Doctorate in Educational and Child Psychology Abigail Miranda Case Study 1: An Evidence-Based Practice Review Report Theme: Interventions for children with Special Educational Needs How effective is Parent Child Interaction Therapy (PCIT) for improving externalising behaviours and compliance in children with conduct disorder and associated externalising behaviour disorders? Summary Parent Child Interaction Therapy (PCIT) is an intervention targeted at parents and children displaying externalising behaviour disorders. It is implemented by trained professionals, and is unrestricted by a set number of sessions but ends when parents achieve mastery of learnt skills and the child scores below clinical cut-off points on measures for conduct disorder. This literature review aimed to evaluate the efficacy of PCIT on externalising behaviours in children with conduct disorder and associated externalising behaviour disorders. A systematic literature review was undertaken using three databases. After application of inclusion and exclusion criteria, five studies were selected for review using Kratchowill’s (2003) coding protocol, and the weight of evidence was subsequently calculated using Gough’s (2007) Weight of Evidence Framework. The review found that PCIT can be efficacious in improving disordered externalising behaviours in children. The implications for minority group participants are discussed, as are methodological limitations. The intervention would be preferable to no treatment in most cases; however, to control for expectancy effects, further research using alternative interventions is recommended. 1 Doctorate in Educational and Child Psychology Abigail Miranda Introduction What is PCIT? PCIT is an intervention developed by Sheila M. Eyberg in the USA which uses a didactic approach. A therapist or PCIT trained professional uses direct coaching with the parent, as research suggests that modifications in behaviour of parents can have an effect on a child’s behaviour. A meta-analytic review by Serketich and Dumas (1996) found that parental behaviour modification programmes had positive effects in the short term on the child’s behaviour. Conversely, a study by Strassberg, Dodge, Pettit, and Bates, (1994), found that parental aggressive behaviour in the form of context-inappropriate spanking increased the aggressive behaviour of children towards their peers. The latter suggests that negative parental behaviours influence the behaviours of a child, and the former indicates that positive behavioural changes in the parent can have beneficial effects on the child’s behaviour. There are two distinct phases in PCIT, which are Child Directed Interactions (CDI) and Parent Directed Interactions (PDI). The former is a play-based relationship building phase, whilst the latter consists of reinforcement of pro-social behaviours whilst ignoring negative behaviours. The intervention consists of instructional sessions, followed by direct coaching sessions using a one-way mirror and ear piece. This set-up enables the therapist to instruct and coach the parent without imposing on the interaction between the parent and child. Aside from sessions with the therapist, parents are given home exercises to complete. PCIT is unique in that the number of sessions given is based on parent performance and reduction in problem behaviours in the child. As well as pre and post measures, the 2 Doctorate in Educational and Child Psychology Abigail Miranda intervention is evaluated after each session. This enables the therapist to cease the intervention when the child’s behaviour falls within the normal range. Research by Goldfine et al., (2008) indicates that the start-up cost is approximately $14,000, which translates to approximately $1000 per parent and child dyad. Psychological Basis PCIT is an intervention designed to treat conduct disordered behaviour in children. The programme aims to enable the carer to establish a nurturing and secure relationship with their child. It is dyadic in the sense that it treats the parent in order to increase pro-social behaviours in the child whilst decreasing negative behaviours. Due to the circular relationship presented between parent and child behaviours, this review focusses on the externalising behaviours presented by the child, as well as compliance to adult requests. The programme draws knowledge from Baumrind’s research on differing parenting styles, including her 1967 study on the positive outcomes associated with authoritative parenting. This parenting style is inclusive of nurturing parenting and setting limits for the child. The idea of the importance of a nurturing parenting style comes from Bowlby’s (1978) Attachment Theory. The theory highlights the importance of an attentive and responsive parent, who acts as a safe base for the child to explore their surroundings. The child directed interaction (CDI) phase of PCIT serves as play-based therapy for the parent to practice skills involved in creating a secure attachment relationship with the child. Bandura’s (1977) Social Learning theory suggests that children learn through what they observe, and are likelier to imitate behaviours of those perceived to be similar to them. They also use responses from others when learning whether to maintain or cease behaviours. This is supported by Skinner’s (1938) Operant Conditioning theory, in which the response to behaviour can serve as either a deterrent or reinforcement. The former 3 Doctorate in Educational and Child Psychology Abigail Miranda and latter theories both suggests that if parental attention is the reward, then externalising behaviour may function as a tool to illicit attention; positive or negative. Looking at both Attachment theory and Social Learning theory, Patterson (1991) proposed an ‘Early Starter Model,’ which suggests that a child’s maladaptive behaviours can be learnt and reinforced by the behaviour of the parent. This theory suggests that change in the parent is instrumental in altering disordered behaviour in the child. The parent directed interaction phase (PDI) allows for anteceding behaviours to be altered, in an attempt to improve negative behavioural outcomes for the child. Rationale Externalising behaviours are an area of focus within the field of both psychology and politics, as children displaying conduct disordered behaviour were found by Scott, Knapp, Henderson, and Maughan, (2001) to cost society up to ten times more than typically developing children. Within the field of education, disruptive behaviour, physical assault, verbal abuse and threatening behaviour account for four in ten permanent exclusions and half of all fixed term exclusions (DfE, 2011). Aside from school exclusion, there are other associated negative educational outcomes for children with externalising behaviour. A longitudinal study by Fergusson and Horwood (1998) found that children displaying externalizing behaviours at 8 years old were significantly likelier to leave school without qualifications and be unemployed at the age of 18. There are several factors thought to be involved in the maintenance of conduct disorders; however singular causal links have not been established. Murray and Farrington (2010) named several environmental as well as within child factors that had been identified via longitudinal studies. Powell et al (2006) noted that research suggests that challenging behaviour displayed by young children is usually interconnected with the quality and consistency of the relationships and interactions the child has with their caregiver. Early 4 Doctorate in Educational and Child Psychology Abigail Miranda intervention for conduct disorder is preferential, as toddlers who display early signs of conduct disorder and externalising behaviours are at risk of poor educational experiences, negative peer relationships and impaired family functioning. This can lead to a cyclical process that negatively impacts on their social and emotional development. Critical periods in brain development have come to the fore in recent years, as Allen (2013) suggests that between the ages of nought and three, a child’s brain is more malleable and therefore early intervention is recommended in order to prevent problems from escalating. Parenting programmes focus on an area of risk where change is possible, and metaanalytic studies have demonstrated the efficacy of parenting programmes. Comer, Chow, Chan, Cooper-Vince and Wilson, (2013) concluded in a meta-analysis that psycho-social treatments for conduct disorder should be used as a first option, as supported by the quantitative data analysed in the review. PCIT therefore has potential importance to the field of Educational Psychology in enabling children displaying externalising behaviours to reduce the frequency and amount of disordered behaviours, therefore reducing the risk of school exclusion and failing to receive qualifications whilst in education. Review Question How effective is Parent Child Interaction Therapy (PCIT) for improving externalising behaviours and compliance in children with conduct disorder and associated externalising behaviour disorders? Critical Review of the Evidence base Literature Search Searches were conducted between January and February 2014. PsychINFO, ERIC and Medline databases were screened using a multi-field search. Search terms were input in all fields, abstract and/ or title. For a detailed breakdown of search strategies, refer to 5 Doctorate in Educational and Child Psychology Abigail Miranda Table 1. All abstracts were screened according to the criteria in Table 2. After screening for duplicates, this initial search produced 19 papers for review of the full text. Table 1. Search terms used in PschINFO, ERIC and Medline database *= wildcard term Including and Excluding Studies Table 2 displays the criterion that was applied to each paper found in the database searches. In many cases, setting limits described in the table allowed papers that did not meet the inclusion criteria to be screened out immediately. Table 2. Inclusion and exclusion criteria Inclusion Criteria Exclusion Criteria 1. Type of Publication The study must be in a peer reviewed journal. This ensures methodological rigour, as peer reviewed journals have been assessed using stringent criteria. The material is not in a peerreviewed journal (including books and conference papers). 2. Language The study is written in English, due to unavailability of resources for translation. The study is not written in English. Multi- field search terms Limits PCIT (abstract) and conduct disorder (all) English language, peer reviewed journal, 2007- current Parent Child Interaction Therapy (abstract) and conduct disorder (all fields) English language, peer reviewed journal, 2007- current PCIT (title) and conduct disorder (all fields) English language, peer reviewed journal, 2007- current Parent Child Interaction Therapy (title) and conduct disorder (all fields) English language, peer reviewed journal, 2007- current PCIT (abstract) and behavio* (abstract) 3. Type of Study English language, peer reviewed journal, 2007- current The study must contain primary The study does not contain empirical data, to ensure primary empirical data e.g. 6 Doctorate in Educational and Child Psychology Abigail Miranda originality of findings. Literature reviews and metaanalyses 4. Intervention The study implements the full and unabbreviated PCIT to individuals in the experimental group as part of the study. The study implements an abbreviated version of PCIT alone, a modified version of PCIT, or does not implement the full programme. 5. Design The study must be a group design, to allow greater generalisability of findings. The study must utilise a clinical control group of participants who did not receive PCIT The study is a single case design. 6. Dependent variables (measures) Analysis The study measures the efficacy of PCIT in relation to child behaviour. The study reports means and standard deviations of behavioural measures. The study does not measure the efficacy of PCIT in relation to child behaviour. The study does not report means and standard deviations of behavioural measures. 7. Year of Publication The study is published post 2007, and has not been included in the previous 2007 systematic review. The study is published pre 2007, and/ or it has been included in the previous 2007 systematic literature review. 