Abstract

Background:

Caregiver-mediated interventions (CMIs) for young autistic children are supported by mounting evidence of efficacy. Attempts to identify child-level factors that predict treatment response have yielded inconsistent findings, with very few such studies focused on the toddler age range.

Objective:

Secondary analysis of data from a randomized control trial (NCT03215394) was conducted to explore predictive relationships between toddlers’ communication-related abilities and response to treatment in a CMI program. Participants. Sixty-seven toddler-caregiver dyads (62 mothers, 4 fathers, 1 grandmother) participated across three Canadian sites. Participating toddlers (52 boys, 15 girls) had a diagnosis of autism or early signs thereof and ranged in age from 18 to 32 months (M = 25.95 months).

Methods:

All dyads participated in a 12-week CMI (the Social ABCs). Communication-related skills were assessed at baseline using a direct standardized assessment of expressive language, a caregiver interview to capture receptive and expressive communication, and a parent-report inventory to document words used and understood. Analyses. A series of regressions with simple mediation analyses were performed to identify communication-related predictors of intervention change scores for caregiver fidelity and toddler outcomes. In all cases, toddler age was entered as a mediator to explore any influence on these relationships. Results. Toddlers’ functional communication inversely predicted their caregivers’ mastery of program strategies. Specifically, lower toddler functional receptive and expressive communication predicted larger gains in caregivers’ strategy use. Conversely, higher baseline (receptive but not expressive) communication-related skills predicted greater gains in toddlers’ responsivity. Trends emerged for prediction of toddler smiling and social orienting. The clinical measure selected for this study did not predict outcomes.

Conclusion:

Caregiver-reported information about toddler communication-related skills was informative for predicting the primary program outcomes. Findings may be helpful in informing clinical triage decisions and personalizing intervention approaches.

Introduction

Early intervention for toddlers on the autism spectrum (formally, autism spectrum disorder / ASD) can lead to developmental progress in a range of domains including language, communication, social skills, and cognitive abilities (Dawson et al., 2010; Sandbank et al., 2020, 2023). To date, most empirical studies and systematic reviews have focused on group-level outcomes, with only a small number of studies exploring whether (or which) child-level factors predict treatment response, an evidence gap initially highlighted over a decade ago (Vivanti et al., 2014). Given substantial heterogeneity in autism presentation across individuals, it stands to reason that particular intervention elements or approaches might vary in terms of “fit” (Lai et al., 2020) across children with varying patterns of strengths, learning needs, and preferences, and at different developmental stages. For example, Koegel and colleagues proposed that “interventions may be more appropriate for subpopulations of individuals with ASD depending on their verbal levels at intake” (2020, p. 2958). The heterogeneity of presentations across the autism spectrum, together with the pressing need to provide evidence-based supports as soon as first signs emerge, establishes the importance of examining factors that predict response to intervention in the toddler years. Moreover, understanding child-level factors that predict which children will benefit most from which approaches may be informative for clinical triage processes, program planning, and advocacy (see Lord et al., 2022).

Amongst baseline child-level characteristics, cognition (IQ), adaptive functioning, and autism characteristics have been demonstrated to predict response to intervention across several studies (e.g., Frazier et al., 2021; Grzadzinski et al., 2023; Koegel et al., 2020; Landa, 2018; Perry et al., 2011; Smith et al., 2015; Vivanti et al., 2014). Pre-intervention language skills or related communication abilities also hold promise as predictors, but findings to date have been mixed. One challenge with interpretation across studies is the use of different measures that capture diverse aspects of communication. Communication entails a broad array of skills involved in sharing information with another person, including through expressive language (use of words, signs, and gestures to express oneself), receptive language (understanding another person’s message), and other modes of sharing information such as facial expressions, gaze, or tone of voice. Following Koegel et al. (2020) the term “communication-related skills” will be used in the current paper to refer broadly to language and other forms of communication. However, when discussing particular research findings or measures, the relevant specific terminology will be used wherever this level of detail is available.

