近幾年來，對於ROC曲線指標，有愈來愈多的文獻開始對於相似性及非劣性檢定方法提出相關的研究。例如：在醫學上一種新的非侵入性診斷方法與一種已被認定為標準的侵入性診斷方法欲做比較，在診斷的效率上不同時，當經過相似性及非劣性的檢定，得到新方法與標準的診斷方法是相似的，若新方法在成本、使用方便性及安全性的考量下較優於標準方法時，將以新方法取代標準的方法。而本篇論文主要是利用無母數法、最大概似法、動差估計法、標準化差評估法、調和相關係數法及拔靴法等六種方法評估相似性及非劣性的檢定方法，並以模擬結果比較這六種方法在常態分配假設、指數分配假設及離散資料假設下的優劣，並於最後提出一實例加以說明。

　　In recent years, for ROC curve, more and more issues have focused on the equivalence/non-inferiority test, e.g. in comparing diagnostic efficacy of a non-invasive alternative diagnostic (test) procedure to an invasive (reference) method. If the diagnostic accuracy of the non-invasive alternative procedure is equivalent to the invasive method, we may replace the invasive procedure by the non-invasive alternative diagnostic procedure because of its easy administration, its better safety profile or its reduced cost. In this paper, we compare the equivalence/non-inferiority tests based on six methods, which are the non-parametric method, method based on MLE, the moment estimation, the standardized difference, concordance correlation coefficient, and bootstrap method. A large simulation study was conducted empirically to investigate the size and power of these six methods for various combinations of distributions, data types, sample size, difference correlation and equivalence limit. A real published dataset is used to illustrate proposed rocedures.

[1] Bamber, D., The area above the ordinal dominance graph and the area below the receiver operation graph, J. Math. Psych., Vol.12, 387-415, 1975.
[2] Blackwelder, W. C., Proving the null hypothesis in clinical trials, Controlled Clinical Trials, Vol.3, 345-353, 1982.
[3] Chan, I. S. F., Proving non-inferiority or equivalence of two treatments with dichotomous endpoints exact methods, Statistical Methods in Medical Research, Vol.12, 37-56, 2003.
[4] Chen, B. C., On the equivalence and non-inferiority for the areas of two ROC curves, Unpublished Master Thesis of the Department of the Statistics of National Cheng Kung University, 2003.
(http://etdncku.lib.ncku.edu.tw/ETD-db/)
[5] DeLong, E., DeLong, D., and Clarke-Pearson, D., comparing the areas under two or more correlated receiver operation characteristic curves:A nonparametric approach, Biometrics, Vol.44, 37-845, 1988.
[6] Dorfman, D. D. and Alf, Jr., E., maximum-likelihood estimation of parameters of signal-detection theory─a direct solution, Psychometrika,Vol.33, 117-124, 1968.
[7] Dunnett, C. W. and Gent, M., Significance testing to establish equivalence between treatments with special reference to data in the form of tables, Biometrics, Vol.33, 593-602, 1977.
[8] Efron, B. and Tibshirani, R. J., Bootstrap methods for standard errors,confidence intervals, and other measures of statistical accuracy, Statistical Science, Vol.1, 54-77, 1986.
[9] Efron, B. and Tibshirani, R. J., An Introduction to the Bootstrap, Chapman and Hall, U.S.A., 1993.
[10] Epstein, A. M., Begg, C. B., and McNeil, B. J., The use of ambulatory testing in prepaid and fee-for-service group practices, N. Engl. J. Med., Vol.314, 1089-1094, 1986.
[11] Farrington, C. P., and Manning, G., Test statistics and sample size formulae for comparative binormal trials with null hypothesis of non-zero risk difference of non-inferiority relative risk, Statistics in Medicine, Vol.9, 1447-1454, 1990.
[12] Hoeffding, W., A class of statistics with asymptotically normal distribution, Annals of Mathematical Statistics, Vol.19, 293-325, 1948.
[13] Hsueh, H. M., Liu, J. P. and Chen, J. J., Unconditional exacts tests for equivalence or non-inferiority for paired binary endpoints, Biometrics, Vol.57, 478-483, 2001.
[14] Liao, J., An improved concordance correlation coefficient, Pharmaceutical Statistics, Vol.2, 253-261, 2003.
[15] Lin, L., A concordance correlation coefficient to evaluate reproducibility, Biometrics, Vol. 45, 225-268, 1989.
[16] Liu, J. P., Hsueh, H. M., Hsieh, E., and Chen, J. J., Tests for equivalence or non-inferiority for paired binary data, Statistics in Medicine, Vol.21, 231-245, 2002.
[17] Lu, Y., and Bean, J. A., On the sample size for one-sided equivalence of sensitivities based upon McNemar’s test, Statistics in Medicine, Vol.14, 1831-1839, 1995.
[18] Luchenbruch, P. A., and Lynch, C. J., Assessing screening tests: extensions of McNemar’s test, Statistics in Medicine, Vol.17, 2207-2217, 1998.
[19] Lusted, L. B., Signal detectability and medical decision-making, Science, Vol.171, 1217-1219, 1971.
[20] Masaryk, A. M., Ross, J. S., DiCello, M. C., Modic, M. T., Paranandi, L., and Masaryk, T. J., 3DFT MR angiography of the carotid bifurcation: Potential and limitations as a screening examination, Radiology, Vol.179, 797-804, 1991.
[21] McNeil, B. J. and Adelstein, S. I., Determining the value of diagnostic and screening tests, J. Nucl. Med., Vol.17, 439-448, 1976.
[22] Moran, P. A. P., Testing for correlation between non-negative variates, Biometrika, Vol.54, 385-394, 1967.
[23] Nam, J-m., Establishing equivalence of two treatments and sample size requirements in matched-pairs design, Biometrics, Vol.17,1422-1430, 1997.
[24] Obuchowski, N. A., Nonparametric analysis of clustered ROC curve data, Biometrics, Vol. 53, 567-578, 1997.
[25] Obuchowski, N. A., and McClish, D. K., Sample size determination for diagnostic accuracy studies involving binormal ROC curve indices, Statistic in medicine, Vol. 16, 1529-1542, 1997.
[26] Serfling, R. J., Approximation Theorem of Mathematical Statistics, John Wiley and Sons, New York, 1980.
[27] Sox, Jr., H. C., Blatt, M. A., Higgins, M. C., and Marton, K. I., Medical decision making, Batterworths-Heinemann, Boston, 1989.
[28] Tu, D., A comparative study of some statistical procedures in establishing therapeutic equivalence of nonsystemic drugs with binary endpoints, Drug Information Journal, Vol.31, 1291-1300, 1997.
[29] Zhou, X. H., Obuchowski, N. A., and McClish D. K., Statistical Methods in Diagnostic Medicine, John Wiley and Sons, New York, 2002.