Socioeconomic Biases in Education Explained: How Teachers, Parents, and Pupils Contribute to Tracking Decisions

In several educational systems, pupils are assigned to a specific school according to their ability (between-school tracking). This decision has significant implications for pupils’ future educational trajectories, including whether they can attend university. As track allocation is often based on the recommendation of a teacher, teachers play a role in the social reproduction within these systems. Studies across various European contexts have found that teacher tracking recommendations are biased against pupils from lower socioeconomic backgrounds (SEBs). Bias here is defined as the discrepancy in tracking recommendations for pupils from different SEBs with equivalent performance. This means that despite similar performance levels, pupils of higher SEBs were placed in higher ability tracks (overestimated), while those of lower SEBs were placed in lower ability tracks (underestimated). What remains less clear, however, is where this bias stems from. Teachers may consider both pupils and parents when making recommendations. Hence, I suggest disentangling the non-cognitive characteristics of the pupil (e.g., motivation) from the resources of the parents (e.g., parental involvement) in affecting teacher tracking recommendations. Using the Kitagawa-Oaxaca-Blinder (KOB) decomposition, I examine the extent to which the socioeconomic gap in tracking recommendations can be attributed to pupils’ characteristics versus parental resources, while adjusting for performance levels. Moreover, using this counterfactual decomposition, I examine whether the bias stems from different returns of characteristics/resources across SEBs (return effects), or rather from different levels of characteristics/resources across SEBs (endowment effects). To test these mechanisms, I use data from the PRIMary to Secondary education (PRIMS) project, merged with register data from the Netherlands Cohort Study on Education (NCO).[1]

[1] I have obtained preliminary results, but due to data confidentiality, I cannot report them yet. The outputs will be available for sharing well before the conference.