Available online at www.sciencedirect.com

ScienceDirect Behavior Therapy 46 (2015) 365 – 378

www.elsevier.com/locate/bt

Parenting Skills and Parent Readiness for Treatment Are Associated With Child Disruptive Behavior and Parent Participation in Treatment Brendan F. Andrade Centre for Addiction and Mental Health University of Toronto Dillon T. Browne Ontario Institute for Studies in Education, University of Toronto Alex R. Naber Centre for Addiction and Mental Health

Parent management training programs for parents of children with disruptive behaviors are efficacious treatments; however, in order to maximize efficiency it is necessary to develop approaches to understand which parents are most likely to participate in treatment. Accordingly, the present study used a person-centered methodology to determine clinically relevant parenting profiles that capture the breadth of parents’ readiness to engage in parenting treatment, and their selfreported parenting skills. Further, identified profiles were compared on the severity of children’s behavior problems and used to predict participation in parent management training. One hundred and forty-three parents completed assessments at an urban children’s mental health clinic. Parents were given measures to assess personal readiness to participate in parenting treatment, parenting skills, and child behavior. A subset of these parents participated in parent management training. We would like to sincerely thank the children, parents, and clinicians who participated in this study. Without their participation the work would not be possible. This research was funded in part by the Ontario Mental Health Foundation, Canadian Child Health Clinician Scientist Program, and the University of Toronto Faculty of Medicine Dean’s Fund New Staff Grant awarded to Brendan F. Andrade. Dillon T. Browne was supported by a Canadian Vanier Graduate Scholarship from the Canadian Institutes of Health Research. Address correspondence to Brendan F. Andrade, Ph.D., C.Psych., Centre for Addiction and Mental Health, 80 Workman Way, Toronto, Canada M6J 1H4; e-mail: [email protected]. 0005-7894/© 2015 Association for Behavioral and Cognitive Therapies. Published by Elsevier Ltd. All rights reserved.

Three profiles emerged that differed in parents’ treatment readiness and level of skills. Forty-one percent of parents were classified as “ready.” They showed relatively higher rates of inconsistent discipline, but also somewhat higher levels of positive parenting. Thirty-nine percent of parents were classified as “less in need.” These parents reported relatively less inconsistent discipline and poor supervision skills and greater positive parenting. Finally, approximately 20% of parents were classified as “almost ready.” They showed high levels of inconsistent discipline and poor supervision skills, and low levels of positive parenting. Almost ready and ready parents reported the most problems with their children’s behavior. Further, parents classified as less in need participated in the fewest treatment sessions. Consideration of parent readiness and skills, in addition to symptom severity, may inform clinical decision making and screening procedures.

Keywords: child disruptive behavior disorders; parent management training; readiness for treatment

PARENT MANAGEMENT TRAINING PROGRAMS for parents of children with disruptive behavior are widely used and efficacious interventions (Lochman & Wells, 2004; Markie-Dadds & Sanders, 2006; WebsterStratton & Reid, 2003). Rooted in behavioral and social learning theories, these interventions help parents develop specific skills in order to better

366

andrade et al.

manage their child’s challenging behavior and build behavioral competencies (Kazdin, Siegel, & Bass, 1992). Although widely used, these interventions are not satisfactory for many parents (Mendez, Carpenter, LaForett, & Cohen, 2009; Miller & Prinz, 2003). A large proportion of parents choose not to participate in these treatments, and those who do show concerning rates of dropout (Frankel & Simmons, 1992; Kazdin & Wassell, 2000). Although effective for many, it is imperative that strategies be developed to better understand which parents are most and least likely to participate in parenting treatment, in order to develop intervention approaches that best engage these parents. To date, a limited number of studies have examined variables that predict parental participation in group treatment (Mendez et al., 2009; Nock & Ferriter, 2005). The majority of these have focused on barriers to access that include logistical factors such as transportation, lack of available time, responsibility for other children, and other important factors (Cunningham et al., 2000; Kazdin, Holland, & Crowley, 1997). Others have focused on parental variables, such as parental cognitions and mental illness that predict engagement and participation in group or parent–child mental health treatment (Morrissey-Kane & Prinz, 1999). Reducing barriers has been shown to improve parent attendance in interventions (Nock & Kazdin, 2005; Shepard, Armstrong, Silver, Berger, & Seifer, 2012). From a clinical service perspective, the need for more efficient and effective approaches for decision making is relatively clear. Children’s mental health clinics are increasingly challenged by greater demands for services with diminishing resources (Lin, Goering, Offord, Campbell, & Boyle, 1996). Efficient and effective methods for screening, assessment, and treatment are necessary to maximize client benefits while minimizing costs. Groupbased parent interventions provide a structured and cost-effective approach to treatment (Van De Wiel, Mathys, Cohen-Kettenis, & Van Engeland, 2003; Webster-Stratton, Hollinsworth, & Kolpacoff, 1989). Although a wide array of evidence-based parent management interventions exist, most can be distilled to a set of core behavioral and skillsdevelopment components (Chorpita, Becker, & Daleiden, 2007; Kazdin, 2011). These include positive parenting approaches (e.g., play-based, praise), strategies to build consistent parenting (i.e., natural and logical consequences, communication strategies, reinforcement schedules), and strategies for authoritative parenting (i.e., time-out, removal of privileges, response–cost programs). Other programs also target adjunctive parental cognitions that influence behaviors (Lochman & Wells, 2004;

Markie-Dadds & Sanders, 2006). Although somewhat flexible in their application, parent interventions include a set of skills-building modules in a prescribed order. Parents who present with their children for treatment show a range in experiences and skills. Given this, it is quite unclear whether this pretreatment heterogeneity is well suited to a structured approach to treatment (Nock & Ferriter, 2005; Shepard et al., 2012). Parents recommended for group-based treatments may vary considerably in their understanding of effective parenting strategies and in their ability to implement these approaches. As such, parents may feel more or less suited to an intervention that requires a multiweek commitment and skillsbased discussion. Moreover, parents may not share the clinician’s belief that their skills are in need of bolstering. As such, parents may show variability in their readiness to engage in group treatment and have a different perception of their readiness for treatment than the clinician (Cunningham et al., 2000; Kazdin & Wassell, 2000), which may contribute to decisions not to participate in treatment or to drop out early. To the clinician’s dismay (and possible frustration), many parents choose not to engage in evidence-based parenting groups, or choose not to continue once treatment has begun (Nock & Ferriter, 2005). However, the factors that most determine a parent’s willingness to participate in treatment are still unclear. Elucidating these factors may contribute to refined approaches to fit parents with a suitable treatment, rather than a general approach in which all parents are recommended a single evidence-based treatment (regardless of their level of skills and beliefs). Client-centered approaches to treatment selection and engagement may contribute to better outcomes (Nock & Kazdin, 2005; Shepard et al., 2012).

