A Correlation Goodness-of-Fit Test for Randomly Censored Data
针对随机删失数据，提出一种类似Shapiro-Francia统计量的相关性统计量，用于检验复合拟合优度假设；在轻度删失下对指数性检验具有渐近稳健性，并通过前列腺癌数据示例应用。
For testing a composite goodness-of-fit hypothesis with randomly censored data, a correlation statistic which is an analogue of the Shapiro-Francia statistic is presented. The test for exponentiality, in case of light censoring, is asymptotically robust against departures from the Koziol-Green model of random censorship. The power of the test is compared with the total-time-on-test procedure and the new-better-than-used test using Monte Carlo simulation. As an example, the test is applied to some data with prostate cancer.