A RESPONSE TO SPROTT'S "USE OF CHI SQUARE".

Issue is taken with Sprott's criticisms of the use of chi square in birth-order studies, including the author's recent paper on alcoholism, birth order, and family size. Sprott's argument that the chi-square test is inappropriate is discussed, but further statistical tests not involving chi square yielded the same conclusion, i.e., no birth-order effect. Also, interactions between birth-order effects and family sizes are unimportant for the alcoholism data and difficult to evaluate in any birth-order study where an overall test is required. The corrected family-size analyses are defended as the best available. Sprott's suggested changes in the family-size analyses do not affect the conclusion that persons from large families are overrepresented among alcoholics. [ABSTRACT FROM AUTHOR]