Matthew A Kraft and colleagues conducted a meta-analysis of the causal evidence on the effect of teacher coaching on teaching and learning. Their paper, published in the Review of Educational Research, reviewed 60 studies on teacher coaching programmes conducted after 2006 that measured the impact of teacher coaching on either teaching (measured using tools such as the Classroom Assessment Scoring System or the Early Language and Literacy Classroom Observation) or pupil academic performance (measured by standardised tests).
Their results found that sustained coaching improves both classroom teaching and pupil achievement, with pooled effect sizes of +0.49 standard deviations for teaching and +0.18 standard deviations for academic achievement.
However, the effectiveness of a teacher coaching programme seems to be determined by the number of participants. When studies were divided into programmes that had fewer than 100 participants and those that had more than 100 participants, the impact on teaching was nearly double for the smaller programmes than for programmes with more than 100 participants. For pupil achievement, the smaller programmes showed an impact of nearly three times that of the larger programmes.
Source: The effect of teacher coaching on instruction and achievement: A meta-analysis of the causal evidence (February 2018), Review of Educational Research, Vol 88, Issue 4