Type of outcome measure Data to extract Standardised mean difference (d) Standard error of standardised mean difference
Continuous reported as mean or mean difference
Means (M1 and M2), standard deviations (S1 and S2) and sample size per group (n1 and n2)
\[\frac{M_{1}-M_{2}}{S}\] where S\(=\sqrt{\frac{{{(n}_{1}-1)}^{2}S_{1}+{{(n}_{2}-1)}^{2}S_{2}}{n_{1}+n_{1}-2}}\)= \[\sqrt{\frac{1}{n_{1}}+\frac{1}{n_{2}}+\frac{d^{2}}{n_{1}+n_{2}-2}}\]
Binary reported using odds ratios Natural logarithm of odds ratio (lnOR) and standard error of log odds ratio SElnOR. This can be obtained from a 95% confidence interval for the odds ratio by taking natural logs and dividing by 2x1.96 \(\frac{\sqrt{3}}{\pi}\text{lnOR}\) \(\frac{\sqrt{3}}{\pi}\text{SE}_{\text{lnOR}}\)
Raw binary data
Raw binary data (c1/n1 and c2/n2) where c1 and c2 are the number of participants complying with the behaviour of interest by group.
OR = \(\frac{\frac{c_{1}}{n_{1}-c_{1}}}{\frac{c_{2}}{n_{2}-c_{2}}}\) Take natural log and continue as above \[\text{SE}_{\text{lnOR}}=\sqrt{\frac{1}{c_{1}}+\frac{1}{{n_{1-}c}_{1}}+\frac{1}{c_{2}}+\frac{1}{{n_{2-}c}_{2}}}\] Continue as above