1

Question:

Is there an equivalent of Nested -ANOVA for data which fails ANOVA requirement? Is it possible to perform in R?

I am doing an analysis of large data set.

  • I have three groups (Genotypes: Wild type, Heterozygous , Homozygous mutant).
  • Each group has data from >10 animals.
  • For each animal, I have 500+ independent observations (example: Volume of 500 different retinal cells).

Based on Levene's test between the three groups, I see I cannot do ANOVA. I have decided to do a Welch ANOVA and I do see significant difference. (I used the oneway.test() function.)

As my 500+ observations are nested inside individual animals, I think I should perform nested Welch ANOVA for my data set.

How can I do nested Welch ANOVA in R?

Could you please show me an example with a similar data?

mydata <- read.table(header=TRUE,
text=
"Tech Rat Protein
Janet 1   1.119 
Janet 1   1.2996 
Janet 1   1.5407 
Janet 1   1.5084 
Janet 1   1.6181 
Janet 1   1.5962 
Janet 1   1.2617 
Janet 1   1.2288 
Janet 1   1.3471 
Janet 1   1.0206 
Janet 2   1.045 
Janet 2   1.1418 
Janet 2   1.2569 
Janet 2   0.6191 
Janet 2   1.4823 
Janet 2   0.8991 
Janet 2   0.8365 
Janet 2   1.2898 
Janet 2   1.1821 
Janet 2   0.9177 
Janet 3   0.9873 
Janet 3   0.9873 
Janet 3   0.8714 
Janet 3   0.9452 
Janet 3   1.1186 
Janet 3   1.2909 
Janet 3   1.1502 
Janet 3   1.1635 
Janet 3   1.151 
Janet 3   0.9367 
Brad  5   1.3883 
Brad  5   1.104 
Brad  5   1.1581 
Brad  5   1.319 
Brad  5   1.1803 
Brad  5   0.8738 
Brad  5   1.387 
Brad  5   1.301 
Brad  5   1.3925 
Brad  5   1.0832 
Brad  6   1.3952 
Brad  6   0.9714 
Brad  6   1.3972 
Brad  6   1.5369 
Brad  6   1.3727 
Brad  6   1.2909 
Brad  6   1.1874 
Brad  6   1.1374 
Brad  6   1.0647 
Brad  6   0.9486 
Brad  7   1.2574 
Brad  7   1.0295 
Brad  7   1.1941 
Brad  7   1.0759 
Brad  7   1.3249 
Brad  7   0.9494 
Brad  7   1.1041 
Brad  7   1.1575 
Brad  7   1.294 
Brad  7   1.4543 
")
kjetil b halvorsen
  • 77,844
Savani
  • 11
  • 3
  • 1
    I'd suggest using package nlme and fit a mixed effects model. The package supports fitting variance structures and can thereby account for heteroscedasticity. – Roland May 24 '16 at 10:49
  • @Roland. Thanks a lot. model1=lm(Volume~genotype*Fish_ID, data=dat ) summary(model1) Is this the way? – Savani May 25 '16 at 10:12
  • This looks like the example from the Handbook of Biological Statistics. That page links to the same example in R, which uses a mixed model with nlme. It doesn't specifically address dealing with heteroscedasticity. – Sal Mangiafico Dec 07 '17 at 17:27
  • 1
    There's also a possibility of using a heteroscedasticity-corrected covariance matrix. With either a lm model or a lme model, the Anova function in the car package will work. I don't know if this approach would be recommended in your case or not. library(car); Anova(model, white.adjust=TRUE) – Sal Mangiafico Dec 07 '17 at 17:37

0 Answers0