dwplot() for lsmeans in R

Question

I would like to plot the output from more than one regression.

MWE:

df <- data.frame(
  y1 = runif(100),
  y2 = runif(100),
  x = c(rep("A",times=50), rep("B",times=50))
)

If I want to plot the intercept and coefficients, I can use dwplot():

m0 <- lm(y1 ~ factor(x), data = df)
m1 <- lm(y2 ~ factor(x), data = df)
dwplot(list(m0, m1), show_intercept=T) + theme_bw()

However, I am interested in the ls means (computed with the package library(lsmeans)). For this case, dwplot() gives me an error.

lsm0 <- lsmeans(m0,~ factor(x), data = df)
lsm1 <- lsmeans(m1,~ factor(x), data = df)
dwplot(list(lsm0, lsm1), show_intercept=T) + theme_bw()

What are possible solutions?

score 1 · Accepted Answer · answered Apr 07 '23 at 11:29

You could use the results from both object and create one dataframe to make your own ggplot with geom_pointrange like this:

library(dotwhisker)
library(lsmeans)

m0 <- lm(y1 ~ factor(x), data = df)
m1 <- lm(y2 ~ factor(x), data = df)

lsm0 <- lsmeans(m0, ~ factor(x), data = df)
lsm1 <- lsmeans(m1, ~ factor(x), data = df)

library(tidyverse)
as.data.frame(lsm0) %>%
  mutate(model = "model 1") %>%
  bind_rows(., as.data.frame(lsm1) %>%
              mutate(model = "model 2")) %>%
  ggplot(aes(x = lsmean, y = x, color = model, xmin = lower.CL, xmax = upper.CL)) +
  geom_pointrange(position = position_dodge(width = 0.2)) +
  theme_bw()

^{Created on 2023-04-07 with reprex v2.0.2}

dwplot() for lsmeans in R

1 Answers1