library(splithalfr)

This vignette shows examples of six methods for splitting data. Each method can be used by passing the arguments shown in each example to by_split, but for illustration we’ll use the lower-level functions stratum_split and strata_split

Example data

We’ll use this example dataset of eight trials, each of which has a condition and rt variable.

data.frame(
  condition = rep(c("a", "b"), each = 4)
  rt = 100 * 1 : 8
)

First-second splitting

Odd-even splitting assigns trials of the first half of rows to one part and trials with the second half of rows to the other (Green et al., 2016; Webb, Shavelson, & Haertel, 1996; Williams & Kaufmann, 1996). For this splitting technique, set method to first_second.

stratum_split(ds, method = "first_second")

Odd-even splitting

Odd-even splitting assigns trials with an odd row number to one part and trials with an even row number to the other (Green et al., 2016; Webb, Shavelson, & Haertel, 1996; Williams & Kaufmann, 1996).For this splitting technique, set method to odd_even.

stratum_split(ds, method = "odd_even")

Permutated splitting

Permutation splitting is also known as bootstrapped splitting (Parsons, Kruijt, & Fox, 2019) and random sample of split halves (Williams & Kaufmann, 1996). It assigns trials to each part via random sampling without replacement. This splitting technique is the default, but you can make it explicit by setting method to random.

stratum_split(ds, method = "random")

Monte Carlo splitting

Monte Carlo splitting assigns trials to each part by sampling with replacement (Williams & Kaufmann, 1996). For constructing parts that of any length, use the split_p argument and set replace to TRUE. The example below constructs two parts of the same length as the original dataset by setting split_p to 1.

stratum_split(ds, method = "random", replace = TRUE, split_p = 1)

Stratified splitting

If a split is stratified by a variable, then trials are separately assigned to each part for each level of that variable (Green et al., 2016). For example, if splits are stratified by ds$condition, the trials with condition a and b are separately split. Below are illustrations of first-second, odd-even, permutated, and bootstrapped splitting, stratified by condition.

ds_stratified <- stratify(ds, ds$condition)
strata_split(ds_stratified, method = "odd_even")
strata_split(ds_stratified, method = "first_second")
strata_split(ds_stratified, method = "random")
strata_split(ds_stratified, method = "random", replace = TRUE, split_p = 1)

Sub-sampled splitting

In a sub-sampled split, part of the trials is randomly sampled without replacement and then split. Sub-sampling only works well with splitting methods that uses random sampling (permutated and Monte Carlo). Since the sub-sampling procedure already randomizes the trials selected for splitting, Splitting methods that assign trials to part based on their row number, such as first-second and odd-even, should give results similar to permutation splitting. Any stratifications are applied both to the sub-sampling and splitting.

stratum_split(ds, method = "random", subsample_p = 0.5)
stratum_split(ds, method = "random", subsample_p = 0.5, replace = TRUE, split_p = 1)
strata_split(ds_stratified, method = "random", subsample_p = 0.5)
strata_split(ds_stratified, method = "random", subsample_p = 0.5, replace = TRUE, split_p = 1)