8. Participants Participants have been selected on the basis that they have externalising behaviour disorders and/ or conduct disorder. Participants must not be selected on the basis that they have been subjected to, or at risk of abuse due to possible coercive nature of intervention Participants selected do not have externalising behaviour disorders and/ or conduct disorder. The study did not have a clinical control group of participants who did not receive PCIT. Participants have been selected on the basis that they have been subjected to, or at risk of abuse. 7 Doctorate in Educational and Child Psychology Abigail Miranda Flowchart 1 displays the details of the number of papers screened at each stage of review. Flowchart: Application of inclusion and exclusion criteria PsycInfo Eric Medline Total n = 309 n = 61 n = 14 n = 384 Papers for review of title and abstract 384 Papers excluded: Duplicates n = 68 Inclusion criteria not met n = 297 Papers for review of full text 19 Studies excluded: Inclusion criteria 1 Inclusion criteria 2 Inclusion criteria 3 Inclusion criteria 4 Inclusion criteria 5 Inclusion criteria 6 Inclusion criteria 7 Inclusion criteria 8 Total excluded 14 n=0 n=0 n=1 n=3 n=6 n=3 n=0 n=1 n= Studies Included 5 8 Doctorate in Educational and Child Psychology Abigail Miranda After the screening process was complete, five studies met the inclusion criteria and were then subject to a critical review. Included studies are listed in Table 3. Full details of excluded studies, as well as reasons for exclusion are listed in Appendix 1. Table 3. Included studies Included Studies Bagner, D. M., & Eyberg, S. M. (2007). Parent–child interaction therapy for disruptive behavior in children with mental retardation: A randomized controlled trial. Journal of Clinical Child and Adolescent Psychology, 36(3), 418-429. Leung, C., Tsang, S., Heung, K., & Yiu, I. (2009). Effectiveness of Parent—Child interaction therapy (PCIT) among Chinese families. Research on Social Work Practice, 19(3), 304-313. McCabe, K., & Yeh, M. (2009). Parent–child interaction therapy for Mexican Americans: A randomized clinical trial. Journal of Clinical Child & Adolescent Psychology, 38(5), 753759. Bagner, D. M., Graziano, P. A., Jaccard, J., Sheinkopf, S. J., Vohr, B. R., & Lester, B. M. (2012). An initial investigation of baseline respiratory sinus arrhythmia as a moderator of treatment outcome for young children born premature with externalizing behavior problems. Behavior therapy, 43(3), 652-665. Bagner, D. M., Sheinkopf, S. J., Vohr, B. R., & Lester, B. M. (2010). Parenting intervention for externalizing behavior problems in children born premature: an initial examination. Journal of developmental and behavioral pediatrics: JDBP, 31(3), 209. Comparison of Selected Studies Five studies were summarised (see Appendix 2) and then compared on various aspects of design and outcomes. Out of the five studies that were critically appraised, four studies were randomised control trials (RCT’s) and one study was a non-randomised between participants block design. Three out of the five studies explicitly stated the randomisation of participants to the point where it would be repeatable. All five of the studies were critically examined using the UCL Educational Psychology Literature Review Coding Protocol, which was adapted from the APA Task Force on Evidence Based Interventions 9 Doctorate in Educational and Child Psychology Abigail Miranda in School Psychology (Kratochwill, 2003). The use of the protocol facilitated a systematic analysis of included studies, based on the same criterion. Areas of focus for analysis were quality of measures, comparison group and analysis. All included studies contained quantitative data, therefore were all analysed using this protocol. The Weight of Evidence framework (Gough, 2007) was subsequently used as a systematic tool to analyse the effectiveness of the studies in answering the proposed research question. Table 4 outlines the aspects analysed using the framework. An overall Weight of Evidence relevant to the research question was formulated scoring studies on specific facets: Quality of Methodology, Relevance of Methodology and Relevance of Evidence to the Review Question. Scores on the three dimensions were averaged in order to obtain an Overall Weight of Evidence score. Details of weightings given to each study can be found in Table 5, and Appendix 4 contains the raw scores from which the Overall Weight of Evidence was calculated. Table 4. Weight of Evidence Framework (Gough, 2007) Weight of Weight of Weight of Evidence Evidence Evidence A B C Generic judgement Review-specific Review-specific about the judgement about judgement about coherence and appropriateness of the relevance of integrity of the the evidence for the focus of the evidence answering the evidence for the (Quality of review question review question Methodology) (Relevance of (Relevance of Methodology) evidence to the review question) Weight of Evidence D Overall assessment of the extent to which the study contributes evidence to answer a review question (Overall weight of evidence) 10 Doctorate in Educational and Child Psychology Abigail Miranda Critical Review Participants As per the inclusion criteria, all participants were selected on the basis that they reached clinical cut-off points for conduct disorder on parent report measures at the screening stage. Only one study (Bagner and Eyberg, 2007) used DSM-IV criteria as a formal diagnosis of oppositional defiant disorder (ODD). The study by Leung, Tsang, Heung, and Yiu, (2009),used clinical cut-off points which were found to be relevant with the target population of Chinese children and their families, which happened to be under the standardised cut-off points on the ECBI scale. Two out of the four studies excluded participants on the basis of co-morbid disorders which may have added confounding variables to the intervention. All studies conducted statistical analyses to compare homogeneity of groups. Participants were recruited from a variety of sources such as health professionals, selfreferral and schools. Abuse cases were screened out during the inclusion process due to the coerced nature of involvement in the study, which could act as a confounding variable. Studies were carried out in USA and China. The demographics of participants varied; however three out of the five studies appeared to have a cultural mixture of participants. The remaining two studies were conducted on homogenous cultural groups, although the participants in the McCabe and Yeh (2009) study would be considered as a minority group, being Mexican-Americans in a USA study. The range of participants between studies suggests that PCIT does have applications for use when working with parents and families from diverse backgrounds; however caution should be taken when interpreting co-morbid disorders, as these may not always be diagnosed or picked up by the screening measures used in the study. Loeber and Keenan (1994) highlighted the 11 Doctorate in Educational and Child Psychology Abigail Miranda relationship between conduct disorder and co-morbid disorders that are commonly associated with the condition. Research Design Control Group All included studies utilised clinical control groups to form a comparison with the intervention group. Three out of the five studies operated using a randomised control or blocked design, which was explicitly stated and replicable. The Leung et al., (2009) study used a quasi- experimental design, which was non-randomised as the control group participants were recruited and allocated to their experimental condition from different sources to the intervention group. All studies either established group equivalence by random assignment, or by post-hoc tests. The McCabe and Yeh (2009) study was the only one that used an alternative intervention control group. In this study, control participants received treatment as usual from a therapist. Ethically, this study is sounder in that participants were given immediate access to treatment. Additionally, the alternative intervention acted as a control for expectancy effects, as the effects of improvement from merely receiving additional support was considered in designing this experiment. Intervention All interventions were carried out in clinical settings. This was necessary due to the requirement for a one-way mirror. The absence of sound longitudinal data meant that it was unclear whether the results generalised to long-term gains for participants. Studies utilising PCIT in natural or group settings would be delivering a modified version of the 12 Doctorate in Educational and Child Psychology Abigail Miranda programme, however it may be beneficial to examine the effects of the naturalistic setting on results. Interventions were carried out by therapists and trained professionals. Two studies explicitly stated the role and qualifications of the therapists delivering the intervention, and the remaining two studies made reference to a ‘therapist,’ although it was unclear who this individual was in terms of role and experience. PCIT is a manualised programme, and all studies reviewed reported that interventions were carried out by professionals trained in delivering PCIT. Measures All five studies used at least one of the following measures: Child Behavior Checklist for 11 2 to 5 Year Olds (CBCL; Achenbach and Rescorla, 2000), Eyberg Child Behavior Inventory (ECBI; Eyberg and Pincus, 1999) and the Dyadic Parent–Child Interaction Coding System (DPICS; Eyberg, Nelson, Duke and Boggs, 2004). These measures were chosen for further inspection due to their relevance to the review question, as they captured parent reported and observed externalising behaviours. The ECBI was deemed suitable for the review as it measured the number of externalising behaviours, as well as the number of times they occurred. This measure yielded reliability coefficients ranging from .80 - .95 in the five studies that used it, suggesting that it is a highly reliable measure. The CBCL consists of both an internalising and externalising behaviour scale. For relevance to the review question, where applicable only scores for the externalising component of the scale were analysed. The CBCL yielded reliability coefficients ranging from .54 - .94 in the three studies that used it (Bagner and Eyberg, 2007; McCabe and Yeh, 2009 and Bagner, D. M., Sheinkopf, S. J., Vohr, B. R., and 13 Doctorate in Educational and Child Psychology Abigail Miranda Lester, B. M. (2010).. For studies that used either the ECBI or the CBCL as a standalone measure, the subjectivity of a parental report can be questioned due to the vested interest of the reporter. All studies reported the reliability coefficients of all measures used, bar the Leung (2009) study which did not report reliability for the DPICS. This was an aspect reflected in the low weighting given for measures. All other studies received a medium rating for measures. The DPICS was used as an observational measure in all studies bar one (Bagner, Graziano, Jaccard, Sheinkopf., Vohr & Lester, 2012). This measure yielded reliability coefficients of .50 - .89 in the three studies that reported this data. Although the manual is very specific about how to code indirect and direct commands, with families who speak English as an additional language (EAL) linguistic subtleties may be incorrectly coded, and this is not discussed in terms of the validity of the observational tool; nor is prior exposure of coders to diverse groups. Given the cultural diversity in many of the studies, this would require some acknowledgement in future studies. Four out of the five studies used two or more methods of report, from two or more sources. This triangulation formed part of the medium to high ratings on Weight of Evidence C, for all but two studies (Leung et al, 2009 and Bagner et al, 2012). Analysis Four out of the five studies used appropriate methods of analysis for results. The study that did not, McCabe and Yeh (2009), calculated within group effect sizes for pre and post data collection. A between group comparison would have been more relevant to the 14 Doctorate in Educational and Child Psychology Abigail Miranda hypotheses stated in the paper. Two studies (Bagner and Eyberg, 2007 and Leung et al., 2009) used the analysis of covariance (ANCOVA), which was considered the most suitable statistical test used in the selected studies as it is