Observations vs Statistics
Ted E Feldman · JAMA · 1985
To the Editor.— In a recent editorial1it is stated that "many papers submitted for publication inJAMAare rejected as a direct result of statistical evaluation. If a study has not been designed and conducted in a manner such that the investigator's hypotheses can be tested by the selected methodological approach, then no statistical revisions will suffice to correct this basic flaw.... Support for publication can be given only to well-designed studies that incorporate sound research methods." Some statistically "sound" experiments are flawed by unsound logic, while many of the great studies in medicine have come from careful observation without the benefit of statistical analysis. Dr Vaisrub does not discuss observational studies such as Withering's description of the clinical use of digitalis,2Prinzmetal and co-workers'3recognition of coronary spasm, and the recent description of acquired immunodeficiency syndrome (AIDS)4(these being only a few among many classic