The natural log of the OR (logged odds ratio, LOR) was used in the meta-analysis. The LOR has a normal distribution with a mean of 0, with LOR<0 indicating reduced odds and LOR>0 indicating increased odds (Lipsey & Wilson, 2001). Advantages of using the LOR include a symmetric distribution, with equal “weighting” on either side of 0, no recoding, and the ability to calculate the variance. An LOR=0 denotes equivalent odds, with symmetry around 0; therefore a 95% CI which includes 0 indicates lack of differences in odds between the two groups. This differs from the odds ratio which ranges between 0 and infinity, with 1 indicating a lack of differences in odds, and non-symmetry around 1. Both LOR and the converted OR (exp(LOR)) are provided in the Results, since most readers are familiar with interpreting the OR and its corresponding CI. The random effects model was used to estimate the overall effect size of the LOR. An analysis of variance (ANOVA) model implemented in SAS was used to test if percent male, race, mean age, or other demographic measures