effective when data has been collected multiple times for the same measures. High weight of evidence ratings were given to these studies for the analysis component. The remaining two studies (Bagner et al, 2010 and Bagner et al., 2012), used calculations of effect sizes and a multiple regression to analyse results. Two studies (Leung et al, 2009 and McCabe and Yeh, 2009) used post hoc tests to control for experimenter error rate; Bonferroni correction and Tukey’s post hoc tests respectively. Findings The majority of the studies found medium to large effects between intervention and control groups, suggesting that PCIT had some efficacy in reducing externalising behaviours. These results must be interpreted with caution, as three out of the five studies (McCabe and Yeh, (2009), Bagner and Eyberg, (2007) and Bagner et al., (2010), some effects were negative. This is a particularly pertinent finding in the McCabe and Yeh (2009) study, which used an active comparison group. The negative effect sizes, displayed in Table 5 for these three studies indicate that the intervention had a negative impact on self-reported data, as measured by the CBCL and ECBI. The Bagner and Eyberg (2007) results displayed a medium sized negative effect, 15 Doctorate in Educational and Child Psychology Abigail Miranda as described by Cohen (1988) on the number of externalising behaviours on the ECBI. This suggests that parents reported fewer externalising behaviours before engaging with the PCIT intervention. McCabe and Yeh (2009) reported results displaying medium to large negative effect sizes on the CBCL Externalising Behaviours subscale, small to medium negative effect sizes on the ECBI Frequency of externalising behaviours, and small negative effect sizes in the number of externalising behaviours on the ECBI. This suggests that in comparison with the alternative intervention of individualised sessions with a therapist, parents who undertook PCIT reported a lesser range and frequency of externalising behaviours before the intervention in comparison to after the intervention. Similarly, the Bagner et al., (2012) study displayed a large negative effect within the ECBI Intensity subscale, which indicated that parents reported more externalising behaviours after the intervention in comparison to before. Whilst this finding may be interpreted as negative, it could also be argued that PCIT increases parental awareness of what externalising and negative behaviours are, which in turn allows them to identify and report these behaviours more accurately. Studies that triangulated self-report data with use of the DPICS would indicate that this was the case, as the DPICS effect sizes were positive and large in the studies that used it as a measure (Bagner & Eyberg, 2007; Leung et al., 2009; McCabe & Yeh, 2009; Bagner et al., 2010). These findings display improvements in observed, coded behaviour of children whose parents undertook PCIT, comparing pre and post intervention behaviours. This is also the case in the McCabe and Yeh (2009) study, where an alternative intervention was used; however as improvements were also made with the alternative intervention, this suggests that PCIT is more focussed on improving child behaviours than an individualised therapeutic intervention. The remaining three studies found positive small to large effects of the intervention when compared to the control. The results and overall quality ratings can be found in Table 5. 16 Doctorate in Educational and Child Psychology Abigail Miranda Table 5. Summary of Effect sizes and overall quality of studies Study Bagner & Eyberg, 2007 Measures CBCL (Achenbach and Rescorla, 2000). ECBI (Eyberg and Pincus, 1999). DPICS (Eyberg, Nelson, Duke andBoggs, 2004). Leung et al., ECBI 2009 (Eyberg and Pincus, 1999). Effect sizes Externalising behaviour, Cohen’s d = .97 (large) Overall Quality Rating Medium Frequency Cohen’s d = 1.5 (large) Number Cohen’s d = -.66 (medium) Compliance Cohen’s d = 1.53 (large) Frequency Cohen’s d = 1.59, CI 1.15 - 1.71 (large) Low Number Cohen’s d = 1.52, CI 1.08 - 1.43 (large) McCabe & Yeh, 2009 DPICS (Eyberg, Nelson, Duke and Boggs, 2004). Compliance Partial eta-squared = .49 (large) CBCL (Achenbach and Rescorla, 2000). Externalising behaviour, Cohen’s d = -.79, (medium to large) ECBI (Eyberg and Pincus, 1999). Frequency Cohen’s d = -.49 (small to medium) High Number Cohen’s d = -.36 (small) Bagner et DPICS (Eyberg, Nelson, Duke andBoggs, 2004). Cohen’s d = .35 (small) ECBI Intensity Low 17 Doctorate in Educational and Child Psychology Abigail Miranda al., 2012 (Eyberg and Pincus, 1999). Cohen’s d = -1.10 (large) Bagner et al., 2010 CBCL (Achenbach and Rescorla, 2000). Externalising behaviours Cohen’s d = 2.3 (large) Medium Frequency Cohen’s d = 2.3 (large) ECBI (Eyberg and Pincus, 1999). Number Cohen’s d = 1.4 (large) DPICS (Eyberg, Nelson, Duke and Boggs, 2004). Compliance Cohen’s d = .90 (large) Note: CBCL=Child Behavior Checklist, ECBI=Eyberg Child Behavior Inventory, DPICS=Dyadic Parent Child Interaction Coding System. Table 6. Descriptors of small, medium and large effect sizes (Green and Salkind, 2008) Type of effect size Partial Eta Square (Cohen, 1988) Cohen's d (Cohen, 1988) Small Medium 0.01 0.06 0.20 0.50 Large 0.14 0.80 Three out of the five studies showed that PCIT was effective when compared to a control group. Out of these three studies, two studies were found to be of medium overall quality. These two studies (Bagner and Eyberg, 2007 and Bagner et al., 2010) found large positive effects of PCIT when compared to a waitlist control group. These findings suggest that there is some evidence for the use of PCIT as a treatment for externalising behaviours. 18 Doctorate in Educational and Child Psychology Abigail Miranda The Bagner and Eyberg, (2007) and Bagner et al., (2010) studies were given a medium rating in regards to relevance to the research question. This was due to their use of a waitlist control group, and triangulation of data. Both studies were judged to be of medium overall quality. Positive large effect sizes of .97 and 2.3 were found on the CBCL. The ECBI frequency scale yielded positive large effect sizes of 1.5 and 2.3. For the ECBI intensity scale, a medium negative effect of -.66 was found in the Bagner and Eyberg (2007) study, whereas the Bagner et al., (2010) study noted an effect size of 2.3 for this subscale. The effect sizes on the DPICS were 1.53 and .90, and this measure was considered to be the most objective used throughout the studies. Despite the promising findings, it must be noted that the Bagner and Eyberg (2007) study was a first generation study, as the PCIT author was also an author of the study. The McCabe and Yeh (2009) study was the only study found to be of high relevance to review, as it utilised an active control group. It also received a high overall rating, as it established group equivalency and found 75% or more significant results. Considering that the study was well designed, the negative effect sizes are all the more pertinent. ECBI effect sizes were -.49 and -.36 on the frequency and intensity scales. The CBCL effect size for externalising behaviour was -.79. A small positive effect of .35 was found on the DPICS. These results indicate that although parents in the intervention condition did not find significant positive improvements between groups, observers did find a small positive between group effect. The final two studies, Bagner et al., (2012) and Leung et al., (2009) were regarded as being of low relevance to the review question due to a lack of triangulation of data and implementation of the intervention by a person other than a therapist. The studies also received low overall ratings, largely due to a lack of establishing group equivalency and equivalent mortality. 19 Doctorate in Educational and Child Psychology Abigail Miranda ECBI intensity effect sizes were large but negative in the Bagner et al., (2012) study at 1.10, and large but positive in the Leung et al., (2009) study at 1.52. Bagner et al., (2012) did not report any other relevant measures. Leung et al., (2009) found a large positive effect of 1.59 on the ECBI frequency subscale, and a large partial eta-squared effect of .49 for compliance on the DPICS, as described by Cohen (1988). Control group data was not available to calculate the Cohen’s d effect size for standardisation purposes. Although the results were promising, the low overall weighting and relevance to the review meant that these findings were considered less robust. Conclusion and Recommendations The studies reviewed that received medium overall ratings (Bagner and Eyberg, 2007 and Bagner et al., 2010) were deemed as efficacious on both parent report and observational measures when compared to a non-active control group. The main factor that compromised ratings was the absence of an active control group. It was also noted that the author of the programme was also an author of the evaluative study, suggesting an element of vested interest; however, the method outlined steps taken to reduce bias, such as inter-rater coding of sessions to check fidelity, and only the first author being involved in the delivery of PCIT. Four out of the five studies used random assignment to treatment conditions, and only one study used an active control group. This meant that expectancy effects were not controlled for in the majority of studies analysed, and it was not apparent that these were adjusted for via statistical analyses. In the McCabe and Yeh (2009) study that used an 20 Doctorate in Educational and Child Psychology Abigail Miranda active comparison group, although the parent report effects between groups were negative, the observation effects between groups showed a small positive effect. This may have been due to parents receiving non-directional one-to-one support from a therapist in the control condition, therefore leading to an increased sense of control over the treatment being compatible with their perceived needs. Although highly prescriptive in the description of coding observed behaviours, the studies that used the DPICS did not note any steps taken to ensure cultural sensitivity in the application of this measurement instrument. It has been noted in a meta-analysis by Van Ijzendoorn and Kroonenberg, (1988) that observational methods such as Ainsworth and Bell’s (1970) Strange Situation procedure are subject to cross cultural differences in the presentation of behaviours. Although some of the studies undertook inter-rater codings to control for observer bias, if the coders all had the same perspectives and experience of diverse ethnic groups, it could be argued that their ability to note subtle differences may be similarly limited. The studies were all clinical trials; therefore there is limited information as to how results translate into real world settings. In cases where children were in educational settings, it would have been feasible to collect pre and post teacher reported data in order to further triangulate wider effects of the programme. This would heighten the case for validity, whilst making a contribution to the field of Educational Psychology. Recommendations There are several recommendations for future research and implementing PCIT as a therapeutic intervention. The use of the Dyadic Parent–Child Interaction Coding System (DPICS) would be recommended in future studies, due to increased subjectivity when 21 Doctorate in Educational and Child Psychology Abigail Miranda using a trained coder blind to experimental conditions, and the ability to assess inter-rater reliability. This would counterbalance the subjectivity of parental report methods used. Further to this recommendation, the use of active control groups is recommended. A comparison could then be made of the efficacy of PCIT with other treatments and would subsequently inform best practice in terms of evidence based and cost effective interventions. As PCIT consists of two parts which have distinctive psychological underpinnings, PCIT would be useful in future research to examine which aspects are successful in treating the holistic profile of conduct disorder. This would then inform evidence based practice targeting the components that make up an authoritative parent. In weighing up start-up and delivery costs alongside the information presented in this review, PCIT would only be recommended where no other alternative was available at this present time, and for cases where parents required directive solutions, such as court or government required parenting programmes in cases of physical abuse. Future studies examining longitudinal or school based effects of PCIT in comparison to alternative treatments are needed to find out the maintenance of the effects, although it is granted that this research will be not without difficulty given the short-term attrition demonstrated in the studies examined. 