I have a within-subjects dataset for which I am running pairwise comparisons on a dependent variable that is be separated into four groups. I have the data organized long-ways and wide-ways:
Example long-ways:
| direction | count |
|---|---|
| l2l | 44 |
| l2l | 49 |
| r2r | 100 |
| r2r | 582 |
| r2l | 24 |
| r2l | 1000 |
| l2r | 3 |
| l2r | 244 |
Example wide-ways:
| l2l_count | r2r_count | r2l_count | l2r_count |
|---|---|---|---|
| 44 | 100 | 24 | 3 |
| 49 | 582 | 1000 | 244 |
I am interested in looking at every possible pairwise comparison on the variable "count" based on the grouping variable "direction" in R. However, I cannot figure out how to do this in R in an automated fashion.
With a data.frame called hcp_dir, I have tried using the following with my long-ways data:
pairwise.wilcox.test(hcp_dir$count, hcp_dir$direction, paired = TRUE, p.adj="bonf")
However, this only gives me a matrix of corrected p-values for every possible comparison, and does not yield a test statistic, which I need.
I have also tried using the following, which gives me bootstrapped confidence intervals, and an effect size, but no test statistic with long-ways oriented data:
wilcox_effsize(hcp_dir, count~direction, paired = TRUE, alternative = "two.sided", mu = 0, ci = TRUE, conf.level = 0.95, ci.type = "perc", nboot = 10000)
The only way I can figure out how to get a test statistic for the Wilcoxon sign-rank test is to use the following code, which requires the data to be organized wide-ways. It is ultimately with this script that I cannot figure out how to iterate through every possible direction pairing automatically.
wilcox.test(hcp_dir$l2l_count, hcp_dir$r2r_count, paired = TRUE, alternative = "two.sided")
Note that this is the only way I am able to get a V statistic. I need to figure out how to iterate across all possible comparisons for efficiency.