Problems of omission in communications

Abstract

The paper describes six difficulties that arise from omissions in reporting on data or their analysis, For each a remedy is suggested: (1) Randomization. Care in reporting the method and execution of randomiration would remedy this omission. (2) Informed consent. Reporting how informed consent was obtained would remind us that it was obtained, and the cumulative experience might also improve our ability to handle the matter both to our own satisfaction and that of many critics. (3) Analytie techniques. Reporting what analytic technique was used gives the reader a chance to consider whether the appropriate method was used or whether some other approach would have been preferable. (4) Descriptive statistics. If the reader is to understand the data analysis, some descriptive statistics are required. These data may be of much more value to the future of medicine than the significance test performed, because later use of the data cannot be anticipated. It is time to consider ways to make data available for those who request it and to examine the conditions for such transactions. (5) Power. Reporting the power of a test will tell us something about the strength of the investigation. It should be encouraged by editors. At present, it is possible to report the power of only the simpler tests. (6) Selectivity and multiplicity. The variety of methods used and the method of selecting the statistics presented in the final paper need to be presented if the reader is to get an honest appraisal of the uncertainty of the study.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 24 citations exist upstream.