2016 年 37 巻 2 号 p. 101-118
In this article, we propose an alternative criterion for evaluating parallelism in parallel-line assays at the subject level. Unlike another recently proposed criterion (“ISP”; Uehara et al., 2016), this is a single criterion concerning the quantile of intrasubject slope ratio, which is straightforward to interpret. Two one-sided tolerance limits for a linear combination of the control and test slopes are developed as metrics for the evaluation of discrepancy from the ideal intrasubject parallelism. The proposed metrics are exemplified using data from two ex vivo parallel-line experiments, and their characteristics are investigated through Monte Carlo simulations.