Coronary-Prone Behavior Pattern and Coronary Heart Disease: Methodological Considerations

Abstract

The central logical problem of nonexperimental epidemiological research is to marshal scientifically persuasive arguments showing that a statistically significant association is not an artifact but is due to a causal relationship between the antecedent condition and the disease under investigation. The problem is usually approached by presenting evidence that tends to rule out likely alternate explanations, such as biased sampling procedures, systematic errors in measurement, or the influence of covariates. When a causal mechanism linking the antecedent condition with the disease is not readily apparent or involves many intermediate steps, as is the case with the Type A behavior pattern and coronary heart disease (CHD), the difficulty of persuasively arguing that a causal relationship exists is magnified because the opportunity for error is greater.