Transtheoretical Model and Parent Readiness for Parenting Treatment Although not well understood with respect to parenting interventions, much research has applied health-belief and decision-making models to better understand client engagement in other domains of health care (DiClemente, Schlundt, & Gemmell, 2004; Prochaska, DiClemente, & Norcross, 1992; Siqueira, Rolnitzky, & Rickert, 2001). These models attempt to clarify the factors that predict a client’s readiness to participate in an intervention, with consideration of personal, family, environmental, and social factors. One well-established model of health-related beliefs that predicts client engagement in treatment is the transtheoretical model (Prochaska et al., 1992). Since its inception, this model has been tested with a variety of health populations, including diabetes education and treatment, smoking cessation

367

parenting profiles programs, and alcohol and drug intervention programs (DiClemente et al., 2004; Siqueira et al., 2001). In each of these domains of health care, the transtheoretical model has been valuable for understanding how to best therapeutically engage with clients, and how to best conceptualize their readiness for treatment. The model includes a series of fluid stages, each of which is informed by the client’s current health beliefs, behaviors, and functioning. Stages in the original model include precontemplation, contemplation, planning, action, and maintenance (Prochaska et al., 1992). Other iterations of the model have included fewer stages that are specifically relevant to the clinical population of interest (Breston, Ondersma, Simpson, & Gurwitch, 1999; Littell & Girvin, 2005). For example, with families involved in child protection services the precontemplation, contemplation, and action stages have been shown to be most relevant (Littell & Girvin, 2005). The precontemplation stage is characterized by low readiness for engagement. A person who is best characterized at this stage of readiness perceives mainly barriers to engagement and feels unable (or unwilling) to pursue the indicated treatment. Contemplation is understood as a transition from precontemplation to action. A person characterized as contemplative is actively considering the “pros and cons” of the treatment and the barriers and solutions to possible engagement. Action is understood as “ready” for change. A person characterized in the action stage is in a position of readiness to engage in the indicated treatment and has considered relevant barriers and benefits. Following action, maintenance is the stage at which behavior changes accomplished during treatment are solidified. The stages that comprise the model are nonlinear. As such, a person can have varying degrees of beliefs consistent with each of the stages. The fluidity of the model better reflects real-life behavior and decision making than would a purely categorical model.

Person-Centered Approach Findings from the aforementioned studies provide insight into variables that limit parent involvement in group treatment. However, a broader understanding of these factors remains necessary in order to develop comprehensive screening procedures and efficient strategies to mitigate identified barriers. Existing studies have primarily used a variablecentered approach to understand potential barriers to parent participation in treatment. Although results from these investigations are important, these studies typically separate parents into groups based on variables. As such, variable-centered approaches do not consider patterns of overlap among characteristics within a person. Alternatively, person-centered

methodology, as used in the present study, can classify people based on their characteristics on a number of measured domains (Lanza & Rhoades, 2013). The utility of person-centered approaches has been emphasized because the patterning of multiple constructs can be examined within individuals to determine overlap of characteristics, permitting the identification of client-centered profiles that may most closely represent the breadth of clinical risk and protective factors. This methodology has been successfully applied to domains of child psychopathology (Andrade, Sorge, Na, & Wharton-Shukster, 2014; Kuny et al., 2013; Ostrander, Herman, Sikorski, Mascendaro, & Lambert, 2008). Following this logic, the application of person-centered methodology to parents who present to children’s mental health clinics may capture the range of characteristics on which parents are similar. Such profiles may be of heuristic utility (i.e., inform clinical decision making), if they are related to important dimensions of child behavior and treatment outcomes.

Present Study Although some work has been done to understand factors that interfere with parents’ participation in treatment, these studies have primarily sought to identify variables that predict better or worse treatment engagement (Frazier, Abdul-Adil, Atkins, Gathright, & Jackson, 2007; Miller & Prinz, 2003; Nock & Ferriter, 2005). Parent readiness for treatment has been associated with less favorable treatment outcomes and higher rates of dropout in parents involved with child protection services and in some clinical populations (Frankel & Simmons, 1992; Kazdin & Wassell, 2000). The objective of this study was to apply person-centered methodology to determine clinically meaningful profiles, which include measured parent readiness for treatment and parent skills that are targeted by parent management training, to provide a novel method to understand how a combination of factors may be associated with parent participation in parenting treatment. Latent profile analysis (LPA), a person-centered analytic approach, was used to determine profiles based on parents’ perception of readiness (i.e., precontemplation, contemplation, and action), as well as three domains of parenting skills (i.e., positive parenting, inconsistent discipline, and poor supervision). Because LPA is a “bottom-up” and data-driven methodology, hypotheses are often not provided for profile identification. However, based on previous studies and empirical review, it was hypothesized that profiles characterized by higher degrees of readiness would be associated with parent report of relatively greater child behavior problems (i.e., parents possibly more motivated to engage in treatment; Kazdin &

368

andrade et al.

Wassell, 2000; Nock & Ferriter, 2005). Conversely, parents who perceived themselves as possessing more skills and who had children with relatively fewer behavior problems would be those least ready to engage in treatment. Finally, it was hypothesized that parents with higher levels of readiness and relatively lower levels of perceived parenting skills would participate in more parent management training sessions (i.e., higher participation rate than those with lower readiness and better-developed skills; Kazdin et al., 1997; Nock & Kazdin, 2005).

Method participants and procedures One hundred and forty-three parents (or legal guardians) and their children who presented for assessment at a specialized children’s mental health clinic participated in this study. The clinic is located within a large mental health hospital in Toronto, Canada, and serves a diverse group of children and families. The clinic offers psychiatric assessment and group and individualized child and parent treatment for children with disruptive behavior and associated mental health challenges. Children are typically referred for assessment because of behavioral challenges by their parents and primary care physicians. Participants were advised of the study prior to the onset of the assessment and agreed to complete measures as part of their clinical assessment. As such, participants are representative of parents and children who routinely present to this urban hospital. Approximately 90% of parents who were invited to participate in the research agreed. Parents provided free and informed consent and children provided their assent to participate in the study. Parents were asked to complete a number of measures that included those used for the present study and those used for broader clinical assessment. The clinical measures were scored and used by clinicians during the families’ second appointment to guide more indepth assessment to inform treatment planning. The hospital’s research ethics board approved all procedures for this study. Of the 143 parents/guardians, 118 (82.5%) were biological mothers, 12 (8.4%) were biological fathers, 10 (7%) were adoptive mothers, and the remaining 3 (2.1%) were either biological grandparents or identified as “other.” Among parents and caregivers, 93 (65%) reported completing some postsecondary education or higher, and 11 (7.7%) parents declined to report their education level. Among the participating children, the average age was 9.12 years (SD = 1.63, range = 6.27 to 12.81), and the majority of them were male (N = 113, 79%). Fifty-two (36%) parents or guardians reported that their child was taking some form of psychotropic

medication at intake, and 7 (4.9%) parents did not answer. Caregivers identified approximately half of these children as being White (N = 76, 54.3%), 32 (22.9%) were identified as being mixed background, 11 (7.9%) were identified as Black, and the remainder identified as one of following categories, consistent with the multicultural setting in Toronto: Aboriginal, East Asian, South Asian, Latin American, or other. Of children, 125 (89.3%) were identified as being born in Canada. Further clinical characteristics are presented in Table 1. Based on established clinical measures and cutoffs, the vast majority of children enrolled in the current study fell within the “clinical” range of severity and impairment associated with disruptive behavior (i.e., between 65.7% and 91.6%, depending on the measure).