22 Doctorate in Educational and Child Psychology Abigail Miranda Reference and Appendices References Allen, G. (2011). Early intervention: the next steps, an independent report to Her Majesty's government by Graham Allen MP. TSO Shop. Bagner, D. M., & Eyberg, S. M. (2007). Parent–child interaction therapy for disruptive behavior in children with mental retardation: A randomized controlled trial. Journal of Clinical Child and Adolescent Psychology, 36(3), 418-429. Bagner, D. M., Graziano, P. A., Jaccard, J., Sheinkopf, S. J., Vohr, B. R., & Lester, B. M. (2012). An initial investigation of baseline respiratory sinus arrhythmia as a moderator of treatment outcome for young children born premature with externalizing behavior problems. Behavior therapy, 43(3), 652-665. Bagner, D. M., Sheinkopf, S. J., Vohr, B. R., & Lester, B. M. (2010). Parenting intervention for externalizing behavior problems in children born premature: an initial examination. Journal of developmental and behavioral pediatrics: JDBP, 31(3), 209. Barker, C., Pistrang, N., & Elliott, R. (2003). Research methods in clinical psychology: An introduction for students and practitioners. John Wiley & Sons. Baumrind, D. (1967). Child care practices anteceding three patterns of preschool behavior. Genetic psychology monographs. 23 Doctorate in Educational and Child Psychology Abigail Miranda Bessmer, J. L. (1998). The Dyadic Parent-Child Interaction Coding System II (DPICS II): Reliability and validity (Doctoral dissertation, ProQuest Information & Learning). Boggs, S. R., Eyberg, S., & Reynolds, L. A. (1990). Concurrent validity of the Eyberg child behavior inventory. Journal of Clinical Child Psychology, 19(1), 75-78. Bowlby, J. (1978). Attachment theory and its therapeutic implications. Adolescent psychiatry. Comer, J. S., Chow, C., Chan, P. T., Cooper-Vince, C., & Wilson, L. A. (2013). Psychosocial treatment efficacy for disruptive behavior problems in very young children: a meta-analytic examination. Journal of the American Academy of Child & Adolescent Psychiatry, 52(1), 26-36. Department for Education. (2011). Statistical First Release PERMANENT AND FIXED PERIOD EXCLUSIONS FROM SCHOOLS. DfE: London. Fergusson, D. M., & Horwood, L. J. (1998). Early conduct problems and later life opportunities. Journal of Child Psychology and Psychiatry, 39(8), 1097-1108. Goldfine, M. E., Wagner, S. M., Branstetter, S. A., & Mcneil, C. B. (2008). Parent-Child Interaction Therapy: An Examination of Cost-Effectiveness. Journal of Early & Intensive Behavior Intervention, 5(1). Gough, D. (2007). Weight of Evidence: A Framework for the Appraisal of the Quality and Relevance of Evidence. Research Papers in Education, 22, 213-228. 24 Doctorate in Educational and Child Psychology Abigail Miranda Kazdin, A.E. (2005a). Evidence-based assessment for children and adolescents: issues in measurement development and clinical application. Journal of Clinical Child &Adolescent Psychology: The Official Journal for the Society of Clinical Child & Adolescent Psychology, American Psychological Association, Division 53, 34(3), 548–558. Kazdin, A. E. (2005). Parent management training: Treatment for oppositional, aggressive, and antisocial behavior in children and adolescents. New York: Oxford University Press. Leung, C., Tsang, S., Heung, K., & Yiu, I. (2009). Effectiveness of Parent—Child interaction therapy (PCIT) among Chinese families. Research on Social Work Practice, 19(3), 304-313. Loeber, R., & Keenan, K. (1994). Interaction between conduct disorder and its comorbid conditions: Effects of age and gender. Clinical Psychology Review, 14(6), 497-523. McCabe, K., & Yeh, M. (2009). Parent–child interaction therapy for Mexican Americans: A randomized clinical trial. Journal of Clinical Child & Adolescent Psychology, 38(5), 753759. Murray, J., & Farrington, D. P. (2010). Risk factors for conduct disorder and delinquency: key findings from longitudinal studies. Canadian journal of psychiatry. Revue canadienne de psychiatrie, 55(10), 633-642. Patterson, G. R., DeBaryshe, B. D., & Ramsey, E. (1989). A developmental perspective on antisocial behavior (Vol. 44, No. 2, p. 329). American Psychological Association. 25 Doctorate in Educational and Child Psychology Abigail Miranda Patterson, G. R., Capaldi, D., & Bank, L. (1991). An early starter model for predicting delinquency. In An earlier draft of this chapter was presented at the Earlscourt Conference on Childhood Aggression, Toronto, Canada, Jun 1988.. Lawrence Erlbaum Associates, Inc. Pearl, E. S. (2009). Parent management training for reducing oppositional and aggressive behavior in preschoolers. Aggression and Violent Behavior, 14(5), 295-305. Powell, D., Dunlap, G., & Fox, L. (2006). Prevention and intervention for the challenging behaviors of toddlers and preschoolers. Infants & Young Children, 19(1), 25-35. Scott, S., Knapp, M., Henderson, J. and Maughan, B. (2001). Financial cost of social exclusion: Follow‐up study of antisocial children into adulthood. British Medical Journal, 323(7306): 1–5. 191 Serketich, W. J., & Dumas, J. E. (1996). The effectiveness of behavioral parent training to modify antisocial behavior in children: A meta-analysis. Behavior therapy, 27(2), 171-186. Strassberg, Z., Dodge, K. A., Pettit, G. S., & Bates, J. E. (1994). Spanking in the home and children's subsequent aggression toward kindergarten peers. Development and psychopathology, 6(03), 445-461. Thomas, R., & Zimmer-Gembeck, M. J. (2007). Behavioral outcomes of parent-child interaction therapy and Triple P—Positive Parenting Program: A review and metaanalysis. Journal of abnormal child psychology, 35(3), 475-495. Van Ijzendoorn, M. H., & Kroonenberg, P. M. (1988). Cross-cultural patterns of attachment: A meta-analysis of the strange situation. Child Development, 147-156. 26 Doctorate in Educational and Child Psychology Abigail Miranda Appendix 1 Excluded articles from PsychINFO, ERIC and Medline Studies Abrahamse, M. E., Junger, M., Chavannes, E. L., Coelman, F. J., Boer, F., & Lindauer, R. J. (2012). Parent–child interaction therapy for preschool children with disruptive behaviour problems in the Netherlands. Child and adolescent psychiatry and mental health, 6(1), 19. Rationale for exclusion Criteria 5 – does not have a clinical control group 2 Bagner, D. M., Rodriguez, G. M., Blake, C. A., Rosa-Olivares, J. (2013). Home-based preventive parenting intervention for at-risk infants and their families: An open trial. Cognitive and Behavioral Practice. 20(3), 334-348. Criteria 5 – does not have a control group 3 Chase, R. M., & Eyberg, S. M. (2008). Clinical presentation and treatment outcome for children with comorbid externalizing and internalizing symptoms. Journal of Anxiety Disorders, 22(2), 273-282. Criteria 5 – does not have a control group 4 Fernandez, M. A., & Eyberg, S. M. (2009). Predicting treatment and follow-up attrition in parent–child interaction therapy. Journal of Abnormal Child Psychology, 37(3), 431-441. Criteria 6 - The study does not measure the efficacy of PCIT in relation to child behaviour. 5 Fernandez, M. A., Butler, A. M., & Eyberg, S. M. (2011). Treatment outcome for low socioeconomic status African American families in parent-child interaction therapy: A pilot study. Child & Family Behavior Therapy, 33(1), 32-48. Criteria 4 - The study implements a modified version of PCIT 6 Graziano, P. A., Bagner, D. M., Sheinkopf, S. J., Vohr, B. R., Lester, B. M. (2012). Evidencebased intervention for young children born premature: Preliminary evidence for associated changes in physiological regulation. Infant Behavior & Development, 35(3), 417-428. Criteria 6 - The study does not measure the efficacy of PCIT in relation to child behaviour. 7 Lanier, P., Kohl, P. L., Benz, J., Swinger, D., Moussette, P., Drake, B. (2011). Parent-child interaction therapy in a community setting: Examining outcomes, attrition, and treatment setting. Research on Social Work Practice, 21(6), 689-698. Criteria 5 - The study did not have a clinical control group of participants who did not receive PCIT. 1 27 Doctorate in Educational and Child Psychology Abigail Miranda 8 Lyon, A. R., & Budd, K. S. (2010). A community mental health implementation of Parent–child Interaction Therapy (PCIT). Journal of child and family studies, 19(5), 654-668. Criteria 5 - The study did not have a clinical control group of participants who did not receive PCIT. 9 Matos, M., Bauermeister, J. J., & Bernal, G. Criteria 4 - The study implements a modified (2009). Parent‐Child Interaction Therapy for Puerto Rican Preschool Children with ADHD and version of PCIT Behavior Problems: A Pilot Efficacy Study. Family process, 48(2), 232-252. 10 Naik-Polan, A. T., & Budd, K. S. (2008). Stimulus Generalization of Parenting Skills during ParentChild Interaction Therapy. Journal of Early & Intensive Behavior Intervention, 5(3). . Criteria 8 - Participants have been selected on the basis that they have been subjected to, or at risk of abuse. 11 Phillips, J., Morgan, S., Cawthorne, K., & Barnett, B. (2008). Pilot evaluation of Parent–Child Interaction Therapy delivered in an Australian community early childhood clinic setting. Australian and New Zealand Journal of Psychiatry, 42(8), 712-719. Criteria 5 - The study is a single case design. 12 Solomon, M., Ono, M., Timmer, S., & GoodlinCriteria 4 - The study Jones, B. (2008). The effectiveness of parent– implements a modified child interaction therapy for families of children on version of PCIT the autism spectrum. Journal of Autism and Developmental Disorders, 38(9), 1767-1776. 13 Tiano, J. D., Grate, R. M., & McNeil, C. B. (2013). Comparison of Mothers' and Fathers' Opinions of Parent–Child Interaction Therapy. Child & Family Behavior Therapy, 35(2), 110-131. Criteria 5 - The study did not have a clinical control group of participants who did not receive PCIT. 14 Timmer, S. G., Zebell, N. M., Culver, M. A., & Urquiza, A. J. (2010). Efficacy of Adjunct InHome Coaching to Improve Outcomes in Parent—Child Interaction Therapy. Research on Social Work Practice, 20(1), 36-45. Criteria 6 - The study does not report means and standard deviations of behavioural measures. 28 Doctorate in Educational and Child Psychology Abigail Miranda Appendix 2 Summary of Studies Author and Aim (relevant to review) Sample Intervention Bagner, D. M., and Eyberg, S. M. (2007). USA Treatment group Quasireceived PCIT once experimental a week, 1 hour per design week. QuasiEach family was Experimental: seen by two Untreated control therapists; Bagner group design with (2 cases) and 8 dependent pretest graduate clinical and posttest and child samples psychology students with prior Randomly assigned training and to WL/ IT group experience as a using two computer PCIT therapist. generated random numbers lists, one Waitlist control for boys and one group received no for girls. The list input from the study was controlled by at the time of pre the second author and post measures. who was not To compare the effectiveness of PCIT on families with children displaying significant externalising behaviour problems, with comorbid learning needs (described as mental retardation in the study), with a waitlist control group. 3 – 6 years Treatment group mean age: 52.40m WL control mean age: 55.87m 77% boys 30 participants All diagnosed with ODD. 60% Mild Mental Retardation (IQ score WISC) Referral 80% Paediatrician 10% Teacher 10% self-referral Design Measures Outcomes Child Behavior Checklist for 11 2 to 5 Year Olds (CBCL; Achenbach and Rescorla, 2000). ANCOVA used to analyse data. Eyberg Child Behavior Inventory (ECBI; Eyberg and Pincus, 1999). Dyadic Parent– Child Interaction Coding System (DPICS; Eyberg, Nelson, Duke and Boggs, 2004). Compliance – pre intervention means 63.88 (19.22) differed significantly from post intervention means, f (1, 18) = 9.68, p <.05 ECBI Intensity – pre and post means differed significantly, f (1, 19) = 13.00, p < .05 ECBI Frequency – pre and post means did not differ significantly, 29 Doctorate in Educational and Child Psychology involved in recruitment or assessment. Odd numbers were assigned to treatment group, even numbers were assigned to the waitlist control group. Excluded ASD, major sensory impairment, suspected child abuse cases 67% Caucasian, 17% African American, 13% biracial, and 3% Hispanic,with a mean Hollingshead (1975) score of 41.30 (14.14) Leung, C., Tsang, S., Heung, K., and Yiu, I. (2009). 