Cutoff determination for increasing number of randomizations

The tie_cutoff function calculates the mean Euclidean distance between commodity worth values. This becomes relevant when ties are present in the data. Depending on how ties are resolved (see prefLimit argument in the function), the items' position will change a lot. Since their relative positions are a function of the number of ties, more randomizations will stabilize their means and thus commodity positions. Increasing the number of randomizations usually leads not only to a stabilized mean but also to smaller confidence intervals. By defining a relative cutoff (e.g., 5 or 10%) for the range of the CIs regarding the maximum range in the data, a cutoff for the number of randomizations can be found.

tie_cutoff(
  data = tiefightR::mouse,
  R = 50,
  ciLvl = 0.95,
  cutoff = 0.1,
  cpus = 2,
  RF = NULL,
  CF = NULL,
  id = NULL,
  RV = NULL,
  ord = NULL,
  prefLimit = 50,
  compstudy = NULL,
  default = NULL,
  showplot = FALSE,
  showCutoff = FALSE
)

Arguments

R	number of maximum randomization steps
cutoff	Percent cutoff level (default: 0.10) - means CI range < than cutoff value
cpus	No. of used local CPUs for parallel computing (you should have more than 2)
RF	name of the reference fluid variable
CF	name of the combination fluid variable
id	subject IDs
RV	name of the response variable
ord	item category order
prefLimit	preference limit for binarization threshold
compstudy	label of the compiled sub study (used for filtering)
default	default item in worth value estimation (usually the lowest worth value)
showplot	show the plot for randomization cutoff determination
showCutoff	show vertical line of the cutoff
dat	imported raw data (should be binary, if not, will be binarized automatically)
standardize	standardize on the maximum CI value?

Value

Exports cutoff value and plots