Outcome/mediator | N | A (SE) | P | R^{2} (a) | B (SE) | P | R^{2} (b) | c’ (SE) | P | IE | MC 95% CI | CFI | RMSEA |

Risk of depression | |||||||||||||

Mindfulness skills | 7865 | 0.02 (0.01) | 0.050 | 0.001 | −0.28 (0.02) | <0.001 | 0.23 | 0.06 (0.32) | 0.844 | −0.01 | −0.010 to −0.001 | 0.971 | 0.029 |

Executive function | 7683 | −0.02 (0.02) | 0.388 | 0.000 | 0.13 (0.01) | <0.001 | 0.23 | 0.25 (0.33) | 0.440 | −0.003 | −0.008 to 0.003 | 0.994 | 0.014 |

Social–emotional–behavioural function | |||||||||||||

Mindfulness skills | 7755 | 0.03 (0.02) | 0.055 | 0.001 | −0.15 (0.01) | <0.001 | 0.31 | 0.09 (0.16) | 0.555 | −0.004 | −0.008 to 0.000 | 0.979 | 0.030 |

Executive function | 7575 | −0.12 (0.03) | <0.001 | 0.002 | 0.09 (0.01) | <0.001 | 0.33 | 0.19 (0.15) | 0.226 | −0.01 | −0.015 to −0.006 | 0.997 | 0.012 |

Well-being | |||||||||||||

Mindfulness skills | 7835 | −0.01 (0.01) | 0.380 | 0.000 | 0.17 (0.02) | <0.001 | 0.19 | 0.16 (0.27) | 0.554 | −0.002 | −0.004 to 0.001 | 0.965 | 0.029 |

Executive function | 7654 | 0.01 (0.02) | 0.818 | 0.000 | −0.09 (0.01) | <0.001 | 0.20 | 0.05 (0.28) | 0.871 | 0.000 | −0.004 to 0.003 | 0.991 | 0.015 |

N is the number of students (number of clusters (schools)=84). R

^{2}(a) is the proportion of the variation in the mediator that is accounted for by the independent variable. R^{2}(b) is the proportion of the variation in the dependent variable that is accounted for by the mediator. a is the unstandardised path ‘a’ (see figure 1E); b is the unstandardised path ‘b’ (see figure 1E); c’ is the unstandardised direct effects after controlling for the mediator and covariates (see figure 1E). IE is the unstandardised maximum likelihood robust-based regression coefficient for the mediational models, reflecting the IE (a×b) of group allocation on follow-up scores in the primary outcomes of risk of depression, social–emotional–behavioural functioning, and well-being (controlling for the baseline levels of the primary outcomes, cohort, country, school size, school sex, and adjusted for the clustering of observations) through prechanges to postchanges in the process outcomes of self-reported mindfulness skills or executive function. MC 95% CI is the 95% confidence interval for the IE based on a Monte Carlo simulation of the joint distribution of the corresponding slopes using 20000 random draws from the parameter estimates and their associated asymptotic variances and covariance.CFI, comparative fit index; IE, indirect effect; RMSEA, root mean square error of approximation.