使用Field在发现统计数据时使用R（pag 561）提出的单向重复测量方法ANOVA：
Participant <- gl(8, 4, labels = c("P1", "P2", "P3", "P4", "P5", "P6", "P7","P8" ))
Animal <- gl(4, 1, 32, labels = c("Stick_Insect", "Kangaroo_Testicle", "Fish_Eye", "Witchetty_Grub"))
Retch <- c(8, 7, 1, 6, 9, 5, 2, 5, 6, 2, 3, 8, 5, 3, 1, 9, 8, 4, 5, 8, 7, 5, 6, 7, 10, 2, 7, 2, 12, 6, 8, 1)
longBush<-data.frame(Participant, Animal, Retch)
pairwise.t.test(longBush$Retch, longBush$Animal, paired=TRUE, alternative="two.sided", p.adjust.method="none")
结果是：
Stick_Insect Kangaroo_Testicle Fish_Eye
Kangaroo_Testicle 0.00202 - -
Fish_Eye 0.00094 0.92007 -
Witchetty_Grub 0.22673 0.29867 0.40204
改变一下'pairwise.t.test'，dof是（主题数是8）：
Stick_Insect Kangaroo_Testicle Fish_Eye
Kangaroo_Testicle 7 - -
Fish_Eye 7 7 -
Witchetty_Grub 7 7 7
到目前为止很好=）
让我们模拟一个双向重复测量ANOVA：
Participant <- gl(8, 8, labels = c("P1", "P2", "P3", "P4", "P5", "P6", "P7","P8" ))
Animal <- gl(4, 1*2, 64, labels = c("Stick_Insect", "Kangaroo_Testicle", "Fish_Eye", "Witchetty_Grub"))
Alive <- gl(2, 1, 64, labels = c("Y", "N"))
Retch <- sample(c(1:12), 64, replace = TRUE)
longBush2 <- data.frame(Participant, Animal, Alive, Retch)
pairwise.t.test(longBush2$Retch, longBush2$Animal, paired=TRUE, alternative="two.sided", p.adjust.method="none");
pairwise.t.test结果是：
Stick_Insect Kangaroo_Testicle Fish_Eye
Kangaroo_Testicle 0.022 - -
Fish_Eye 0.202 0.584 -
Witchetty_Grub 0.374 0.041 0.433
和自由度：
Stick_Insect Kangaroo_Testicle Fish_Eye
Kangaroo_Testicle 15 - -
Fish_Eye 15 15 -
Witchetty_Grub 15 15 15
相反，Alive因子有31自由度。有趣。
所以，我的问题：为什么dof被错误地改变了？我做错了什么？
提前谢谢
干杯