measures Parent Readiness for Change Scale (PRCS) Parent readiness to engage in treatment was measured by the PRCS (Littell & Girvin, 2005). The 28-item scale contains questions on a 5-point Likert scale ranging between strongly disagree and strongly agree. Five items load on the Precontemplation scale (e.g., “As far as I’m concerned, I don’t need to change how I take care of my child”), six items on the Contemplation scale (e.g., “I think I might be ready to improve how I take care of my child”), and six items on the Action scale (e.g., “I am trying to learn how to take care of my child better”). Another 12 items do not load onto these subscales and were not used for the purpose of this study. Internal consistencies of the Precontemplation, Contemplation, and Action subscales in this study were α = .70, α = .82, and α = .73, respectively. Alabama Parenting Questionnaire–Short Form (APQ-S) The APQ-S is a parent-reported measure of parenting skills (Shelton, Frick, & Wootton, 1996). Responses Table 1

Clinical Characteristics of Sample (N = 143) Characteristic

Clinical cutoff

Impulsive–Overactive

≥ 11 (K–Gr. 3), ≥ 9 (Gr. 4–5) ≥ 9 (K–Gr. 3), ≥ 6 (Gr. 4–5) ≥ 90 ≥3

Oppositional–Defiant Total Difficulties (percentile) Overall Impairment

N (%) above cutoff

94 (65.7%) 128 (89.5%) 110 (76.9%) 131 (91.6%)

Note. K = Kindergarten, Gr. = Grade. Impulsive–Overactive and Oppositional–Defiant are subscales of the IOWA Conners rating scale; Total Difficulties is a subscale of the Strength and Difficulties Questionnaire; Overall Impairment is a scale from the Impairment Rating Scale.

parenting profiles to nine questions are on a 5-point Likert scale ranging between never and always. Items independently load onto three subscales: Positive Parenting, Inconsistent Discipline, and Poor Supervision. Internal consistencies of these subscales in this study were α = .89, α = .71, and α = .71, respectively. Strength and Difficulties Questionnaire (SDQ) The SDQ is a brief screening questionnaire used with parents (Bourdon, Goodman, Rae, Simpson, & Koretz, 2005; Goodman, Ford, Simmons, Gatward, & Meltzer, 2003). The questionnaire inquires about 25 attributes that are evenly divided among five behavioral dimensions (i.e., five items per behavioral dimension): prosocial skills, emotional symptoms, conduct problems, hyperactivity–inattention, and peer problems and a Total Difficulties composite scale. Subscales do not overlap, and each produces a total score. The parent-reported Total Difficulties scale was used in this study because it is a composite of the four problem-functioning subscales. Each item is rated on a 3-point Likert scale ranging from not true to somewhat true to certainly true. Subscale raw scores can be converted to percentile ranks based on normative data available at www.sdqinfo.com. Clinical percentile scores were used in the present analysis. The scale shows strong reliability, internal consistency, and validity in past studies (Bourdon et al., 2005; Stone, Otten, Engels, Vermulst, & Janssens, 2010). In this study, internal consistency of the Total Difficulties composite scale was α = .79. Impairment Rating Scale (IRS) The IRS is a parent-reported measure of impairment in seven clinical domains. Parents report how children’s behavioral difficulties impair their peer relationships, relationships with brothers and sisters, relationships with parents, academic progress, selfesteem, family functioning, and overall severity of impairment. Parents’ responses are recorded on a visual analogue scale, which is then transposed to numeric values. Scores of 3 or greater indicate a potential need for clinical attention. The overall severity of behavior rating was used in this study. Cronbach’s alphas are not computed because each scale comprises a single dimensional item; however, the validity of the IRS has been documented in past research (Fabiano et al., 2009). IOWA Conners Rating Scale The IOWA is 10-item parent report of child behavior. The scale includes five questions that examine inattentive–impulsive–overactive (IO) domains and five questions that examine oppositional defiant (OD) domains of behavior. Items that measure IO behaviors include “excitable, impulsive”; “fidgeting”; “hums and makes other odd noises”; “inattentive, easily

369

distracted”; and “fails to finish things he or she starts.” Items measuring OD behaviors include “quarrelsome,” “acts ‘smart’,” “temper outburst—behavior explosive and unpredictable,” “defiant,” and “uncooperative.” Items are rated on a 4-point Likert scale ranging between 0 (not at all) and 3 (very much). The psychometric properties of the IOWA have been demonstrated (Waschbusch & Willoughby, 2008). The IO and OD scales were used in this study to provide description of the severity of behavioral difficulties of this clinic-referred sample of children at assessment.

parent management training A subset of 37 parents who participated in this study had children between the ages of 9 and 12 years and were also invited to participate in a randomized and controlled trial comparing a clinic-modified version of the multicomponent Coping Power program (Lochman & Wells, 2003, 2004) to individualized parent–child treatment. These parents were invited to participate in the trial because their children met inclusion criteria (i.e., child between the ages of 9 and 12 years with clinically elevated disruptive behavior) and did not demonstrate any of the exclusion criteria (i.e., autism spectrum disorder, impaired cognitive functioning). Data used in this study were from those parents who participated in the parent management training group in the Coping Power arm of the trial. This manualized cognitive-behavioral parent program helps parents further develop their understanding of the social–cognitive contributions to their child’s behavior problems, behavioral management, problem-solving, and family communication skills. Groups consist of approximately 8–12 parents and combine didactic and discussion-based approaches for instruction. The parent’s group is 15 weeks in duration and runs concurrently with the children’s group. The reader is referred to other publications that describe the intervention in more detail (Lochman & Wells, 2003). primary analysis LPAs were conducted on the aforementioned parental characteristics and behaviors. Standardized parent variables were treated as continuous observed indicators, which were regressed onto a single categorical latent variable. Models were fit that ranged from two to five profiles. Recommendations for selecting the best model fit were followed (Asparouhov & Muthen, 2012). The best-fitting model was selected based on information criteria, statistical hypothesis testing, model entropy, and clinical interpretability. Models that have lower values of the Akaike information criterion (AIC), Bayesian information criterion (BIC), and sample size adjusted Bayesian