53 (treatment) + 77 (control) Ages 2- 8 To compare the effectiveness of PCIT on Chinese families with children displaying significant externalising behaviour problems, with a waitlist control group. Referred by hospitals, social service agencies, preschools, or primary schools, or who were selfreferred because of concerns about their child’s behaviour. Participants had to Abigail Miranda Treatment sessions were conducted once per week and lasted approximately 1 hr. . Administered by PCIT trained social workers, in Cantonese who adhered to CAARE manual guidelines. Sessions were implemented until parents achieved ‘mastery’ over their Quasiexperimental design QuasiExperimental: Untreated control group design with dependent pretest and posttest samples Post hoc tests to control for participant differences. F (1, 19) = 2.68, p < .05. These results suggest that PCIT was effective in improving child compliance and intensity of problem behaviours, but not the number of problem behaviours. Eyberg Child Behavior Inventory (ECBI; Eyberg and Pincus, 1999). ANCOVA used to analyse data. Bonferroni correction applied. Dyadic Parent– Child Interaction Coding System: Abbreviated Version ECBI Intensity – pre and post means differed significantly F(1, 106) = 60.90, p < .001 ECBI Problem– pre and post means differed significantly F(1, 106) = 70.55, p < .001 30 Doctorate in Educational and Child Psychology reach the clinical cut-off point on the ECBI, which was adjusted in this case to match the cultural norms of the target population. Abigail Miranda skills, and the child’s behaviour fell within the normal range on parent self-report methods. Compliance F(2, 26) = 31.84, p < .05 These results suggest that PCIT was effective in improving child compliance, intensity and number of problem behaviours. Significant between group differencecs in gender, χ2(1, N = 110) = 4.44, p < .05, with more females in the intervention than control group. McCabe, K., and Yeh, M. (2009). USA: 58 3-7 year olds. To compare the effectiveness of PCIT, a culturally adapted version of PCIT and treatment as usual, for a sample of Mexican American children with clinically significant behavioural problems. PCIT – N = 19 (mean age = 48.9 months) Control – N = 18 (mean age = 55.1 months) Participants had to reach the clinical cut-off point on the ECBI. Groups did not PCIT – Therapists actively coached parents using detailed session checklists outlined in the PCIT treatment manual. Therapists were allowed an unlimited number of sessions. Treatment as usual – parents were assigned to therapists who Randomised controlled design. Pre and post data collection Alternative intervention control group compared to intervention group Child Behavior Checklist for 11 2 to 5 Year Olds (CBCL; Achenbach and Rescorla, 2000). Eyberg Child Behavior Inventory (ECBI; Eyberg and Pincus, 1999). Dyadic Parent– Child Interaction Coding System (DPICS; Eyberg, Effect sizes were calculated for pre and post measures for each condition, however this was a within groups comparison rather than a between groups comparison. See coding protocol for calculated effect sizes between groups. Tukey post hoc 31 Doctorate in Educational and Child Psychology differ significantly in their demographic characteristics. PCIT M= 13.42, (8.03); Control M=10.94, 10.01), F(2, 57)=62, p<.50 were not trained in PCIT who used approaches described as ‘person centred cognitive behavioural,’ ‘trauma focussed cognitive behavioural,’ and ‘family systems.’ Therapists were allowed freedom in the approach that they used. Therapists were allowed an unlimited number of sessions. Bagner, D. M., Graziano, P. A., Jaccard, J., Sheinkopf, S. J., Vohr, B. R., and Lester, B. M (2012) 28 mothers and their 20- to 60monthold child who was born < 37 weeks gestation. PCIT – one session per week, one hour in length. To compare the effectiveness of PCIT on families with children displaying significant externalising 82% White, 10% Biracial, 4% African American, and 4% Asian, and 21% of children were Hispanic. Sessions were videotaped, and 50% were randomly selected and coded by a separate therapist for integrity and fidelity. Abigail Miranda Randomised controlled design. Waitlist control group as a base for comparison. Two computergenerated random numbers lists, one for boys and one for girls, were maintained by a statistician Nelson, Duke and Boggs, 2004). tests were applied. Eyberg Child Behavior Inventory (ECBI; Eyberg and Pincus, 1999). Intensity scale Multiple regression analysis carried out. ECBI intensity scores suggest a significant difference within groups at pre and post treatment t =14 7.01, p <.01 32 Doctorate in Educational and Child Psychology behaviour problems who were born premature, with a waitlist control group. Hollingshead (1975) score M= 43.39 (13.21), middle range of socio economic status Abigail Miranda uninvolved in recruitment, intervention delivery, and data collection PCIT – n = 14 Age 39.7 months (14.2) WL – n = 14 Age 36.5 months (13.0) Referred by neonatal clinic director, health professionals or self referred. For study inclusion, mothers had to rate their child above the clinically significant range on the externalizing problems scale of the Child Behavior Checklist (CBCL), and be able to speak and understand English Excluded from study if a major 33 Doctorate in Educational and Child Psychology Abigail Miranda sensory impairment was present, significant motor impairments, oxygen dependence for chronic lung disease, and autism. Bagner, D. M., Sheinkopf, S. J., Vohr, B. R., and Lester, B. M (2010) To compare the effectiveness of PCIT on families with children displaying significant externalising behaviour problems who were born premature, with a waitlist control group. 28 mothers and their 18- to 60monthold child who was born < 37 weeks gestation. PCIT – n = 14 Age 39.7 months (14.2) WL – n = 14 Age 36.5 months (13.0) Referred by neonatal clinic director, health professionals or self-referred. For study inclusion, mothers had to rate their child above the clinically significant range (T score _60) on the PCIT – one session per week, one hour in length. 5 Child Directed Intervention sessions Received treatment until the child achieved within ½ SD on the ECBI Randomised controlled design. Waitlist control group as a base for comparison. Two computergenerated random numbers lists, one for boys and one for girls, were maintained by a statistician uninvolved in recruitment, intervention delivery, and data collection Child Behavior Checklist for 11 2 to 5 Year Olds (CBCL; Achenbach and Rescorla, 2000). Eyberg Child Behavior Inventory (ECBI; Eyberg and Pincus, 1999). Dyadic Parent– Child Interaction Coding System (DPICS; Eyberg, Nelson, Duke and Boggs, 2004). Cohen’s d effect sizes were calculated. Results indicated significant improvements on the ECBI intensity and frequency scales, as well as Child compliance (see coding protocol for more details). 34 Doctorate in Educational and Child Psychology Abigail Miranda externalizing problems scale of the Child Behavior Checklist (CBCL) 35 Doctorate in Educational and Child Psychology Appendix 3 Coding Protocols 1- 5 [adapted from Task Force on Evidence-Based Interventions in School Psychology, American Psychology Association, Kratochwill, T.R. (2003)] Coding Protocol Name of Coder:___AM______________________ Date:__01.02.14__ Full Study Reference in proper format Bagner, D. M., & Eyberg, S. M. (2007). Parent– child interaction therapy for disruptive behavior in children with mental retardation: A randomized controlled trial. Journal of Clinical Child and Adolescent Psychology, 36(3), 418-429. Intervention Name (description of study):_Parent Child Interaction Therapy (PCIT)____ Study ID Number:__01______ Type of Publication: Book/Monograph Journal Article Book Chapter Other (specify): 1.General Characteristics A. General Design Characteristics A1. Random assignment designs (if random assignment design, select one of the following) Completely randomized design Randomized block design (between participants, e.g., matched classrooms) Randomized block design (within participants) – blocked on sex Randomized hierarchical design (nested treatments A2. Nonrandomized designs (if non-random assignment design, select one of the following) Nonrandomized design Nonrandomized block design (between participants) Nonrandomized block design (within participants) Nonrandomized hierarchical design Optional coding for Quasi-experimental designs A3. Overall confidence of judgment on how participants were assigned (select on of the following) Very low (little basis) Low (guess) Moderate (weak inference) High (strong inference) Very high (explicitly stated) N/A Unknown/unable to code B Participants 36 Doctorate in Educational and Child Psychology Total size of sample (start of study): _30__ Intervention group sample size:__15__ Control group sample size:__15__ C. Type of Program Universal prevention program Selective prevention program Targeted prevention program Intervention/Treatment Unknown D. Stage of Program Model/demonstration programs Early stage programs Established/institutionalized programs Unknown E. Concurrent or Historical Intervention Exposure Current exposure Prior exposure Unknown Section 2 Key Features for Coding Studies and Rating Level of Evidence/Support A Measurement (Estimating the quality of the measures used to establish effects) (Rating Scale: 3= Strong Evidence, 2=Promising Evidence, 1=Weak Evidence, 0=No Evidence) A1 The use of the outcome measures produce reliable scores for the majority of the primary outcomes (see following table for a detailed breakdown on the outcomes) Yes No Unknown/unable to code A2 Multi-method (at least two assessment methods used) Yes No N/A Unknown/unable to code A3 Multi-source (at least two sources used self-reports, teachers etc.) Yes parent report and observation No N/A Unknown/unable to code A4 Validity of measures reported (well-known or standardized or norm-referenced are considered good, consider any cultural considerations) Yes validated with specific target group In part, validated for general population only No Unknown/unable to code 37 Doctorate in Educational and Child Psychology Overall Rating of Measurement: 3 2 1 0 B Comparison Group B1 Type of Comparison group Typical intervention Attention placebo Intervention element placebo Alternative intervention Pharmacotherapy No intervention Wait list/delayed intervention Minimal contact Unable to identify type of comparison B2 Overall rating of judgment of type of comparison group Very low Low Moderate High Very high – explicitly stated in text Unable to identify comparison group B3 Counterbalancing of change agent (participants who receive intervention from a single therapist/teacher etc were counter-balanced across intervention) By change agent Statistical (analyse includes a test for intervention) Other Not reported/None B4 Group equivalence established Random assignment Posthoc matched set Statistical matching Post hoc test for group equivalence – Levene’s test. This was not significant for all but one subtest, indicating that differences in results between groups were due to the non-homogeneity of the groups. B5 Equivalent mortality Low attrition (less than 20 % for post) Low attrition (less than 30% for follow-up) Intent to intervene analysis carried out? Findings:_Attrition was 33% in the treatment group, 20% for WL group (not significant difference between groups). Intent to treat analysis showed significant differences in the ECBI scores of IT & WL groups, F(1, 29) = 5.79, p = .023, d = .67. This showed that disruptive behaviour scores were lower in the IT group._ Overall Level of Evidence _2_ 3= Strong Evidence 2=Promising Evidence 1=Weak Evidence 0=No Evidence 38 Doctorate in Educational and Child Psychology C Primary/Secondary Outcomes – Complete the Outcome Tables before completing this section C1 Evidence of appropriate statistical analysis for Primary