Methodological concerns for meta-analyses of meditation: Comment on Sedlmeier et al. (2012).

Reanalysis of a prior meta-analysis on meditation found that the global index used to compare meditation types was confounded because it aggregated different sets of variables for each group. When composite indices were limited to variables for which each group had at least three studies, effect sizes for the Transcendental Meditation technique were significantly larger than those for mindfulness meditation or other meditations. Quantitative tests did not indicate publication bias for any meditation type, and effect sizes did not differ between studies from researchers affiliated with a TM organization and those from other universities. The reanalysis also identified 35 missed studies on TM and several unreviewed studies on job performance, substance abuse, and prison recidivism.

Study at a glance

Abstract

We commend Sedlmeier et al. (2012) for their significant undertaking of meta-analysis of all meditation types on all psychological variables, but additional analyses may modify some of their conclusions. Whereas they suggest from visual inspection of funnel diagrams that there may be publication bias of underreporting low-effect studies on the Transcendental Meditation (TM) technique, quantitative tests do not indicate the presence of bias for any type of meditation. We additionally found that there was no significant difference in effect sizes between studies originating from researchers affiliated with a TM organization and studies from other universities. We found that comparison of different types of meditation on their global index was confounded because their global index aggregated different sets of variables for the different groups. That is, using composite indices that only aggregated variables for which each group had at least 3 studies confirmed the authors' conclusion that effect sizes for different research designs were not different, but found that effect sizes for the TM technique were significantly larger than effect sizes for mindfulness meditation or other meditations. We also located 35 studies on the TM technique that appear to meet the authors' inclusion criteria that were missed by their meta-analysis, and several others on important psychosocial behavioral variables, such as job performance, substance abuse, and prison recidivism that were not reviewed. In addition, we suggest that future meta-analyses on psychological variables include cross-validating physiological studies.