370

andrade et al. to the small percentages, it was not expected that missingness would have any impact on results. Indeed, Little’s test of missingness was nonsignificant, indicating that data are missing completely at random (MCAR; i.e., the MCAR null hypothesis was not rejected), χ 2 (70) = 56.96, p = .87. However, in order to avoid losing participants, descriptive statistics are reported using estimation maximization in SPSS, whereas profiles were derived in MPlus using full information maximum likelihood estimation (see Graham, 2009). Descriptive statistics and intercorrelations are presented in Table 2.

information criterion (aBIC) are better fitting. Higher entropy scores (approaching 1.00) are indicative of better class assignment. Also, a significant LoMendell-Rubin (LMR) likelihood ratio test indicates that the k class solution is a better fit than the k-1 class solution, where k is the number of classes of the model in question. After selecting the bestfitting model, random starts and optimizations were increased, starting values were systematically manipulated, and the parametric bootstrapped likelihood ratio test (BLRT) was employed in order to replicate the best-fitting solution. In order to validate profiles, participants were assigned to the profile to which they had the greatest probability of belonging. Profiles were compared on the SDQ Total Difficulty scale and the IRS using a one-way MANCOVA, where mean child behavior problems were compared across identified profiles adjusting for child medication status. The assumption of homogeneity of slopes was established before proceeding with analyses. Effect sizes are reported as partial eta squared. Following the multivariate and univariate tests, pairwise group comparisons were conducted using the least significant difference test. For descriptive purposes, profiles were also compared on medication status (currently taking medication or not) and gender using chi-square goodness-of-fit tests and client age using a one-way ANOVA. All tests of statistical significance are two-tailed at the p b .05 level.

supplementary analysis: probability of profile membership and treatment participation In a supplementary step to validate the profile, a subgroup analysis was conducted whereby the relationship between probability of profile membership and client attrition from treatment was examined. Data were available on 37 of the 143 families that had children between the ages of 9 and 12 years and as such were eligible for, and participated in, the group treatment arm of a randomized and controlled trial operating in the clinic. Thus, given that we only have treatment participation data on a small proportion of the sample, it was inappropriate to include this step in the primary analytic plan. As such, this analysis is intended to provide preliminary information to further inform the potential clinical utility of identified profiles. We first tested whether this subgroup was representative of the larger sample. In order to do this, a dummy indicator was created that identified whether or not treatment participation data were available for each of the study participants. This variable was subsequently correlated with probability of class membership for each of the classes. There were no

missing data and descriptive statistics Primary analyses are based on all 143 parents/ guardians and children. There was a very small amount of missing data on study variables (0 to 3.5%). Missingness was attributable to a few parents returning partially incomplete questionnaires. Due

Table 2

Descriptive Statistics and Variable Intercorrelations

A B C D E F G H

Precontemplation Contemplation Action Pos. Parent. Inconsistent Discipline Poor Supervision SDQ Total Difficulties Impairment

B

C

-.54⁎⁎

-.37⁎⁎ .73⁎⁎

D

E

.04 .02 .00

-.21⁎ .19⁎ .04 -.25⁎⁎

F

G

H

K

M

SD

-.09 .08 .05 -.17⁎ .21⁎

-.19⁎ .34⁎⁎ .22⁎⁎

-.22⁎⁎ .40** .25⁎⁎

.08 .11 .23⁎⁎

.03 -.02 .13 .56⁎⁎

-.11 .17⁎ .13 -.07 .06 .24⁎⁎ .47⁎⁎ .42⁎⁎

9.31 25.64 25.14 12.66 7.59 4.02 92.78 4.56

2.61 3.27 3.12 1.96 2.42 1.84 10.07 1.311

Note. Pos. Parent. = Positive Parenting; SDQ = Strength and Difficulties Questionnaire. Scores on the Contemplation and Action scales range between 5 and 30 and the Precontemplation scale between 5 and 25. Scores on the Positive Parenting, Inconsistent Discipline, and Poor Supervision scales range between 3 and 15. The SDQ Total Difficulties score ranges between 30 and 100 and the Impairment score between 0 and 6. * p b .05, ** p b .01.

371

parenting profiles Table 3

Model Fit Information for Latent Profile Analysis of Parental Characteristics for Models Ranging From Two to Five Profiles Index

Two Class

Three Class

Four and Five Class a

AIC BIC aBIC Entropy p(LMR)

2,189.67 2,245.97 2,185.85 0.71 0.05

2,111.85 2,188.89 2,106.62 0.81 0.09

– – – – –

Note. AIC = Akaike information criterion; BIC = Bayesian information criterion, aBIC = sample size adjusted BIC, p(LMR) = the p value for the Lo-Mendell-Rubin likelihood ratio test, p(BLRT) = p value for the bootstrapped likelihood ratio test. a The log likelihood in these models was not replicated due to local maxima. Random starts were increased to 5,000 with 500 optimized starts, though there were still convergence problems, likely due to overextraction resulting in model nonidentification. Parameters in these models are not reported and the three-class solution was selected as the best fit.

significant relationships between data availability and probability of class membership, suggesting the presence of a nonbiased subgroup in terms of class probability. Additionally, there were no significant associations between data availability and child gender or parental education (postsecondary or higher vs. no postsecondary), respectively, based on 2 × 2 chi-square goodness-of-fit tests. Thus, probability of membership in each of the identified classes was saved for the 37 available parents. Second, we determined the relationship between probability of class membership and client attrition during treatment. The probability of membership in each of the identified classes was treated as predictor in a regression where “number of sessions missed out of 15” was the outcome. Negative binomial regres-

sion was used, given that the outcome was an overdispersed count variable. Analyses controlled for parental education level, child gender, and child severity of behavioral impairment.

Results latent profile analysis of parental readiness for treatment and skills Fit information for models with two to five requested latent profiles are presented in Table 3. Based on these statistics, it was determined that the three-profile model fit the data best. Compared with the two-profile model, the three-profile model demonstrated a substantial reduction in AIC (–77.19), BIC (–57.08), and aBIC (–79.23). Additionally, the three-profile model had an acceptable entropy statistic (approaching 1.00), suggesting that profile assignment is relatively unambiguous. The LMR likelihood ratio test was marginally significant (p b .10) in preference of the three-profile solution. Moreover, the three-profile solution was chosen because it had greater heuristic and clinical value than the two-profile solution. The reader will note that fit values for the four and five class solutions are not reported in Table 3. The log likelihood in these models was not replicated due to local maxima. Random starts were increased to 5,000 with 500 final stage optimizations, though there were still convergence problems, likely due to overextraction resulting in model nonidentification. Parameters in these models are not reported (because they are untrustworthy) and the three-profile solution was retained as the best fit. The optimal log likelihood of this model (–1,029.93) was robust to an increase in random starts and optimizations. Thus, the BLRT was conducted and was significant across five successful

FIGURE 1 Standardized mean scores of parent readiness or skills as a function of latent profile membership.