Outcomes Appropriate unit of analysis- ANCOVA Familywise/expermenter wise error rate controlled when applicable Sufficiently large N – to get an effect size of <.97, a sample size n=14 per condition is needed. Calculated using Gpower. C2 Percentage of Primary Outcomes that are significant Proportion of significant primary outcomes out of the total primary outcome measures for each key construct. at least 75% 50-74% 25%-49% less than 25% C3 Evidence of appropriate statistical analysis for Secondary Outcomes Appropriate unit of analysis Familywise/expermenter wise error rate controlled when applicable Sufficiently large N C4 Percentage of Secondary Outcomes that are significant Proportion of significant primary outcomes out of the total secondary outcome measures for each key construct. at least 75% 50-74% 25%-49% less than 25% 39 Doctorate in Educational and Child Psychology Significant Outcomes Outcome 1 Externalising behaviours Outcome 2 Frequency of problem behaviours Outcome 3 Child compliance during parent-child social interaction Primary vs Secondary Primary Secondary Unknown Primary Secondary Unknown Primary Secondary Unknown Who Changed What Changed Source Child Teacher Parent/Sig.A Ecology Other Unknown Behaviour Attitude Knowledge Other Unknown Self Report Parent Report Teacher Report Observation Test Other Unknown Child Teacher Parent/Sig.A Ecology Other Unknown Behaviour Attitude Knowledge Other Unknown Self Report Parent Report Teacher Report Observation Test Other Unknown Child Teacher Parent/Sig.A Ecology Other Unknown Behaviour Attitude Knowledge Other Unknown Self Report Parent Report Teacher Report Observation Test Other Unknown Treatment Information Intervention vs. waitlist control Intervention vs. waitlist control Intervention vs. waitlist control Outcome Measure Used Reliability Effect Size Child Behavior Checklist (CBCL) Test- retest .87- .90 Internal consistency .82 (externalising) and .95 (total) Test- retest .80 - .85 (12 weeks) & .75 .75 (10 months) Cohen’s d = .97, p < 0.05 Eyberg Child Behavior Inventory (ECBI) Intensity scale Dyadic Parent Child Interaction Coding System (DPICS) Internal consistency .90 Inter rater reliability kappa .55- .89. Effect size data taken from the study. Cohen’s d = 1.50, p < 0.05 Effect size data taken from the study. Cohen’s d = 1.53, p >. 0.05 Effect size data taken from the study. 40 Doctorate in Educational and Child Psychology Non-Significant Outcomes Outcome 1 Number of problem behaviours Primary vs Secondary Primary Secondary Unknown Who was targeted for change Child Teacher Parent/Sig.A Ecology Other Unknown What was targeted for change Behaviour Attitude Knowledge Other Unknown Source Self Report Parent Report Teacher Report Observation Test Other Unknown Note null/ negative outcomes Outcome Measure Used Reliability Effect Size Cohen’s d = -.66, p = 0.12 Eyberg Child Behavior Inventory (ECBI) Problem scale Test- retest .80 - .85 (12 weeks) & .75 - .75 (10 months) Cohen’s d = -.66, p > 0.05 There was no significant difference between the Intervention & control groups for the number of problem behaviours, post intervention Internal consistenc y .91 Effect size data taken from the study. 41 Doctorate in Educational and Child Psychology [adapted from Task Force on Evidence-Based Interventions in School Psychology, American Psychology Association, Kratochwill, T.R. (2003)] Coding Protocol Name of Coder:___AM______________ Date:__01.02.14__ Full Study Reference in proper format: Leung, C., Tsang, S., Heung, K., & Yiu, I. (2009). Effectiveness of Parent—Child interaction therapy (PCIT) among Chinese families. Research on Social Work Practice, 19(3), 304-313. Intervention Name (description of study):__Parent Child Interaction Therapy (PCIT)____ Study ID Number:__02____ Type of Publication: Book/Monograph Journal Article Book Chapter Other (specify): 1.General Characteristics A. General Design Characteristics A1. Random assignment designs (if random assignment design, select one of the following) Completely randomized Randomized block design (between participants, e.g., matched classrooms) Randomized block design (within participants) Randomized hierarchical design (nested treatments A2. Nonrandomized designs (if non-random assignment design, select one of the following) Nonrandomized design Nonrandomized block design (between participants) – non randomised control trial as groups were recruited from different sources Nonrandomized block design (within participants) Nonrandomized hierarchical design Optional coding for Quasi-experimental designs A3. Overall confidence of judgment on how participants were assigned (select on of the following) Very low (little basis) Low (guess) Moderate (weak inference) High (strong inference) Very high (explicitly stated) N/A Unknown/unable to code B Participants Total size of sample (start of study): _130_ Intervention group sample size:_48_ Control group sample size:__62__ 42 Doctorate in Educational and Child Psychology C. Type of Program Universal prevention program Selective prevention program Targeted prevention program Intervention/Treatment Unknown D. Stage of Program Model/demonstration programs Early stage programs Established/institutionalized programs Unknown E. Concurrent or Historical Intervention Exposure Current exposure Prior exposure Unknown Section 2 Key Features for Coding Studies and Rating Level of Evidence/Support A Measurement (Estimating the quality of the measures used to establish effects) (Rating Scale: 3= Strong Evidence, 2=Promising Evidence, 1=Weak Evidence, 0=No Evidence) A1 The use of the outcome measures produce reliable scores for the majority of the primary outcomes (see following table for a detailed breakdown on the outcomes) Yes – however, no data on reliability for DPICS noted. Therefore, reliability = .94 + .93/ 3 = .62 No Unknown/unable to code A2 Multi-method (at least two assessment methods used) Yes No N/A Unknown/unable to code A3 Multi-source (at least two sources used self-reports, teachers etc.) Yes parent report and observation No N/A Unknown/unable to code A4 Validity of measures reported (well-known or standardized or norm-referenced are considered good, consider any cultural considerations) Yes validated with specific target group – validated in a previous study by the authors on the target population In part, validated for general population only No Unknown/unable to code Overall Rating of Measurement: 3 2 1 0 43 Doctorate in Educational and Child Psychology B Comparison Group B1 Type of Comparison group Typical intervention Attention placebo Intervention element placebo Alternative intervention Pharmacotherapy No intervention Wait list/delayed intervention Minimal contact Unable to identify type of comparison B2 Overall rating of judgment of type of comparison group Very low Low Moderate High Very high – explicitly stated in text Unable to identify comparison group B3 Counterbalancing of change agent (participants who receive intervention from a single therapist/teacher etc were counter-balanced across intervention) By change agent Statistical (analyse includes a test for intervention) Other Not reported/None B4 Group equivalence established Random assignment Posthoc matched set Statistical matching Post hoc test for group equivalence B5 Equivalent mortality Low attrition (less than 20 % for post) – 15.38% attrition overall Low attrition (less than 30% for follow-up) Intent to intervene analysis carried out? Findings_____________ Overall Level of Evidence _2_ 3= Strong Evidence 2=Promising Evidence 1=Weak Evidence 0=No Evidence 44 Doctorate in Educational and Child Psychology C Primary/Secondary Outcomes – Complete the Outcome Tables before completing this section C1 Evidence of appropriate statistical analysis for Primary Outcomes Appropriate unit of analysis Familywise/expermenter wise error rate controlled when applicable – Bonferroni correction Sufficiently large N – required 40 participants overall to get an effect size of <.49. Calculated using Gpower C2 Percentage of Primary Outcomes that are significant Proportion of significant primary outcomes out of the total primary outcome measures for each key construct. at least 75% 50-74% 25%-49% less than 25% C3 Evidence of appropriate statistical analysis for Secondary Outcomes Appropriate unit of analysis Familywise/expermenter wise error rate controlled when applicable Sufficiently large N C4 Percentage of Secondary Outcomes that are significant Proportion of significant primary outcomes out of the total secondary outcome measures for each key construct. at least 75% 50-74% 25%-49% less than 25% 45 Doctorate in Educational and Child Psychology Significant Outcomes Outcome 1 Frequency of problem behaviours Outcome 2 Number of problem behaviours Outcome 3 Child compliance during parent-child social interaction Primary vs Secondary Primary Secondary Unknown Primary Secondary Unknown Primary Secondary Unknown Who Changed What Changed Source Treatment Information Intervention vs. waitlist control Child Teacher Parent/Sig.A Ecology Other Unknown Behaviour Attitude Knowledge Other Unknown Self Report Parent Report Teacher Report Observation Test Other Unknown Child Teacher Parent/Sig.A Ecology Other Unknown Behaviour Attitude Knowledge Other Unknown Self Report Parent Report Teacher Report Observation Test Other Unknown Intervention vs. waitlist control Child Teacher Parent/Sig.A Ecology Other Unknown Behaviour Attitude Knowledge Other Unknown Self Report Parent Report Teacher Report Observation Test Other Unknown Intervention vs. waitlist control Outcome Measure Used Eyberg Child Behavior Inventory (ECBI) Intensity scale Reliability Effect Size Internal consistency .94 Cohen’s d = 1.59, CI 1.15 1.71 Eyberg Child Behavior Inventory (ECBI) Problem scale Internal consistency .93 Dyadic Parent Child Interaction Coding System (DPICS) Inter-rater reliability not noted Effect size data taken from the study. Cohen’s d = 1.52, CI 1.08 1.43 Effect size data taken from the study. Partial etasquared = .49, p < 0.05 Effect size data taken from the study. Data for control group was unavailable to calculate Cohen’s d effect size. 46 Doctorate in Educational and Child Psychology [adapted from Task Force on Evidence-Based Interventions in School Psychology, American Psychology Association, Kratochwill, T.R. (2003)] Coding Protocol Name of Coder:____AM________ Date:__01.02.14____ Full Study Reference in proper format: McCabe, K., & Yeh, M. (2009). Parent–child interaction therapy for Mexican Americans: A randomized clinical trial. Journal of Clinical Child & Adolescent Psychology, 38(5), 753-759. Intervention Name (description of study):___ Parent Child Interaction Therapy (PCIT)__ Study ID Number:__03__ Type of Publication: Book/Monograph Journal Article Book Chapter Other (specify): 1.General Characteristics A. General Design Characteristics A1. Random assignment designs (if random assignment design, select one of the following) Completely randomized design Randomized block design (between participants, e.g., matched classrooms) Randomized block design (within participants) Randomized hierarchical design (nested treatments A2. Nonrandomized designs (if non-random assignment design, select one of the following) Nonrandomized design Nonrandomized block design (between participants) Nonrandomized block design (within participants) Nonrandomized hierarchical design Optional coding for Quasi-experimental designs A3. Overall confidence of judgment on how participants were assigned (select on of the following) Very low (little basis) – text says ‘randomly assigned’ but doesn’t have further information on how this assignment was carried out Low (guess) Moderate (weak inference) High (strong inference) Very high (explicitly stated) N/A Unknown/unable to code B Participants Total size of sample (start of study): _58_ Intervention group sample size:_19__ 47 Doctorate in Educational and Child Psychology Control group sample size:_18__ C. Type of Program Universal prevention program Selective prevention program Targeted prevention program Intervention/Treatment Unknown D. Stage of Program Model/demonstration programs Early stage programs Established/institutionalized programs Unknown E. Concurrent or Historical Intervention Exposure Current exposure Prior exposure Unknown Section 2 Key Features for Coding Studies and Rating Level of Evidence/Support A Measurement (Estimating the quality of the