Does Homework Perpetuate Inequities in Education?

PISA 2012 and PISA 2003 country samples; 38 countries and economies with comparable trend data

15-year-olds across OECD and partner countries

PISA in Focus No. 46, December 2014, OECD Publishing. A four-page descriptive brief drawing on PISA tables IV.3.48, IV.3.27/28 and IV.1.8c. Grade D because every estimate is a cross-sectional between-student or between-school association with no causal design; its value to this topic is as a source of DOSE and DISTRIBUTION facts - how much homework children actually get, and who gets it - which the experimental literature never establishes.

Dose in the real world: the OECD average is 'almost 5 hours per week' of homework for 15-year-olds, with Shanghai at 14 hours, Ireland, Italy, Kazakhstan, Romania, the Russian Federation and Singapore at 7 or more, and Finland and Korea at 'less than three hours per week'. Homework is FALLING: the OECD average was 5.9 hours in 2003, an hour more than in 2012, and the time fell in 31 of 38 countries with comparable data, with only Australia and Austria showing significant increases. DISTRIBUTION: in EVERY participating country advantaged students do more homework than disadvantaged students, with an OECD-average gap of 1.6 hours a week (5.7 versus 4.1) and gaps of 3.5 hours or more in some countries. ON DOSE-RESPONSE the brief reports that 'evidence from PISA 2009 suggests that after around four hours of homework per week, the additional time invested in homework has a negligible impact on performance.' On the level-of-analysis question it gives the school-level association: comparing students of similar background in similarly resourced schools, those in schools where students do more homework perform better, with a net pay-off of '17 score points or more per extra hour of homework' in Hong Kong, Japan, Macao and Singapore. AND THE FINDING THAT DEFLATES ALL OF IT: 'PISA also finds that the average number of hours that students spend on homework or other study set by teachers tends to be UNRELATED to the school system's overall performance.' A variable that predicts differences within systems and nothing between them is behaving like a positional or selection marker rather than a production input.

High and unaddressed. Advantaged families produce both more homework hours and higher scores; nothing in a cross-sectional survey separates the two. The socio-economic gradient in homework time is a description of who does homework, not evidence about what homework does.

The four-hours-a-week plateau is broadly consistent with Fernández-Alonso and colleagues' finding that beyond about 70 minutes a day the marginal return is about 3% of a standard deviation, and with Cooper's report that the junior-high association 'disappeared entirely after students reported doing between 1 and 2 hours each night'. The between-system null is corroborated by the fact that the two highest-performing PISA systems in this brief, Finland and Korea, sit at the bottom of the homework distribution.