Skip to contents

Robust effect size, median error and Winsorized regression

Source: R/robust_wilcox.R
morie_akp_effect_size.Rd

morie_akp_effect_size is the trimmed-mean analogue of Cohen's d of Algina, Keselman and Penfield (2005). morie_median_se is the McKean-Shrader (1984) standard error of the median, which Wilcox warns is unreliable with ties. morie_winsorized_regression solves the normal equations built from Winsorized covariances.

Usage

morie_akp_effect_size(x, y, tr = 0.2, equal_variance = TRUE)

morie_median_se(x)

morie_winsorized_regression(X, y, tr = 0.2, n_iter = 20, tol = 1e-04)

Arguments

x, y

numeric vectors

tr

amount of trimming or Winsorizing

equal_variance

pool the Winsorized variances

X

Argument X; see Usage.

n_iter

Argument n_iter; see Usage.

tol

Argument tol; see Usage.

Value

a list; see each method's description

Examples

V <- c(1, 2, 3, 4, 5, 6, 7, 8)
morie_akp_effect_size(V, V)
#> $effect_size
#> [1] 0
#> 
#> $cterm
#> [1] 0.6419398
#> 
#> $pooled_sd
#> [1] 2.070197
#> 
#> $n1
#> [1] 8
#> 
#> $n2
#> [1] 8
#>