372

andrade et al.

bootstrap draws, 2 × Log likelihood = 91.82, df = 7, p b .0001, confirming the preference of the threeprofile solution over the two-profile solution. See Figure 1 for the presentation of latent profiles. There was one profile with high scores on the Precontemplation scale, and low scores on the Contemplation and Action scales. These parents also had the highest scores on Positive Parenting, though this subscale did not differentiate profiles very well. However, this profile of parents rated themselves as having the lowest levels of Inconsistent Discipline and low levels of Poor Supervision. This profile was called “less in need” and made up 39.2% (N = 56) of the sample. Another profile had the lowest scores on Precontemplation, along with the highest scores on Contemplation and Action. They report moderate levels of Positive Parenting and low levels of Poor Supervision, but slightly higher levels of Inconsistent Discipline. This group was called “ready” and made up 40.6% (N = 58) of parents. Finally, there was a profile that reported average levels of Precontemplation, Contemplation, and Action. Additionally, they reported slightly lower levels of Positive Parenting, higher levels of Inconsistent Discipline, and very high levels of Poor Supervision. This profile was called “almost ready” and made up the smallest group, at 20.2% (N = 29) of the sample. To get a further understanding of derived profiles, gender composition and medication status were examined as a function of profile membership using chi-square goodness-of-fit tests. Additionally, average age was compared across profiles using a one-way ANOVA. Results for these analyses are presented in Table 4. There were no significant differences in age or gender as a function of profile. However, there were significant differences in the

Table 4

Age, Gender, and Medication Status of Children as a Function of Parenting Profile Outcome

Male Medication

Age

Less in need

Ready

Almost ready

N = 56

N = 58

N = 29

N = 47 (84%) N = 20 (36%)

N = 42 (72%) N = 16 (28%)

N = 24 (83%) N = 16 (55%)

M (SD)

M (SD)

M (SD)

9.03 (1.48)

9.16 (1.60)

9.21 (1.98)

Statistic (df)

χ 2(2) = 2.59 χ 2(2) = 6.38 ⁎

F(2,140) = .15

Note. Percentages are displayed as a proportion of column totals. ⁎ p b .05.

proportion of children taking medication across profiles. This effect was explored by examining the expected cell frequency based on the assumption of there being no systematic relationship between profile and medication status versus the observed cell frequency. There were more individuals on medication in the almost ready profile than one would expect by chance (unstandardized residual = + 5.5), whereas there were fewer children on medication in the ready profile than one would expect by chance (–5.1). Conversely, the expected and observed frequencies for individuals in the less in need group were almost identical (–.4).

child behavior problems across profiles of parental readiness and skills A one-way MANCOVA was used to examine differences in child outcomes (Total Difficulties on the SDQ and Total Impairment on the IRS) across profiles (less in need, almost ready, and ready), while controlling for child medication status (on medication vs. not on medication). The multivariate test was significant, Wilks λ = .90, F(4, 276) = 3.77, p = .005, η 2p = .05, indicating significant differences in child outcomes across profiles of smallto-medium magnitude. Specifically, there were significant differences across profiles in terms of Total Difficulties, F(2, 139) = 4.00, p = .021, η 2p = .05, and in Total Impairment, F(2, 139) = 6.93, p = .001, η 2p = .09. Both of these differences were of smallto-medium effect size. Means and standard errors are presented across profiles in Figure 2. For Total Difficulties, children of parents in the less in need profile had significantly lower scores than children of parents in the ready profile, Δ = –3.74, p = .043, and in the almost ready profile, Δ = –5.90, p = .010. The Total Difficulties of children in the ready profile did not significantly differ from the almost ready profile, Δ = –2.16, p = .344. For Total Impairment, children of parents in the less in need profile had significantly lower scores than children of parents in the ready profile, Δ = –.75, p = .002, and in the almost ready profile, Δ = –.89, p = .003. The Total Impairment of children in the ready profile did not significantly differ from the almost ready profile, Δ = –.14, p = .627. Overall, parents of children in the almost ready and ready profile reported the greatest child problems, followed by those in the less in need profile. Thus, although parents who were more ready for parenting treatment reported greater problems with their children’s behaviors, there was not a simple one-to-one correspondence between readiness for change and child difficulties. That is, parents in the almost ready group may not be quite ready for the therapeutic process of a parenting group despite having children with high levels of problems.

parenting profiles

373

FIGURE 2 Means and standard errors of parent-reported child problems across identified latent profiles of parent readiness and skills. Percentile ranks for the Strength and Difficulties Questionnaire and raw scores for the Impairment Rating Scale are indicated. Significant contrasts are identified at the p b .05 level. Note. SDQ = Strengths and Difficulties Questionnaire. SDQ Total = SDQ Total Problems.

supplementary analyses probability of profile membership and treatment participation In order to ensure that there was no systematic over- or underrepresentation of a particular class in the parent management training component of the study, a chi-square goodness-of-fit test was conducted examining the relationship between profile and participation (participated vs. did not participate). Among the less in need profile, 17/56 participants (30.4%) were involved. Among the ready profile, 16/58 (27.6%) were involved. Among the almost ready profile, 4/29 participants (13.8%) were involved. Results of the chi-square test were not statistically significant, χ 2(2) = 2.88, p = .237, indicating that there wasn’t an enrollment bias in the intervention component of the study as a function of profile. The distribution for number of sessions missed was positively skewed and overdispersed, indicative of a negative binomial distribution. That is, the majority

of families missed few sessions, but a small number of families missed several sessions. Both the median and modal number of sessions missed was three. Negative binomial regression was used to examine the relationship between probability of profile membership and the rate of missed sessions. The chi-square goodnessof-fit test was computed and nonsignificant, deviance χ 2(30) = 22.12, p = .85, indicating that there is not a significant difference between the negative binomial form and the model data (i.e., there was good model of fit). Results are presented in Table 5. Note that there is not a coefficient corresponding to the almost ready profile, because it was used as a reference category. A greater probability of being assigned to the less in need profile corresponded to a higher rate of missing sessions (B = .81, SE = .42, p = .05), though this effect is marginally outside the conventional level of significance. Conversely, there was no association between probability of membership in the ready profile and sessions missed (B = .41, SE = .38, p = .27). Thus, preliminary findings from

374

andrade et al.

