Understanding the Factors Associated with Academic Achievement

Portrait of young schoolboy leaning at desk with teacher teaching in background

Background

Current measures of deprivation, like free school meal (FSM) eligibility, are able to identify a large proportion of children in the UK who may benefit from additional school funding (for instance in the form of the Pupil Premium) but they do not capture the entire range of potential deprivation circumstances in which children may live. Other indicators may be better suited to describe these circumstances and to explain underachievement related to socioeconomic background factors.

Goals

The Department for Education (DfE) commissioned RAND Europe and the Faculty of Education at the University of Cambridge to assess the quality of the current measure of socioeconomic deprivation used by DfE. Specifically, the work aimed to assess the relationship between FSM eligibility, pupil achievement at the end of Key Stage 2 (age 11) and Key Stage 4 (age 16, when most children take GCSEs), and measures that may act as proxies for socioeconomic status (SES), to answer three questions:

  1. Can FSM histories be improved on as a proxy for social deprivation?
  2. What alternative (practical) proxy measures of SES can be used that better capture variation in achievement?
  3. Do alternative proxy measures better enable us to identify pupils at risk of low achievement?

Findings

Our findings are available in a series of public reports on www.gov.uk.

  • There was a high degree of consistency in the pattern of results between KS2 & KS4.
  • Parent/carer education and parental occupation were ‘better’ socioeconomic status (SES) predictors of pupil achievement, but the gains over FSM were marginal.
  • Therefore, FSM was the best-performing ‘practical’ proxy for SES.
  • Regional variations in attainment were found at the end of both primary and secondary school. But accounting for prior attainment largely eliminates the regional differences found.

The socioeconomic gaps found — even when accounting for a host of other factors — were stark and substantial. However, these gaps may have been even larger if there had not been a long-running redistributive and compensatory system aimed at alleviating disadvantage in place. This highlights why it is crucial to identify poor/disadvantaged pupils at risk of underachievement as early as possible – in order that additional resources can be targeted at this group in particular.

Design and Methods

Given the focus on statistical modelling and on the identification of proxies for socioeconomic disadvantage that are associated with underachievement, the project was based around the estimation of two types of models:

  1. A first set of models, to include pupil demographic characteristics (including gender, age and ethnicity) and rich socioeconomic controls derived for children (Index of multiple deprivation, for example), for parents (including parents’ aspirations for their children, household income), for the neighbourhood in which the pupils live (for instance, low participation in higher education) and for the school they attend (including measures of overall school deprivation and quality)
  2. A second set of models, to include the same pupil demographic characteristics and a series of varied proxies for socioeconomic status.

The aim was to compare the predictive power of the rich controls in the first set of models with those of the proxies in the second one and to compare the features of the individuals being identified as disadvantaged by the two models.

These models are intended for use as predictive tools and as such will not provide evidence of causation. However, given the aims of the project, they will provide tools to more accurately identify children at risk of underachievement and pinpoint the exact factors whose combination results in lower-than-expected academic outcomes.

Project Leader

Alex Sutherland