The relationship between teacher punishment and student academic achievement: a meta-analysis

Teachers' use of "zero tolerance" policies in student discipline has attracted widespread social attention. However, due to the insufficient professional training and guidance, some teachers adopt unscientific punitive methods, such as corporal punishment, suspension, or expulsion, which may be associated with adverse outcomes in students' academic achievement, cognitive development, and physical and mental health.

This study conducted a comprehensive meta-analysis of empirical research conducted over the past two decades to examine the relationship between teacher punishment and secondary school student's academic achievement, with a focus on identifying moderating variables. A distinctive feature of this meta-analysis is its categorization of moderators into participant characteristics, study characteristics, and measurement characteristics, thereby clarifying the multifaceted factors associated with the relationship.

The literature search identified 18 independent effect sizes involving 6,052,011 participants. The heterogeneity test supported the use of a random-effects model. The pooled effect size indicated a weak but statistically significant negative association between teacher punishment and student academic achievement (r = -0.183, 95% CI [-0.221, -0.145], p < 0.001). Moderator analyses showed that gender composition, achievement type, and punishment type contributed to variations in the association, whereas grade level and research design did not show significant moderating effects. Publication bias tests did not detect statistically significant evidence of publication bias.

The study found a weak negative association between teacher punishment and student academic achievement. This association was moderated by gender composition, achievement type, and punishment type, but not by grade level or research design. Given the high heterogeneity and the observational nature of the included studies, the findings should be interpreted cautiously as correlational evidence.