Table 5

Negative Binomial Regression Output Examining Relationship Between Probability of Class Membership, Postsecondary Education, and Parent Gender With Number of Sessions Missed Parameter

B

SE

Wald χ2

p

Intercept Probability “less in need” Probability “ready” No postsecondary Female Impairment

.96 .81 .41 .81 -.46 .29

.54 .42 .38 .26 .28 .33

3.16 3.70 1.20 10.00 2.69 0.95

.08 .05 .27 b .01 .10 .38

Note. Probability of membership in the “almost ready” class serves as the reference category.

this small subset of parents show that membership in the less in need profile is associated with missing a greater number of parent management training sessions. However, there are no treatment participation differences between the almost ready and ready profiles.

Discussion This study identified person-centered parenting profiles based on parents’ report of readiness for treatment and self-perception of parenting skills. We hypothesized that parents categorized into profiles with higher levels of readiness would report greater severity of child behavior problems and would participate in more parent management training sessions than parents in the other profiles. Three relatively discrete profiles were determined, which included parents who varied in degree of readiness and perceived skills. Parents more ready for treatment reported higher levels of inconsistent discipline and poor supervision; however, the three identified profiles were only partly associated with severity of children’s behavior difficulties, as expected. Findings from this study demonstrate the importance and potential utility of person-centered approaches like LPA to understanding the overlap and breadth of important factors that may influence clinical practice (Lanza & Rhoades, 2013). Use of this methodology allowed us to identify that approximately 41% of parents could be classified as “ready for treatment” because of their high levels of contemplation and action beliefs, and high perception of both their positive parenting skills and their inconsistent discipline. A second profile, representing approximately 39% of the parents, was classified as “less in need” of intensive parenting treatment because of high levels of precontemplative beliefs, low levels of contemplation and action beliefs, high perception of positive parenting skills, and relatively lower belief that their parenting is inconsistent or that

they lack supervision skills. A third profile, representing approximately 20% of parents, was classified as “almost ready for treatment” because of moderate levels of precontemplation, contemplation, and action beliefs; relatively low perceptions of positive parenting skills; and high levels of perceived inconsistent discipline and poor supervision. LPA methodology grouped parents together based on similar characteristics, and as such identified profiles of parents that could be used to understand the clinical needs of parents with similar parenting characteristics. These findings are clinically relevant and raise a number of important questions, which are in need of further investigation. First, a large proportion of these parents were classified as less in need of intensive parenting treatment. This finding is concerning because these parents presented to a children’s mental health service and, as is common in many agencies, may be considered for a parenting intervention. As such, clinical resources may be allocated to providing intensive parenting intervention to parents who may not be inclined to participate, or may benefit more from alternate or briefer forms of intervention (Kazdin et al., 1997). In fact, preliminary findings from the present study using a small subgroup of participants show that parents characterized by the less in need profile participated in the fewest group treatment sessions. Despite the fact that these parents reported a relatively high degree of positive parenting and a somewhat lower perception of inconsistent discipline and poor supervision, these parents did report clinically significant concern with many aspects of their child’s behaviors (albeit less severe than the other two profiles). As such, these parents would likely benefit from some form of clinical intervention, possibly a brief treatment or other targeted treatment to help build on their identified parenting strengths. Another possibility is that, although these findings highlight possible consistency between these parents’ perceptions of readiness for treatment and their skills, these parents may also experience a disconnect between their perceived parenting skills and severity of their child’s behavior problems. Some research has demonstrated how parental attributions for the cause of their children’s behavior problems, and other parental characteristics, may add important information to clarify parental readiness for treatment (Hoza et al., 2000; Milner, 2003; Morrissey-Kane & Prinz, 1999). Determining which characteristics and circumstances clarify readiness for treatment for these parents (in addition to skills and severity of child behavior) is an area in need of further study. Moreover, further study of treatment models that best match these parents’ level of need is necessary in order to potentially

parenting profiles maximize parent participation in treatment that is most suitable for them. Second, about 20% of parents were classified as almost ready for treatment. This profile was associated with parents’ perception of relatively low positive parenting skills and high inconsistent discipline and poor supervision. Parents categorized into this profile reported relatively more severe difficulties with their children’s behavior and conduct. Given that these parents presented with concerns to a children’s mental health clinic, it seems likely that they would benefit from some form of intervention; however, their profiles indicate that they may yet not be ready to participate in standard parenting treatment. This knowledge is important because it can inform implementation and testing of therapeutic approaches, including motivational enhancement strategies such as motivational interviewing and other structured interventions prior to intensive parent group treatment, to foster participation by parents who may not be ready to engage in treatment (Nock & Kazdin, 2005; Sanders et al., 2004; Shepard et al., 2012). These pretreatment interventions may provide the scaffolding these almost ready parents may require to effectively engage in a longer and more intensive clinical service and maximize benefits. As such, approaches are tailored to best meet these parents’ level of readiness. Although possible, these assertions are in need of further investigation (Frankel & Simmons, 1992; Kazdin et al., 1997). Moreover, preliminary analyses with a small subset of parents in the present study show that despite reporting more challenging child behaviors, parents characterized by the almost ready profile did not differ from those characterized by the ready profile in their rate of participation in parent management treatment. However, these data should be interpreted as preliminary, given the small sample and that all of these parents (whether classified as ready, almost ready, or less in need) agreed to participate in the randomized treatment trial from which the treatment participation data were collected. Further investigation to specify differences between the almost ready and ready parents, and compare their participation in treatment using larger clinical samples of participants, would be informative. A third question raised by this study pertains to the association among parent readiness for treatment, parent-reported skills, and severity of childhood behavior difficulties. The degree of parent readiness was somewhat related to the relative severity of children’s behavior difficulties. Parents classified as almost ready and ready for treatment reported that their children had relatively more difficulties compared with parents classified as less in need of parenting treatment. As such, the severity

