CHARACTERIZING PREDICTORS AND MODERATORS OF PARENT PARTICIPATION ENGAGEMENT IN EARLY AUTISM INTERVENTION By Hannah C. Tokish A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Psychology – Master of Arts 2024 ABSTRACT Parent engagement in early intervention supports child and family outcomes but is currently lacking in community service settings. The current study examined parent and provider factors that influence parent participation engagement (PPE), defined as active, independent, and responsive contribution to treatment, in early autism intervention in the Part C Early Intervention (EI) system, which provides publicly funded services for developmentally delayed children under three and emphasizes the use of parent coaching. While most studies have examined influences on proxies of PPE (e.g., attendance, retention, homework completion), the current study utilized objective observational assessment to examine parent and provider factors that predict PPE. Multilevel modeling was used to analyze PPE across 215 EI session recordings that involved 113 independent parents and 61 independent providers. Results suggested that a) single parents face greater barriers to PPE, b) provider quality of parent coaching may promote higher PPE, c) provider quality of parent coaching may not moderate the influence of single parent status on PPE, and d) the quality of specific parent coaching strategies influences PPE. Provider use of high-quality Reflection & Problem Solving (i.e., eliciting parent concerns and working through barriers to at-home intervention strategy use) predicted significantly higher PPE while provider use of high-quality Demonstration (i.e., explaining and demonstrating intervention techniques to parents) predicted lower PPE. Implications for increased provider training in particular parent coaching domains to promote PPE in early autism intervention are discussed. ACKNOWLEDGEMENTS I would like to sincerely thank my mentor, Dr. Brooke Ingersoll, without whom I could not have conducted this research. Thank you for helping me turn my interests and questions into a feasible research study, for digging into the data with me, and for advocating for me at every step of the way. Thank you to my amazing team of undergraduate research assistants, who coded over 250 intervention recordings for the current study and continue to code for future analyses: Emily Assemany, Alyssa Bonikowski, Claire Bongiorno, Olivia Bowman, and Grace Kim. I also appreciate the help of Naomi Alvarado and Jessie Greatorex, who volunteered their time to assist with data management, coding assignments, and many additional study tasks that arose. In addition, I am incredibly grateful to the larger RISE Research Network, including Dr. Alice Carter, Dr. Brooke Ingersoll, Dr. Wendy Stone, and Dr. Allie Wainer, for their support in conducting this research and for allowing me to incorporate the rich data that many RISE researchers have worked so hard to collect. Thank you to Lara Cunningham, Dr. Chris Sheldrick, Dr. Kyle Frost, and Anthuanet Espinel, who oriented me to the RISE data, assisted with data management, and provided valuable feedback along the way. In addition, thank you to Anamiguel Pomales Ramos, who provided valuable mentorship and support as a friend and more senior graduate student in the lab. Last but certainly not least, thank you to my wonderful family and friends, whose support throughout this process has truly meant the world to me. This work was funded by the Blue Cross Blue Shield of Michigan Foundation Student Award Program (2023020029.SAP) and the National Institute of Mental Health (NIMH) of the National Institutes of Health (NIH) by grant numbers R01MH122725-01, R01 MH122726-01, R01 MH122727-01, and R01 MH122728-01. iii TABLE OF CONTENTS INTRODUCTION ......................................................................................................................1 METHODS ............................................................................................................................... 10 RESULTS ................................................................................................................................. 21 DISCUSSION ........................................................................................................................... 31 CONCLUSION ......................................................................................................................... 41 REFERENCES ......................................................................................................................... 42 iv Parent Participation Engagement INTRODUCTION Autism is a neurodevelopmental disorder characterized by difficulties with social communication and interaction as well as restricted and repetitive behaviors, interests, and activities (American Psychological Association, 2022). Early and intensive intervention (i.e., targeting fundamental developmental skills in the first three years of life) is key to supporting developmental outcomes and quality of life for children with autism (Althoff et al., 2019; Eunice Kennedy Shriver National Institute of Child Health and Human Development, 2019). Increasingly, best practice guidelines for early intervention are focused on parenti engagement to support child outcomes and family well-being (Casagrande & Ingersoll, 2017). Parent participation engagement (PPE) is defined as a caregiver’s active, independent, and responsive contribution to treatment (Haine-Schlagel & Walsh, 2015). PPE is characterized by both in-session behaviors, such as asking questions, sharing perspectives, and participating in therapy activities, and between-session behaviors that involve following through on provider recommendations (Haine-Schlagel & Walsh, 2015; Guan et al., 2019). Meta-analyses suggest that parent engagement in intervention sessions is related to improvements in both parent and child outcomes as well as long-term retention of families in treatment (Dowell & Ogles, 2010, Guan et al., 2019). In child treatment, PPE is particularly important given the role that parents play in seeking treatment, facilitating attendance, following through with provider recommendations at home, and supporting efforts to adjust child behavior (Haine-Schlagel & Walsh, 2015). However, parent engagement is currently lacking in community service settings, i In this proposal, parent refers to the wide variety of caregiving roles such as biological parent, stepparent, adoptive parent, grandparent, foster parent, and other guardians. 1 including the state-funded early intervention (EI) system which provides care for toddlers on the autism spectrum (Haine-Schlagel & Walsh, 2015). Notably, much of the current research literature on parent engagement has focused on session attendance, homework completion, and therapist and parent-reported engagement rather than in-session PPE. A recent review of treatment engagement interventions found that attendance was the most common indicator of treatment engagement and was measured in over 90% of studies (Lakind et al., 2021). Research suggests that these measures are not truly equivalent to active engagement, which remains relatively understudied and not well-understood, particularly in autism populations (Becker et al., 2015). Few studies have examined PPE using more objective methods, such as observational coding schemes of parent engagement behaviors during sessions. To our knowledge, only one study has examined PPE in autism-specific intervention (Guan et al., 2019). Thus, observational assessment of PPE is needed to inform methods of promoting in-session engagement to improve outcomes for children on the autism spectrum and their families. Parent Characteristics and Parent Engagement The majority of research on predictors of parent engagement has focused on how child, parent, and family factors affect proxies including attendance, homework completion, and self- reported participation in child psychotherapy more broadly (Haine-Schlagel & Walsh, 2015). For example, single parent status has consistently been associated with reduced parent attendance (Haine-Schlagel & Walsh, 2015). In addition, parents with high stress levels, low sense of self- efficacy, and low motivation to participate in treatment often demonstrate lower attendance and participation (King et al., 2014; Kurzrok et al., 2021; Stadnick et al., 2016; Solish & Perry, 2008; Nock & Kazdin, 2005; Haine-Schlagel & Walsh, 2015). Two studies have utilized observational coding schemes to examine how parent motivation to participate in therapy influences PPE with 2 mixed findings (Stadnick et al., 2016; Haine-Schlagel et al., 2019). While both studies found that parent motivation as measured by the Parent Motivation Inventory (PMI) did not significantly predict PPE, Stadnick et al. (2016) found that the PMI ability to change subscale, which measures parent beliefs that they can adjust their parenting behaviors in treatment, was significantly associated with increased PPE. Furthermore, parents of minoritized and/or low socioeconomic status (SES) backgrounds often exhibit lower levels of attendance and reported participation than non-Latine White and high SES families due to factors including obstacles to healthcare access, lack of resources, cultural stigma, and language barriers (Haine-Schlagel & Walsh, 2015; Haine-Schlagel et al., 2022; Holly et al., 2019; Stadnick et al., 2016; Dickson et al., 2017; Guan et al., 2019; Tomczuk et al., 2022; Miller et al., 2008; Fawley-King et al., 2013; Garland et al., 2012; Pereira et al., 2016; Hansen & Warner, 1994). To our knowledge, three articles have reported ethnic disparities in PPE using the Parent Participation Engagement in Child Psychotherapy: Observational Coding Manual, in which coders rate videos of psychotherapy or intervention sessions according to the quality of specific behaviors indicating PPE. These articles found reduced levels of PPE for Hispanic/Latine parents compared to their non-Hispanic/Latine counterparts (Dickson et al., 2017; Guan et al., 2019; Stadnick et al., 2016). Guan et al. (2019) examined PPE in the context of a randomized controlled trial (RCT) of An Individualized Mental Health Intervention for Children with Autism Spectrum Disorder (AIM HI), a parent-mediated intervention for children between 5 and 13 years of age with autism and challenging behaviors, in publicly funded outpatient and school-based mental health programs in Southern California. In a sample of 39 parents randomly assigned to the intervention condition (51% English-speaking non-Latine White, 31% English-speaking Latine, and 18% Spanish-speaking Latine), the authors found 3 reduced PPE (defined by a composite score of asking questions, participating in session activities, and showing commitment to therapy) for Spanish-speaking Latine parents and parents with lower household income (Guan et al., 2019). The other two articles examined PPE in the context of a randomized pilot study of the Parent and Caregiver Active Participation Toolkit (PACT), a set of strategies designed to promote PPE, in Southern California community-based child psychotherapy services for children between the ages of 4 and 13 years with disruptive behavior problems. Using a mean PPE score of 5 engagement behaviors (e.g., sharing general perspectives, sharing home action perspectives, showing enthusiasm about home actions, asking questions, and demonstrating commitment to therapy), Stadnick et al. (2016) found reduced PPE for a sample of 8 Hispanic parents compared to their non-Hispanic White counterparts. In item- level analyses using the same sample, Dickson et al. (2017) reported that the Hispanic parents exhibited reduced general perspective sharing, home action perspective sharing, and enthusiasm about home actions specifically, finding no differences in question asking or commitment to therapy in Hispanic versus non-Hispanic parents. These findings suggest that decreased engagement (i.e., greater barriers to active engagement) may exacerbate existing disparities in access to quality intervention services for underserved families (Haine-Schlagel et al., 2022; Dickson et al., 2017). Thus, there