Higher- and lower-level processing in strategic reading: Reconceptualising the Survey of Reading Strategies (SORS)

Abstract The Survey of Reading Strategies (SORS) is widely used to assess English-as-a-second-language (ESL) students’ metacognitive regulation of reading across three strategy dimensions (Global, Problem-Solving, Support), yet this three-factor structure has rarely undergone confirmatory validation. This study introduces HiLoLev , a theory-driven two-factor measurement reconceptualisation that distinguishes higher-level integrative/monitoring strategies from lower-level decoding/support strategies. Using confirmatory factor analysis (CFA) on data from 554 Swedish upper-secondary ESL students, legacy three-factor specifications (oblique, higher-order) were contrasted with the proposed theory-driven HiLoLev model based on SORS items. The original structures showed poor fit with high interfactor correlations and substantial cross-loadings; by contrast, an 11-item HiLoLev model fit well (RMSEA = 0.051, SRMR = 0.045) and offered clearer interpretability. While the full SORS remains pedagogically useful, its subscales should not be assumed to represent stable latent constructs for inferential analyses. For research applications (e.g., regression models, mediation analyses, or group comparisons), confirm the structure in-sample via CFA or consider using the 11-item HiLoLev model as a compact, cognitively grounded latent specification of SORS items.