Unravelling gender composition effects on rule-breaking at school: a focus on study attitudes

Abstract

Previous research on consequences of schools' gender composition has mostly investigated students' socio-emotional well-being and achievement, while students' academic attitudes and behavioural outcomes – including school deviancy – have been studied less. Moreover, most studies compared single-sex and coeducational schools, and did not focus on the proportion of girls at school. Starting from reference group theory, we hypothesise that boys attending schools with a higher proportion of girls adopt the latter's positive study attitudes, rendering them less susceptible to disruptive behaviour. Conversely, girls in schools with more boys are expected to adopt the latter's negative study attitudes, consequently being more likely to misbehave. Multilevel analyses on data from the Flemish educational assessment, consisting of 5961 girls and 5638 boys in 81 schools, showed that both boys and girls valued studying more and were less likely to misbehave at school when proportionally more girls attended their school. Implications are discussed.

Keywords:

• gender composition
• school misconduct
• study attitudes
• reference group theory
• multilevel analysis
• gender gap

Acknowledgements

The authors are grateful to Simon Boone and the two anonymous reviewers for their helpful feedback on this manuscript.

Notes

Skewness measures the degree of asymmetry in the distribution of a variable. This statistic is negative when the distribution is skewed to the left, and positive when skewed to the right.

We used HLM6 to perform overdispersed Poisson models with constant exposure, which yielded the same basic image as the linear multilevel model, the results of which are shown in Tables 2 and 3.

In an unconditional ‘null’ model, no determinants are specified. This enables to partition the total variance of a dependent variable across the various levels in the multilevel analysis. In this case, we may partition the variance in the dependent variable across the school and the individual level. The σ²-statistic gives the variance for the individual level, the τ ₀-statistic gives the variance for the school level. The p-statistic shows whether the variance at the school level is significant. Dividing the τ ₀-statistic by the sum of the τ ₀-statistic and the σ²-statistic gives the percentage of the total variance situated at the school level.

As in OLS-methods, the coefficients of multilevel regression analyses may be presented unstandardised (notation γ). However, if one wants to compare the sizes of the effects, one must standardise the coefficients (notation γ*). This is done by calculating the following formula: γ* = (γ × SD_{independent variable})/SD_{dependent variable}.