1 Pediatric Pulmonary Unit, Massachusetts General Hospital, Boston, MA 02114
In Reply.
Although the contingency tables noted by Dr LeBlanc each contain one cell with an expected frequency of less than 5, the Yates correction, which applies to all 2 by 2 tables irrespective of expected frequencies, does not address this issue. Furthermore, we agree with Grizzle1 that this correction produces an extremely conservative test and should not be used. Fisher exact test, however, is germane to the problem of small numbers. The results of this test are in good agreement with our published values.