Two Measurement Comparison of Sensitivity, Confidence and Prediction Interval, Sample Size Planning

The statistical methods used to evaluate and compare different methods of measurement are a vital common component of all methods of scientific research, focusing on measurement error and variance components. Measurements of the performance of a trained panel obtain most attention especially in the field of medicine and education. In addition to the research on sensitivity of experts of judges, the sensitivity of questionnaires is another main concern for practitioners. The present study will develop the test statistic of equivalence for two sensitivity comparison by using the general distribution of the ratio of two F noncentral variates for the first year. Since it is not possible to provide comprehensive tables for the numerous combinations of possible degree-of-freedom and probability levels, intensive computation will be performed and the R programs will be provided. Moreover, the correspondent sample sizes will be derived based on the nominal alpha and statistical power. For the second year, the confidence interval and prediction interval will be derived. For the third year, the study will be extended to the case that the sample size for one study is fixed. Finally, we will use the data from Pediatric Quality of Life Inventory to compare the sample size needed in different experiments.