R devel - Jan 2011 - normality and equal variance testing

I currently have a program that automates 2-way ANOVA on a series of endpoints, 
but before the ANOVA is carried out I want the code to test the assumptions of 
normality and equal variance and report along with each anova result in the 
output file.  How can I do this? 


I have pasted below the code that I currently use. 

 
library(car)
numFiles = x #
<--<--<--<--<--<--<--<--<--<--<-- supply the
number of files
containing the source data here
for (iIndx in 1:numFiles) {
       sinkFilePath = paste("C:/AnovaData/2WayAnovaForProteins_Set",
iIndx,
".txt", sep="")
 sink(sinkFilePath)
 sourceFilePath = paste("C:/AnovaData/ProteinsFor2WayAnova_Set",
iIndx, ".txt",
sep="")
 dataSet = read.delim(sourceFilePath)  
 numProteins = ncol(dataSet)
 nameProteins = colnames(dataSet)
 for (i in 3:numProteins) {
  fla = as.formula(paste(nameProteins[i],"~","Trt*Dose"))
  mod = lm(fla, data = dataSet , contrasts = list(Trt = contr.sum, Dose = 
contr.sum))
  ano = Anova(mod, type = "III")
  print.noquote("")
  print.noquote("")
  print.noquote("--------------------------------------------")
  print.noquote(paste("--------- Analysis of Spot: 
",nameProteins[i],"------------"))
  print.noquote("--------------------------------------------")
  print(ano)
 }
}


Thanks in advance.
Karthi


	[[alternative HTML version deleted]]

Karthi Subramanian <karthi_subramanian <at> yahoo.ca> writes:
> 
> I currently have a program that automates 2-way ANOVA on a series of
> endpoints, but before the ANOVA is carried out I want the code to
> test the assumptions of normality and equal variance and report along
>  with each anova result in the 
> output file.? How can I do this? 
> 
 [snip]

  I think this is the wrong list for this question; please see
<http://www.r-project.org/posting-guide.html#which_list>

  See ?shapiro.test, ?fligner.test, nortest package ...

also library(sos); findFn("{normality test}"), etc.

Why do you want to test for normality and equal variances?

If those are really a concern then you should use a method up front that is
robust against those.  Those tests are usually testing a hypothesis that is
different from what you are actually interested in and generally have low power
to guide further tests.

-- 
Gregory (Greg) L. Snow Ph.D.
Statistical Data Center
Intermountain Healthcare
greg.snow at imail.org
801.408.8111

> -----Original Message-----
> From: r-devel-bounces at r-project.org [mailto:r-devel-bounces at r-
> project.org] On Behalf Of Karthi Subramanian
> Sent: Monday, January 24, 2011 10:06 AM
> To: R-devel at r-project.org
> Subject: [Rd] normality and equal variance testing
> 
> I currently have a program that automates 2-way ANOVA on a series of
> endpoints,
> but before the ANOVA is carried out I want the code to test the
> assumptions of
> normality and equal variance and report along with each anova result in
> the
> output file.? How can I do this?
> 
> 
> I have pasted below the code that I currently use.
> 
> 
> library(car)
> numFiles = x #
<--<--<--<--<--<--<--<--<--<--<-- supply the
number of
> files
> containing the source data here
> for (iIndx in 1:numFiles) {
> ?????? sinkFilePath =
paste("C:/AnovaData/2WayAnovaForProteins_Set",
> iIndx,
> ".txt", sep="")
> ?sink(sinkFilePath)
> ?sourceFilePath = paste("C:/AnovaData/ProteinsFor2WayAnova_Set",
iIndx,
> ".txt",
> sep="")
> ?dataSet = read.delim(sourceFilePath)
> ?numProteins = ncol(dataSet)
> ?nameProteins = colnames(dataSet)
> ?for (i in 3:numProteins) {
> ??fla =
as.formula(paste(nameProteins[i],"~","Trt*Dose"))
> ??mod = lm(fla, data = dataSet , contrasts = list(Trt = contr.sum, Dose
> > contr.sum))
> ??ano = Anova(mod, type = "III")
> ??print.noquote("")
> ??print.noquote("")
> ??print.noquote("--------------------------------------------")
> ??print.noquote(paste("--------- Analysis of Spot:
> ",nameProteins[i],"------------"))
> ??print.noquote("--------------------------------------------")
> ??print(ano)
> ?}
> }
> 
> 
> Thanks in advance.
> Karthi
> 
> 
> 	[[alternative HTML version deleted]]