I'm running a posthoc Tukey HSD on my data, which has ten factor levels. The table is massive and I was hoping to just present the p-values to the reader, in a pairwise table, leaving the 45 row-ed table for the appendix.
Here is an example dataset:
set.seed(42)
x <- rnorm(100,1,2)
category <- letters[1:10]
data <- cbind.data.frame(x, category)
summary(data.aov <- aov(x~category, data = data))
data.hsd<-TukeyHSD(data.aov)
data.hsd.result<-data.frame(data.hsd$category)
data.hsd.result
The result is a table of 45 rows. Instead, I'd like a table with the factor levels as row and column names, with the p-value in the cell, showing if the two are significantly different. Xs or underscores or whatever could represent repeated or unnecessary comparisons. Something like this:
a b c d e f ... j
a X 0.97 1 0.99 0.89 0.99 ... 0.99
b X X 0.99 0.89 0.94 0.92 ... 0.97
c X X X 0.85 0.93 0.96 ... 0.98
| ... ... ... ... ... ... ... ...
i X X X X X X ... 0.84
and so on.
Is there a way to produce a table like this automatically?
You want the p-values in the upper-triangular matrix form. That's a bit unnatural for R since it fills its matrices by column, but it easy enough to fix. First check that you are getting the correct order:
> resm <- matrix(NA, 10, 10)
> resm[lower.tri(resm) ] <- rownames(data.hsd.result)
> resm
[,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10]
[1,] NA NA NA NA NA NA NA NA NA NA
[2,] "b-a" NA NA NA NA NA NA NA NA NA
[3,] "c-a" "c-b" NA NA NA NA NA NA NA NA
[4,] "d-a" "d-b" "d-c" NA NA NA NA NA NA NA
[5,] "e-a" "e-b" "e-c" "e-d" NA NA NA NA NA NA
[6,] "f-a" "f-b" "f-c" "f-d" "f-e" NA NA NA NA NA
[7,] "g-a" "g-b" "g-c" "g-d" "g-e" "g-f" NA NA NA NA
[8,] "h-a" "h-b" "h-c" "h-d" "h-e" "h-f" "h-g" NA NA NA
[9,] "i-a" "i-b" "i-c" "i-d" "i-e" "i-f" "i-g" "i-h" NA NA
[10,] "j-a" "j-b" "j-c" "j-d" "j-e" "j-f" "j-g" "j-h" "j-i" NA
> t(resm)
[,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10]
[1,] NA "b-a" "c-a" "d-a" "e-a" "f-a" "g-a" "h-a" "i-a" "j-a"
[2,] NA NA "c-b" "d-b" "e-b" "f-b" "g-b" "h-b" "i-b" "j-b"
[3,] NA NA NA "d-c" "e-c" "f-c" "g-c" "h-c" "i-c" "j-c"
[4,] NA NA NA NA "e-d" "f-d" "g-d" "h-d" "i-d" "j-d"
[5,] NA NA NA NA NA "f-e" "g-e" "h-e" "i-e" "j-e"
[6,] NA NA NA NA NA NA "g-f" "h-f" "i-f" "j-f"
[7,] NA NA NA NA NA NA NA "h-g" "i-g" "j-g"
[8,] NA NA NA NA NA NA NA NA "i-h" "j-h"
[9,] NA NA NA NA NA NA NA NA NA "j-i"
[10,] NA NA NA NA NA NA NA NA NA NA
So it's just:
resm <- matrix(NA, 10, 10)
resm[lower.tri(resm) ] <-round(data.hsd.result$p.adj, 3)
t(resm)
[,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10]
[1,] NA 0.974 1.00 1 0.885 0.997 0.985 0.673 0.559 1.000
[2,] NA NA 0.99 1 1.000 1.000 1.000 0.999 0.997 0.999
[3,] NA NA NA 1 0.938 0.999 0.995 0.772 0.666 1.000
[4,] NA NA NA NA 0.990 1.000 1.000 0.921 0.856 1.000
[5,] NA NA NA NA NA 1.000 1.000 1.000 1.000 0.988
[6,] NA NA NA NA NA NA 1.000 0.991 0.974 1.000
[7,] NA NA NA NA NA NA NA 0.998 0.993 1.000
[8,] NA NA NA NA NA NA NA NA 1.000 0.914
[9,] NA NA NA NA NA NA NA NA NA 0.846
[10,] NA NA NA NA NA NA NA NA NA NA
Adding row and column names to a matrix is rather trivial with the functions: rownames<- and colnames<-. See their shared help page for worked examples.