Table 1

 Multilevel studies reporting school effects on pupil outcomes

StudyOutcome measuresSchool effectIntraclass correlation; between school variance explained by school level variables in a full model
*Calculations made by the author.
Smoking habits and alcohol use
Johnson and Hoffmann, USA, 2000Daily cigarette start at baseline and two year follow up (binary)In west region v other regions pupils smoked less; OR = 0.66Explained variance  =  15% (grade 8) and 40% (grade10)
In Catholic schools v non-Catholic schools pupils smoked more; OR = 1.46
In schools with a competitive climate pupils smoked more at baseline; OR = 1.17, and at follow up OR = 1.22
Maes and Lievens, Belgium, 2003Regular weekly smoking (binary)High teacher work load increased pupils’ smoking; OR = 1.23ICC smoking habits  = 9%*
Alcohol use (binary)Strong school policy decreased pupils’ smoking and alcohol consumption; OR = 0.83 and 0.87 respectivelyICC alcohol consumption  = 12%*
Female school administrator decreased pupils’ alcohol consumption; OR = 0.66Explained variance  =  13%–19%*
Moore et al,Daily smoking (binary)Weak school policy increased daily smoking; OR = 2.77ICC = 7%*
Great Britain, 2001Weekly smoking (binary)Medium school policy increased daily smoking; OR = 2.04Explained variance with reference to smoking policy  = 14%*
Weak enforcement increased daily smoking; OR = 1.52Explained variance with reference to enforcement of pupils  = 4%*
Pinilla et al,Daily smoking (binary)High compliance antismoking rules increased daily smoking; OR = 0.46ICC  = 12%*
Spain, 2002
Reeder et al, New Zealand, 1999Daily smoking (binary)School level variables (socioeconomic status and sex composition) were not significantly associated with smoking habits
Current (⩾ weekly) smoking (binary)
Rountree P, USA, 1999Alcohol use (continuous)Strong drug subculture increased pupils alcohol useICC  = 10.3%
Explained variance  = 4.9%*
School achievement
Battistich V, USA, 1995Mathematics and reading achievement (continuous)Low poverty improved pupils’ mathematics achievement; effect size  =  −0.968Mathematics achievement ICC = 28%
Low poverty improved reading achievement; effect size  = −0.922Reading achievement ICC = 28%
Explained variance; mathematics achievements =  75%
Explained variance; reading achievements = 74%
Blau J, USA, 2001Gains in social studiesSocioeconomic inequality decreased pupils’ gains in social studiesICC = 13%
(continuous)Explained variance = 12%
Webster B, Australia, 2000Achievements in mathematics and science (continuous)High average SES improved pupils’ mathematics achievementICC mathematics achievement  =  7.6%
High average SES improved pupils’ science achievementICC science achievement  = 2.4%
Rural location decreased pupils’ mathematics achievementExplained variance, mathematics achievement  = 91%*
Rural location decreased pupils’ science achievementExplained variance, science achievement  = 98%*
Young D, Australia, 1998Achievements in mathematics and science (continuous)High average SES improved pupils’ mathematics and science achievementICC mathematics achievement  =  5.13%
Rural location decreased pupils’ mathematics and science achievementICC science achievement  = 1.36%
Explained variance, by average SES; mathematics achievement  = 16.1%
Explained variance, by location; mathematics achievement;  = 37.6%
No variance in science achievement explained by school level variables
Problem behaviour/wellbeing
Konu A, Finland, 2002Wellbeing (continuous)High parental education increased pupils’ wellbeingICC  = 1%.
Low bullying at school increased pupils’ wellbeing
High proportion of teachers treating pupils fairly increased pupils’ wellbeing
High proportion of pupils with plans for future education increased pupils’ wellbeing
Mooij T, Netherlands, 1998Aggressive behaviourNot having a school exchange programme increased aggressive behaviour in pupilsICC  = 2.6%–8.1%
Victim behaviour (continuous)High proportion of girls in school increased victim behaviour in pupilsExplained variance  = 5%–8% *
George R, USA, 2000Victimisation (binary)Large v small and medium schools increased victimisation in pupils (grade 8)
Suburban v urban and rural schools increased victimisation in pupils (grade 8)
Public v private schools increased victimisation in pupils (grade 8)
Low SES increased victimization in pupils (grade 8)
Roeger L, Australia, 2001Depressive symptoms (continuous)No multilevel regression models presentedICC grade 8  = 0.87
ICC grade 9  = 1.48
ICC grade 10  = 1.73%
van den Oord E, Netherlands, 1999Problem behaviour (aggression, restlessness, fear/uncertainty)Schools having a speech therapist increased pupils’ wellbeing (+)ICC problem behaviour  = 0%–3%
ICC wellbeing  = 4%
Self reported wellbeing (continuous)High average parent education increased fear/uncertainty in pupils
Low proportion of pupils with plans for future education increased fear/uncertainty in pupils
Wilcox P, USA, 2001Carrying weapon at school (binary)High proportion free lunch increased pupils weapon carryingICC  = 25%*
Explained variance  = 19%*
Physical activity
Zhu W, USA, 1997Cardiovascular endurance (continuous)Schools with physical education programme improved cardiovascular endurance in pupilsICC  = 22.35%
Schools with fitness tests improved cardiovascular endurance in pupilsExplained variance  = 10%