Varied associations have been reported between communication-related skills and treatment response in autism intervention research. Namely, some studies report associations in a positive direction, wherein individuals who begin with stronger skills make larger gains (e.g., Casenhiser et al., 2011 – total language; Tager-Flusberg and Kasari, 2013 – spoken words; Virues-Ortega et al., 2013 – expressive language), whereas others report the opposite (e.g., Siller et al., 2013 found a larger intervention response for children with lower expressive language skills at baseline). Other studies have failed to demonstrate such an association (Green et al., 2010; Hardan et al., 2015) or have identified complex interactions between variables. For example, Fossum et al. (2018) identified a profile predictive of positive response to intervention that included preschoolers with high expressive language ability, positive affect, and appropriate toy contact, combined with reduced social avoidance and repetitive vocalizations. Thus, although baseline communication-related skills have potential as informative predictors, consistent patterns have yet to be firmly established. Moreover, prediction of outcomes in the younger years remains under-studied, resulting in an evidence gap with respect to interventions for toddlers.

Naturalistic developmental behavioral intervention (i.e., NDBI; Schreibman et al., 2015) approaches, which have gained traction over the past decade, provide some insights into the role of communication-related skills in early intervention with toddlers. A solid body of evidence demonstrates group-level developmental progress across studies that use NDBI approaches (e.g., see Sandbank et al., 2020, 2023; Song et al., 2025, for evidence syntheses). In efforts to understand which children are best served by NDBI models, some researchers have examined outcomes specifically for young children who enter intervention with relatively limited expressive language abilities. Two recent studies (Gengoux et al., 2019; McDaniel et al., 2020) report on outcomes from autistic children aged 2–5 years who also presented with expressive language delays (i.e., with scores 1–3 standard deviations below the mean, depending on the child’s age, as measured on the Preschool Language Scale (PLS-5); Zimmerman et al., 2011). In both studies, children participated in a 12-week NDBI, called the ‘Pivotal Response Treatment Package’ (PRT-P) which is an in-home program delivered by therapists, with associated parent training. Group-level increases were reported for children’s word use and vocal responsiveness immediately or 12 weeks following program completion (Gengoux et al., 2019; McDaniel et al., 2020, respectively), when compared to controls receiving community treatment as usual. Although these findings demonstrate positive outcomes for young children with delayed language, the predictive power of different language profiles was not examined.

Barrett et al. (2020) similarly examined outcomes from a 6-month PRT-based NDBI program (‘Pivotal Response Intervention for Social Motivation’ or ‘PRISM’ intervention) with children aged 1.5–4.5 years. Compared to waitlist controls, children in the treatment condition demonstrated gains in social responsiveness to their caregivers and increased spoken language (mean length of utterance). Notably, the intervention was found to be particularly effective for a small subgroup that the authors defined as “minimally verbal” based on having used fewer than 30 functional spoken words during an initial play interaction – this subgroup experienced particularly large increases in responsivity and word use following the intervention, demonstrating a relative advantage for children starting with fewer spoken words. These findings mirror the earlier findings by Siller et al. (2013), wherein young children with baseline expressive language abilities below a 12-month level exhibited the largest expressive language gains in response to a caregiver “capacity building” program designed to foster coordinated play between caregiver and child (‘Focused Playtime Intervention’; FPI). Taken together, these findings suggest that NDBI approaches may be particularly well-suited to children at relatively early stages of expressive language development.

