Background Working memory (WM) is the ability to store and process information over short time periods. WM is a strong predictor of educational attainment; which is important for health and wellbeing across the lifecourse. There is controversy about whether or not different aspects of WM are affected by socioeconomic position, and very little known about how ethnicity may shape these relationships. We studied these interrelationships in a longitudinal study of children in Bradford, a multi-ethnic city with high levels of deprivation.

Methods Born in Bradford (BiB) is a prospective birth cohort study following the lives of over 13,500 children and their families. As part of BiB, children completed three tasks of WM when they were aged 7–10 years. Ethnicity data were collected through classroom records (n=14,076), and socioeconomic data were linked from BiB baseline (at birth) (n=4916). Linear regression was used for all analyses. WM was analysed by age to provide an interpretation of the magnitude of the effect between socioeconomic and ethnic groups. We next analysed WM by (1) a latent-class measure of socioeconomic position at birth (with least deprived as baseline) and (2) nine different ethnic groups (White British – ethnic majority - as baseline). Finally, WM scores were presented by ethnic-specific groups of socioeconomic position for the ethnic majority and largest ethnic minority group (White British and Pakistani).

Results The difference between the least and most deprived socioeconomic groups was equivalent to at least a 1-year age difference (B=-6.02 [95% CI -7.51 to -4.54]). In comparison to White British children, Gypsy/Irish Traveller children had the lowest WM scores (B=-9.58 [-11.93 to -7.23]) (equivalent to a two-year age difference). Most other ethnic minority children scored higher than White British children for ≥1 task(s), for example, Pakistani children had higher scores on the forwards digit recall task (B= 3.12 [2.53 to 3.71]) (equivalent to a 9-month age difference). Finally, there was a social gradient in WM for White British children, but not for Pakistani children.

Conclusion Given the strong associations between WM and learning ability and the potential consequences for lifelong trajectories of health and wellbeing, these large socioeconomic and ethnic group differences in children’s WM are concerning. This study also found socioeconomic disadvantage was more detrimental for WM among ethnic majority children than for ethnic minority children; this may suggest that the negative effects of disadvantage are buffered by other factors for ethnic minorities, such as social support and own ethnic density.