Rank transform F statistic in a 2$\times$2 factorial design

  • Park, Young-Hun (Department of Statistics, Hanshin University, 411 Yangsna-dong, Osan, 447-791)
  • Published : 1994.06.01

Abstract

For a $2 \times 2$ factorial design without the restriction of a linear model or without regard to error terms having homoscedasticity, under the null hypothesis of no interaction we can have the rank transformed F statistic for interaction converge in distribution to a chi-squared random variable with one degree of random if and only if there is only main effect.

Keywords

References

  1. Communications in Statistics, Part B- Simulation and Computation v.16 Limitations of the Rank Transform Statistic in Tests for Interactions Blair,R.C.;Sawilowsky,S.S.;Higgins,J.J.
  2. Communications in Statistics v.A5 On Some Alternative Procedure Using Ranks for the Analysis of Experimental Designs Conover,W.J.;Iman,R.L.
  3. Theory and Application of the Linear Model Graybill,F.A.
  4. The Annals of Mathematical Statistics v.39 Asymptotic Normality of Simple Linear Rank Statistics under Alternatives H$\'{a}$jek,J.
  5. Theory of Rank Tests H$\'{a}$jek,J.;Sid$\'{a}$k,Z.
  6. Journal of the American Statistical Association v.79 The F Statistic in the Two-Way Layout with Rank-Score Transformed Data Hora,S.C.;Conover,W.J.
  7. Journal of the American Statistical Association v.83 Asymptotic Relative Efficiencies ot the Rank Transformation Procedure in Randomized Complete Block Design Hora,S.C.;Iman,R.L.
  8. Journal of the American Statistical Association v.79 Comparison of Asymptotically Distribution Free Procedures for the Analysis of Complete Blocks Iman,R.L.;Hora,S.C.;Conover,W.J.
  9. Biometrika v.78 no.3 A Note on the Rank Transform for Interactions Thompson,G.L.