is a need to increase understanding of the family and sociodemographic factors that influence PPE using objective observational coding schemes of parent engagement behaviors within early autism intervention sessions. Part C Early Intervention, Parent Coaching, and Parent Engagement Given the general lack of parent engagement observed in community service settings, where many children receive care, it is important to understand how to promote parent engagement in these types of settings (Haine-Schlagel & Walsh, 2015). The current study uniquely occurs within the context of the Part C Early Intervention (EI) System, which serves 4 children under age three with or at increased likelihood of developmental delays or disabilities, including autism and related social communication delays. The Part C EI system is publicly funded, available in all states, and serves approximately 3.7% of all children under three years old (Michigan Department of Health and Human Services, 2023; U.S. Department of Education, 2023). Part C serves a diverse population of children, and it is currently estimated that approximately 50% of children receiving services are of a racially and/or ethnically minoritized background (U.S. Department of Education, 2023). Within Part C, it is currently considered best practice to emphasize parent consultation and/or parent coaching, in which providers teach parents intervention techniques to use at home with their child and provide feedback on parent strategy use, to empower parents to support their children’s development (Kuhn et al., 2023; Michigan Department of Health and Human Services, 2023; Casagrande & Ingersoll, 2017; Ingersoll & Wainer, 2013). In addition, Part C emphasizes the provision of family-centered services, which are described as practices that treat families with dignity and respect, involve families in choices to strengthen functioning, and are individualized, flexible, and responsive to each family’s unique needs. Part C emphasizes creating parent-provider collaboration, building family knowledge and skills to enhance parenting self-efficacy, and providing services in the home and community to the extent possible (Division for Early Childhood, 2014). Generally, parents tend to report high satisfaction with Part C EI and perceive it to be family-centered, which may promote greater engagement given previously established associations between treatment satisfaction and engagement (Noyes-Grosser et al., 2018; Bailey et al., 2004; Hebbeler et al., 2007; Fawley-King et al., 2013; Haine-Schlagel & Walsh, 2015). Since Part C is a unique service system, it is vital to examine factors that may influence PPE in order to understand who 5 is at increased likelihood for reduced engagement and to enhance the effectiveness of Part C EI for children on the autism spectrum. Little research has focused on how provider characteristics or behaviors during intervention sessions may influence parent engagement (Garland et al., 2012). Some studies suggest that provider-reported training in parent-mediated interventions (PMIs) and greater use of evidence-based practices are associated with higher parent attendance at treatment sessions (Dickson et al., 2017; Garland et al., 2012; Stevens et al, 2006). One study found that parents whose providers received training in the Parent and Caregiver Active Participation Toolkit (PACT), which incorporates therapist training in increasing parent engagement opportunities, exhibited higher PPE than parents whose providers were in the treatment as usual group (TAU; Haine-Schlagel et al., 2018). Further analyses from this study found that provider-reported training in PMIs prior to the study was also associated with increased PPE as measured by the same behavioral coding scheme (Stadnick et al., 2016). Thus, there is evidence that provider training in PMIs and engagement techniques, as well as adherence to evidence-based practices, may be related to parent engagement. Furthermore, specific techniques used by providers during intervention sessions may influence parent engagement. Studies indicate that supportive and facilitative communication increased parental compliance with the provider while lack of directive skill-building strategies and limited involvement in shared decision-making were associated with lower parent-reported participation (Patterson & Forgatch, 2001; Baker-Ericzén et al., 2013). One study that utilized an observational coding scheme of PPE indicated that therapist use of engagement strategies including Collaboration (e.g., offering suggestions, seeking and incorporating parent input, 6 involving parent in therapeutic activities, working with the parent on parent-focused homework plans), Empowerment (e.g., recognizing strengths and efforts, addressing barriers to parent participation), and Psychoeducation (e.g., providing information about child problems and treatments) increased parent perspective sharing about strategy use at home while Alliance (e.g., active listening, conveying parent-therapist partnership, communicating positive regard) did not (Martinez & Haine-Schlagel, 2018). These findings warrant an increased focus on the nature and quality of provider intervention techniques (i.e., parent coaching strategies) to promote PPE. While the Part C EI system emphasizes parent coaching, which may increase PPE, research suggests that EI providers in Part C spend little time actively coaching parents during sessions and that “working with the child without explanation” is common (Sawyer & Campbell, 2017; Peterson et al., 2007; Romano & Schnurr, 2020). A recent study by Pellecchia et al. (2023) examined the use of parent coaching in Part C and found overall low use of coaching techniques and significant variability in the quality of coaching across providers. The authors also found that when providers used parent coaching, they tended to only use a few strategies (e.g., collaboration and in-vivo feedback). As parent coaching may promote greater parent engagement through the incorporation of explicit opportunities for engagement, it is vital to understand how provider parent coaching within Part C influences PPE to inform provider training and strategies in order to increase PPE in EI, thereby strengthening the quality of care for children on the autism spectrum. Present Study Since certain family and sociodemographic characteristics may reduce parent engagement in treatment while provider training in parent coaching may increase engagement, it is possible that the quality of provider coaching of parents moderates the influence of parent personal and sociodemographic characteristics on parent engagement. Higher quality parent 7 coaching by providers may encourage higher levels of engagement for families who might otherwise demonstrate lower engagement related to high stress, low self-efficacy, and low motivation as well as barriers associated with single parenthood, low SES background, and minoritized racial/ethnic identity. Further research is needed to examine how providers’ coaching of parents in PMIs may reduce barriers to engagement for underserved families in particular. As early intervention is key to supporting outcomes for children on the autism spectrum and parent engagement is vital for intervention efficacy, the proposed study will also examine how the quality of provider coaching of parents may broadly increase parent engagement in autism-specific intervention. This study addresses existing gaps in the literature (i.e., use of proxies of engagement as outcome measures, little research on engagement in autism-specific intervention, and lack of understanding of how provider behaviors influence engagement) by applying an objective observational coding scheme of parent engagement, the Parent Participation Engagement (PPE) in Child Psychotherapy Observational Coding Manual (Haine- Schlagel & Martinez, 2014), to recordings of early autism intervention sessions (Haine-Schlagel & Walsh, 2015; Stadnick et al., 2016; Garland et al., 2012). By increasing understanding of factors that predict active parent engagement and reduce barriers to engagement for underserved families, the findings of this study will inform improved provider training and bolster the quality of autism intervention, thereby improving outcomes for children on the autism spectrum. The first aim of the current study was to examine whether parent characteristics predict PPE in Part C early autism intervention. Based on previous research, we hypothesized that parents with single marital status, minoritized racial and/or ethnic identity, low SES background, high stress, low-self-efficacy, and low motivation will demonstrate reduced PPE (King et al., 2014; Kurzrok et al., 2021; Stadnick et al., 2016; Nock & Kazdin, 2005; Haine-Schlagel & 8 Walsh, 2015; Haine-Schlagel et al., 2022; Miller et al., 2008; Holly et al., 2019; Dickson et al., 2017; Guan et al., 2019; Tomczuk et al., 2022; Fawley-King et al., 2013; Garland et al., 2012, Pereira et al., 2016; Hansen & Warner, 1994). The second aim of the current study was twofold. First, we planned to investigate how overall provider quality of parent coaching influences PPE and we expected that higher provider coaching quality will be associated with increased PPE. If parent characteristics and overall provider quality of parent coaching predicted PPE as expected, we then planned to examine whether provider quality of parent coaching moderates the effect of parent characteristics on PPE in Part C early autism intervention. We hypothesized that higher quality parent coaching by providers would reduce the association of parent characteristics with PPE such that parents with single marital status, minoritized racial and/or ethnic identity, low SES background, high stress, low-self-efficacy, and low motivation would demonstrate higher levels of parent participation engagement. The third aim of the current study was to explore how the quality of specific provider parent coaching strategies influences parent engagement in Part C early autism intervention. We hypothesized that high quality Collaboration (i.e., working together with the parent during the session and co-creating goals) and In-Vivo feedback (i.e., giving feedback as the parent practices intervention techniques during sessions) would be related to higher parent participation engagement based on literature suggesting that collaboration, supportive and facilitative communication, shared decision-making, and directive skill-building strategies are associated with greater participation in child mental health treatment sessions (Martinez & Haine-Schlagel, 2018; Patterson & Forgatch, 2001; Baker-Ericzén et al., 2013). 9 Study Design METHODS The current study recruited participants from the ongoing Reciprocal Imitation and Social Engagement (RISE) Study, a National Institute of Mental Health (NIMH)-funded multi-site randomized controlled trial (RCT) across four states (Michigan, Massachusetts, Washington, and Illinois). The RISE study is examining the effectiveness of caregiver-implemented Reciprocal Imitation Teaching (CI-RIT), an evidence-based naturalistic developmental behavioral intervention (NDBI) for young children with social communication delays, in the Part EI System. EI providers are recruited and randomly assigned to undergo comprehensive training in delivering CI-RIT or to join the waitlist control group, in which providers receive CI-RIT training after the conclusion of their study participation. Families on their caseload whose children are 18-30 months and show early signs of autism, including social communication delays, are recruited to participate. Both English and Spanish-speaking families are recruited. Families are assessed at 3 timepoints: intake (T1), post-intervention (4 months after intake; T2), and follow-up (9 months after intake; T3). Each provider-parent dyad is required to have weekly intervention sessions during the duration of their 4-month active study participation between T1 and T2, and EI sessions can occur either in-person in the family’s home, in-person in a community space, or via telehealth. An EI session is video recorded approximately every four weeks between T1 and T2 and up to three EI session recordings are collected for each family as part of the study. Participants Participants in the current study included 61 community providers and 113 parents that participated in the larger RISE study. Each provider-parent dyad that completed at least one in- 10 person intervention session recording were included. Telehealth sessions were excluded for the purpose of the current analyses, as little research has examined potential differences in provider parent coaching methods administered in-person versus via telehealth. Both English and Spanish-speaking families were included if their EI sessions were conducted in English. 