The studies discussed above report on NDBI approaches that are delivered by therapists with some involvement from family caregivers (i.e., concomitant parent training is a cornerstone of NDBI models; Smith et al., 2015). However, in some NDBI models, the intervention is delivered exclusively by a family caregiver (usually a parent) who is trained to implement the intervention strategies – such a difference might uncover unique predictors. These caregiver-mediated intervention (CMI) approaches often focus on toddlers under 3 years old (see Zhao et al., 2025, for a review), with some studies enrolling children as young as 1 year of age (Kasari et al., 2010; Rogers et al., 2012, 2019; Stahmer et al., 2020; Yoder et al., 2021), allowing for exploration of early developmental factors. Across studies, CMI approaches appear to be acceptable and feasible (e.g., see Pellecchia et al., 2024, for a review), and they yield group-level gains in various aspects of toddler development and skill acquisition (e.g., responsiveness to caregiver, adaptive and social communication skills; see review by Zhao et al., 2025). However, as with studies in older children, the examination of individual predictors is scant in this younger age range. Moreover, “language-level” groupings that are conceptualized in studies of older children are not likely to be applicable to toddlers. For example, using a cut-off of <30 functional spoken words to identify a “minimally verbal” subgroup may not be meaningful in the first 2 years of life because “nearly all toddlers younger than about 18 months of age are NV [non-verbal] or MV [minimally verbal] … However, one could argue that limited verbal ability during the toddler years is meaningfully different from that during later childhood due to the developmental trajectory of language” (Koegel et al., 2020, p. 2958). As such, the extent to which emerging communication-related abilities predict treatment response in autistic toddlers is particularly understudied and remains poorly understood. Extrapolating from older samples may not yield accurate information to guide clinical decisions about who will benefit from which interventions and when. Direct examination of predictors in the toddler years has become possible with the rise of interventions developed for autistic toddlers.

One such CMI program, the Social ABCs, was developed specifically for use with toddlers (≤ age 3 years) with emerging or confirmed autism, and is delivered exclusively by a family caregiver. The standard version of the program entails 12 weeks of individual parent-coaching (tapering over time) supported by a Parent Manual. The program entails coaching caregivers in the use of 10 specific strategies to enhance child social attention and vocal communication directed to the parent, as well as positive affect-sharing between caregiver and child. Each caregiver receives live, in-the-moment coaching while interacting with their toddler, an approach that has been identified to be particularly effective (referred to as “real-time” “triadic intervention” in Sone et al., 2021). Research findings, across pilot, randomized control trial, and community implementation studies, demonstrate that participation in the Social ABCs yields increased use of enhancing strategies by caregivers (i.e., implementation fidelity) and group-level gains in toddlers’ early social communication skills (i.e., vocal responsivity, social orienting; see Brian et al., 2016, 2017, 2022a, for further details). An abbreviated, 6-week, group-based adaptation of the program has also demonstrated positive findings from a large pilot evaluation (Brian et al., 2022b). An attempt to identify predictors did not yield any informative family- or treatment-level factors, including marital status, number of languages spoken in the home, frequency of English spoken in the home, which caregiver was coached, their ethnicity, occupation, educational attainment, or the number of coaching sessions attended (note that all had attended a minimum of 5 sessions; Brian et al., 2022b). The predictive role of children’s communication-related skills before starting the program has not been explored.

Purpose of the current study

The primary purpose was to understand the role of toddlers’ communication-related abilities on their response to the standard, 12-week, 1:1 version of the Social ABCs. Given the importance of caregiver fidelity in CMI models (i.e., all therapeutic components are delivered by the caregiver, and previous findings show associations between fidelity and child progress; Brian et al., 2022a), we also examined caregiver fidelity as an outcome. Program enrollment in the Social ABCs places no restrictions on minimal or maximal developmental levels (i.e., there were no exclusion criteria based on toddlers’ expressive language, other communication-related skills, or cognitive abilities), resulting in the inclusion of toddlers across a wide range of levels at program entry. This enables the examination of possible associations between early communication-related skills and outcomes. In light of challenges with defining a “minimally verbal” subgroup in this young age range, the current study explored communication-related skills as continuous, rather than categorical, variables. Moreover, information from multiple sources was used to capture different aspects of communication-related skills across participants (details below).

Hypotheses

Despite variability across findings and limited evidence from toddlers, the following hypotheses were proposed: Toddlers’ baseline expressive language skills, expressive communication, and word use will be inversely associated with change in (1) caregiver fidelity and (2) toddler vocal responsivity. Examination of receptive communication-related predictors and secondary (non-verbal) Social ABCs outcomes (social orienting and shared smiling) was exploratory.

