A Nonparametric Test for the Parallelism of Regression Lines Based on Kendall's Tau

Kendall의 Tau에 의한 회귀직선의 평행성에 관한 비모수 검정

  • Published : 1978.06.01

Abstract

For testing $\beta_i=\beta, i=1,...,k$, in the regression model $Y_{ij} = \alpha_i + \beta_ix_{ij} + e_{ij}, j=1,...,n_i$, a simple and robust test based on Kendall's tau is proposed. Its asymptotic distribution is proved to be chi-square under the null hypthesis and noncentral chi-square under an appropriate sequence of alternatives. For the optimal designs, the asymptotic relative efficiency of the proposed procedure with respect to the least squares procedure is the same as that of the Wilcoxon test with respect to the t-test.

Keywords

References

  1. Ann. Math. Statist. v.38 Estimates of Regression Parameters Based on Rank tests Adichie,J.N.
  2. Ann. Statist. v.2 Rank Score Comparison of Several Regression Parameters Adichie,J.N.
  3. Ann. Math. Statist. v.27 The Efficiency of Some Non-parametric Competitors of the T-Test Hodges,J.L.;Lehmann,E.L.
  4. J. Amer. Statist. Assoc. v.65 A Distribution-free Test for Parallelism Hollander,M.
  5. Elements of Non-parametric Statistics Noether,G.
  6. Ann. Statist. v.2 A Non-parametric Test of Whether Two Simple Regression Lines are Parallel Potthoff,R.F.
  7. J. Amer. Statist. Assoc. v.63 Estimates of the Regression Coefficients based on Kendall's tau Sen,P.K.
  8. Ann. Math. Statist. v.40 On a Class of Rank Order Tests for the Parallelism of Several Regression Lines Sen,P.K.