On parallelism of regression lines (Q2913244)

The authors discuss the testing of parallelism of regression lines. The classical \(F\)-test is compared with a rank test proposed by \textit{P. K. Sen} [Ann. Math. Stat. 40, 1668--1683 (1969; Zbl 0184.22401)]. A rank based multiple comparisons rule for detecting regression lines with different slopes is constructed and the asymptotic distribution of the underlying statistic is derived. The proposed rule is compared by means of simulations with the rule derived from the assumption that the random fluctuations are Gaussian. The simulations indicate that the classical \(F\)-test and the proposed rule are better for symmetric distributions with light tails while Sen's approach is preferable for symmetric heavy tailed and asymmetric distributions.