7 EI session recordings were conducted in Spanish and were excluded from the current analyses due to a lack of Spanish-speaking coders. Measures Table 1. Description of Study Measures Measure Time- point Informant Construct Family Demographic Information Form (FDIF) Intake Parent Provider Practices Survey (PPS) Intake Provider Parent Motivation Inventory (PMI) Readiness to Change Subscale Intake Parent Parent and child demographics Provider demographics Parent motivation to change parenting behavior Parenting Stress Index-4 Short Form (PSI-4 SF) Intake Parent Parenting stress Parenting Efficacy Survey (PES) Intake Parent Parent Empowerment and Coaching in Early Intervention (PEACE) Caregiver Coaching Fidelity Tool T1-T2 Observational Coding Parent Participation Engagement (PPE) in Child Psychotherapy: Observational Coding Manual T1-T2 Observational Coding Parenting efficacy Quality of provider parent coaching strategies Parent participation engagement in EI sessions Family and Provider Demographic Questionnaires Family and provider demographics were collected at intake via questionnaires about child and parent characteristics (i.e., age, sex, race, ethnicity, marital status, education level) as well as 11 provider characteristics and professional experience (i.e., sex, race, ethnicity, type of degree, professional background, years in profession, caseload). The Parenting Stress Index-4 Short Form The Parenting Stress Index-4 Short Form (PSI-4 SF; Abidin 1990) was administered at intake to measure stress specifically related to parenting, including personal distress, dissatisfaction regarding parent-child interactions, and perceptions of child self-regulatory skills. Parents rated 36 items (e.g., “I often have the feeling that I cannot handle things very well”, “My child’s behavior is more of a problem than I expected”, etc.) on a scale of 1 (Strongly Disagree) to 5 (Strongly Agree), with higher scores indicating higher parenting stress. The PSI-4 SF includes 3 subscales (Parental Distress, Parent-Child Dysfunctional Interaction, Difficult Child), which are summed to produce a Total Stress score (score range: 36-180). To account for missing items, a prorated Total Stress score was calculated and used in all analyses. The prorated Total Stress score for each participant was calculated as the Total Stress sum score multiplied by the total number of items (36), divided by the total number of items (36) minus the number of missing items. In the current sample, the internal consistency of the Total Stress scale was found to be excellent according to published guidelines (36 items; α = .93; Cicchetti 1994). The Parent Motivation Inventory The Parent Motivation Inventory (PMI) was administered to parents at intake as a measure of parent motivation to participate in intervention (Nock & Photos, 2006). The PMI includes 25 items rated by parents on a 1 (strongly disagree) to 5 (strongly agree) scale incorporating three aspects of motivation including 1. Desire for child change, 2. Readiness to change parenting behavior and 3. Perceived ability to change parenting behaviors. This study used only the “Readiness to Change” subscale of 14 items (e.g., “I am willing to change my 12 current parenting techniques and try new ones”, “I am motivated to practice the techniques I will learn in session at home with my child”) as a measure of parent motivation. Items are summed to create a Parent Readiness to Change score (score range: 14-70), with higher scores indicating greater readiness to change parenting behavior in the intervention process. To account for missing items, a prorated Parent Readiness to Change score was calculated and used in the current analyses. The prorated score for each participant was calculated as the Parent Readiness to Change sum score multiplied by the total number of items (14), divided by the total number of items (14) minus the number of missing items. In the current sample, the internal consistency of the PMI Readiness to Change subscale was found to be excellent (14 items; α = .94; Cicchetti 1994)). The Parenting Efficacy Survey The Parenting Efficacy Survey (PES) was administered at intake. The PES is a 10-item parent-report survey that asks parents to assess their parenting abilities and quality as a parent (e.g., “When your child is upset, fussy, or crying, how good are you at soothing him or her?”, “In general, how good a parent do you feel you are?”; Teti & Gelfand, 1991). The PES items are scored on scale from 1 (Not Good at All) to 4 (Very Good) and the items are summed to get a Total Efficacy score (score range: 10-40), with a higher score indicating a greater sense of parenting self-efficacy. To account for missing items, a prorated Total Efficacy score was calculated and used in all analyses. The prorated score for each participant was the Total Efficacy sum score multiplied by the total number of items (10), divided by the total number of items (10) minus the number of missing items. In the current sample, internal consistency of the PES was found to be fair based on published guidelines (10 items; α = .77; Cicchetti 1994). 13 The Parent Participation Engagement in Child Psychotherapy: Observational Coding Manual The Parent Participation Engagement in Child Psychotherapy: Observational Coding Manual was used to measure active parent engagement during recorded EI session videos (Haine-Schlagel & Martinez, 2014; Appendix A). The PPE observational coding manual was adapted from the PRAC Therapy Process Observational Coding System for Child Psychotherapy- Strategies (PRAC TPOCS-S) and “An Individualized Mental Health Intervention for ASD” (AIM HI) observational coding manual as a behavioral coding scheme to score parents’ participation engagement observed in recordings of psychotherapy sessions for children with disruptive behavior and autism, respectively (Garland, Brookman-Frazee, & Mcleod, 2008; Brookman-Frazee & Chlebowski, 2013; Haine-Schlagel & Martinez, 2014). It is also applicable to early intervention and has been previously used in a study of a parent-mediated autism intervention (Guan et al., 2019). Coders rated parents on the frequency and thoroughness of six behaviors indicating active engagement. Items included 1. general perspective sharing (e.g., sharing perspectives about child skills/progress, the intervention in general, etc.), 2. perspective sharing about home actions (i.e., sharing perspectives about practicing strategies at home outside of intervention sessions), 3. expressing agreement with or enthusiasm about home actions, 4. asking questions (e.g., administrative or clinically relevant questions), 5. participating in therapy activities (e.g., participating in in-session practice of strategies with the provider present), and 6. demonstrating commitment to therapy. Items were rated on a 1 (absent) to 5 (present and high quality) scale. If no perspectives about home actions were shared, the two home action items were rated as not applicable (N/A). If there was no opportunity to participate in a therapy activity in session, the therapy activity participation quality was rated as not applicable (N/A). All scored items were averaged for an overall PPE score, with higher scores indicating higher PPE. In the current 14 sample, internal consistency of the PPE coding scheme was found to be relatively low (6 items; α = .67). Undergraduate research assistants (RAs) were trained to reliability (>80%) prior to independently coding sessions and remained blind to condition. One EI session recording from each provider-parent dyad was randomly selected and prioritized for coding. Once all parent- provider dyads had one session coded, further EI session recordings were randomly selected for coding. All RAs participated in monthly consensus coding meetings to prevent drift. Approximately 20% of the 215 total videos (n=46) were coded by two independent observers. Inter-rater reliability was assessed with intraclass correlations and their 95% confidence intervals based on an average measures one-way random effects model (n = 46, ICC = .79, 95% CI [.61, .88]). Inter-rater reliability was assessed to be excellent according to published guidelines (Cicchetti 1994). The Parent Empowerment and Coaching in Early Intervention (PEACE) Caregiver Coaching Fidelity Observational Coding Scheme The Parent Empowerment and Coaching in Early Intervention (PEACE) Caregiver Coaching Fidelity tool was used to measure providers’ quality of parent coaching during the same recorded EI sessions (Pellecchia et al., 2022; Appendix B). The PEACE Caregiver Coaching Fidelity tool evaluates parent coaching fidelity/quality in five domains linked to adult learning theory and best practice in caregiver coaching. Adapted from the Triadic Intervention and Evaluation Scale (TIERS) and the Project ImPACT Fidelity of Implementation for Coaching Form, this intervention-agnostic behavioral coding scheme focuses on provider behaviors that indicate the quality of their parent coaching strategies. It was designed for use with any PMI and 15 has been used to evaluate provider coaching quality within the Part C EI system (Basu et al., 2010; Ingersoll & Dvortcsak, 2010; Pellecchia et al., 2022; Pellecchia et al., 2023). Coders rated the quality of 21 provider coaching behaviors on a scale from 1 (Never) to 5 (Almost Always) regarding whether the provider used that strategy when given the opportunity (e.g., “Let caregivers make some decisions and lead parts of the intervention session”, “Observe ongoing interactions and comment on specific strategies that are working well”, etc.). Of the 215 total videos coded with the PPE coding scheme, 181 were also coded by the PEACE coding scheme (i.e., approximately 84% of the videos were coded with both coding schemes). In the current sample, internal consistency of the PEACE coding scheme was found to be fair according to published guidelines (21 items; α = .78; Cicchetti 1994). This measure includes an overall provider quality of parent coaching score, which is the average of all 21 items, with higher scores indicating higher quality of coaching (score range: 1- 5). The measure also includes subscale scores for each of the five domains of parent coaching strategies. The General domain involves organizing aspects of intervention session structure (e.g., “Set the plan for today’s session with caregiver”, “Arrange aspects of the environment to promote parent-child interaction”; 5 items; α = .23). The Collaboration domain involves working with the parent and co-creating goals (e.g., “Let caregivers make some decisions and lead parts of the intervention session”, “The coach and caregiver collaboratively set goals for the child”; 4 items; α = .53). The Demonstration domain incorporates explaining and demonstrating techniques for the parent (e.g., “Explicitly teach a strategy to the caregiver”, “Demonstrate/model techniques that promote caregiver-child interaction”; 3 items; α = .83), while the In-Vivo Feedback domain focuses on giving feedback as the parent practices techniques (e.g., “Allow sufficient time for the caregiver to practice strategies during session”, 16 “Observe ongoing interactions and comment on specific strategies that are working well [positive feedback]”; 3 items; α = .76). Finally, the Reflection and Problem-Solving domain centers on helping the parent work through obstacles to using intervention strategies at home (e.g., “The coach asks the caregiver about possible barriers to practice and discusses solutions”, “Helps the caregiver work through any obstacles in the implementation of the techniques using reflective strategies”; 6 items; α = .27). Items in each domain are averaged for domain-specific quality scores, with higher scores indicating higher quality of the coaching strategy (score range: 1-5). Undergraduate research assistants (RAs) were trained to reliability (>80%) prior to independently coding sessions and remained blind to condition. All RAs participated in monthly consensus coding meetings to prevent drift. Approximately 20% of the 181 videos were coded by two independent observers (n=34). Inter-rater reliability of the total score and each subscale score was assessed by calculating intraclass correlations and their 95% confidence intervals based on average measures one-way random effects models. According to published guidelines, (Cicchetti 1994), inter-rater reliability was assessed to be excellent for the total score (ICC=.92, 95% CI [.84, .96]). Inter-rater reliability was found to be good to excellent for all domains/coaching strategies: General (ICC=.73, 95% CI [.47, .87]), Collaboration (ICC=.68, 95% CI [.37, .84]), Demonstration (ICC=.93, 95% CI [.85, .96]), In-Vivo Feedback (ICC=.77, 95% CI [.54, .88]), Reflection & Problem Solving (ICC=.82, 95% CI [.64, .91]). 