Use statistic software to verify the given estimates of (unweighted) kappa coefficients and their variances in Example3.6 for the two individual tables in Table3.5

Transcribed Image Text:

Example 3.6 For the Detection of Depression in a Primary Care Setting Study (DDPC), consider testing whether there is a difference in agreement between probands and informants on depression diagnosis stratified by the gender of the informants. The stratified information is given in Table 3.5. TABLE 3.5: Depression diagnosis, stratified by informant gender Informant Informant Proband 0 1 2 Total Proband 0 1 2 Total 0 44 8 5 57 0 22 5 1 28 1 24 12 10 46 1 12 4 0 16 2 9 7 19 35 2 5 5 8 18 Total 77 27 34 138 Total Female informants 39 14 9 Male informants 62 0.2887 0.2663 0.009 08 The estimated (unweighted) kappa coefficients for the two tables are 0.2887 and 0.2663, with the corresponding variances 0.00384 and 0.009 08. Thus, the estimated overall kappa is (0.00384 3)/ (0.00384 + 0.00908) = 0.2820 and the estimated asymptotic variance is 1/(0.00384 +0.009 08) = 0.002699. A 95% confidence interval is (0.1801, 0.3839). The test of the null hypothesis of equal kappa coefficients has a p-value 0.8435, indicating that agreement between probands and informants is similar for informant females and males.