Objectives: The purpose of the present study was to factor analyze the 41 items comprising the four PIA-CV domains assessing social-communication development (i.e., Social Relating, Nonverbal Communication, Language Understanding, and Imitation) to examine the presence of an underlying factor structure.
Methods: An exploratory factor analysis was conducted on PIA-CV item scores from 223 children under age 48 months (mean CA=30.3mo, range=16.4–46.8mo) with or at risk for an autism diagnosis. Derived factors were then validated using data from 58 toddlers from the initial assessment of A Multi-Site Clinical Randomized Trial of the Hanen More than Words Intervention (mean CA=21.5 mo, range=15.5–25.0 mo) who met a predetermined cutoff on the Screening Tool for Autism in Two-Year-Olds (STAT) and had a clinical presentation consistent with ASD. A nomological network approach (Cronbach & Meehl, 1955) was utilized to assess the construct validity of proposed factors. Construct validity for individual factors was examined using the Mullen, Vineland, STAT, Infant/Toddler Social and Emotion Assessment (ITSEA), Early Social Communication Scales (ESCS), Parenting Stress Index (PSI) and Maternal Efficacy Scale (MES).
Results: Factor analysis yielded three factors, together accounting for 43% of the variance. Factor 1 contains 14 items (alpha=.89) representing language understanding, imitation, and joint attention. It was predicted to be associated with language (from the Mullen and Vineland, imitation (from the STAT and ITSEA), and directing attention (from the STAT and ESCS). Factor 2 contains 15 items (alpha=.88) reflecting instrumental communication and affection. It was predicted to be associated with requesting (from the STAT and ESCS). Factor 3 contains 11 items (alpha=.78) representing social engagement and enjoyment. It was predicted to be associated with social relatedness (from the ITSEA), interpersonal relationships (from the Vineland), parent child-interactions (from the PSI), and parenting efficacy (from the MES).
As predicted, Factor 1 correlated with language subscales on the Mullen (receptive r=.33, p=.02; expressive r=.39, p=.006) and Vineland (receptive r=.56, p<.001; expressive r=.48, p=.002). Factor 1 also correlated with the ITSEA imitation/play subscale (r=.69, p<.001) and with joint attention on the STAT (r=-.29, p=.04), but not with the ESCS. Correlations between Factor 2 and requesting measures were non-significant. Factor 3 was correlated with ITSEA social relatedness (r=.68, p<.001), Vineland interpersonal relationships (r=.64, p<.001), MES parenting efficacy (r=.72, p<.001), and PSI parent-child interactions (r=-.74, p<.001).
Conclusions: New PIA-CV factors show promise for assessing different domains of social-communicative development. Factors 1 and 3 were correlated with related constructs, as measured across observational, questionnaire, and interview methods. Future research will determine the utility of these factors for assessing response to the More Than Words intervention in young children with autism symptoms.