17 Table 2. Internal Consistency and Interrater Reliability of Study Measures Parenting Stress Index (PSI) Parental Distress (PD) Subscale Parent-Child Dysfunctional Interaction (PCDI) Subscale Difficult Child (DC) Subscale Parenting Efficacy Survey (PES) Parent Motivation Inventory Readiness to Change Subscale Parent Participation Engagement (PPE) Observational Coding Mean 1. General Perspective Sharing 2a. Any Home Action Perspectives? (Yes/No) 2b. Home Action Perspective Sharing 2c. Enthusiasm about Home Actions 3. Question Asking 4a. Any Therapy Activities? (Yes/No) 4b. Therapy Activity Participation 5. Overall Commitment Parent Empowerment and Coaching in Early Intervention (PEACE) Observational Coding General Domain Collaboration Domain Demonstration Domain In-Vivo Feedback Domain Reflection & Problem-Solving Domain Statistical Analysis No. of Items 36 12 12 12 10 14 6 1 1 1 1 1 1 1 1 21 5 4 3 3 6 α .93 .89 .79 .85 .77 .94 .67 - - - - - - - - .78 .23 .53 .83 .76 .27 ICC - - - - - - .79 .62 .62 .62 .62 .85 .88 .88 .67 .92 .73 .68 .93 .77 .82 ICC 95% CI - - - - - - .61-.88 .32-.79 .31-.79 .31-.79 .31-.79 .74-.92 .79-.93 .79-.93 .41-.82 .84-.96 .47-.87 .37-.84 .85-.96 .54-.88 .64-.91 Due to the repeated and nested nature of the EI session recordings, intraclass correlations were calculated through fitting the null model to inform the appropriate number of levels to include in multilevel models. ICCs indicated that 35.1% of the variance in PPE in the current sample was due to provider-level differences at Level 3, 29.8% of the variance was explained by parent-level differences at Level 2, and 35.1% of the variance was attributable to the EI session recording timepoint at Level 1. Thus, 3-level multilevel models were run with EI session recording timepoint (Level 1) nested within parents (Level 2), nested within providers (Level 3). 18 Descriptive statistics were run in SPSS and all multilevel analyses were run using the nlme package in R (IBM Corp., 2021; Pinheiro et al., 2023). EI session recording timepoint did not significantly predict PPE and was therefore not included as a fixed effect (t(100)= -1.47, p = .15). Since data were drawn from an ongoing RCT, intervention condition (CI-RIT versus waitlist-control) was included as a covariate in all analyses, as in previous intervention studies (Stanick et al., 2016; Dickson et al., 2017). Mean PPE was included as the outcome variable in all analyses and the residuals were approximately normally distributed. Correlational analyses indicated that the vast majority of correlations between predictor variables were less than .50 and examination of variance inflation factors (VIFs) indicated that multicollinearity was not a concern, as all VIF values were below 5. Pearson product-moment correlation matrices for all aims are presented in Tables 3 and 4. Table 3. Correlation Matrix of Aim 1 and Aim 2 Predictor and Outcome Variables l a t i r a M t n e r a P d e z i t i r o n i M s u t a t S s u t a t S t n e r a P n o i t a c u d E t n e r a P g n i t n e r a P s s e r t S g n i t n e r a P y c a c i f f E n o i t a v i t o M t n e r a P C P r e d i v o r P y t i l a u Q E P P 1 -.09 -.07 .13 1 -.08 -.08 .07 -.15 1 -.38* .34* .12 -.12 .05 Parent Marital Status Parent Minoritized Status Parent Education Parenting Stress Parenting Efficacy Parent Motivation Provider PC Quality EI Recording 1 EI Recording 2 EI Recording 3 PPE .30* EI Recording 1 .28* EI Recording 2 .11 EI Recording 3 *p < .05; PC = Parent Coaching; PPE = Parent Participation Engagement -.19 -.35* -.12 1 -.52* -.01 -.20 -.22 -.27* .25* .24 .22 .06 .26* .11 -.04 .04 .02 -.08 -.16 -.06 .05 .01 .02 -.24* -.16 -.03 -.14 -.32* -.12 1 .05 1 -.07 -.24 .05 .19 -.02 .08 1 1 .36* .49* .41* 19 Table 4. Correlation Matrix of Aim 3 Predictor and Outcome Variables n o i t a r o b a l l o C 1 .46* .50* .34* .29* n o i t a r t s n o m e D & n o i t c e l f e R - m e l b o r P g n i v l o S n o i t a p i c i t r a P t n e m e g a g n E t n e r a P k c a b d e e F o v i V - n I 1 .51* .46* .27* 1 .38* .25* 1 .46* 1 Collaboration Quality Demonstration Quality In-Vivo Feedback Quality Reflection & Problem-Solving Quality Parent Participation Engagement *p < .05 Continuous parent-level predictors were grand mean centered prior to analysis (e.g., parenting stress, parenting efficacy, parent motivation). Continuous provider-level variables, including provider overall quality of parent coaching in addition to quality of individual coaching domains (Collaboration, Demonstration, In-Vivo Feedback, and Reflection & Problem- Solving), were also grand mean centered. Parent marital status, parent minoritized status, and parent education level were dichotomously coded prior to analysis. Marital status was coded 0 for single parents and 1 for parents living with a spouse or partner. Minoritized status was coded 0 for White, Non-Hispanic/Latine parents and 1 for parents of a racial and/or ethnic minoritized background. Education level was coded 0 for parents who completed education below a 4-year college degree and 1 for parents who completed a 4-year college degree or above. Neither parent age (t(52) = 1.03 , p = .31) nor parent gender (t(52) = -1.01, p = .32) significantly predicted PPE and thus were not included as covariates. 20 RESULTS The current sample included 215 observations of EI session recordings at Level 1, 113 parents at Level 2, and 61 providers at Level 3. Observations/participants at each level were approximately evenly split between the intervention and TAU conditions. The EI session recordings were approximately evenly split across recording timepoints 1, 2, and 3 with slightly more recordings at timepoint 1. The average session recording length was around 48 minutes (SD = 11 minutes, 9 seconds). Full descriptive statistics for EI session recordings are presented in Table 5. Table 5. Early Intervention Session Recording Characterization (n = 215) n (%) or Mean (SD) Study Condition RIT-Now RIT-Later EI Session Recording Timepoint Timepoint 1 Timepoint 2 Timepoint 3 Video Length (mm:ss) Observational Coding Scheme Applied Parent Participation Engagement (PPE) Parent Empowerment and Coaching in Early Intervention (PEACE) 109 (50.7%) 106 (49.3%) 84 (39.1%) 70 (32.6%) 61 (28.4%) 47:45 (11:09) 215 (100.0%) 181 (84.2%) Parents were predominantly mothers living with a spouse or partner and were between 22 and 55 years of age (Mean = 34.15, SD = 6.12). Their children were predominantly males between the ages of 16 and 34 months (Mean = 26.71, SD = 3.56). Most parents identified as White, non-Hispanic/Latine, and English-speaking; however, when race and ethnicity were combined into minoritized status, approximately 60% of the parent sample were of a minoritized racial and/or ethnic background. There was considerable variability in education level and slightly over half the sample reported having received below a 4-year college degree. Parent and 21 child demographics are presented in Table 6 and full descriptive statistics for the parent measures are listed in Table 7. Table 6. Parent and Child Demographics (n = 113) Study Condition RIT-Now RIT-Later Age: Mean (SD) Gender Male Female Parent Relationship to Child Mother Father Foster Mother Grandmother Parent Marital Status Lives with spouse/partner Does not live with spouse/partner Race Asian Black/African American Indigenous/Native Alaskan Native Hawaiian/Other Pacific Islander White More than one race Other/Not listed Prefer not to answer Ethnicity Hispanic/Latine Not Hispanic/Latine Prefer not to answer Parent Preferred Language English Spanish Parent Minoritized Status Minoritized Non-Hispanic/Latine White Unknown Parent: n (%) Child: n (%) 60 (53.1%) 53 (46.9%) 60 (53.1%) 53 (46.9%) 34.15 (6.12) years 26.71 (3.56) months 13 (11.5%) 100 (88.5%) 77 (68.1%) 36 (31.9%) 99 (87.6%) 12 (10.6%) 1 (0.9%) 1 (0.9%) 82 (72.6%) 31 (27.4%) 11 (9.7%) 25 (22.1%) 1 (0.9%) 2 (1.8%) 55 (48.7%) 8 (7.1%) 3 (2.7%) 8 (7.1%) 33 (29.2%) 76 (67.3%) 4 (3.5%) 110 (97.3%) 3 (2.7%) 71 (62.8%) 38 (33.6%) 4 (3.5%) 22 - - - - - - 8 (7.1%) 22 (19.5%) 1 (0.9%) 0 (0.0%) 53 (46.9%) 18 (15.9%) 3 (2.7) 8 (7.1%) 36 (31.9%) 74 (65.5%) 3 (2.7%) - - - - - Table 6 (cont’d) Parent Highest Education 9th-11th grade High school diploma/GED Trade/vocational school Associate’s degree/2-year degree Courses toward college Bachelor’s degree/4-year degree Master’s degree Professional degree (MD, PhD, JD) Prefer not to answer Parent College Degree College degree or above Below college degree Unknown 6 (5.3%) 19 (16.8%) 8 (7.1%) 12 (10.6%) 17 (15.0%) 31 (27.4%) 16 (14.2%) 3 (2.7%) 1 (0.9%) 50 (44.2%) 62 (54.9%) 1 (0.9%) Table 7. Parent Characterization - - - - - - - - - - - - Parenting Stress Index (PSI) Total Stress Raw Score Mean (SD) 81.10 (21.05) Parenting Efficacy Scale (PES) Total Efficacy Score 31.07 (4.22) Parent Motivation Inventory Readiness to Change Score 64.07 (7.54) Parent Participation Engagement (PPE) Score 3.02 (0.91) Parent Participation Engagement (PPE) Items 1. Perspective Sharing in General 2a. Were any home actions discussed?: n (%) Yes No 2b. Perspective Sharing about Home Actions 2c. Agreeement with/Enthusism about Home Actions 3. Asking Questions 4a. Did any therapy activities take place?: n (%) Yes No 4b. Participation in Therapy Activities 5. Demonstrating Commitment 3.41 (1.07) - 163 (75.8%) 52 (24.2%) 3.37 (1.12) 3.12 (1.17) 2.07 (1.17) - 114 (53.0%) 101 (47.0%) 3.96 (0.92) 3.21 (1.17) EI providers were predominantly female, White, non-Hispanic Latine, and English- speaking and reported ages between 26 and 64 years (Mean = 40.24, SD = 9.88). The majority of providers received a master’s degree and reported a professional background as a speech 23 language pathologist, occupational therapist, or developmental therapist. Providers had a an average of approximately 12 years of professional experience (Mean = 12.59, SD = 10.63) and about half had worked at their Part C agency for over 5 years. The average total caseload reported by providers in the past month was approximately 20 (SD = 8.80), with an average of about 7 children with social communication delays (SD = 5.45). Complete provider demographics are presented in Table 8. Descriptive statistics of the average overall quality of parent coaching and of each parent coaching strategy are listed in Table 9 and reveal relatively lower quality of Collaboration in comparison to the other parent coaching domains/strategies (Mean = 1.97, SD = 0.76). Table 8. Early Intervention Provider Demographics (n = 61) Study Condition RIT-Now RIT-Later Age (years): Mean (SD) Gender Male Female Missing Race Asian Black/African American Indigenous/Native Alaskan Native Hawaiian/Other Pacific Islander White More than one race Other/Not listed Prefer not to answer Missing Ethnicity Hispanic/Latine Not Hispanic/Latine Prefer not to answer Missing 24 n (%) 61 34 (55.7%) 27 (44.3%) 40.24 (9.88) 4 (6.6%) 56 (91.8%) 1 (1.6%) 4 (6.6%) 3 (4.9%) 0 (0.0%) 0 (0.0%) 47 (77.0%) 3 (4.9%) 2 (3.3%) 1 (1.6%) 1 (1.6%) 1 (1.6%) 57 (93.4%) 2 (3.3%) 1 (1.6%) Table 8 (cont’d) Fluent Languages English Spanish Highest Degree Earned Bachelor’s Degree Master’s Degree Doctoral Degree Other Years of Professional Experience: Mean (SD) Total Years Years with Autism/Social Communication Delays Caseload in Past Month: Mean (SD) Total Number of Children Number of Children with Autism Diagnosis Number of Children with Social Communication Delay Number of Children with Suspected Autism Professional Background Speech Language Pathologist (SLP) Occupational Therapist Physical Therapist Early Childhood Educator Special Educator Developmental Therapist Social Worker Other Amount of Time at Part C Agency 0-3 months 3-6 months 6-12 months 1-3 years 3-5 years 5+ years 61 (100.0%) 1 (1.6%) 6 (9.8%) 52 (85.2%) 2 (3.3%) 1 (1.6%) 12.59 (10.63) 12.22 (8.96) 19.43 (8.80) 2.03 (3.56) 7.03 (5.45) 3.64 (2.03) 19 (31.1%) 11 (18.0%) 1 (1.6%) 2 (3.3%) 8 (13.1%) 15 (24.6%) 3 (4.9%) 2 (3.3%) 3 (4.9%) 3 (4.9%) 6 (9.8%) 12 (19.7%) 9 (14.8%) 28 (45.9%) Table 9. Provider Quality of Parent Coaching Characterization PEACE Overall Quality of Parent Coaching PEACE Quality of Individual Parent Coaching Strategies General Collaboration Demonstration In-Vivo Feedback Reflection & Problem-Solving 25 Mean (SD) 2.76 (0.68) 3.12 (0.76) 1.97 (0.76) 2.43 (1.33) 2.47 (1.26) 3.40 (0.80) Aim 1: The Influence of Parent Characteristics on PPE Initially, separate 3-level random intercept models were run to examine the independent effect of each predictor on PPE. Not controlling for any other parent factors, marital status significantly predicted PPE (t(52) = 2.89, SE = .15, p = .006, b = .44) and parent minoritized status approached significance (t(50) = -1.81, SE = .15, p = .077, b = -0.27). Parent education, parenting stress, parenting efficacy, and parent motivation did not significantly predict PPE in separate models. Provider quality of coaching marginally significantly predicted PPE without controlling for any parent factors (t(72) = 1.90, SE = .19, p = .06, b = .24).ii For Aim 1, a 3-level random intercepts model with only parent-level predictors (e.g., parent marital status, parent minoritized status, parent education level, parenting stress, parenting efficacy, and parent motivation) was run (Table 8). Parent marital status significantly predicted PPE when controlling for other parent characteristics (t(42)=3.04, SE = .17, p = .004, b = .52), indicating that parents living with a spouse or partner demonstrated higher PPE than single parents. The model did not indicate statistically significant effects of any other parent characteristics on PPE.iii ii In 3-level models with all parent characteristics included, Level 2 slopes were also individually modeled as random for each parent-level characteristic separately. While the variance of most random slopes ranged from 3.70e-06 to 0.003, the variance of the random slope for minoritized status was 0.29. The inclusion of minoritized status as a random effect versus a fixed effect did not significantly influence results, so all parent characteristics were modeled as fixed effects in final analyses to run the most parsimonious models. iii Given prior research suggesting that Hispanic/Latine parents exhibit reduced PPE compared to their non- Hispanic/Latine counterparts, we also examined the influence of ethnicity (coded 1 for Hispanic/Latine and 0 for non-Hispanic/Latine) on PPE without controlling for other parent characteristics. Hispanic/Latine ethnicity marginally predicted reduced PPE (t(49)=-1.84, b=-.30, p=.07). The full Aim 1 model was also run with parent ethnicity as a fixed effect rather than parent minoritized status, controlling for other parent characteristics (Stadnick et al., 2016; Dickson et al., 2017; Guan et al., 2019). Findings were virtually unchanged (Hispanic/Latine: t(41)=- .61, b=-.10, p=.55), so minoritized status was used instead of ethnicity in all models moving forward. 26 Table 10. Aim 1 Multilevel Model: The Influence of Parent Characteristics on Parent Participation Engagement in Early Autism Intervention Fixed Effects Intercept Parent Lives with Spouse/Partner Parent Minoritized Status Parent College or Higher Parent Stress Parent Self-Efficacy Parent Motivation Random Effects 0.14* Level 3 (Provider) Intercept 0.17* Level 2 (Parent) Intercept Residual 0.29* *Significance demonstrated with 95% confidence intervals Unstandardized Coefficients 2.39 0.52 -0.07 -0.06 0.00 -0.02 0.01 Variance 95% CI Parent Participation Engagement Standard Error 0.21 0.17 0.15 0.15 0.00 0.02 0.01 Standard Deviation 0.37 0.41 0.54 1.98-2.80 0.18-0.85 -0.37-0.22 -0.35-0.23 -0.01-0.01 -0.05-0.02 -0.01-0.03 95% CI of Variance 0.05-0.36 0.07-0.42 0.21-0.40 P- Value .00 .004 .63 .68 .99 .32 .21 Aim 2: The Influence of Provider Quality of Parent Coaching on PPE For Aim 2, a 3-level random intercepts model was first run with all parent characteristics plus the added fixed effect of provider quality of parent coaching (Table 9). Parent marital status remained a significant predictor of PPE when controlling for provider quality of parent coaching (t(38) = 2.92, SE = .17, p = .006, b = .51). Provider quality of coaching was a marginally significant predictor of PPE (t(67) = 1.93, SE = .13, p = .058, b = .25), indicating that higher provider quality of parent coaching may be related to higher PPE. A log-likelihood ratio test was used to compare the fit of this model compared to the Aim 1 model and revealed that the current model including provider quality of coaching was a significantly better-fitting model than the Aim 1 model with only parent characteristics (χ2(1)= 54.94, p < .0001). 27 Table 11. Aim 2 Multilevel Model 1: The Influence of Provider Quality of Coaching and Parent Characteristics on Parent Participation Engagement in Early Autism Intervention Fixed Effects Intercept Parent Lives with Spouse/Partner Parent Minoritized Status Parent College or Higher Parent Stress Parent Self-Efficacy Parent Motivation Provider PC Quality Random Effects Level 3 (Provider) Intercept Level 2 (Parent) Intercept Residual 95% CI Unstandardized Coefficients 2.49 0.51 -0.04 -0.05 -0.00 -0.03 0.01 0.25 Parent Participation Engagement Standard Error 0.22 0.17 0.16 0.15 0.00 0.02 0.01 0.13 Standard Deviation 0.36 0.35 0.58 2.06-2.92 0.17-0.85 -0.35-0.28 -0.35-0.26 -0.01-0.01 -0.06-0.01 -0.01-0.03 -0.00-0.50 95% CI for Variance 0.05-0.37 0.03-0.53 0.24-0.49 0.13* 0.12* 0.34* Variance P- Value .00 .006 .82 .76 .76 .17 .18 .058 *Significance demonstrated with 95% confidence intervals; PC = Parent Coaching A final 3-level random-intercepts model was run for Aim 2 with an added interaction term between parent marital status and provider quality of parent coaching (Table 10)iv. The model did not demonstrate a significant interaction (t(66) = 1.48, SE = .23, p = .14, b = .33). According to a log-likelihood ratio test, it was not a significantly better-fitting model compared to the previous model that included only the main effects of parent characteristics and provider quality of coaching (χ2(1)= 2.26, p = .13). iv Another random-intercepts model was run including only parent marital status, provider quality of parent coaching and their interaction (provider quality*parent marital status). This was a significantly poorer-fitting model than the Aim 2 interaction model reported in the main text that included all parent characteristics (χ2(5)= 29.78, p < .0001). This was also a significantly poorer-fitting model compared to the first Aim 2 model that included only parent characteristics and provider parent coaching quality without an interaction term (χ2(4)= 27.52, p < .0001). 28 Table 12. Aim 2 Multilevel Model 2: The Influence of Provider Quality of Parent Coaching on the Relationship Between Parent Marital Status and Parent Participation Engagement in Early Autism Intervention Parent Participation Engagement 95% CI Fixed Effects Unstandardized Coefficients 2.45 0.51 -0.01 -0.02 -0.00 -0.03 0.01 0.02 0.33 Intercept Parent Lives with Spouse/Partner Parent Minority Status Parent College or Higher Parent Stress Parent Self-Efficacy Parent Motivation Provider PC Quality Provider Quality*Parent Spouse 2.02-2.88 0.17-0.85 -0.33-0.31 -0.32-0.29 -0.01-0.01 -0.07-0.01 -0.01-0.03 -0.36-0.41 -0.10-0.77 95% CI for Variance 0.04-0.36 Level 3 (Provider) Intercept 0.03-0.52 Level 2 (Parent) Intercept Residual 0.24-0.48 *Significance demonstrated with 95% confidence intervals; PC = Parent Coaching 0.12* 0.13* 0.34* Random Effects Variance P- Value 0.00 0.005 0.95 0.92 0.61 0.14 0.18 0.90 0.14 Standard Error 0.22 0.17 0.16 0.16 0.00 0.02 0.01 0.20 0.23 Standard Deviation 0.34 0.36 0.58 Aim 3: The Influence of Provider Parent Coaching Strategies on PPE For Aim 3, one 3-level random intercepts model including provider quality of Collaboration, Demonstration, In-Vivo Feedback, and Reflection & Problem-Solving as predictors was run (Table 11). The model indicated a significant effect of Reflection & Problem Solving on PPE (t(69) = 4.00, SE = .08, p = .0002, b = .33), such that higher quality Reflection & Problem Solving predicted higher PPE by an average of .33. Demonstration also significantly predicted PPE (t(69) = -2.20, SE = .06, p =.03, b = -.13) such that higher quality Demonstration predicted slightly lower PPE by an average of .13. Quality of Collaboration approached significance (t(69) = 1.74, SE = .08, p = .087, b = .15), with higher quality Collaboration trending towards predicting higher PPE. In-Vivo Feedback did not significantly predict PPE (t(69) = -.11, SE = .05, p = .91, b = -.006). 29 Table 13. Aim 3 Multilevel Model: The Influence of the Quality of Provider Parent Coaching Strategies on Parent Participation Engagement in Early Autism Intervention Fixed Effects Intercept Quality of Collaboration Quality of Demonstration Quality of In-Vivo Feedback Quality of Reflection & PS Random Effects Unstandardized Coefficients 2.68 0.15 -0.13 -0.01 0.33 Variance 95% CI Parent Participation Engagement Standard Error 0.12 0.08 0.06 0.05 0.08 Standard Deviation 0.34 0.40 0.56 2.43-2.92 -0.02-0.31 -0.24-0.01 -0.11-0.10 0.17-0.50 95% CI for Variance 0.04-0.34 0.06-0.45 0.22-0.45 P- Value .00 .09 .03 .91 .0002 Level 3 (Provider) Intercept Level 2 (Parent) Intercept Residual *Significance demonstrated with 95% confidence intervals; PS = Problem-Solving 0.12* 0.16* 0.31* 30 DISCUSSION The current study examined the influence of parent characteristics and provider parent coaching behaviors on parent participation engagement (PPE) in early autism intervention in the Part C Early Intervention (EI) system using an observational coding scheme of PPE. The Influence of Parent Characteristics on PPE Examination of parent characteristics including marital status, minoritized status, education, parenting stress, parenting efficacy, and motivation to change parenting behaviors through intervention indicated that only parent marital status significantly predicted PPE. This suggests that single parents face greater barriers to active, in-session PPE. This finding is consistent with previous research indicating that single parents demonstrate lower attendance and participation more broadly (Haine-Schlagel & Walsh, 2015). Contrary to our hypotheses, no other parent characteristics significantly predicted PPE. There are a number of possible explanations for the lack of replication of previous research suggesting that parents with minoritized racial/ethnic identities, low SES backgrounds, high stress, low self-efficacy, and low motivation exhibit reduced parent engagement in child treatment. First, the differences in findings may be related to measurement characteristics. For example, the current study used the Parenting Stress Index-4 Short Form as a proxy for parent stress, but it may be the case that parenting stress does not impact parent engagement in the same way as broader life stressors. Interestingly, another study examining parent involvement in Intensive Behavioral Intervention for autism did not find a relationship between parent scores on the PSI and either parent-reported or therapist-reported involvement in their child’s intervention (Solish & Perry, 