A randomised pilot study with a waiting-list design was undertaken to investigate the sensitivity of various outcome measures to detect changes in strength, balance and quality of life in people with multiple sclerosis participating in a Tai Chi intervention. One group received a 20-week Tai Chi class intervention while the other group waited over this time before receiving the same intervention. Three assessments of outcome measures were undertaken for both groups; one at baseline, and another two after each group had completed their Tai Chi classes. A case study in analytical planning for these data will be presented, exploring the benefits of possible techniques (independent samples t-test, paired t-test, ANCOVA, and so on), and the ultimate bringing together of these benefits into one single random effects analysis with appropriate covariance structure. Commentary will be given on how this approach to detect a signal differs to what would have been undertaken if the data had been from a trial in search of conclusive answers.