Methods

The current study entailed secondary analysis of data from a randomized control trial (RCT) evaluating the Social ABCs with an attention training co-intervention (Clinicaltrials.gov identifier NCT03215394). Primary intervention outcomes paralleled previous findings (Brian et al., 2017, 2022a), revealing significant group-level gains in caregiver fidelity (i.e., strategy use), and toddler responsivity, smiling, and social orienting (Brian et al., 2025). Toddlers were assessed prior to program entry (at baseline; BL) using one domain of a standardized direct clinical assessment measure, a standardized parent interview of children’s everyday communication, and a parent questionnaire of vocabulary (measures described below). Video-coded data were used to characterize the following outcomes: Caregiver implementation fidelity (primary parent outcome), and toddler responsivity (primary toddler outcome), initiations, orienting toward caregiver, and social smiling. The change scores (post- vs. pre-intervention) for these target skills were used analyses.

Attention training co-intervention

Prior to beginning the Social ABCs, participants were randomly assigned to complete 4 weeks of computer-based Attention Training (AT) or a control condition (viewing child-appropriate video segments) to explore whether attention training would impact (i.e., “prime”) participants’ response to the Social ABCs. The methods and results of the AT phase are described in Sacrey et al. (2025). The main Social ABCs outcomes and the impact of AT on Social ABCs were among the primary analyses in the RCT (reported by Brian et al., 2025) but are not the focus of the current paper.

Before conducting the main analyses of interest, preliminary analyses examined whether AT had any influence on Social ABCs change scores, using non-parametric K-independent samples, to explore appropriateness for inclusion in regression analyses. In cases where regressions used change scores that differed between AT and control conditions, condition would be entered as a covariate. The impact of AT on the variables of interest was assessed using Mann–Whitney U. There was no significant difference between the two AT conditions on either of the six Social ABCs change scores (all ps > 0.05). As such, AT condition was not included as a covariate in the following analyses and is not included in the results and discussion below. All participants received the Social ABCs following the AT (or control) phase.

Participants

See Table 1 for family demographic information and Table 2 for toddlers’ baseline characteristics. Sixty-seven families (62 mothers, 4 fathers, 1 grandmother) participated in a randomized control trial (RCT) of the Social ABCs across three sites in Canada (Toronto, Ontario; Edmonton, Alberta; and Halifax, Nova Scotia). Toddlers (52 boys, 15 girls) ranged in age from 18–32 months (M = 25.95 months), were born at 37–42 weeks’ gestation weighing ≥ 2,500 g, and had a confirmed diagnosis (n = 32) or clinically significant early signs of autism. Early signs were based on caregiver-reported developmental concerns and clinical judgement informed by scores on the assessment measures described below, which were conducted prior to beginning the Social ABCs (i.e., following the AT co-intervention). Ten-minute video segments were collected at two key time points: (1) During a baseline free-play interaction between caregiver and toddler, and (2) following the final week 12 Social ABCs coaching session. No coaching took place during the video collection sessions and caregivers were instructed to play or interact with their child as they typically do. All participants received the 12-week, individual Social ABCs intervention (most in-home, with a small number completing the program virtually due to COVID-19 isolation measures). Table 1 shows caregiver ethnicity (41.8% identified as Black, Indigenous, or a Person of Color (BIPOC), 44.8% White, and 11.9% did not disclose); educational attainment (high school or less: 19.4%; college or trade school: 22.4%, undergraduate university degree: 26.9%, graduate school: 13.4%); and marital status (61.2% married, 13.4% in a common law relationship, 10.5% separated or never married – none were divorced). 39.1% of families had two children, and 44.9% of the toddlers were first-born.