2008). Furthermore, the current analyses utilized the Parent Readiness to Change subscale of the Parent Motivation Inventory (PMI), while other studies have used the 31 total PMI sum score and/or different subscales to measure parent motivation in treatment (Stadnick et al., 2016; Haine-Schlagel et al., 2019). PPE may also be qualitatively different for parents involved in Part C EI. Given the family-centered focus of Part C and the emphasis on provider use of parent coaching, parents may naturally exhibit higher PPE regardless of sociodemographics and/or stress, self-efficacy, and motivation than in previously studied community- and school-based child mental health services without the same emphasis on parent and family priorities. In addition, while previous studies have been predominantly conducted in populations of older children with disruptive behavior disorders and attention-deficit/hyperactivity disorder (ADHD), the current study examined PPE in a population of young children under three years old with early signs of autism for whom intervention may naturally require greater parent involvement due to age and diagnosis. Indeed, studies suggest that parents are more likely to receive psychotherapy incorporating parent involvement for younger children and that parents of children on the autism spectrum are similarly more likely to be involved in EI for younger versus older children (Dowell & Ogles, 2010; Solish & Perry, 2008). Moreover, because most Part C EI providers administer services in families’ homes, this aspect of the Part C system may allow greater PPE during sessions for parents who would otherwise face increased barriers to attending clinic-based sessions or transferring clinic-based strategies to the home context. Thus, aspects of the Part C EI context and young ASD population may lead to different levels of PPE during EI sessions as well as different influences of parent characteristics on PPE levels. It may also be the case that parent characteristics influence different aspects of parent engagement differently. While parent characteristics seem to influence session attendance, homework completion, and parent-reported participation, they may not influence in-session 32 active PPE in the same way. When comparing the current findings to previous studies that have utilized the same PPE Observational Coding Scheme, it is worth highlighting a few notable differences. The three articles that examined the effect of Hispanic/Latine ethnicity on PPE were conducted with two samples from Southern California, while the current study’s Hispanic/Latine population was recruited from multiple regions of the United States (e.g., Michigan, Illinois, Washington, and Massachusetts; Stadnick et al., 2016; Dickson et al., 2017; Guan et al., 2019). It is likely that the Hispanic/Latine sample in the current study is more heterogeneous than the Southern California Hispanic/Latine parents included in other studies. In addition, Hispanic/Latine parents in the current sample were predominantly English-speaking while Dickson et al. (2017) and Stadnick et al. (2016) included predominantly Spanish-speaking Hispanic/Latine parents. Guan et al. (2019) also only found reduced PPE for Spanish-speaking Hispanic/Latine parents compared to non-Hispanic Latine White and English-speaking Hispanic/Latine parents. Since the current study excluded EI sessions conducted in Spanish, there are notable differences in the population of parents in the current study compared to previous publications. Furthermore, while the outcome variable included in the current analyses was the PPE score averaged across all individual PPE behaviors, other studies have used scores derived differently from the PPE coding scheme. For example, the composite PPE score used by Guan et al. (2019) only included three of the PPE items (e.g., asking questions, participating in session activities, showing commitment to therapy) and Dickson et al. (2017) examined the effect of ethnicity on each PPE behavior individually rather than an average score. It is important to note that the current study focused primarily on minoritized status, finding a marginal effect of minoritized status on PPE when not controlling for other parent characteristics. Exploratory 33 analyses similarly found a marginal effect of parent ethnicity on PPE when not controlling for other parent characteristics. In the current sample, both parent minoritized status and ethnicity were correlated with marital status and education, which may explain why minoritized status and ethnicity did not significantly predict PPE when controlling for other parent characteristics. It is also possible that we would be more likely to find relationships between these parent characteristics and PPE in a larger sample. The Influence of Provider Quality of Parent Coaching on PPE The addition of provider quality of parent coaching to the model indicated that higher quality parent coaching by providers may promote greater PPE even after controlling for parent characteristics, but this finding was only marginally significant. Marital status remained a significant predictor even after controlling for provider coaching behavior. However, provider quality of coaching did not moderate the association of single parent status with reduced PPE, suggesting that increasing the quality of parent coaching may not allow single parents to overcome barriers to PPE. This suggests provider quality of parent coaching may increase PPE broadly for families, but that that further strategies are needed to reduce barriers to PPE for single-parent families. The Influence of Provider Parent Coaching Strategies on PPE Aim 3 analyses indicated that the quality of providers’ Reflection & Problem Solving with families may have the greatest impact on PPE compared to other provider parent coaching strategies (e.g., Collaboration, Demonstration, In-Vivo Feedback). Interestingly, the current study found that greater quality of provider Demonstration predicted decreased PPE, potentially indicating that greater use of Demonstration provides fewer opportunities for parents to engage. When providers are modeling and explaining techniques to the parent, they are typically working directly with the child rather than the parent and may provide less opportunity for parents to 34 actively participate in those parts of the session. However, Demonstration may still be an important coaching strategy to encourage parent learning and ultimate fidelity to the intervention strategies. These findings suggest that in order to promote PPE in early autism intervention, providers should engage in reflection about family experiences of barriers to practicing and using intervention strategies at home as well as help parents to problem solve methods of overcoming any barriers faced. It may also be beneficial during EI sessions to balance the amount of Demonstration of techniques with other strategies that may more effectively engage parents. Given that Pellecchia et al. (2022) found that most EI providers tended to only use Collaboration and In-Vivo Feedback, we echo their call for targeted training for providers in specific parent coaching strategies, particularly Reflection & Problem-Solving, to increase PPE and support child and family outcomes in early autism intervention. Reflection & Problem-Solving may be a particularly important and helpful strategy for families facing increased barriers to active in- session PPE, such as single parent families. Limitations and Future Directions It is possible that the current analyses were underpowered to find small effects of parent characteristics on PPE. While we are unaware of a simulation study that perfectly matches the current study’s data structure, a similar simulation study indicates that the current analyses may be underpowered. In a psychotherapy simulation study utilizing a 3-level multilevel model with measurement occasion (Level 1) nested within patient (Level 2), nested within therapist (Level 3), the authors found that the necessary sample size to obtain sufficient power (i.e., a power of 0.80) with 5 measurements per patient included 85 therapists with two patients each (i.e., 170 total patients; de Jong et al., 2010). This study only examined 5, 11, or 21 measurements per patient, which does not correspond to the approximately 2 EI session recordings per parent in the current study (de Jong et al., 2010). However, research suggests that while increasing sample 35 size at Level 2 and Level 3 increases power, increasing the number of measurements does not substantially influence power, suggesting that this study accurately approximates the sample size needed in the current study for a power of 0.80 despite differences in the number of measurement occasions (Lee & Hong, 2021). Notably, simulation studies also suggest that a sample size of at least 30-50 of the highest cluster (i.e., providers in the current study) is sufficient for unbiased fixed effect estimates, which suggests that the current regression coefficients and standard errors are likely stable despite slightly reduced power (Maas & Hox 2004, Maas & Hox, 2005, Lee & Hong, 2021; Clark and Wheaton, 2007). Behavioral research (e.g., psychotherapy and educational research) simulation studies have found reliable estimates even with smaller Level 3 (e.g., therapist and school) samples, with one psychotherapy study suggesting at least 10 therapists were necessary for unbiased standard error estimates (e.g., therapists and schools; Magnusson et al., 2018; Falkenström et al., 2020; McNeish & Wentzel, 2017). It is also unlikely that we were powered to detect moderation, as power for a moderator effect is typically lower than for a main effect, particularly for a Level 3 moderator (Dong et al., 2018). Given the potential for Type II error, the lack of significant findings do not necessarily mean that parent characteristics do not meaningfully influence PPE and that provider overall quality of coaching does not moderate the effect of parent marital status on PPE. Thus, there may be effects of parent sociodemographic characteristics, stress, self-efficacy, and motivation that we could not detect in the current sample due to power constraints. Coding of EI sessions with both the PPE and PEACE observational coding schemes is ongoing to increase power in future analyses. 36 While all analyses controlled for study intervention condition, the RISE RCT is ongoing and has not yet released the intervention condition data for publication. In future analyses, it seems theoretically likely that provider training in CI-RIT may increase provider quality of parent coaching compared to providers in the waitlist-control condition. If future analyses confirm this hypothesis, we plan to conduct mediational models to examine whether CI-RIT training influences PPE through improved provider quality of parent coaching. Although the demographics of our sample are representative of the general demographics of families receiving services in Part C, current findings should be interpreted with caution when generalizing to families in EI. Given studies indicating that parents who choose to participate in research studies often demonstrate higher motivation and engagement than families in standard care settings, (Haine-Schlagel & Walsh, 2015), our sample likely included parents with higher levels of motivation and PPE, which may have influenced the observed relationships of parent characteristics, provider behaviors, and parent engagement. In addition, the current study excluded EI sessions conducted in Spanish due to a lack of Spanish-speaking coders. Since only 7 sessions were excluded for this reason, it is unlikely that results were biased but the current findings are nonetheless based on an English-speaking sample. As observational coding of parent engagement continues, Spanish-speaking research assistants will be recruited to code Spanish EI sessions for future analyses. The current findings may also not apply to parent coaching conducted via telehealth since the study sample was limited to in-person EI session recordings. Next steps include adapting the PPE Observational Coding Scheme for telehealth EI sessions to better