The evaluation of a multiple strategies approach to teach upper primary students with high-functioning autism to draw social inferences from text

Abstract

The ability to draw inferences is essential for reading comprehension. This is because inferences connect textual information with readers' background knowledge, enabling them to move beyond surface reading and uncover what is implicitly stated in the text (Elleman, 2017; Graesser et al., 1994). However, research shows that learners with High-Functioning Autism (HFA) often struggle with processing inferences (Norbury & Nation, 2011; Westerveld et al., 2021). In particular, they face pervasive challenges with making social inferences about the emotions and intentions of characters (Bodner et al., 2015; Happé, 1994; White et al., 2009). This study had two aims. The first was to investigate whether an intervention combining explicit and visually cued instruction could help upper primary students with HFA improve their social inferential reading comprehension performance. The second aim was to develop an item bank of statistically equated reading passages to objectively measure students’ social inferential reading comprehension performance throughout the different experimental phases. In the first phase, an item bank of narrative passages was created by writing passages based on the ‘Strange Stories’ by Happé (1994). Test administration procedures and the calibration of difficulty levels for the passages followed this. The 101 passages were administered to 117 ten-year-old students from a mainstream primary school in Singapore. The difficulty level of passages was determined using Rasch analysis before creating 15 tests of comparable difficulty. Findings indicated that the logit range of the tests was between -1.43 and -1.68. The mean difficulty was 1.55. The standard deviation was 0.07. This suggested that the tests were statistically similar in difficulty. In the second phase, the multiple probe design was adopted as it enabled close monitoring of students’ performance, allowed replication across participants, reduced test fatigue, and did not require the treatment to be withdrawn. A multiple strategies intervention approach, comprising explicit and visually cued instruction, was applied across four ten-year-old students with HFA from three mainstream primary schools in Singapore to help them draw social inferences from short stories. Explicit instruction using a step-by-step modelling approach and error correction prompts addressed Weak Central Coherence (WCC) difficulties by ensuring students focus on the story context to draw social inferences. Visually cued instruction with the graphic organiser worksheet addressed Executive Function (EF) deficits, specifically those related to working memory impairments, helping students organise information to reduce cognitive load and enable them to focus on social inference-making. The graphic organiser also addressed Theory of Mind (ToM) deficits by breaking down the characters’ emotions and intentions, thereby facilitating social inferential reading comprehension. The effectiveness of the intervention was based on the improvement in students’ social inferential reading comprehension scores. There was an average improvement of 4.18 points (52.25%) in raw scores from baseline to intervention. Rasch analysis showed a mean of 2.73 logits growth between baseline and intervention. The non-parametric measure of effect size for the Percentage of Nonoverlapping Data (PND), Percentage of Data Points Exceeding the Median (PEM), Percentage of All Non-Overlapping Data (PAND) and Tau-U values suggested a medium to high effect. The randomisation test conducted using the Single-Case Randomisation Tests (SCRT-R) software yielded a p-value of 0.008, indicating that the intervention was highly effective.