Caregiver/family characteristicN (%)EthnicityBIPOC1: 28 (41.8)
White: 30 (44.8)
Not disclosed: 8 (11.9)
Missing: 1 (1.5)Educational attainmentHighschool or less: 13 (19.4)
College2/Trade school: 15 (22.4)
University undergraduate degree: 18 (26.9)
Graduate degree: 9 (13.4)
Missing: 12 (17.9)Marital statusMarried: 41 (61.2)
Common law: 9 (13.4)
Separated: 3 (4.5)
Never married: 4 (6.0)
Missing: 10 (14.9)Birth order of participantFirstborn: 31 (46.3)
Second: 20 (29.8)
Third or later: 8 (11.9)
Missing: 8 (11.9)Number of children in familyOne: 17 (25.4)
Two: 27 (40.3)
Three or more: 14 (20.9)
Missing: 9 (13.4)

Family and caregiver characteristics.

1BIPOC: Black, Indigenous, Person of Colour; 2College: In the Canadian context, “College” typically refers to post-high school diploma or certificate programs (usually ≤ 2 years), whereas “University” typically refers to a Bachelor’s degree program (usually 3–4 years).

Baseline characteristicNMeanSDRangeToddler chronological age (CA, months)6725.953.6518–32ADOS-Toddler total score6219.905.923–28PROCESS1 total score5922.909.114–43MSEL2 EL – Age equivalent (AE, months)6313.796.974–36MSEL EL – Developmental quotient (DQ)36352.8723.7915–120MSEL VR – AE (months)6317.816.636–46MSEL VR – DQ6369.5825.9823–178CDI4 – Words understood60145.80113.290–393CDI – Words spoken6051.4595.970–377VABS5 RC – AE (months)6011.426.070–27VABS EC – AE (months)6111.885.534–26

Toddler baseline characteristics.

1Parent-Rated Observation of Communication, Emotion and Social Skills (previously APSI). 2Mullen Scales of Early Learning (EL: Expressive Language; VR: Visual Reception). 3DQ = AE/CA × 100. 4MacArthur Communicative Development Inventory – Words and Gestures. 5Vineland Adaptive Behavior Scale-3rd edition (VABS-3); RC: Receptive Communication; EC: Expressive Communication.

MeasuresVideo-coded variables

Video-coded data (based on 10-min toddler-caregiver free-play segments) were used to characterize caregiver implementation fidelity as well as toddler outcomes (defined below). All video coding was conducted using well-established coding procedures developed for and used in previous studies, blinded to study time point, with interrater reliability established for 20% of videos (see Brian et al., 2016, 2017, 2022a, 2022b).

Implementation fidelity

As outlined by Koegel and Koegel (2006) and consistent with previous work (Brian et al., 2016, 2017, 2022b), caregiver implementation fidelity was calculated via continuous interval coding, consisting of ten, 1-min intervals. Each interval was coded as correct or incorrect / not used across 10 program techniques for which the caregiver received coaching: Child choice, child attending, clear opportunity, contingent reinforcement, natural reinforcement, and reinforcement of attempts, shared control, pace, recast, and positive emotion (see Brian et al., 2017 for definitions of these strategies). Implementation fidelity is reported as percentage (%) of intervals during which the caregiver demonstrated appropriate / correct use of techniques.

Toddler and dyad outcomes

For the current study, we used change scores (post - pre) for the following behaviors as the dependent variables: (1) Toddler responsivity to caregiver-provided communication opportunities. This included any toddler vocalization, directed to the caregiver, that occurred immediately following the caregiver’s single-word cue (“model-prompt”), calculated as: % Responsivity = [(number of responses ÷ number of caregiver language opportunities) × 100]. (2) Toddler initiations – coded as number (rate/min) of toddler-led vocal overtures directed to the caregiver; only considered an initiation when there was no preceding directive action from the caregiver (cue, request, question) within the prior 3 s. (3) Social orienting – interval-coded as present when the toddler’s head was oriented toward the caregiver’s torso or head/face at any time during a 5-s coding interval. This code did not require eye-to-eye gaze. (4) Smiling – coded separately for toddler and caregiver, as presence or absence of a clear smile per 5-s coding interval (based on detailed video coding instructions including visual representations of smiling codes); Shared smiling between caregiver and toddler entailed a count of all intervals where toddler and child were both smiling. Inter-rater reliability was moderate (Kappa = 0.50 for toddler responsivity) to substantial (Kappa = 0.70, 0.66, and 0.63 for adult smile, child smile, and orienting, respectively).