375

of children’s behavior problems may be an important factor motivating parents to engage in treatment. However, contrary to hypotheses, parents who were almost ready to engage in treatment did not report behavior problems that were lower than the ready profile. This finding is somewhat confusing given that one might expect parents’ motivation for treatment to be mainly driven by their child’s degree of concern. Moreover, within children’s mental health centers it is commonly assumed that parents will make a decision to engage in treatment based on their report of the types and severities of their parenting challenges. This study showed that parents more ready for treatment reported higher levels of inconsistent discipline and poor supervision. As such, parents who identified fewer skills were more ready for treatment. However, of note, those parents almost ready for treatment reported the highest levels of poor supervision. Supervision skills may represent an important domain that differentiates parent readiness for treatment. Additionally, it is possible that parenting skills and severity of child behavior represent two important domains contributing to readiness for treatment; however, other factors are also likely important. An analogy can be taken from the obesity literature. The fact that a person is obese and appreciates that he or she has associated lifestyle difficulties related to his or her obesity may not predict whether a person will choose to exercise or choose an alternate diet (Mastellos, Gunn, Felix, Car, & Majeed, 2014). In the parenting domain, in addition to perceived parenting skills and severity of child behaviors, factors that may also influence decision making could include (a) parent characteristics (e.g., mental health, parental competencies, parental attributions and cognitions), (b) social challenges (e.g., socioeconomic status), (c) family factors (e.g., supportive co-parent), and (d) cultural practices and other domains that impact health decision making (Mendez et al., 2009; Nock & Ferriter, 2005; Williford, Graves, Shelton, & Woods, 2009). Including some of these variables in largerscale studies to determine a broader scope of factors that influence parents’ decisions to participate in treatment may clarify distinctions between the almost ready and ready groups identified in this study.

study limitations Although findings from this study have important research and clinical implications, some limitations should be considered. First, as is common in children’s mental health agencies, this study had a disproportionate number of mothers presenting at assessment compared with fathers. As such, findings

376

andrade et al.

can only realistically be generalized to maternal caregivers. Further study of treatment readiness with fathers would be informative. Second, information was gathered from one rater and, consequently, a common method variance bias may have inflated correlations between variables. Moreover, we cannot rule out that a third variable may account for some of the variation among the measures in the study. Third, all parents who participated in the study presented to an urban children’s mental health clinic (within a larger mental health hospital). Findings therefore are most relevant to this population of parents. Fourth, the study utilized a self-report measure of parent readiness and a brief measure of parenting skills. Although clinically useful and efficient, further studies with multimodal measures of parenting may be important (Hoza, Johnston, Pillow, & Ascough, 2006; Kazdin & Wassell, 2000). Further, the present study did not include measures of parent cognition, mental health, or other important domains of functioning in the analysis. Future studies that incorporate these domains into readiness profiles using person-centered approaches would be informative. Fifth, we don’t have information with regard to previous interventions in which parents have participated. This information would have been helpful to further explain findings from the current study. Finally, as indicated previously, this study preliminarily tested treatment participation by profile in a small subset of participants. Testing intervention outcomes using a larger sample of parents categorized into profiles based on these and other characteristics may be additionally informative. Future research may benefit from the use of this methodology to determine treatment outcomes as an alternate to variable-centered grouping approaches. Using person-centered approaches in intervention studies may result in findings that are most applicable to “real-life” clients because each profile accounts for multiple overlapping characteristics.

clinical implications With growth in demand for services and reductions in resources, it is imperative that screening procedures for children’s mental health interventions be developed that are sensitive and realistic. While it is extremely important to assess levels of skills, disruptive behaviors, and other parent and child characteristics, it may also be important to assess client readiness and desire to participate in an intervention (Kazdin & Wassell, 2000). The best intervention can only be effective if a client is willing to participate in the process. Matching a client to a suitable treatment should extend to include his or her beliefs of readiness and motivation for treatment to best match him or

her to a suitable intervention. The best first intervention for the client may not be the one with the most evidence for effectiveness but the one most suited to the client’s perception of skills, his or her level of functioning, and current readiness for the treatment intensity and process. Conflict of Interest Statement The authors declare that there are no conflicts of interest.

References Andrade, B. F., Sorge, G., Na, J., & Wharton-Shukster, E. (2014). Determination of clinically-relevant profiles of children with disruptive behavior based on callous-unemotional traits, emotional difficulties and conduct problems. Child Psychiatry and Human Development. http: //dx.doi.org/10.1007/s10578-014-0497-8. Asparouhov, T., & Muthen, B. (2012). Using Mplus Tech11 and Tech14 to test the number of latent classes. MPlus Web Notes, 14, 1–17. Bourdon, K. H., Goodman, R., Rae, D. S., Simpson, G., & Koretz, D. S. (2005). The Strengths and Difficulties Questionnaire: U.S. normative data and psychometric properties. Journal of the American Academy of Child and Adolescent Psychiatry, 44(6), 557–564. Breston, E. V., Ondersma, S. J., Simpson, S. M., & Gurwitch, R. (1999). Application of stage of change theory to parenting behavior: Validating the parent readiness to change scale. Paper presented at the Florida Conference on Child Health Psychology, FL. Chorpita, B. F., Becker, K. D., & Daleiden, E. L. (2007). Understanding the common elements of evidence-based practice: Misconceptions and clinical examples. Journal of the American Academy of Child and Adolescent Psychiatry, 46(5), 647–652. Cunningham, C. E., Boyle, M., Offord, D., Racine, Y., Hundert, J., Secord, M., & McDonald, J. (2000). Tri-ministry study: Correlates of school-based parenting course utilization. Journal of Consulting and Clinical Psychology, 68(5), 928–933. DiClemente, C. C., Schlundt, D., & Gemmell, L. (2004). Readiness and stages of change in addiction treatment. American Journal on Addictions, 13, 103–119. Fabiano, G. A., Chacko, A., PelhamJr., W. E., Robb, J., Walker, K. S., Wymbs, F., . . . Pirvics, L. (2009). A comparison of behavioral parent training programs for fathers of children with attention-deficit/hyperactivity disorder. Behavior Therapy, 40, 190–204. Frankel, F., & Simmons, J. Q., III. Parent behavioral training: Why and when some parents drop out. Journal of Clinical Child Psychology, 21(4), 322–330. Frazier, S. L., Abdul-Adil, J., Atkins, M., Gathright, T., & Jackson, M. (2007). Can’t have one without the other: Mental health providers and community parents reducing barriers to services for families in urban poverty. Journal of Community and Social Psychology, 35(4), 435–446. Goodman, R., Ford, T., Simmons, H., Gatward, R., & Meltzer, H. (2003). Using the Strengths and Difficulties Questionnaire (SDQ) to screen for child psychiatric disorders in a community sample. British Journal of Psychiatry, 177, 534–539.

