Grammar for Writing: Evaluation Report and Executive Summary

2,510 pupils in 53 schools

Year 6 pupils (aged 10-11) in English primary schools, in the transition phase before secondary school

EEF-funded EFFICACY trial of the University of Exeter Grammar for Writing programme, adapted for the end of primary schooling. Within-school design: one class per school was allocated to the intervention and one to control. Within intervention classes, children expected to score between Level 3c and Level 4b at KS2 were FURTHER randomised to whole-class teaching only versus whole-class teaching plus small-group teaching. Four-week version of the programme. Outcome was the GL Progress in English assessment - a published standardised test, which makes this the measure-type-cleanest of the grammar trials. NOTE ON PROVENANCE: the figures recorded here are taken from the detailed primary description of this trial in the background section of the independent 2019 EEF effectiveness-trial report (Tracey et al. 2019, pp. 6-7), not from the 2014 report itself.

Only limited effects. The whole-class intervention produced a small and statistically NON-SIGNIFICANT effect of ES = 0.10. The additional small-group condition produced ES = 0.24 - but the evaluators' own caveat is the important part, and it is fatal for reading this as evidence about grammar: the small-group children were "not much higher than those taught in larger groups (i.e. whole-class teaching) in either the control or intervention conditions", so as Torgerson et al. state at p.33, "this difference could have been as a result of teaching children in small groups per se rather than as a result of the programme". This is the single most relevant finding in the archive for anyone reading a small-group effect as a subject-matter effect. The trial was also criticised for its timing - delivered in the transition phase after KS2 assessments - and for the short (4-week) dose, which motivated the longer 2019 effectiveness trial that then returned a flat null.

Low. Randomisation at class level within school plus a further within-class randomisation for the small-group contrast; the small-group comparison is the best-identified group-size contrast in this literature precisely because it was randomised inside the intervention classes.