Skip to main content

Archived Comments for: The effectiveness and cost-effectiveness of a participative community singing programme as a health promotion initiative for older people: protocol for a randomised controlled trial

Back to article

  1. Randomization, sample size and comparability.

    Xavier Rojano i Luque, Ageing Institute of the Universitat Autònoma de Barcelona

    15 March 2011

    The nature of the study is quite original, but is not clear to me the reason to inflate the sample size for clustering (I will be grateful is someone can provide some reference to justify this). As far as I know, this is only necessary when the unit of randomization (ie:venue) is not the unit of analysis (person), but it does not seem the case, which in this aspect looks like a multicentre clinical trial in which the unit of randomization is the person. As authors appoint, groups of friends would want to go together in the same group. This issue can be solved randomizing by blocks (the authors should ask volunteers in order to put all friends in the same block), but this option complicates the realization of strata (you may have groups of friends of only men, only women or both men and women).
    As authors appoint, the participants are self selected, so it could be interesting to obtain other variables to known until which point they are representative of the overall population (marital status, degree of education, variables available on health surveys...) in order to explore the feasibility of the intervention outside the study.

    Competing interests