Hasan Ö DER 1* Zeynel CEBECİ2

Genellikle hipotezin ve/veya model parametrelerinin testi için F ve t-testleri kullanılır. Parametrik testler parametrik olmayan karşıtlarına göre daha etkili olsa da, pek çok veri seti için gerekli olan model varsayımlarının sağlanamadığı durumlarda, etkilerini yitirmektedirler. Bu durumda, varsayımlardan etkilenmeyen Permütasyon testleri parametrik olmayan bir yöntem olarak uygulanabilmektedir. Bu çalışmada, ham verinin permütasyonu, kalıntıların tam permütasyonu, kalıntıların kısmi permütasyonu yöntemleri, çoklu doğrusal regresyon, tesadüf parselleri, tesadüf blokları ve Latin kare deneme desenleri için I. tip hata olasılıkları bakımından karşılaştırılmıştır. Yöntemlerin karşılaştırılmasında hayvancılık verileri kullanılmıştır. Sonuç olarak, I Tip hata olasılığı bakımından Permütasyon testlerinin parametrik yöntemlere göre daha güvenilir sonuçlar ürettiği ve daha yüksek I. Tip hatadan kaçınmak için önerilebilecekleri görülmüştür

USE AND COMPARISON OF PERMUTATION TESTS IN LINEAR MODELS

F and t-test are generally used to test significance of hypothesis and/or model parameters. Although parametric tests are considerably effective, they can be ineffective when the assumptions needed by model are not provided, which is a usual situation for many data sets. In this case, permutation test not affected by the assumptions can be applied as a non-parametric method. In this study, permutation tests such as permutation of raw data, permutation of residuals under full model and permutation of residuals under restricted model are compared for multiple linear regression, completely randomized designs, randomized block design and Latin square design in terms of the Type I error rates, and performance of each tests are studied via animal science data. Results from this study indicate that permutation tests yields more reliable results than parametric tests in terms of Type I error rate, and permutation tests are recommended in order to reduce Type I errors.

Keywords:

Linear models Resampling methods, Permutation tests,

PDF

___

Anderson, M.J., Legendre, P. 1999. An Empirical Comparison of Permutation Methods for Tests of Partial Regression Coefficients in a Linear Model. J. Statist. Comput. Simul. 62, 271-303.
Anderson, M.J. 2000. NPMANOVA: A FORTRAN Computer Program for Non-Parametric Multivariate Analysis of Variance (For Any Two Factor ANOVA Design) Using Permutation Tests. Department of Statistics, University of Auckland.
Anderson, M.J. 2001. Permutation Tests for Univariate and Multivariate Analysis of Variance and Regression. Can. J. Fish. Aquat. Sci. 58, 626639.
Anderson, M.J., Robinson, J. 2001. Permutation Tests for Linear Models. Aust. N. Z. Stat. 43(1), 75 – 88.
Anderson, M.J. 2003. DISTLM: A FORTRAN Computer Program to Calculate a Distance-Based Multivariate Analysis for a Linear Model. Department of Statistics, University of Auckland.
Darcan, N. 2004. Data obtained from small ruminant research. (Personal communication).
Fisher, R.A. 1935. The Design of Experiments. Oliver and Body, Edinburgh.
Good, P. 2000. Permutation Tests, A Practical Guide to Resampling Methods for Testing Hypotheses. Springer-Werlag New York Inc.
Kleinbaum, D.G., Kupper, L.L., Muller, K.E., Nizam, A. 1998. Applied Regression Analysis and Other Multivariable Methods. Brooks/Cole Publishing Company.
Mielke, P.W., Berry, K.J. 2001. Permutation Methods: A Distance Function Approach. Springer-Werlag New York Inc.
Önder, H. 2007. Using Permutation Tests to Reduce Type I and II Errors for Small Ruminant Research. J. Appl. Anim. Res. 32 (1), 69 – 72.
Pitman, E.J.G. 1937. Significance Tests Which May be Applied to Samples from Any Population. Royal Staistical Society Supplement. Part I. 4, 119-130.
Routledge, R.D. 1997. P-Values from Permutation and F-Tests. Computational Statistics & Data Analysis. 24, 379 – 386.
Scheffé, H. 1959. The Analysis of Variance, John Willey & Sons, Inc. New York.
Serbester, U., Görgülü, M., Kutlu, H.R., Yurtseven, S, Arieli, A., Kowalski, Z.M. 2005. The Effects of Sprinkler+Fan, Fish Meal or Dietary Fat on Milk Yield and Milk Composition of Dairy Cows in Mid Lactation During Summer. Journal of Animal and Feed Sciences 14, 639 – 653.
Tanizaki, H. 2001. On Small Sample Properties Permutation Tests: An Independence Test Between Two Samples and A Significance Test For Regression Models. (http://ht.econ.kobeu.ac.jp/~tanizaki/cv/working/permute.pdf Last visit: 12/03/2007).