The study examines whether the average performance of the population with dyslexia in a working memory measure can be inferred dimensionally from the characteristics of the typical population. Specifically, we focused on Associative Phonological Working Memory (APWM), an ability that we predicted being impaired in dyslexia due to the relationship of reading with both associative learning and working memory (WM). Study 1a revealed a linear relationship between APWM and reading ability in 438 typically-developing (TD) children, after accounting for fluid intelligence and phonological WM. In Study 1b a simulation procedure was used to calculate the deficit in APWM expected in children with dyslexia, based on the set of correlations found in Study 1a. This prediction was compared with the actual performance of 26 children with dyslexia. A deficit in APWM was confirmed, and its extent was in line with that simulated from the correlational structure seen in the TD population.