Balancing Realism and Purity

It is difficult to give an objective assessment of the validity of a statistical analysis. Rather like medicine, statistics is an art, and there is rarely a unique correct approach. Rather judgement is necessary to select one of a number of possible analyses, each with their own advantages and limitations. In my review I encountered many analyses which I would have tackled differently, but where in my judgement the analysis as presented was perfectly acceptable. In writing this article I have attempted to confine my criticisms to points where I believe that the vast majority of statisticians would agree that the approach adopted was not acceptable, and indeed many statisticians would probably take a harder line than I have. My statistical philosophy leans towards being a realist rather than a purist, and my research interests lie in the area of how to obtain the least biased results possible in areas where perhaps for ethical or practical reasons randomization is not possible, or where missing data abound.