, posttest, follow-up design. Journal of Clinical Child and Adolescent Psychology, 32(3), 467–486. https://doi.org/10.1207/S15374424JCCP3203_15

51.

Reynolds

C. R.

Richmond

B. O.

(1978). What I think and feel: A revised measure of children’s manifest anxiety. Journal of Abnormal Child Psychology, 6(2), 271–280. https://doi.org/10.1007/BF00919131

52.

Reynolds

Wilson

Austin

Hooper

(2012). Effects of psychotherapy for anxiety in children and adolescents: A meta-analytic review. Clinical Psychology Review, 32(4), 251–262. https://doi.org/10.1016/j.cpr.2012.01.005

53.

Schludermann

E. H.

Schludermann

(1970). Replicability of factors in children’s report of parent behavior (CRPBI). The Journal of Psychology, 76(2), 239–249. https://doi.org/10.1080/00223980.1970.9916845

54.

Schludermann

E. H.

Schludermann

S. M.

(1988). Children’s report on parent behavior (CRPBI-108, CRPBI-30 for children and older adolescents, technical report). University of Manitoba.

55.

Settipani

C. A.

O’Neil

K. A.

Podell

J. L.

Beidas

R. S.

Kendall

P. C.

(2013). Youth anxiety and parent factors over time: Directionality of change among youth treated for anxiety. Journal of Clinical Child and Adolescent Psychology, 42(1), 9–21. https://doi.org/10.1080/15374416.2012.719459

56.

Shaffer

Gould

M. S.

Brasic

Ambrosini

Fisher

Bird

Aluwahlia

(1983). A children’s global assessment scale (CGAS). Archives of General Psychiatry, 40(11), 1228–1231. https://doi.org/10.1001/archpsyc.1983.01790100074010

57.

Shrout

P. E.

Bolger

(2002). Mediation in experimental and nonexperimental studies: New procedures and recommendations. Psychological Methods, 7(4), 422–445. https://doi.org/10.1037/1082-989X.7.4.422

58.

Silverman

W. K.

Albano

A. M.

(1996). Anxiety Disorders Interview Schedule for children-IV (Child and parent versions). Oxford University Press.

59.

Silverman

W. K.

Kurtines

W. M.

(1996). Anxiety and phobic disorders: A pragmatic approach. Springer Science & Business Media.

60.

Silverman

W. K.

Kurtines

W. M.

Ginsburg

G. S.

Weems

C. F.

Rabian

Serafini

L. T.

(1999). Contingency management, self-control, and education support in the treatment of childhood phobic disorders: A randomized clinical trial. Journal of Consulting and Clinical Psychology, 67(5), 675–687. https://doi.org/10.1037/0022-006X.67.5.675

61.

Silverman

W. K.

Kurtines

W. M.

Jaccard

Pina

A. A.

(2009). Directionality of change in youth anxiety treatment involving parents: An initial examination. Journal of Consulting and Clinical Psychology, 77(3), 474–485. https://doi.org/10.1037/a0015761

62.

Silverman

W. K.

Marin

C. E.

Rey

Kurtines

W. M.

Jaccard

Pettit

J. W.

(2019). Group- versus parent involvement CBT for childhood anxiety disorders: Treatment specificity and long-term recovery mediation. Clinical Psychological Science, 7(4), 840–855. https://doi.org/10.1177/2167702619830404

63.

Silverman

W. K.

Ollendick

T. H.

(2005). Evidence-based assessment of anxiety and its disorders in children and adolescents. Journal of Clinical Child and Adolescent Psychology, 34(3), 380–411. https://doi.org/10.1207/s15374424jccp3403_2

64.

Silverman

W. K.

Pina

A. A.

Viswesvaran

(2008). Evidence-based psychosocial treatments for phobic and anxiety disorders in children and adolescents. Journal of Clinical Child and Adolescent Psychology, 37(1), 105–130. https://doi.org/10.1080/15374410701817907

65.

Silverman

W. K.

Saavedra

L. M.

Pina

A. A.

(2001). Test-retest reliability of anxiety symptoms and diagnoses with the Anxiety Disorders Interview Schedule for DSM-IV: Child and parent versions. Journal of the American Academy of Child and Adolescent Psychiatry, 40(8), 937–944. https://doi.org/10.1097/00004583-200108000-00016

66.

