Sem título

Página 1

Paired comparison is used to compare the relative financial managerial performance of U.S. and CN firms. Eight hypotheses are tested on the basis of ROA, ROE, ROI, LTDE, TDTE, TAT, RTO, and ITO. In all of these cases the null hypothesis states that there is no difference between U.S. and CN firms with regard to the ratio that is being compared. The alternative hypothesis explains that these ratios are different. Because matched pairs are used, an appropriate test is

124 The Comparative Financial Performance of U.S.& Chinese Firms

the Wilcoxon Matched-pairs Signed-ranked test. This test is ideal because it is a nonparametric test, not requiring a large sample size. The Wilcoxon test gives more weight to pairs that show a large difference than pairs indicating small ones. In this manner the Wilcoxon test is similar to the t-test but it deals with ordinal data. This test is one of the most powerful non-parametric tests. Even for small samples its power is about 95 percent of that of the t-tests (see [10], [11], & [12]). To conduct the Wilcoxon test, first the differences between each pair, with regard to the ratios that are being compared, are computed. Then these differences are ranked on the basis of their absolute values. Next, the sums of the ranks of the positive and negative differences are used as the test statistics T-, and T+, respectively. Finally, the appropriate absolute T value in each table is chosen as the test statistic and is compared to the critical value given in the table of critical values for the Wilcoxon test, for the number of observations (N).

Página 2