understand how different service delivery modalities may influence PPE. Another limitation to consider in the current study is that parent engagement occurs in the context of a bidirectional relationship between the parent and provider, which makes it difficult 37 to parse out which individual is driving the level of engagement observed. For example, it is possible that a parent who demonstrates lower engagement may lead a provider to provide fewer opportunities for the parent to engage in the session, which may further reduce parent engagement. On the other hand, a provider who asks many questions and invites parent perspectives will likely lead to the parent to naturally demonstrate higher engagement as there are more opportunities to engage. To parse apart the bidirectional influences on parent engagement, future research will utilize lag-sequential analysis methods to ensure greater understanding of how parent engagement is influenced sequentially by both parent and provider in-session behaviors. Multidimensionality of Parent Engagement While a unique strength of the current study is its use of an observational coding scheme to objectively measure in-session PPE, this was the only measure of parent engagement included and it is unclear how it relates to other more commonly examined behavioral measures of engagement (e.g., attendance, retention, and homework completion). Conceptual models suggest that parent engagement is multidimensional and have highlighted different aspects of engagement including social (e.g., therapeutic alliance), cognitive (e.g., beliefs about treatment outcomes), affective (e.g., emotions experienced related to treatment), and behavioral (e.g., attendance and participation; Chorpita & Becker, 2022). Despite conceptual models and emerging empirical evidence suggesting that parent engagement is multidimensional, individual studies have tended to only focus on one aspect of engagement (Chorpita & Becker, 2022). A recent review of 52 randomized controlled trials of treatment engagement interventions found that 61.5% of studies measured only one engagement domain and 26.9% measured two domains (Lakind et al., 2021). Given this research suggesting the multidimensionality of parent engagement, future analyses will investigate how parent characteristics and provider coaching 38 strategies influence other aspects of parent engagement to promote broader understanding of multiple engagement measures. Provider Perceptions of Parent Engagement Recent qualitative work suggests that provider perceptions of parent engagement impact provider beliefs and decisions about the fit of parent coaching for individual families (Tomczuk et al., 2021). For example, providers reported that families with a low SES background and competing priorities (i.e., busy work schedules, life stressors), were not a good fit for parent coaching because parents did not have sufficient resources at home and demonstrated low energy during sessions (Tomczuk et al., 2021). Providers also referenced multigenerational or single- parent families, language barriers, and cultural stigma as factors that reduced engagement and made parent-mediated interventions a poor fit. However, providers interpreted competing life stressors as lack of motivation and engagement, using terminology such as “lazy” and “checked out” to describe parents, suggesting that provider biases influence perceptions of parent engagement. Another study by Straiton et al. (2023) similarly indicated that provider perceptions of parent readiness for parent coaching influence providers’ decisions about whether to offer parent training. Providers often perceived parent “readiness” for parent coaching through overt behavioral indicators, such as regular session attendance without missing sessions, independently choosing to closely observe sessions, giving their undivided attention during sessions, following provider recommendations, and asking providers for advice and information. Thus, providers’ perceptions of parent engagement and related intervention fit may be perpetuating ongoing disparities in access to quality evidence-based interventions. Since provider perceptions of engagement and commitment often influence the type of intervention families ultimately receive, it is vital to further understand how providers interpret parent engagement and how they perceive barriers to engagement to address disparities in service access and quality. In addition, increasing 39 provider training in parent coaching and encouraging its use with diverse families may reduce the effect of provider perceptions of engagement on the perceived fit of parent coaching for marginalized families. If providers are encouraged to use parent coaching with all families regardless of initial behavioral engagement indicators, parents may ultimately exhibit increased engagement during sessions in response to provider parent coaching strategies. Increased parent engagement may then encourage continued use of parent coaching by providers and alter their perceptions of the fit of parent coaching for marginalized families, potentially reducing disparities in access to this evidence-based intervention method. Additional research suggests that providers often attribute low treatment engagement to the client rather than the provider or service agency and that many providers overestimate the strength of client engagement (Gearing et al., 2012; O’Keeffe et al., 2019). Moreover, providers tend to mainly observe behavior when assessing client engagement, but the method of assessment was found to differ according to the level of engagement in a recent study (Becker et al., 2021). The majority of providers reported observing attendance when engagement was perceived to be low, but the majority of providers considered in-session participation or homework completion when engagement was perceived to be high. The authors also reported that it was uncommon for providers to report considering therapeutic relationship, client beliefs about treatment, or client understanding of the treatment approach when thinking about engagement (Becker et al., 2021). Given that the findings of the current study suggest the value of provider behaviors to promote parent engagement, future research should investigate provider perspectives on parent engagement through qualitative interviews to better understand how providers are interpreting and reacting to parent engagement to improve provider training and use of engagement strategies. 40 CONCLUSION The current study suggests that while many parent sociodemographic and personal characteristics may not influence active parent participation engagement (PPE) in early autism intervention sessions in the Part C Early Intervention (EI) system, single parents face greater barriers to demonstrating PPE in this context. Though high-quality provider parent coaching in Part C may promote PPE, it does not seem to differentially increase PPE for single parents versus parents living with a spouse or partner. However, provider use of high-quality reflection and problem-solving parent coaching strategies is related to increased PPE. Findings suggest that provider training should incorporate emphasis on the quality of particular parent coaching strategies to promote PPE in early autism intervention and increase the quality of Part C EI services for diverse families. 41 REFERENCES Abidin, R.R. (1990). Parenting Stress Index- Short Form. Pediatric Psychology Press. Althoff, C. E., Dammann, C. P., Hope, S. J., & Ausderau, K. K. (2019). Parent-mediated interventions for children with autism spectrum disorder: A systematic review. The American Journal of Occupational Therapy, 73(3), 7303205010p1-7303205010p13. https://doi.org/10.5014/ajot.2019.030015 American Psychological Association (2022, January 1). Autism spectrum disorder. https://www.apa.org/topics/autism-spectrum-disorder Bailey, D.B., Hebbeler, K., Scarborough, A., Spiker, D., & Mallik, S. (2004). First experiences with early intervention: A national perspective. Pediatrics, 113(4), 887-896. https://doi.org/10.1542/peds.113.4.887 Baker-Ericzén, M. J., Jenkins, M. M., & Haine-Schlagel, R. (2013). Therapist, parent, and youth perspectives of treatment barriers to family-focused community outpatient mental health services. Journal of Child and Family Studies, 22(6), 854-868. https://doi.org/10.1007/s10826-012-9644-7 Basu, S., Salisbury C.L., Thorkildsen, T.A. (2010). Measuring collaborative consultation practices in natural environments. Journal of Early Intervention, 32(2), 127–50. https://doi.org/10.1177/1053815110362991 Becker, K. D., Lee, B. R., Daleiden, E. L., Lindsey, M., Brandt, N. E., & Chorpita, B. F. (2015). The common elements of engagement in children’s mental health services: Which elements for which outcomes? Journal of Clinical Child & Adolescent Psychology, 44(1), 30–43. https://doi.10.1080/15374416.2013.814543 Becker, K.D., Wu, E.G., Hukill, A., Brandt, N., & Chorpita, B.F. (2021). How do mental health providers assess treatment engagement of youth and caregivers? Journal of Child and Family Studies, 30, 2527-2538. https://doi.org/10.1007/s10826-021-02042-x Brookman-Frazee, L.I. & Chlebowski, C. (2013). An Individualized Mental Health Intervention for ASD (AIM HI) Observational Coding System Manual. University of California, San Diego. Casagrande, K.A. & Ingersoll, B.R. (2017). Parent-mediated interventions for social communication in young children with ASD. In J.B. Leaf (Ed.), Handbook of social skills and autism spectrum disorder, Autism and Child Psychopathology Series (pp. 285-312). Springer International Publishing. https://doi.org/10.1007/978-3-319-62995-7_17 Chorpita, B.F. & Becker, K.D. (2022) Dimensions of treatment engagement among youth and caregivers: Structural validity of the REACH framework. Journal of Consulting and Clinical Psychology, 90(3), 258-271. https://doi.org/10.1037/ccp0000711 42 Cicchetti, D.V. (1994). Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychological Assessment, 6(4), 284-290. https://doi/10.1037/1040-3590.6.4.284 Clark, P. & Wheaton, B. (2007). Addressing data sparseness in contextual population research: Using cluster analysis to create synthetic neighborhoods. Sociological Methods & Research, 35(3), 311-351. https://doi.org/10.1177/0049124106292362 de Jong, K., Moerbeek, M., van der Leeden, R. (2010) A priori power analysis in longitudinal three-level multilevel models: An example with therapist effects. Psychotherapy Research, 20(3), 273-284. https://doi.org/10.1080/10503300903376320 Dickson, K.S., Zeedyk, S.M., Martinez, J., & Haine-Schlagel, R. (2017). Examining ethnic disparities in provider and parent in-session participation engagement. Journal of Children’s Services, 12(1), 47-58. https://doi.org/10.1108/JCS-12-2016-0022 Division for Early Childhood of the Council for Exceptional Children. (2014). DEC Recommended Practices. https://www.dec-sped.org/dec-recommended-practices Dong, N., Kelcey, B., Spybrook, J. (2018). Power analyses for moderator effects in three-level cluster randomized trials. The Journal of Experimental Education, 86(3), 489-514. https://doi.org/10.1080/00220973.2017.1315714 Dowell, K. A., & Ogles, B. M. (2010). The effects of parent participation on child psychotherapy outcome: A meta-analytic review. Journal of Clinical Child & Adolescent Psychology, 39(2), 151-162. https://doi.org/10.1080/15374410903532585 Eunice Kennedy Shriver National Institute of Child Health and Human Development (2019, April 19). Early intervention for autism. National Institutes of Health. https://www.nichd.nih.gov/health/topics/autism/conditioninfo/treatments/early- intervention Fawley-King, K., Haine-Schlagel, R., Trask, E. V., Zhang, J., & Garland, A. F. (2013). Caregiver participation in community-based mental health services for children receiving outpatient care. The Journal of Behavioral Health Services & Research, 40(2), 180-190. https://doi.org/10.1007/s11414-012-9311-1 Falkenström, F., Solomonov, N., & Rubel, J.A. (2020). Do therapist effects really impact estimates of within-patient mechanisms of change? A Monte Carlo simulation study. Psychotherapy Research, 30(7), 885-899. https://doi.org/10.1080/10503307.2020.1769875 