A New Revised Version Of McNemar’s Test And Its C Dimension Improvement

McNemar’s test is a test on 2 x 2 classification tables with matched pairs of data. It is widely applied in contemporary research studies, especially in different areas of medical research and advertisements of products. However, it only used the two categories A and D, which represent changes; the retained parts B and C, which represent concordant responses, are not considered. Generally, it would be more reasonable for the significance of the changes to depend not only on A and D, but also on B and C. To develop the test, this paper has used chi-square test, binomial distribution and total dispersion square sum to improve McNemar’s test. The details are as follows.At first, this paper analyzes the process and the reason of unreasonable issue of McNemar’s test. Meanwhile, we use a new method to prove McNemar’s test and derive W’s test which includes all of data in 2 x 2 classification tables. Finally, W’s test has been verified by chi-square test and binomial distribution.Secondly, the same issue as McNemar’s test of Cochran Q test is researched in this paper. Furthermore, this paper uses total dispersion square sum formula to derive Qw test which includes all of data in r x c classification tables.