jolliffe_test {ClustMC} | R Documentation |
Jolliffe test for multiple comparisons
Description
I.T. Jolliffe test for multiple comparisons.
Implements a cluster-based alternative closely linked to the
Student-Newman-Keuls multiple comparison method. Single-linkage cluster
analysis is applied, using the p-values obtained with the SNK test for
pairwise mean comparison as a similarity measure. Groups whose means join
beyond 1 - \alpha
are statistically different. Alternatively, complete
linkage cluster analysis can also be applied.
Usage
jolliffe_test(
y,
trt,
alpha = 0.05,
method = "single",
show_plot = TRUE,
console = TRUE,
abline_options,
...
)
Arguments
y |
Either a model (created with |
trt |
If |
alpha |
Numeric value corresponding to the significance level of the test. The default value is 0.05. |
method |
|
show_plot |
Logical value indicating whether the constructed dendrogram should be plotted or not. |
console |
Logical value indicating whether the results should be printed on the console or not. |
abline_options |
|
... |
Optional arguments for the |
Value
A list with three data.frame
and one hclust
:
stats |
|
groups |
|
parameters |
|
dendrogram_data |
object of class |
Author(s)
Santiago Garcia Sanchez
References
Jolliffe, I. T. (1975). Cluster analysis as a multiple comparison method. Applied Statistics: Proceedings of Conference at Dalhousie University, Halifax, 159-168.
Examples
data("PlantGrowth")
# Using vectors -------------------------------------------------------
weights <- PlantGrowth$weight
treatments <- PlantGrowth$group
jolliffe_test(y = weights, trt = treatments, alpha = 0.1, show_plot = FALSE)
# Using a model -------------------------------------------------------
model <- lm(weights ~ treatments)
jolliffe_test(y = model, trt = "treatments", alpha = 0.1, show_plot = FALSE)