Clinical assessment measures

Vineland Adaptive Behavior Scales – 3rd edition (VABS-3; Sparrow et al., 2016). The VABS-3 is a semi-structured parent interview designed to assess adaptive functioning in everyday life. Adaptive behavior is assessed across four domains – Communication, Daily Living, Socialization, and Motor skills, outlined by typical developmental milestones anchored to specific ages. As a measure of adaptive functioning (vs. capacity), the VABS-3 captures the extent to which an individual engages in activities on a regular basis (“usually,” “sometimes,” or “never”) – the tool asks whether a child does, rather than can do each of the developmentally sequenced items. The Communication domain includes a “Receptive” subdomain that characterizes an individual’s understanding and responses to communication from others, such as looking toward caregivers when they are speaking, responding to gestures and facial expressions, understanding the meaning behind someone’s tone of voice, and understanding spoken language, signs, and other communicative input. The “Expressive” subdomain captures information about making sounds, using words/phrases/sentences (spoken or otherwise), using gestures, and using facial expressions to communicate. Age-equivalent scores from the Communication domain (Receptive and Expressive subdomains) were included in the current paper.

MacArthur Communicative Development Inventories – Words and Gestures (CDI, Infant Scale; Fenson et al., 2007) is a highly reliable and well-validated parent-report measure that provides an inventory of the child’s current abilities. The infant scale covers the period from 8 to 16 months but has also been used to map the developmental trajectories of older autistic children (e.g., Charman et al., 2003; Iverson, 2018). The CDI subscales capture words and phrases understood, words spoken, and gestural communication. The raw scores for total words understood (CDI-WU) and total words spoken (CDI-WS) were included in the current analyses.

Mullen Scales of Early Learning (MSEL; Mullen, 1995) is a standardized direct assessment of children’s developmental abilities from 0 to 68 months across domains, used widely in this population for research and clinical purposes. The measure has adequate-to-good psychometric properties (e.g., internal consistency ≥ 0.90) and good test–retest reliability (≥ 0.80 across 1–2 weeks). Due to the young age of participants and their limited tolerance for lengthy assessment, we prioritized the Expressive Language (EL) and Visual Reception domains of the MSEL. For this paper, we focused on the EL domain, with scores represented as age equivalents (AE; derived from raw scores) and developmental quotients (DQ; calculated as [AE/chronological age (CA)] × 100). We elected not to use the MSEL T-scores as many children (~40% of our sample) had T-scores ≤ 20, yielding inadequate sensitivity to explore the associations of interest (i.e., the scale does not provide specific T-scores when a child’s score is < 20). The Expressive Language domain of the MSEL focuses mostly on vocal and verbal output.

The Autism Diagnostic Observation Schedule (ADOS-2), Toddler Module (Lord et al., 2012) was used to characterize toddlers’ autism-related profiles in the areas of social communication (Social Affect domain; SA) and repetitive /restricted /intense behaviors and/or interests (Restricted and Repetitive Behavior domain; RRB). The Toddler module has strong internal consistency for the SA domain (Cronbach’s α = 0.88–0.90) but poor for the RRB domain (α = 0.50); as reported by McCrimmon and Rostad (2014). The ADOS was used to characterize the sample but not examined as a predictor.

The Parent-Rated Observation of Communication, Emotions, and Social Skills (PROCESS©; formerly APSI; Sacrey et al., 2018; Bryson et al., 2007) was used to capture parent-reported autism-related characteristics of toddlers at the beginning of the study. This is a 26-item forced-choice questionnaire (“yes – sometimes – no”), with higher scores representing more frequent and/or marked autism-related characteristics, and age-specific cut-offs. Sensitivity and specificity for total scores at 18 months are 0.65 and 0.72 (based on a cutoff of 9; Sacrey et al., 2018). The PROCESS© questionnaire was used to characterize the sample but not examined as a predictor.

