Abstract In two recent articles in the periodical "The Accounting Review," Beaver and Dukes (B⁄D), have used market association tests to assess the relative information content of accounting earnings numbers under various tax allocation principles. This article is a brief comment on two of their measures of association, "Percent Correct" and "Composite Average Price Index (API)." Correct interpretation of these measures is important for policy makers who might rely on their results and for researchers who may want to use their research method. The reinterpretation of the ⁄D data does not change the conclusions they draw regarding the consistency of different earnings numbers with the set of information used in setting security prices. Rather, its significance lies in the added confidence a policy maker can put in these results. There seems to be no ex ante reason to expect negative composite APl's or especially that the proportion of times that positive forecast errors are associated with negative unexpected price changes should be statistically significant.
Sundem et al. (Thu,) studied this question.