measures used to establish effects) (Rating Scale: 3= Strong Evidence, 2=Promising Evidence, 1=Weak Evidence, 0=No Evidence) A1 The use of the outcome measures produce reliable scores for the majority of the primary outcomes (see following table for a detailed breakdown on the outcomes) Yes No Unknown/unable to code A2 Multi-method (at least two assessment methods used) Yes No N/A Unknown/unable to code A3 Multi-source (at least two sources used self-reports, teachers etc.) Yes No N/A Unknown/unable to code A4 Validity of measures reported (well-known or standardized or norm-referenced are considered good, consider any cultural considerations) Yes validated with specific target group – (Eyberg & Pincus, 1999; Garcia-Tornel et al., 1998) In part, validated for general population only No Unknown/unable to code Overall Rating of Measurement: 3 2 1 0 48 Doctorate in Educational and Child Psychology B Comparison Group B1 Type of Comparison group Typical intervention Attention placebo Intervention element placebo Alternative intervention Pharmacotherapy No intervention Wait list/delayed intervention Minimal contact Unable to identify type of comparison B2 Overall rating of judgment of type of comparison group Very low Low Moderate High Very high – explicitly stated in text Unable to identify comparison group B3 Counterbalancing of change agent (participants who receive intervention from a single therapist/teacher etc were counter-balanced across intervention) By change agent Statistical (analyse includes a test for intervention) Other Not reported/None B4 Group equivalence established Random assignment Posthoc matched set Statistical matching Post hoc test for group equivalence B5 Equivalent mortality Low attrition (less than 20 % for post) Low attrition (less than 30% for follow-up) Intent to intervene analysis carried out? Findings_no differences between groups apart from PPQ scores, non-significant after Familywise error rate control (Bonferroni correction)__ Overall Level of Evidence _2_ 3= Strong Evidence 2=Promising Evidence 1=Weak Evidence 0=No Evidence 49 Doctorate in Educational and Child Psychology C Primary/Secondary Outcomes – Complete the Outcome Tables before completing this section C1 Evidence of appropriate statistical analysis for Primary Outcomes Appropriate unit of analysis – p values were not reported. Within group effect sizes were calculated in the paper, rather than between groups which would have been more relevant to the hypotheses. Familywise/expermenter wise error rate controlled when applicable Sufficiently large N C2 Percentage of Primary Outcomes that are significant Proportion of significant primary outcomes out of the total primary outcome measures for each key construct. at least 75% 50-74% 25%-49% - however the effect is negative less than 25% C3 Evidence of appropriate statistical analysis for Secondary Outcomes Appropriate unit of analysis Familywise/expermenter wise error rate controlled when applicable Sufficiently large N C4 Percentage of Secondary Outcomes that are significant Proportion of significant primary outcomes out of the total secondary outcome measures for each key construct. at least 75% 50-74% 25%-49% less than 25% 50 Doctorate in Educational and Child Psychology Significant Outcomes Outcome 1 Externalising behaviours Primary vs Secondary Primary Secondary Unknown Who Changed Child Teacher Parent/Sig.A Ecology Other Unknown What Changed Behaviour Attitude Knowledge Other Unknown Source Self Report Parent Report Teacher Report Observation Test Other Unknown Treatment Information Intervention vs. waitlist control Outcome Measure Used Child Behavior Checklist (CBCL) Reliability Effect Size Internal consistency .91 - .94 Cohen’s d = -.79, p > 0.05 Effect size calculation= between groups mean difference/poo led standard deviation 51 Doctorate in Educational and Child Psychology Non-Significant Outcomes Outcome 2 Frequency of problem behaviours Outcome 3 Number of problem behaviours Outcome 4 Child compliance during cleanup Primary vs Secondary Primary Secondary Unknown Primary Secondary Unknown Primary Secondary Unknown Who Changed What Changed Source Treatment Information Intervention vs. waitlist control Child Teacher Parent/Sig.A Ecology Other Unknown Behaviour Attitude Knowledge Other Unknown Self Report Parent Report Teacher Report Observation Test Other Unknown Child Teacher Parent/Sig.A Ecology Other Unknown Behaviour Attitude Knowledge Other Unknown Self Report Parent Report Teacher Report Observation Test Other Unknown Intervention vs. waitlist control Child Teacher Parent/Sig.A Ecology Other Unknown Behaviour Attitude Knowledge Other Unknown Self Report Parent Report Teacher Report Observation Test Other Unknown Intervention vs. waitlist control Outcome Measure Used Eyberg Child Behavior Inventory (ECBI) Intensity scale Reliability Effect Size Internal consistency .86 Cohen’s d = -.49, p < 0.05 Eyberg Child Behavior Inventory (ECBI) Problem scale Internal consistency .80 Dyadic Parent Child Interaction Coding System (DPICS) Internal consistency Kappa coefficients (pretreatment) .59 - .85 Effect size calculation= between groups mean difference/poo led standard deviation Cohen’s d = -.36, p < 0.05 Effect size calculation= between groups mean difference/poo led standard deviation Cohen’s d = .35, p < 0.05 Effect size calculation= between groups mean difference/poo led standard deviation 52 Doctorate in Educational and Child Psychology [adapted from Task Force on Evidence-Based Interventions in School Psychology, American Psychology Association, Kratochwill, T.R. (2003)] Coding Protocol Name of Coder:__AM__ Date:__01.02.14__ Full Study Reference in proper format: Bagner, D. M., Graziano, P. A., Jaccard, J., Sheinkopf, S. J., Vohr, B. R., & Lester, B. M. (2012). An initial investigation of baseline respiratory sinus arrhythmia as a moderator of treatment outcome for young children born premature with externalizing behavior problems. Behavior therapy, 43(3), 652-665. Intervention Name (description of study):_ Parent Child Interaction Therapy (PCIT)____ Study ID Number:__04___ Type of Publication: Book/Monograph Journal Article Book Chapter Other (specify): 1.General Characteristics A. General Design Characteristics A1. Random assignment designs (if random assignment design, select one of the following) Completely randomized design Randomized block design (between participants, e.g., matched classrooms) Randomized block design (within participants) Randomized hierarchical design (nested treatments A2. Nonrandomized designs (if non-random assignment design, select one of the following) Nonrandomized design Nonrandomized block design (between participants) Nonrandomized block design (within participants) Nonrandomized hierarchical design Optional coding for Quasi-experimental designs A3. Overall confidence of judgment on how participants were assigned (select on of the following) Very low (little basis) Low (guess) Moderate (weak inference) High (strong inference) Very high (explicitly stated) N/A Unknown/unable to code B Participants Total size of sample (start of study): _28__ Intervention group sample size:_14_ 53 Doctorate in Educational and Child Psychology Control group sample size:_14_ C. Type of Program Universal prevention program Selective prevention program Targeted prevention program Intervention/Treatment Unknown D. Stage of Program Model/demonstration programs Early stage programs Established/institutionalized programs Unknown E. Concurrent or Historical Intervention Exposure Current exposure Prior exposure Unknown Section 2 Key Features for Coding Studies and Rating Level of Evidence/Support A Measurement (Estimating the quality of the measures used to establish effects) (Rating Scale: 3= Strong Evidence, 2=Promising Evidence, 1=Weak Evidence, 0=No Evidence) A1 The use of the outcome measures produce reliable scores for the majority of the primary outcomes (see following table for a detailed breakdown on the outcomes) Yes No Unknown/unable to code A2 Multi-method (at least two assessment methods used) Yes No N/A Unknown/unable to code A3 Multi-source (at least two sources used self-reports, teachers etc.) Yes No N/A Unknown/unable to code A4 Validity of measures reported (well-known or standardized or norm-referenced are considered good, consider any cultural considerations) Yes validated with specific target group In part, validated for general population only No Unknown/unable to code Overall Rating of Measurement: 3 2 1 0 54 Doctorate in Educational and Child Psychology B Comparison Group B1 Type of Comparison group Typical intervention Attention placebo Intervention element placebo Alternative intervention Pharmacotherapy No intervention Wait list/delayed intervention Minimal contact Unable to identify type of comparison B2 Overall rating of judgment of type of comparison group Very low Low Moderate High Very high – explicitly stated in text Unable to identify comparison group B3 Counterbalancing of change agent (participants who receive intervention from a single therapist/teacher etc were counter-balanced across intervention) By change agent Statistical (analyse includes a test for intervention) Other Not reported/None B4 Group equivalence established Random assignment Posthoc matched set Statistical matching Post hoc test for group equivalence B5 Equivalent mortality Low attrition (less than 20 % for post) Low attrition (less than 30% for follow-up) Intent to intervene analysis carried out? Findings__ No significant difference between Time 1 ECBI Intensity score, t(26)=0.43, p=.668.__ Overall Level of Evidence _2_ 3= Strong Evidence 2=Promising Evidence 1=Weak Evidence 0=No Evidence 55 Doctorate in Educational and Child Psychology C Primary/Secondary Outcomes – Complete the Outcome Tables before completing this section C1 Evidence of appropriate statistical analysis for Primary Outcomes Appropriate unit of analysis Familywise/expermenter wise error rate controlled when applicable Sufficiently large N - minimum sample of 9 per group was sufficient to detect expected differences. C2 Percentage of Primary Outcomes that are significant Proportion of significant primary outcomes out of the total primary outcome measures for each key construct. at least 75% 50-74% 25%-49% less than 25% C3 Evidence of appropriate statistical analysis for Secondary Outcomes Appropriate unit of analysis Familywise/expermenter wise error rate controlled when applicable Sufficiently large N C4 Percentage of Secondary Outcomes that are significant Proportion of significant primary outcomes out of the total secondary outcome measures for each key construct. at least 75% 50-74% 25%-49% less than 25% 56 Doctorate in Educational and Child Psychology Significant Outcomes Outcome 1 Frequency of problem behaviours Primary vs Secondary Primary Secondary Unknown Who Changed Child Teacher Parent/Sig.A Ecology Other Unknown What Changed Behaviour Attitude Knowledge Other Unknown Source Self Report Parent Report Teacher Report Observation Test Other Unknown Treatment Information Intervention vs. waitlist control Outcome Measure Used Reliability Effect Size Eyberg Child Behavior Inventory (ECBI) Intensity scale Test-retest reliability coefficient of .80 (12 weeks) and .75 (10 months) Cohen’s d = -1.03, CI -1.87 - -0.19 Internal consistency coefficient of .95 Effect size calculation= between groups mean difference/poo led standard deviation 57 Doctorate in Educational and Child Psychology [adapted from Task Force on Evidence-Based Interventions in School Psychology, American Psychology Association, Kratochwill, T.R. (2003)] Coding Protocol Name of Coder:__AM____ Date:__01.02.14__ Bagner, D. M., Sheinkopf, S. J., Vohr, B. R., & Lester, B. M. (2010). Parenting intervention for externalizing behavior problems in children born premature: an initial examination. Journal of developmental and behavioral pediatrics: JDBP, 31(3), 209. Full Study Reference in proper format:__ Intervention Name (description of study):_ Parent Child Interaction Therapy (PCIT)___ Study ID Number:__05__ Type of Publication: Book/Monograph Journal Article Book Chapter Other (specify): 1.General Characteristics A. General Design Characteristics A1. Random assignment designs (if random assignment design, select one of the following) Completely randomized design Randomized block design (between participants, e.g., matched classrooms) Randomized block design (within