With regard to the physiology of meditation, research designs can be divided into 3 categories:

1) Case studies of meditation featuring small numbers of participants in which there is no attempt to control for confounding variables. While these are useful for generating hypotheses, they do not provide scientifically valid insights into meditation’s purportedly unique effects.

2) Own-control studies use participants in time sequential series (ie ab, aba or similar designs) in which the “a” condition is pre-meditation or non-meditation and the “b” condition is meditation. The “a” and “b” conditions are then compared. These studies have generally demonstrated significant differences (interpreted to be in favour of meditation) but they suffer from substantial limitations because they compare meditation to non-meditation and then assume that any differences are due to meditation, whereas they may in fact be due to methodological factors not specific to meditation at all (such as rest, expectancy, researcher demand and environmental issues). Further, this design is vulnerable to the possibility that one condition might be contaminated by carry-over effects from the preceding condition. A further important consideration applies especially to the case of novices — if the meditator is not sufficiently skilled then they may have difficulty in generating physiological changes and the effects (although potentially real) may not become detectable.

3) Experimental control studies are much more reliable as they involve two independent groups in which one meditates and the other engages in a control activity such as rest. There are however, methodological difficulties associated with this approach as well. First, use of novices (who are introduced to meditation during the trial) may mean that the effects of meditation are not large enough to generate a detectable change. Second, while use of advanced meditators (those who have practiced meditation over many years) may ensure that the necessary magnitude of effect is achieved, the question of selection bias becomes a significant consideration.

An ideal strategy would involve an experiment in which participants are randomly assigned from the same sample to either a meditation or a control group. This approach is rarely feasible however, since it would conventionally require many months or even years of practice before the participants achieved competence. With such lengthy time-lines the accumulation of drop-outs may in any case lead to selection bias. Therefore, a reasonable compromise strategy might involve using experienced meditators and comparing them to non-meditators who have either been matched for parameters including interest in meditation, or have been randomly selected from the population. This latter design was selected for the study described in this chapter.

Dr Ramesh Manocha