Analytic plan

The influence of baseline communication-related skills on Social ABCs outcomes was assessed using five putative predictors: MSEL Expressive Language age equivalent (MSEL-EL-AE), VABS-3 Receptive and Expressive Communication age equivalents (VABS-RC-AE and VABS-EC-AE), and MacArthur CDI Words Understood (CDI-WU) and Words Spoken (CDI-WS). A series of regressions with simple mediation analyses were performed using the Hayes (2013) PROCESS macro to identify communication-related predictors of intervention change scores. In all cases, toddler age was entered as a mediator to see if age influenced any of these relationships. The PROCESS macro was chosen over other analytic approaches (such as Mixed Modeling) because it allows for continuous predictor variables, as well as both mediation and covariates. Baseline age was used as a mediator rather than a covariate in order to determine how or why age may influence any particular relationship (mediation) rather than removing an effect of age (covariate). Due to multiple tests, a more conservative p-value of ≤ 0.01 was used for significance testing. Bootstrapping with 5,000 resamples was used to calculate the 95% confidence intervals for the indirect effect. Significant effects on the variables of interest (direct predictors and those that are mediated by age) are described in detail in each section below.

ResultsInfluence of communication-related skills on social ABCs outcomes

The results of the regression analyses are summarized in Table 3. Only the significant results, and sub-threshold trends related to program objectives, are described in detail below.

Social ABCs outcome variablePredictorFidelityResponsivityInitiationsOrientingChild SmilingParent SmilingMullen expressive language – age equivalent (MSEL-EL-AE)Direct−0.57 (0.04)0.01 (0.11)0.002 (0.76)−0.001 (0.80)0.001 (0.63)0.002 (0.55)Age0.18 (0.02)0.16 (0.03)0.17 (0.02)0.18 (0.02)0.18 (0.012)0.18 (0.012)Control Age−0.41 (0.08)0.02 (0.06)0.01 (0.52)−0.001 (0.84)0.003 (0.17)0.002 (0.59)MediationnsnsnsnsnsnsVineland receptive communication (VABS-RC-AE)Direct−0.81 (<0.01)*0.03 (<0.01)*0.001 (0.46)−0.009 (0.03)0.003 (0.16)−0.001 (0.75)Age0.09 (0.31)0.06 (0.52)0.09 (0.31)0.07 (0.40)0.07 (0.40)0.07 (0.40)Control Age−0.72 (0.02)0.03 (<0.01)*0.01 (0.36)−0.001 (0.03)0.004 (0.09)−0.001 (0.76)Mediation21% (<0.01)17% (ns)nsnsnsnsVineland expressive communication (VABS-EC-AE)Direct−1.04 (0.003)*0.02 (0.07)0.006 (0.55)−0.01 (0.06)0.003 (0.29)−0.003 (0.27)Age0.16 (0.10)0.11 (0.26)0.14 (0.12)0.15 (0.11)0.15 (0.11)0.15 (0.11)Control Age−0.90 (0.001)*0.02 (0.05)0.001 (0.41)−0.001 (0.07)0.004 (0.11)−0.003 (0.28)Mediation23% (<0.01)nsnsnsnsnsMacArthur CDI1 Words Understood (CDI-WU)Direct0.03 (0.02)0.002 (0.001)*0.001 (0.07)−0.002 (0.35)0.001 (0.70)0.001 (0.78)Age0.007 (0.17)0.003 (0.52)0.005 (0.29)0.005 (0.32)0.005 (0.32)0.004 (0.32)Control Age−0.03 (0.12)0.002 (0.001)*0.001 (0.05)0.004 (0.36)0.001 (0.43)0.001 (0.78)Mediation13% (ns)24% (<0.01)nsns13% (ns)nsMacArthur CDI Words Spoken (CDI-WS)Direct0.02 (<0.01)*0.001 (0.03)0.001 (0.34)−0.001 (0.81)0.001 (0.84)0.001 (0.70)Age0.02 (0.005)*0.01 (0.011)0.01 (0.004)*0.02 (0.004)*0.02 (0.004)*0.02 (0.004)*Control Age−0.04 (0.06)