1

I have a data structure organized as in next image (two populations and some parameters). I'm getting confused with coding. Which is the way to make an ANOVA test that drops the suitability (F, p value) of each parameter, please?

enter image description here

galtor
  • 147
  • 6
  • Be more specific: Do you want to run one ANOVA for each variable? Can you provide some toy data? – broti Mar 23 '20 at 13:55

1 Answers1

1

You could try something like this, but probably need to account for multiple comparisons:

library(data.table)

DT <- data.table("GROUP" = c("GR1", "GR1", "GR1", "GR2", "GR2")
               , "Weight" = c(78, 85, 84.3, 70, 67)
               , "BloodPres" = c(11, 14, 13, 12, 12)
               , "Heart" = c(140, 130, 142, 135, 120)
               , "Glucose" = c(80, 110, 95, 97, 105))

Outcomes <- c("Weight", "BloodPres", "Heart", "Glucose")

sapply(Outcomes, function(my) {
       f <- as.formula(paste(my, "~GROUP", sep=""))
       summary(aov(f, data=DT))
})

This works for one-way, but you'll need to adjust it for two-way (see Correct use of sapply with Anova on multiple subsets in R).

cgrafe
  • 443
  • 1
  • 6
  • 14
  • 1
    Thanks. I found this https://www.educba.com/anova-in-r/ interesting. At the end of the page, it compares response, vs treatment & BP, similar case to mine – galtor Mar 23 '20 at 15:09