Garland, A.F., Brookman-Frazee, L., & McLeod, B.D. (2008) Scoring Manual for the PRAC Study Therapy Process Observational Coding System for Child Psychotherapy: Strategies Scale. Department of Psychiatry, University of California, San Diego. Garland, A.F., Haine-Schlagel, R., Accurso, E.C., Baker-Ericzen, M.J., & Brookman-Frazee, L. (2012). Exploring the effect of therapists’ treatment practices on client attendance in 43 community-based care for children. Psychological Services, 9(1), 74-88. https://doi.org/10.1037/a0027098 Gearing, R.E., Schwalbe, C.S., Short, K.D. (2012). Adolescent adherence to psychosocial treatment: Mental health clinicians’ perspectives on barriers and promoters. Psychotherapy Research, 22(3), 317-326. https://doi.org/10.1080/10503307.2011.653996 Guan, K., Lau, A.S., Zhang, A., Chlebowski, C., Haine-Schlagel, R., & Brookman-Frazee, L. (2019). In-session caregiver behaviors during evidence-based intervention delivery for children with ASD in community mental health services. Evidence-Based Practice in Child and Adolescent Mental Health, 4(1), 55-71. https://doi.org/10.1080/23794925.2019.1565500 Haine-Schlagel, R. & Martinez, J.I. (2014). Parent Participation Engagement in Child Psychotherapy: Observational Coding Manual. San Diego State University. Haine-Schlagel, R., & Walsh, N. E. (2015). A review of parent participation engagement in child and family mental health treatment. Clinical Child and Family Psychology Review, 18(2), 133-150. https://doi.org/10.1007/s10567-015-0182-x Haine-Schlagel, R., Dickson, K.S., Lind, T., Kim, J.J., May, G.C., Walsh, N.E., Lazarevic, V., Crandal, B.R., & Yeh, M. (2022). Caregiver participation engagement in child mental health prevention programs: A systematic review. Prevention Science, 23, 321-339. https://doi.org/10.1007/s11121-021-01303-x Haine-Schlagel, R., Dickson, K.S., Shapiro, A.F., May, G.C., & Cheng, P. (2019). Parent mental health problems and motivation as predictors of their engagement in community-based child mental health services. Children and Youth Services Review, 104, 104370. https://doi.org/10.1016/j.childyouth.2019.06.005 Haine-Schlagel, R., Martinez, J.I., Roesch, S.C., Bustos, C.E., & Janicki, C. (2018). Randomized trial of the Parent and Caregiver Active Participation Toolkit for child mental health treatment. Journal of Clinical Child & Adolescent Psychology, 47, S150-S160. https://doi.org/10.1080/15374416.2016.1183497 Hansen, D. J., & Warner, J. E. (1994). Treatment adherence of maltreating families: A survey of professionals regarding prevalence and enhancement strategies. Journal of Family Violence, 9(1), 1-19. https://doi.org/10.1007/BF01531965 Hebbeler, K., Spiker, D., Bailey, D., Scarborough, A., Mallik, S., Simeonsson, R., Singer, M., & Nelson, L. (2007). Early intervention for infants and toddlers with disabilities and their families: Participants, services, and outcomes. January 2007. Final report of the National Early Intervention Longitudinal Study (NEILS). SRI International. https://www.sri.com/wp-content/uploads/2021/12/neils_finalreport_200702.pdf Holly, L. E., Fenley, A. R., Kritikos, T. K., Merson, R. A., Abidin, R. R., & Langer, D. A. (2019). Evidence-base update for parenting stress measures in clinical samples. Journal 44 of Clinical Child & Adolescent Psychology, 48(5), 685-705. https://doi.org/10.1080/15374416.2019.1639515 IBM Corp (Released 2021). IBM SPSS Statistics for Windows, Version 28.0. Armonk, NY: IBM Corp. Ingersoll B. & Dvortcsak, A. (2010). Teaching Social Communication to Children with Autism: A Practitioner's Guide to Parent Training. Guilford Press. Ingersoll, B. & Wainer, A. (2013). Initial efficacy of Project ImPACT: A parent-mediated social communication intervention for young children with ASD. Journal of Autism and Developmental Disorders, 43, 2943-2952. https://doi.org/10.1007/s10803-013-1840-9 Karver, M.S., Handelsman, J.B., Fields, S., Bickman, L. (2006). Meta-analysis of therapeutic relationship variables in youth and family therapy: The evidence for different relationship variables in the child and adolescent treatment outcome literature. Clinical Psychology Review, 26(1), 50-65. https://doi.org/10.1016/j.cpr.2005.09.001 King, G., Currie, M., & Petersen, P. (2014). Review: Child and parent engagement in the mental health intervention process: A motivational framework. Child and Adolescent Mental Health, 19(1), 2-8. https://doi.org/10.1111/camh.12015 Kuhn, J., Hickey, E., Lindly, O., Stransky, M., Masaro, M., Patts, G. J., Cabral, H., Crossman, M., Augustyn, M., Feinberg, E., & Broder-Fingert, S. (2023). Parenting stress, child social functioning, and Part C Early Intervention in predominantly low-income families of children with or at high risk for developmental delay from minoritized racial and ethnic groups. Journal of Early Intervention, 45(4), 467-487. https://doi.org/10.1177/10538151231155406 Kurzrok, J., McBride, E., & Grossman, R. B. (2021). Autism-specific parenting self-efficacy: An examination of the role of parent-reported intervention involvement, satisfaction with intervention-related training, and caregiver burden. Autism, 25(5), 1395-1408. Lau, A. S., Gonzalez, J. C., Barnett, M. L., Kim, J. J., Saifan, D., & Brookman-Frazee, L. (2018). Community therapist reports of client engagement challenges during the implementation of multiple EBPs in children’s mental health. Evidence-Based Practice in Child and Adolescent Mental Health, 3(3), 197-212. https://doi.org/10.1080/23794925.2018.1455545 Lee, E. & Hong, S. (2021). Adequate sample sizes for a three-level growth model. Frontiers in Psychology, 12(685496), 1-9. https://doi.org/10.3389/fpsyg.2021.685496 Maas, C.J. & Hox, J.J. (2004). Robustness issues in multilevel regression analysis. Statistica Neerlandica, 58(2), 127-137. https://doi.org/10.1046/j.0039-0402.2003.00252.x Maas, C.J. & Hox, J.J. (2005) Sufficient sample sizes for multilevel modeling. Methodology, 1(3), 86-92. https://doi.org/10.1027/1614-1881.1.3.86 45 Magnusson, K., Andersson, G., & Carlbring, P. (2018). The consequences of ignoring therapist effects in trials with longitudinal data: A simulation study. Journal of Consulting and Clinical Psychology, 86(9), 711-725. http://dx.doi.org/10.1037/ccp0000333 Martinez, J. & Haine-Schlagel, R. (2018). Observational assessment of engagement strategies to promote parent homework planning in community-based child mental health treatment: A pilot study. Journal of Child and Family Studies, 27, 1968-1980. https://doi.org/10.1007/s10826-018-1030-7 McNeish, D. & Wentzel, K.R. (2017). Accommodating small sample sizes in three-level models when the third level is incidental. Multivariate Behavioral Research, 52(2), 200-215. https://doi.org/10.1080/00273171.2016.1262236 Michigan Department of Health & Human Services (2023, Jan 31). Part C: Early On. https://www.michigan.gov/mdhhs/adult-child-serv/childrenfamilies/familyhealth/part-c- early-on_2 Miller, L.M., Southam-Gerow, M.A., & Allin, R.B. (2008). Who stays in treatment? Child and family predictors of youth client retention in a public mental health agency. Child & Youth Care Forum, 37, 153-170. https://doi.org/10.1007/s10566-008-9058-2 Nock, M. K., & Kazdin, A. E. (2005). Randomized controlled trial of a brief intervention for increasing participation in parent management training. Journal of Consulting and Clinical Psychology, 73(5), 872-879. https://doi.org/10.1037/0022-006x.73.5.872 Nock, M.K. & Photos, V. (2006). Parent motivation to participate in treatment: Assessment and prediction of subsequent participation. Journal of Child and Family Studies, 15(3), 345- 358. https://doi/10.1007/s10826-006-9022-4 Noyes-Grosser, D. M., Elbaum, B., Wu, Y., Siegenthaler, K. M., Cavalari, R. S., Gillis, J. M., & Romanczyk, R. G. (2018). Early intervention outcomes for toddlers with autism spectrum disorder and their families. Infants & Young Children, 31(3), 177-199. https://doi.org/10.1097/IYC.0000000000000121 Patterson, G.R. & Forgatch, M.S. (2001). Therapist behavior as a determinant for client noncompliance: A paradox for the behavior modifier. In C.E. Hill (Ed.), Helping skills: the empirical foundation (pp. 271-283). American Psychological Association. https://doi/10.1037/10412-016 Pellecchia, M., Ingersoll, B., Marcus, S.C., Rump, K., Xie, M., Newman, J., Zeigler, L., Crabbe, S., Straiton, D., Chavez, E.C., Mandell, D.S. (2023). Pilot randomized trial of a caregiver-mediated naturalistic developmental behavioral intervention in Part C Early Intervention. Journal of Early Intervention, 1-23. https://doi.org/10.1177/10538151231217462 Pellecchia, M., Mandell, D.S., Beidas, R.S., Dunst, C.J., Tomczuk, L., Newman, J., Zeigler, L., & Stahmer, A.C. (2022). Parent coaching in early intervention for autism spectrum 46 disorder: A brief report. Journal of Early Intervention, 10538151221095860. https://doi.org/10.1177/10538151221095860 Pereira, A. I., Muris, P., Mendonça, D., Barros, L., Goes, A. R., & Marques, T. (2016). Parental involvement in cognitive-behavioral intervention for anxious children: Parents’ in-session and out-session activities and their relationship with treatment outcome. Child Psychiatry & Human Development, 47(1), 113-123. https://doi.org/10.1007/s10578-015-0549-8 Peterson, C. A., Luze, G. J., Eshbaugh, E. M., Jeon, H.-J., & Kantz, K. R. (2007). Enhancing parent- child interactions through home visiting: Promising practice or unfulfilled promise? Journal of Early Intervention, 29(2), 119-140. https://doi.org/10.1177/105381510702900205 Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., & R Core Team (2023). nlme: Linear and Nonlinear Mixed Effects Models, version 3.1-164. https://cran.r- project.org/web/packages/nlme/index.html Romano, M. & Schnurr, M. (2022). Mind the gap- Strategies to bridge the research-to-practice divide in early intervention caregiver coaching practices. Topics in Early Childhood Special Education, 42(1), 64-76. https://doi.org/10.1177/0271121419899163 Sawyer, B. E., & Campbell, P. H. (2017). Teaching caregivers in early intervention. Infants & Young Children, 30(3), 175–189. https://doi.org/10.1097/IYC.0000000000000094 Solish, A. & Perry, A. (2008). Parents’ involvement in their children’s behavioral intervention programs: Parent and therapist perspectives. Research in Autism Spectrum Disorders, 2(4), 728-738. https://doi.org/10.1016/j.rasd.2008.03.001 Stadnick, N.A., Haine-Schlagel, R., & Martinez, J.I. (2016). Using observational assessment to help identify factors associated with parent participation engagement in community- based child mental health services. Child & Youth Care Forum, 45, 745-758. https://doi.org/10.1007/s10566-016-9356-z Stevens, J., Kelleher, K. J., Ward-Estes, J., & Hayes, J. (2006). Perceived barriers to treatment and psychotherapy attendance in child community mental health centers. Community Mental Health Journal, 42(5), 449-458. https://doi.org/10.1007/s10597-006-9048-5 Straiton, D., Frost, K., & Ingersoll, B. (2023). Factors that influence clinical decisions about offering parent coaching for autistic youth served within the Medicaid system. Implementation Research and Practice, 4. https://doi.org/10.1177/26334895231153631 Teti, D.M. & Gelfand, D.M. (1991). Behavioral competence among mothers of infants in the first year: The mediational role of maternal self-efficacy. Child Development, 62(5),918- 929. https://doi.org/10.1111/j.1467-8624.1991.tb01580.x Tomczuk, L., Stewart, R.E., Beidas, R.S., Mandell, D.S., & Pellecchia, M. (2022). Who gets coached? A qualitative inquiry into community clinicians’ decisions to use caregiver coaching. Autism, 26(3), 575-585. https://doi.org/10.1177/13623613211059499 47 U.S. Department of Education (2023). IDEA Section 618 Data Products: Static Tables Part C Child Count and Settings, Tables 2 and 5, 2021-2022. https://www2.ed.gov/programs/osepidea/618-data/static-tables/index.html Wainer, A. L., Pickard, K., & Ingersoll, B. R. (2017). Using web-based instruction, brief workshops, and remote consultation to teach community-based providers a parent- mediated intervention. Journal of Child and Family Studies, 26(6), 1592-1602. https://doi.org/10.1007/s10826-017-0671-2 48