To the Editor.—
In his letter, Dr. Voron states that Dicken et al1 did not consider their data relating diabetes to generalized granuloma annulare as insignificant. This is an interesting interpretation of the data since the authors themselves state in their abstract, "No significant association with other diseases was noted." This statement was the basis of choice (d) being considered as true.To argue that there is a significant association, as Dr. Voron apparently is doing, one must currently depend on the citation of case reports of three or four patients which appear to suggest a significant relationship because they show the presence of these two diseases occurring together. However, a statistician would hesitate to make a statement regarding the "significance" of an association on the basis of such small sample sizes and without the type of sampling technique that renders statistical analysis valid or relevant. The question that