The significance of statistical significance

Currently, much nursing practice is based on limited evidence, for example, small-scale research, case studies and clinical experience. In a mature science this would be undesirable, but nursing is in the early stages of development as a science, and many of its practices depend on relatively informal knowledge. To encourage the spread of potentially valuable ideas, nurses must be willing to share their clinical experience and journal editors should consider publishing this information. High-quality research is essential to the long-term development of 'evidence-based practice', but it is crucial at the present stage of nursing science that we do not become too concerned with perfect research methodology at the expense of good ideas. This particularly applies to tests of statistical significance. If we accept only information that has demonstrated statistical significance, we risk the dismissal of qualitative research and other information which may be extremely valuable but which have not yet been fully investigated. The aim of this paper is to convince practitioners and journal editors that statistical significance is not the only way to judge clinical importance and to suggest that decisions on what should be submitted and accepted for publication should be based on potential clinical relevance as well as statistical analysis.