parenting profiles Graham, J. W. (2009). Missing data analysis: Making it work in the real world. Annual Review of Psychology, 60, 549–576. Hoza, B., Johnston, C., Pillow, D. R., & Ascough, J. C. (2006). Predicting treatment response for childhood attentiondeficit/hyperactivity disorder: Introduction of a heuristic model to guide research. Applied and Preventative Psychology, 11, 215–229. Hoza, B., Owens, J. S., Pelham, W. E., Swanson, J., Conners, C. K., Hinshaw, S. P., . . . Kraemer, H. C. (2000). Parent cognitions as predictors of child treatment response in attention-deficit/hyperactivity disorder. Journal of Abnormal Child Psychology, 28(6), 569–583. Kazdin, A. E. (2011). Evidence-based treatment research: Advances, limitations, and next steps. American Psychologist, 66(8), 685–698. Kazdin, A. E., Holland, L., & Crowley, M. (1997). Family experience of barriers to treatment and premature termination from child therapy. Journal of Consulting and Clinical Psychology, 65(3), 453–463. Kazdin, A. E., Siegel, T. C., & Bass, D. (1992). Cognitive problem-solving skills training and parent management training in the treatment of antisocial behavior in children. Journal of Consulting and Clinical Psychology, 60(5), 733–747. Kazdin, A. E., & Wassell, G. (2000). Predictors of barriers to treatment and therapeutic change in outpatient therapy for antisocial children and their families. Mental Health Services Research, 2(1), 27–40. Kuny, A. V., Althoff, R. R., Copeland, W., Bartels, M., Van Beijsterveldt, C. E. M., Baer, J., & Hudziak, J. J. (2013). Separating the domains of oppositional behavior: Comparing latent models of the Conners' oppositional subscale. Journal of the American Academy of Child and Adolescent Psychiatry, 52(2), 172–182. Lanza, S. T., & Rhoades, B. L. (2013). Latent class analysis: An alternative perspective on subgroup analysis in prevention and treatment. Prevention Science, 14(2), 157–168. Lin, E., Goering, P., Offord, D. R., Campbell, D., & Boyle, M. H. (1996). The use of mental health services in Ontario: Epidemiologic findings. Canadian Journal of Psychiatry, 41, 572–577. Littell, J. H., & Girvin, H. (2005). Caregivers’ readiness for change: Predictive validity in a child welfare sample. Child Abuse and Neglect, 29, 59–80. Lochman, J. E., & Wells, K. C. (2003). Effectiveness of the Coping Power program and of classroom intervention with aggressive children: Outcomes at a 1-year follow-up. Behavior Therapy, 34, 493–515. Lochman, J. E., & Wells, K. C. (2004). The Coping Power program for preadolescent aggressive boys and their parents: Outcome effects at the 1-year follow-up. Journal of Consulting and Clinical Psychology, 72(4), 571–578. Markie-Dadds, C., & Sanders, M. (2006). Self-directed Triple P (Positive Parenting program) for mothers with children at-risk of developing conduct problems. Behavioural and Cognitive Psychotherapy, 34, 259–275. Mastellos, N., Gunn, L. H., Felix, L. M., Car, J., & Majeed, A. (2014). Transtheoretical model for dietary and physical exercise modification in weight loss management for overweight and obese adults. Cochrane Database of Systematic Reviews(2). Mendez, J. L., Carpenter, J. L., LaForett, D. R., & Cohen, J. S. (2009). Parental engagement and barriers to participation in a community-based preventive intervention. American Journal of Community Psychology, 44, 1–14.

377

Miller, G. E., & Prinz, R. J. (2003). Engagement of families in treatment for childhood conduct problems. Behaviour Therapy, 34, 517–534. Milner, J. S. (2003). Social information processing in high-risk and physically abusive parents. Child Abuse and Neglect, 27, 7–20. Morrissey-Kane, E., & Prinz, R. J. (1999). Engagement in child and adolescent treatment: The role of parental cognitions and attributions. Clinical Child and Family Psychology Review, 2(3), 183–198. Nock, M. K., & Ferriter, C. (2005). Parent management of attendance and adherence in child and adolescent therapy: A conceptual and empirical review. Clinical Child and Family Psychology Review, 8(2), 149–166. Nock, M. K., & Kazdin, A. E. (2005). Randomized controlled trial of a brief intervention for increasing participation in parent management training. Journal of Consulting and Clinical Psychology, 73(5), 872–879. Ostrander, R., Herman, K., Sikorski, J., Mascendaro, P., & Lambert, S. (2008). Patterns of psychopathology in children with ADHD: A latent profile analysis. Journal of Clinical Child and Adolescent Psychology, 37(4), 833–847. Prochaska, J. O., DiClemente, C. C., & Norcross, J. C. (1992). In search of how people change applications to addictive behaviors. American Psychologist, 47(9), 1102–1114. Sanders, M. R., Pidgeon, A. M., Gravestock, F., Connors, M. D., Brown, S., & Young, R. W. (2004). Does parental attributional retraining and anger management enhance the effects of the Triple P—Positive Parenting program with parents at risk of child maltreatment? Behavior Therapy, 35, 513–535. Shelton, K. K., Frick, P. J., & Wootton, J. (1996). Assessment of parenting practices in families of school aged children. Journal of Clinical Child Psychology, 25(3), 317–329. Shepard, S., Armstrong, L. M., Silver, R. B., Berger, R., & Seifer, R. (2012). Embedding the family check-up and evidence-based parenting programmes in Head Start to increase parent engagement and reduce conduct problems in young children. Advances in School Mental Health Promotion, 5(3), 194–207. Siqueira, L. M., Rolnitzky, L. M., & Rickert, V. I. (2001). Smoking cessation in adolescents. Archives of Pediatric Adolescent Medicine, 155, 489–495. Stone, L., Otten, R., Engels, R. C., Vermulst, A. A., & Janssens, J. M. (2010). Psychometric properties of the parent and teacher versions of the Strengths and Difficulties Questionnaire for 4–12 year olds: A review. Clinical Child and Family Psychology Review, 13, 254–274. Van De Wiel, N. M. H., Mathys, W., Cohen-Kettenis, P., & Van Engeland, H. (2003). Application of the Utrecht Coping Power program and care as usual to children with disruptive behavior disorders in outpatient clinics: A comparative study of cost and course of treatment. Behavior Therapy, 34, 421–436. Waschbusch, D. A., & Willoughby, M. T. (2008). Parent and teacher ratings on the IOWA Conners rating scale. Journal of Psychopathology and Behavioral Assessment, 30(3), 180–192. Webster-Stratton, C., Hollinsworth, T., & Kolpacoff, M. (1989). The long-term effectiveness and clinical significance of three cost-effective training programs for families with conduct-problem children. Journal of Consulting and Clinical Psychology, 57(4), 550–553.

378

andrade et al.

Webster-Stratton, C., & Reid, J. M. (2003). Treating conduct problems and strengthening social and emotional competence in young children: The Dina Dinasour treatment program. Journal of Emotional and Behavioral Disorders, 11(3), 130–144. Williford, A. P., Graves, K. N., Shelton, T. L., & Woods, J. E. (2009). Contextual risk and parental attributions of children’s behavior as factors that influence the acceptability

of empirically supported treatments. Vulnerable Children and Youth Studies, 4(3), 226–237.

R E C E I V E D : June 20, 2014 A C C E P T E D : January 29, 2015 Available online 8 February 2015