participants) Randomized hierarchical design (nested treatments A2. Nonrandomized designs (if non-random assignment design, select one of the following) Nonrandomized design Nonrandomized block design (between participants) Nonrandomized block design (within participants) Nonrandomized hierarchical design Optional coding for Quasi-experimental designs A3. Overall confidence of judgment on how participants were assigned (select on of the following) Very low (little basis) Low (guess) Moderate (weak inference) High (strong inference) Very high (explicitly stated) N/A Unknown/unable to code B Participants Total size of sample (start of study): _28_ Intervention group sample size:__11_ Control group sample size:__14__ 58 Doctorate in Educational and Child Psychology C. Type of Program Universal prevention program Selective prevention program Targeted prevention program Intervention/Treatment Unknown D. Stage of Program Model/demonstration programs Early stage programs Established/institutionalized programs Unknown E. Concurrent or Historical Intervention Exposure Current exposure Prior exposure Unknown Section 2 Key Features for Coding Studies and Rating Level of Evidence/Support A Measurement (Estimating the quality of the measures used to establish effects) (Rating Scale: 3= Strong Evidence, 2=Promising Evidence, 1=Weak Evidence, 0=No Evidence) A1 The use of the outcome measures produce reliable scores for the majority of the primary outcomes (see following table for a detailed breakdown on the outcomes) Yes - .78 when reliability scores were averaged No Unknown/unable to code A2 Multi-method (at least two assessment methods used) Yes No N/A Unknown/unable to code A3 Multi-source (at least two sources used self-reports, teachers etc.) Yes No N/A Unknown/unable to code A4 Validity of measures reported (well-known or standardized or norm-referenced are considered good, consider any cultural considerations) Yes validated with specific target group In part, validated for general population only No Unknown/unable to code Overall Rating of Measurement: 3 2 1 0 59 Doctorate in Educational and Child Psychology B Comparison Group B1 Type of Comparison group Typical intervention Attention placebo Intervention element placebo Alternative intervention Pharmacotherapy No intervention Wait list/delayed intervention Minimal contact Unable to identify type of comparison B2 Overall rating of judgment of type of comparison group Very low Low Moderate High Very high – explicitly stated Unable to identify comparison group B3 Counterbalancing of change agent (participants who receive intervention from a single therapist/teacher etc were counter-balanced across intervention) By change agent Statistical (analyse includes a test for intervention) Other Not reported/None B4 Group equivalence established Random assignment Posthoc matched set Statistical matching Post hoc test for group equivalence B5 Equivalent mortality Low attrition (less than 20 % for post) Low attrition (less than 30% for follow-up) Intent to intervene analysis carried out? Findings_ No significant differences at time 2, F(1,25) = 24.2, p =.000, d =1.8._ Overall Level of Evidence _2_ 3= Strong Evidence 2=Promising Evidence 1=Weak Evidence 0=No Evidence 60 Doctorate in Educational and Child Psychology C Primary/Secondary Outcomes – Complete the Outcome Tables before completing this section C1 Evidence of appropriate statistical analysis for Primary Outcomes Appropriate unit of analysis Familywise/expermenter wise error rate controlled when applicable Sufficiently large N – power calculation by the authors suggest that a minimum sample of 9 per group was sufficient to detect expected differences. C2 Percentage of Primary Outcomes that are significant Proportion of significant primary outcomes out of the total primary outcome measures for each key construct. at least 75% 50-74% 25%-49% less than 25% C3 Evidence of appropriate statistical analysis for Secondary Outcomes Appropriate unit of analysis Familywise/expermenter wise error rate controlled when applicable Sufficiently large N C4 Percentage of Secondary Outcomes that are significant Proportion of significant primary outcomes out of the total secondary outcome measures for each key construct. at least 75% 50-74% 25%-49% less than 25% 61 Doctorate in Educational and Child Psychology Significant Outcomes Outcome 1 Externalising behaviours Outcome 2 Frequency of problem behaviours Outcome 3 Number of problem behaviours Outcome 4 Child compliance during parent-child social interaction Primary vs Secondary Primary Secondary Unknown Primary Secondary Unknown Primary Secondary Unknown Primary Secondary Unknown Who Changed What Changed Source Treatment Information Intervention vs. waitlist control Child Teacher Parent/Sig.A Ecology Other Unknown Behaviour Attitude Knowledge Other Unknown Self Report Parent Report Teacher Report Observation Test Other Unknown Child Teacher Parent/Sig.A Ecology Other Unknown Behaviour Attitude Knowledge Other Unknown Self Report Parent Report Teacher Report Observation Test Other Unknown Intervention vs. waitlist control Child Teacher Parent/Sig.A Ecology Other Unknown Behaviour Attitude Knowledge Other Unknown Self Report Parent Report Teacher Report Observation Test Other Unknown Intervention vs. waitlist control Child Teacher Parent/Sig.A Ecology Other Unknown Behaviour Attitude Knowledge Other Unknown Self Report Parent Report Teacher Report Observation Test Other Unknown Outcome Measure Used Child Behavior Checklist (CBCL) Reliability Effect Size internal consistency .54 - .81. Cohen’s d = 2.3, p < 0.05 Eyberg Child Behavior Inventory (ECBI) Intensity scale internal consistency .90 Eyberg Child Behavior Inventory (ECBI) Problem scale internal consistency .85. Dyadic Parent Child Interaction Coding System (DPICS) Kappa .50 (noncomply) to 0.78 (praise) Effect size data taken from the study Cohen’s d = 2.3, p < 0.05 Effect size data taken from the study Cohen’s d = 1.4, p < 0.05 Effect size data taken from the study Cohen’s d = .90, p < 0.05 Effect size data taken from the study 62 Doctorate in Educational and Child Psychology Appendix 4 Weighting of Studies A: Methodological Quality The Kratochwill (2003) coding protocol was used to weight experimental and quasi-experimental designs; Studies were weighted on ‘Measures’, ‘Comparison Group’ and ‘Analysis’. Measures Weighting High Medium Low Description All measures used produce reliable scores of 0.85 for the population under study. Data collected using multiple methods AND from multiple sources. A case for validity must be presented. Study uses at least 75% of measures that produce reliable scores of 0.70 for the population under study. Data collected using either multiple methods AND/OR from multiple sources. A case for validity does not need to be presented. Study uses at least 50% of measures that produce reliable scores of 0.50 for the population under study. Data collected using either multiple methods AND/OR from multiple sources, however this is not a requirement. A case for validity does not need to be presented. Comparison Group Weighting Description Uses at least one type of ‘active’ comparison group. Establishes initial group equivalency. High Evidence of counterbalancing of change-agents and demonstrates equivalent mortality and low attrition at the post test. Medium Low Uses at least a ‘no intervention group’ comparison group. Evidence for at least two of: 1) Counterbalancing of change-agents, 2) Group equivalency, 3) Equivalent mortality with low attrition. Conducted an intent-to-intervene analysis if equivalent mortality is not demonstrated, with non-significant results Uses a comparison group. Evidence for at least one of: 1) Counterbalancing of changeagents, 2) Group equivalency, 3) Equivalent mortality with low 63 Doctorate in Educational and Child Psychology attrition. Conducted an intent-to-intervene analysis if equivalent mortality is not demonstrated, with non-significant results Analysis Weighting High Medium Low Description Appropriate statistical analysis conducted including: appropriate unit of analysis, familywise/ experimentwise error rate controlled (if applicable) and a sufficiently large N. Significant outcomes for at least 75% of the total primary outcome measures Three of the following: Appropriate statistical analysis conducted including: 1) appropriate unit of analysis, 2) familywise/ experimentwise error rate controlled (if applicable) and 3) a sufficiently large N. Significant outcomes for at least 50% of the total primary outcome measures Two of the following: Appropriate statistical analysis conducted including: 1) appropriate unit of analysis, 2) familywise/ experimentwise error rate controlled (if applicable) and 3) a sufficiently large N. Significant outcomes for at least 25% of the total primary outcome measures Overall Methodological Quality Overall methodological quality of studies was calculated by assigning scores of: - ‘3’ for ‘High’ weightings - ‘2’ for ‘Medium’ weightings - ‘1’ for ‘Low’ weightings Scores were then averaged. Overall Methodological Quality High Medium Low Average Scores At least 2.5 Between 1.5 and 2.4 Less than 1.4 64 Doctorate in Educational and Child Psychology Studies Measures Comparison Group Analysis Bagner and Eyberg, 2007 Weighting score Medium 2 Medium 2 High 3 Overall Quality of Methodology Medium 2.3 Leung et al., 2009 Weighting score Low 1 Medium 2 High 3 Medium 2 McCabe and Yeh, 2009 Weighting score Medium 2 Medium 2 Low 1 Medium 1.7 Bagner et al., 2012 Weighting score Low 1 Medium 2 Low 1 Low 1.3 Bagner et al., 2010 Weighting score Medium 2 Medium 2 Low 1 Medium 1.7 B: Relevance of Methodology This weighting reviews the appropriateness of the evidence for answering the review question. Weighting High Description Uses random assignment of participants in an attempt to establish group equivalence pre intervention. Uses an active comparison group to measure the efficacy of the intervention in comparison with an alternate intervention. Medium Uses random assignment of participants in an attempt to establish group equivalence pre intervention. Uses a waitlist control comparison group to compare the effects of the intervention with treatment as usual. Low Does not have to use random assignment of participants. Uses a comparison group. Demonstrates group equivalence. C: Relevance of Evidence to Review Question 65 Doctorate in Educational and Child Psychology This weighting is a review-specific judgement about the relevance of the focus of the evidence for the review question. Weighting High Medium Low Description Uses the PCIT intervention delivered by a trained therapist.. Uses children with externalising behaviour problems which are significant at clinical cut-offs on CBCL/ ECBI. Triangulates data using self- report and observation in order to assess observable and perceived behavioural changes. Uses the PCIT intervention delivered by a trained therapist... Uses children with externalising behaviour problems which are significant at clinical cut-offs on CBCL/ ECBI. Triangulates data using self- report and observation in order to assess observable and perceived behavioural changes. Uses the PCIT intervention delivered by a trained professional... Uses children with externalising behaviour problems which are not significant at clinical cut-offs on CBCL/ ECBI. Does not have to triangulate data using self- report and observation D: Overall Weight of Evidence This is an overall assessment of the extent to which the study contributes evidence to answer the review question. It is calculated by giving studies scores: - ‘3’ for ‘High’ weightings - ‘2’ for ‘Medium’ weightings - ‘1’ for ‘Low’ weightings These scores were then averaged to find the overall weight of evidence score. Overall Methodological Quality High Medium Low Average Scores At least 2.5 Between 1.5 and 2.4 Less than 1.4 66 Doctorate in Educational and Child Psychology Studies (C) Relevance (A) (B) of Quality of Relevance of evidence Methodology Methodology to the review question Medium Medium Medium 2.3 2 2 (D) Overall Weight of Evidence Bagner and Eyberg, 2007 Weighting score Medium 2.1 Leung et al., 2009 Weighting score Medium 2 Low 1 Low 1 Low 1.3 McCabe and Yeh, 2009 Weighting score Medium 1.7 High 3 High 3 High 2.6 Bagner et al., 2012 Weighting score Low 1.3 Medium 2 Low 1 Low 1.4 Bagner et al., 2010 Weighting score Medium 1.7 Medium 2 Medium 2 Medium 1.9 67