COMPARISON OF SOME TESTS FOR CORRELATED ERROR IN ANIMAL STUDIES

Yıl 1996, Cilt: 6 Sayı: 1, 191 - 203, 07.06.2016

Hayrettin Okut Tufan Altın Serhat Arslan Ahmet Taşdelen

Öz

Some assumptions should be checkcd . because  of autqçprre.Iation of residuals, when data collected on the same subject at each timtU5:ıJp_aCif Thefo is' 'no problem for testing the treatınents,  whercas  some

assumptfons - should  be  valid  for  testing ' the  period   and  period   freatment

interaction in case of iepeated ·measures . .in this study, . 53 male lambs werc; assigned to th '.İhree tieatmertts and each one was measured with  15 days' ·.;

interval through '70 days.: Four different analysis  methods, standard  analysis.  ' transformed   data  analysis,   auto-regressive   error  model.   and  inuitivariate ·.:: anaiysis· wefrf(pefforiried to comparc for   epeated  measures. ' There we're nô . diffo(ertces :irt terrns  of  imprövement , .of ,F  test  among  standard  analysis. trasfomied'·'; data'    analysis          and       n:iultivariate . arialvsis ·   methods .·   Thc ·

.    (J .,:._ _.,:· ı'·' ff .:.'.·<. )-:  '   ".d  .              .  .                            .          '          -      :        .     . .  . .      .  .,     ..    -

autorgrssıve. error model ımproved thc F:tcst by _%20 ,Whereas assu mptıon ,: . were'not vaıief fo{this analysis:

Anahtar Kelimeler

-

HAYVANCILIK UYGULAMALARINDA HATALARIN BİRBİRİ İLE İLİŞKİLİ OLDUĞU DURUMLARDA BAZI TESTLERİN KARŞILAŞTIRILMASI

Öz

ÖZET

Anahtar Kelimeler

-