Unable to change effsize.ylim

[silasprincipe]

Hi! I'm having trouble in changing the scale of y-axis of the effect size in a two-independent DABEST plot. I have two different plots to construct, and I want them to have the same scale sizes to facilitate understanding.

Here is the code with some data to exemplify:

cont1 <- c(17, 18.6, 21, 27, 13, 26, 19.5, 14.8, 12, 23, 10, 14.6, 23, 24)
treat1 <- c(30, 24, 18, 21, 12, 27.1, 25.5, 29, 45, 22, 29.3, 34, 22, 13)
cont2 <- c(6, 6, 22, 20, 15, 21, 13, 6, 6, 19, 6, 6, 31, 29)
treat2 <- c(6.1, 25, 22, 20, 15, 21, 13, 30, 60, 19, 6, 6, 31, 29)

First dataset

data.a <- tibble::tibble(Control = cont1, Treatment = treat1)
data.a.f <-
data.a %>%
tidyr::gather(key = Group, value = Measurement)
dabest.a <-
data.a.f %>%
dabest(Group, Measurement,
# The idx below passes "Control" as the control group,
# and "Group1" as the test group. The mean difference
# will be computed as mean(Group1) - mean(Control1).
idx = c("Control", "Treatment"),
paired = FALSE)
plot(dabest.a,
rawplot.ylabel = "Concentration",
effsize.ylabel = "Mean difference", palette = "Dark2",
rawplot.markersize = 4, rawplot.groupwidth = 0.2,
rawplot.ylim = c(10, 50),
effsize.ylim = c(0, 30))

Second dataset

data.b <- tibble::tibble(Control = cont2, Treatment = treat2)
data.b.f <-
data.b %>%
tidyr::gather(key = Group, value = Measurement)
dabest.b <-
data.b.f %>%
dabest(Group, Measurement,
# The idx below passes "Control" as the control group,
# and "Group1" as the test group. The mean difference
# will be computed as mean(Group1) - mean(Control1).
idx = c("Control", "Treatment"),
paired = FALSE)
plot(dabest.b,
rawplot.ylabel = "Concentration",
effsize.ylabel = "Mean difference", palette = "Dark2",
rawplot.markersize = 4, rawplot.groupwidth = 0.2,
rawplot.ylim = c(10, 50),
effsize.ylim = c(0,30))

Another question: how do I change the scale interval (for example, by 5, not 10). Thanks!

[josesho]

Hi,

I would suggest using a Cumming plot rather than two Gardner-Altman plots to achieve what you want.

all.data <- tibble::tibble(Control1 = cont1, Treatment1 = treat1,
                           Control2 = cont2, Treatment2 = treat2)

all.data.tidy <-
  all.data %>%
  tidyr::gather(key = Group, value = Measurement)

dabest.all.data <- 
  all.data.tidy %>%
  dabest(Group, Measurement,
         idx = list(c("Control1", "Treatment1"),
                    c("Control2", "Treatment2")),
         paired = FALSE)

plot(dabest.all.data)

If, however, you want two Gardner-Altman plots, then what you have done above (setting rawplot.ylim) works. The nature of the Gardner-Altman plots is such that one can only set the ylims for the rawdata plot (on the left); the axes with the effect size and confidence interval is, by design, aligned to the axes on the left and currently cannot be set.

[josesho]

Another question: how do I change the scale interval (for example, by 5, not 10). Thanks!

To the best of my knowledge, this is effectively not possible now. (We use cowplot to stitch the raw swarmplot with the effect size plot. You might be able to isolate the axes, and then use ggplot's scale_y_continuous to set the tick breaks, but I'm not sure how yet.)

[silasprincipe]

Thanks @josesho !