Latent Change Score Analysis of Writing Development over 12 Months in School-Age Children with Autism Spectrum Disorder

Background: Children with autism spectrum disorder (ASD) experience challenges with literacy development, commonly demonstrating challenges with lower-order (i.e., transcription) and higher-order (i.e., text generation) writing abilities. These difficulties have often been explored cross sectionally, but no studies have investigated the writing abilities of children with ASD using multiple time points.

Objectives: 1) To examine change in writing across two time points spaced roughly 12 months apart and 2) to explore predictors of change in writing abilities.

Methods: Seventy-two children with ASD participated in this study (Table 1). Children were aged 9-17 at the first time point (T1) and 10-18 at the second time point (T2). Time points were spaced roughly 12 months apart for all participants. Community diagnoses of ASD were confirmed with the ADOS-2. Cognitive ability (FIQ) was measured with the WASI-II, and all participants fell within the borderline to above average ranges. Writing performance was assessed using the Test of Written Language, 4^th Edition (TOWL-4), a narrative writing task where participants generate a story about a provided picture. Participants received different but equivalent forms at T1 and T2. Responses were scored for Contextual Conventions (CC; lower-order writing abilities) and Story Composition (SC; higher-order writing abilities). Word count (WC) was also recorded. Two trained research assistants scored all writing samples and demonstrated excellent interrater reliability on a sample subset (n = 34; αs > .90).

Latent change score models (LCSM) were used to examine change and predictors of change between time points. LCSMs model change between two time points as a latent change factor and allow for assessing the average change between time points, the variance in the change factor (the extent to which individuals differ in change over time), and the extent to which change is proportional at T2 based on T1 (and other predictors).

Results: See Table 2 for estimates and p values. All LCSMs were run using Mplus 7.3. Participants demonstrated significant change in CC and WC but not SC. Additionally, participants showed similar amounts of change for CC and WC but varied for SC. T1 CC and SC negatively predicted latent change (i.e., participants who initially scored lower at T1 improved at T2), but T1 WC showed no association to T2. Adding additional predictors showed CC and SC remained negatively predictive, FIQ positively predicted CC but not SC, and age and ADOS-2 were not predictive for either. For WC, no predictors significantly predicted latent change.

Conclusions: Findings demonstrate that children with ASD make distinct changes in their writing abilities across 12 months. Participants showed more change in lower-order writing abilities (i.e., CC and WC) compared to higher-order writing abilities (i.e., SC). However, while initial lower-order and higher-order abilities predicted later change, initial WC did not predict later change. This provides potential further support to the graphomotor challenges experienced by children with ASD. Furthermore, FIQ appeared to positively influence growth for SC but not for CC or WC. Findings suggest further research is needed to understand the developmental trajectories of writing development in children with ASD.