Skip to contents

Statistical testing for SuperPlots

Source: vignettes/stats.Rmd
stats.Rmd

Adding statistical test results to a SuperPlot

SuperPlotR will do statistical testing for you (in a simple way) but is turned off by default. You can turn it on by setting stats = TRUE in your call to superplot().

library(SuperPlotR)
p <- superplot(lord_jcb, "Speed", "Treatment", "Replicate", stats = TRUE)
#> Performing t-test
#> 
#>  Welch Two Sample t-test
#> 
#> data:  x and y
#> t = 1.2907, df = 3.8128, p-value = 0.2695
#> alternative hypothesis: true difference in means is not equal to 0
#> 95 percent confidence interval:
#>  -11.91888  31.89283
#> sample estimates:
#> mean of x mean of y 
#>  31.58355  21.59657

Four types of statistical tests are available: t-test, Wilcoxon rank sum test, ANOVA and Kruskal-Wallis test. Select between these options by setting stats_test to one of the following: "para_unpaired" (default), "para_paired", "nonpara_unpaired" or "nonpara_paired".

If the number of conditions is two, these options result in:

  • "para_unpaired": t-test
  • "para_paired": paired t-test
  • "nonpara_unpaired": Wilcoxon rank sum test a.k.a Mann-Whitney U test
  • "nonpara_paired": Wilcoxon signed rank test

If the number of conditions is greater than two, these options result in:

  • "para_unpaired": ANOVA followed by Tukey’s HSD test
  • "para_paired": Repeated measures ANOVA
  • "nonpara_unpaired": Kruskal-Wallis test
  • "nonpara_paired": direction to perform a Friedman test

Returning to the example data, in the Lord et al. paper they perform a paired t-test on the data. We can do this in SuperPlotR by setting stats = TRUE and stats_test to "para_paired".

p <- superplot(lord_jcb, "Speed", "Treatment", "Replicate", stats = TRUE,
               stats_test = "para_paired")
#> Performing t-test
#> 
#>  Paired t-test
#> 
#> data:  x and y
#> t = 8.1741, df = 2, p-value = 0.01464
#> alternative hypothesis: true mean difference is not equal to 0
#> 95 percent confidence interval:
#>   4.730053 15.243899
#> sample estimates:
#> mean difference 
#>        9.986976

As a further example, we will generate some test data with four treatments.

set.seed(123)
example <- data.frame(meas = rep(rep(c(10, 9, 6, 6), each = 25), 4) + rnorm(400),
                      cond = rep(rep(c("ctrl", "drug 1", "drug 2", "drug 3"), each = 25), 4),
                      expt = rep(c("exp1","exp2","exp3","exp4"), each = 100))
superplot(example, "meas", "cond", "expt", linking = TRUE,
          stats_test = "para_unpaired")

How to add the p-values or stars to the SuperPlot

The only way to do this is manually.

Adding p-values or stars to the SuperPlot is currently not supported due to the number of different ways that users may wish to add them. However, it is possible to use the output and add them manually to the SuperPlot object after it has been generated, either in R or afterwards in Inkscape or similar program.