Supported Input Formats • cld

library(cld)

Overview

The make_cld() function works seamlessly with multiple input formats, making it a versatile tool for creating compact letter displays from various statistical packages and custom data.

Supported Input Types

Input Type	Example Packages	Function Examples
`pairwise.htest`	base R	`pairwise.t.test()`, `pairwise.wilcox.test()`
`PMCMR` / `PMCMRplus`	PMCMR, PMCMRplus	`kwAllPairsConoverTest()`, `dunnTest()`
`data.frame` (rstatix)	rstatix	`games_howell_test()`, `tukey_hsd()`
`PostHocTest`	DescTools	`ConoverTest()`, `DunnettTest()`
`matrix`	Custom	Symmetric p-value matrices
`data.frame`	Custom	Custom comparison data frames
`formula`	Custom	Formula interface for data frames

Base R Pairwise Tests

Pairwise Wilcoxon Test

# Pairwise Wilcoxon rank sum test
result <- pairwise.wilcox.test(chickwts$weight, chickwts$feed, exact = FALSE)
make_cld(result)
#> Compact Letter Display (CLD)
#> Signif. level (alpha):  0.05 
#> Method:  Wilcoxon rank sum test with continuity correction 
#> 
#>      group cld spaced_cld
#>     casein   a        a__
#>  horsebean   b        _b_
#>    linseed  bc        _bc
#>   meatmeal  ac        a_c
#>    soybean   c        __c
#>  sunflower   a        a__

Pairwise t-test

# Pairwise t-test
result2 <- pairwise.t.test(chickwts$weight, chickwts$feed)
make_cld(result2, alpha = 0.01)  # More stringent threshold
#> Compact Letter Display (CLD)
#> Signif. level (alpha):  0.01 
#> Method:  t tests with pooled SD 
#> 
#>      group cld spaced_cld
#>     casein   a        a__
#>  horsebean   b        _b_
#>    linseed  bc        _bc
#>   meatmeal  ac        a_c
#>    soybean   c        __c
#>  sunflower   a        a__

P-value Matrices

You can use symmetric matrices of p-values directly:

# Create a symmetric matrix of p-values
m <- matrix(c(
  1.00, 0.22, 0.05, 0.00,
  0.22, 1.00, 0.17, 0.01,
  0.05, 0.17, 1.00, 0.22,
  0.00, 0.01, 0.22, 1.00
), nrow = 4)
rownames(m) <- colnames(m) <- c("GroupA", "GroupB", "GroupC", "GroupD")

# Generate CLD
make_cld(m, alpha = 0.05)
#> Compact Letter Display (CLD)
#> Signif. level (alpha):  0.05 
#> Method:  matrix 
#> 
#>   group cld spaced_cld
#>  GroupA   a         a_
#>  GroupB   a         a_
#>  GroupC  ab         ab
#>  GroupD   b         _b

PMCMRplus (Non-parametric Tests)

The package supports results from PMCMRplus for non-parametric post-hoc tests:

library(PMCMRplus)

# Kruskal-Wallis post-hoc test
kw_result <- kwAllPairsConoverTest(count ~ spray, data = InsectSprays)
make_cld(kw_result)

# Dunn test
dunn_result <- kwAllPairsDunnTest(count ~ spray, data = InsectSprays)
make_cld(dunn_result)

rstatix (Tidyverse-friendly)

The rstatix package provides tidyverse-compatible statistical tests. Since rstatix returns data frames, you need to specify the column names for groups and p-values:

library(rstatix)

# Games-Howell test (for unequal variances)
gh_result <- games_howell_test(PlantGrowth, weight ~ group)
make_cld(gh_result, gr1_col = "group1", gr2_col = "group2", p_val_col = "p.adj")

# Tukey HSD test
tukey_result <- tukey_hsd(PlantGrowth, weight ~ group)
make_cld(tukey_result, gr1_col = "group1", gr2_col = "group2", p_val_col = "p.adj")

# Pairwise t-test
pwt_result <- pairwise_t_test(PlantGrowth, weight ~ group)
make_cld(pwt_result, gr1_col = "group1", gr2_col = "group2", p_val_col = "p.adj")

DescTools

The DescTools package offers various post-hoc tests:

library(DescTools)

# Conover test
conover_result <- ConoverTest(count ~ spray, data = InsectSprays)
make_cld(conover_result)

# Dunnett test (comparison to control)
dunnett_result <- DunnettTest(weight ~ group, data = PlantGrowth)
make_cld(dunnett_result)

Data Frames (Custom Results)

You can create CLDs from custom comparison results stored in data frames:

# Custom comparison results
comparisons <- data.frame(
  group1 = c("Treatment_A", "Treatment_A", "Treatment_B"),
  group2 = c("Treatment_B", "Treatment_C", "Treatment_C"),
  p.adj  = c(0.9, 0.02, 0.03)
)

make_cld(comparisons, alpha = 0.05)
#> Compact Letter Display (CLD)
#> Signif. level (alpha):  0.05 
#> Method:  data.frame 
#> 
#>        group cld spaced_cld
#>  Treatment_B   a         a_
#>  Treatment_C   b         _b
#>  Treatment_A   a         a_

Custom Column Names

If your data frame uses different column names, specify them:

# Data frame with custom column names
my_comparisons <- data.frame(
  first_group = c("A", "A", "B"),
  second_group = c("B", "C", "C"),
  adjusted_p = c(0.9, 0.02, 0.03)
)

make_cld(my_comparisons,
  gr1_col = "first_group",
  gr2_col = "second_group",
  p_val_col = "adjusted_p",
  alpha = 0.05
)
#> Compact Letter Display (CLD)
#> Signif. level (alpha):  0.05 
#> Method:  data.frame 
#> 
#>  group cld spaced_cld
#>      B   a         a_
#>      C   b         _b
#>      A   a         a_

Formula Interface

The formula interface provides a convenient way to specify p-values and comparison labels. There are two formula formats available:

Single-variable formula (for pre-formatted comparison strings)

No hyphens are allowed in group names when using this format as they are used as separators.

# Using formula for data frames with comparison strings
my_data <- data.frame(
  Comparison = c("A-B", "A-C", "B-C"),
  p_value = c(0.12, 0.001, 0.045),
  p_adjusted = c(0.18, 0.003, 0.068)
)

# Use the adjusted p-values
make_cld(p_adjusted ~ Comparison, data = my_data)
#> Compact Letter Display (CLD)
#> Signif. level (alpha):  0.05 
#> Method:  formula 
#> 
#>  group cld spaced_cld
#>      A   a         a_
#>      B  ab         ab
#>      C   b         _b

# Or use the raw p-values
make_cld(p_value ~ Comparison, data = my_data)
#> Compact Letter Display (CLD)
#> Signif. level (alpha):  0.05 
#> Method:  formula 
#> 
#>  group cld spaced_cld
#>      A   a         a_
#>      B   a         a_
#>      C   b         _b

Two-variable formula (recommended for group names with hyphens)

When your group names contain hyphens or special characters, use the two-variable formula format which specifies groups separately:

# Data with group names containing hyphens
my_data2 <- data.frame(
  group1 = c("Treatment-A", "Treatment-A", "Treatment-B"),
  group2 = c("Treatment-B", "Treatment-C", "Treatment-C"),
  p_adjusted = c(0.18, 0.003, 0.068)
)

# Two-variable formula (handles hyphens automatically)
make_cld(p_adjusted ~ group1 + group2, data = my_data2)