Background: A fundamental aim of standardized educational assessment is to achieve reliable discrimination between students differing in the knowledge, skills and abilities assessed. However, questions of the purity with which these tests index students’ genuine abilities have arisen. Specifically, literacy and numeracy assessments may also engage unintentionally assessed capacities. Aims: The current study investigated the extent to which domain-general processes – working memory (WM) and non-verbal reasoning – contribute to students’ standardized test performance and the pathway(s) through which they exert this influence. Sample: Participants were 91 Grade 2 students recruited from five regional and metropolitan primary schools in Australia. Methods: Participants completed measures of WM and non-verbal reasoning, as well as literacy and numeracy subtests of a national standardized educational assessment. Results: Path analysis of Rasch-derived ability estimates and residuals with domain-general cognitive abilities indicated: (1) a consistent indirect pathway from WM to literacy and numeracy ability, through non-verbal reasoning; (2) direct paths from phonological WM and literacy ability to numeracy ability estimates; and (3) a direct path from WM to spelling test residuals. Conclusions: Results suggest that the constitution of this nationwide standardized assessment confounded non-targeted abilities with those that were the target of assessment. This appears to extend beyond the effect of WM on learning more generally, to the demands of different assessment types and methods. This has implications for students’ abilities to demonstrate genuine competency in assessed areas and the educational supports and provisions they are provided on the basis of these results.