Spielmans

G. I.

Pasek

L. F.

McFall

J. P.

(2007). What are the active ingredients in cognitive and behavioral psychotherapy for anxious and depressed children? A meta-analytic review. Clinical Psychology Review, 27(5), 642–654. https://doi.org/10.1016/j.cpr.2006.06.001

67.

Strube

M. J.

(1991). Small sample failure of random assignment: A further examination. Journal of Consulting and Clinical Psychology, 59(2), 346–350. https://doi.org/10.1037/0022-006X.59.2.346

68.

Strube

M. J.

(2015). Small sample failure of random assignment. In Cautin

R. L.

Lilienfeld

S. O.

(Eds.), The encyclopedia of clinical psychology. John Wiley & Sons. https://doi.org/10.1002/9781118625392.wbecp323

69.

Sun

Rith-Najarian

L. R.

Williamson

T. J.

Chorpita

B. F.

(2018). Treatment features associated with youth cognitive behavioral therapy follow-up effects for internalizing disorders: A meta-analysis. Journal of Clinical Child and Adolescent Psychology, 48(S1), 1537–4424. https://doi.org/10.1080/15374416.2018.1443459

70.

Ten Have

T. R.

Joffe

M. M.

(2012). A review of causal estimation of effects in mediation analyses. Statistical Methods in Medical Research, 21(1), 77–107. https://doi.org/10.1177/0962280210391076

71.

Thulin

Svirsky

Serlachius

Andersson

Ost

L. G.

(2014). The effect of parent involvement in the treatment of anxiety disorders in children: A meta-analysis. Cognitive Behavior Therapy, 43(3), 185–200. https://doi.org/10.1080/16506073.2014.923928

72.

Vaclavik

Buitron

Rey

Marin

C. E.

Silverman

W. K.

Pettit

J. W.

(2017). Parental acculturation level moderates outcome in peer-involved and parent-involved CBT for anxiety disorders in Latino youth. Journal of Latina/o Psychology, 5(4), 261–274. https://doi.org/10.1037/lat0000095

73.

van der Bruggen

C. O.

Stams

G. J. J. M.

Bögels

S. M.

(2008). Research review: The relation between child and parent anxiety and parental control: A meta-analytic review. Journal of Child Psychology and Psychiatry, 49(12), 1257–1269. https://doi.org/10.1111/j.1469-7610.2008.01898.x

74.

van der Sluis

C. M.

van Steensel

F. J.

Bögels

S. M.

(2015). Parenting and children’s internalizing symptoms: How important are parents? Journal of Child and Family Studies, 24(12), 3652–3661. https://doi.org/10.1007/s10826-015-0174-y

75.

Weersing

V. R.

Weisz

J. R.

(2002). Mechanisms of action in youth psychotherapy. Journal of Child Psychology and Psychiatry, 43(1), 3–29. https://doi.org/10.1111/1469-7610.00002

76.

Wilcox

R. R.

(2016). Comparing dependent robust correlations. British Journal of Mathematical and Statistical Psychology, 69(3), 215–224. https://doi.org/10.1111/bmsp.12069

77.

Wood

J. J.

Piacentini

J. C.

Bergman

R. L.

McCracken

Barrios

(2002). Concurrent validity of the anxiety disorders section of the Anxiety Disorders Interview Schedule for DSM-IV: Child and parent versions. Journal of Clinical Child and Adolescent Psychology, 31(3), 335–342. https://doi.org/10.1207/S15374424JCCP3103

78.

Wood

J. J.

Piacentini

J. C.

Southam-Gerow

Chu

B. C.

Sigman

(2006). Family cognitive behavioral therapy for child anxiety disorders. Journal of the American Academy of Child and Adolescent Psychiatry, 45(3), 314–321. https://doi.org/10.1097/01.chi.0000196425.88341.b0

79.

Yzerbyt

Muller

Batailler

Judd

C. M.

(2018). New recommendations for testing indirect effects in mediational models: The need to report and test component paths. Journal of Personality and Social Psychology, 115(6), 929–943. https://doi.org/10.1037/pspa0000132

80.

Zabin

M. A.

Melamed

B. G.

(1980). Relationship between parental discipline and children’s ability to cope with stress. Journal of Behavioral Assessment, 2(1), 17–38. https://doi.org/10.1007/BF01321430

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.25 MB