On the matching tables from the consistency test.docVIP

 PAGE \* MERGEFORMAT 3 On the matching tables from the consistency test [Abstract] presents a matched pair of tables from the consistency test of the test methods. [Keywords:] matching tables from; consistency; the most dominant line inspection The paired count data used in the two tests, comparison of diagnostic methods, as the number of patients in two methods for examination or diagnosis, the same sample in two ways to do testing and so on, which is characterized by all observation units in the sample were handled in two ways. paired count data are often represented as a form of data matching tables from, as shown in Table 1. Table 1 matched pair design of fourfold table A test method In the matching tables from the analysis, an intuitive idea is to use p = b + cn rates that are inconsistent, that is 1-p = a + dn, said the same rate. We know that for a given α, sided hypothesis testing problem H0: pgt; = p0 vs H1: p @, not reject the null hypothesis H0, p can not be considered the same rate Example 2 Example: The A, B and two methods for bacteriological testing, test to compare the effect of the two methods will be divided into two samples each were tested in two ways, by 2042 the sample test results, as shown in Table 2 asked the two tests of the test results have no difference? Solution: H0: pgt; = 0.05 vs H1: p lt;0.05 p = P {x ≤ 14} =  14i = 0Ci20420.05i (1-0.05) 2042-i = 0.000 p lt;0.01, with α = 0.01 level tests reject the null hypothesis of unilateral H0, consistent rate of less than 0.05, the two tests the same rate of 95%, and consistency is very good. Table 2 the results of different methods of bacteriological examination Links to free paper download ..com If using McNemar test this example, p lt;0.01, the positive rate reached inconsistent conclusions. The reason for such a conflict, it is because McNemar test only considers the test results to differ both cases (b and c, in this case 12 and 2), without consi