multiple: Multiple comparisons with plot.

Description

multiple displays results from post-doc analysis and constructs corresponding plot.

Usage

multiple(
  model,
  formula,
  adjust = "hochberg",
  type = "response",
  reverse = TRUE,
  level = 0.95,
  digits = 2,
  ...
)

Value

A list with objects: df A data frame with adjusted p-values, fig_ci a plot with estimates and adjusted confidence intervals and fig_pval a plot comparing adjusted p-values.

Arguments

model: A fitted model supported by emmeans, such as the result of a call to aov, lm, glm, etc.
formula: A formula with shape: ~ y or ~ y|x (for interactions). Where y is the term of the model on which comparisons are made and x is a term interacting with y.
adjust: Method to adjust CIs and p-values (see details).
type: Type of prediction (matching "linear.predictor", "link", or "response").
reverse: Logical argument. Determines the direction of comparisons.
level: Confidence interval significance level.
digits: Number of digits for rounding (default = 2).
...: Further arguments passed to emmeans.

Details

The default adjusting method is "mvt" which uses the multivariate t distribution. Other options are: "bonferroni", "holm", "hochberg", "tukey" and "none". The default option for argument reverse is to make reverse comparisons, i.e., against the reference level matching comparisons from lm and glm.

Examples

Run this code

data(birthwt, package = "MASS")
birthwt$race <- factor(birthwt$race, labels = c("White", "African American", "Other"))

model_1 <- aov(bwt ~ race, data = birthwt)
multiple(model_1, ~race)

multiple(model_1, ~race) |>
  effect_plot(x = estimate, y = contrast, orientation = "y") +
  xlab("Difference in birth weights (g)") +
  ylab("") + geom_vline(xintercept = 0, col = bmj[2], lty = 2)