Split-half reliability of the network edge structure — net_split

Repeatedly splits the sample into two halves, estimates a network on each, and compares their edge-weight vectors. Reports, across splits, the edge-weight correlation between halves plus the mean/median/maximum absolute edge deviation - a psychometric reliability view of the estimated structure.

Usage

net_split_reliability(
  data,
  method = "glasso",
  iter = 100L,
  split = 0.5,
  cor_method = c("pearson", "spearman", "kendall"),
  labels = NULL,
  estimator_args = list(),
  ...
)

Arguments

data: Data frame or matrix (rows = observations), resampled exactly as given (see net_boot()), or a psychnet_group (split-half per level).
method: Estimator (see psychnet()). Default "glasso".
iter: Number of split-half iterations. Default 100.
split: Fraction of rows in the first half. Default 0.5.
cor_method: Correlation method for the between-halves edge comparison: "pearson" (default), "spearman", or "kendall".
labels: Optional node labels.
estimator_args: Named list of estimator arguments. Use this for names consumed by the diagnostic itself, such as estimator threshold.
...: Passed to the estimator.

Value

A tidy data.frame (class psychnet_reliability), one row per metric with columns metric, mean, sd, lower, upper. The per-split draws are carried in attr(x, "iterations") for plot.psychnet_reliability().

Examples

# `method` and `iter` are chosen so the example runs quickly; the defaults
# (method = "glasso", iter = 100) are what a real assessment should use.
net_split_reliability(SRL_Claude, method = "pcor", iter = 10)
#> # split-half reliability: pcor | 10 iterations (50/50 split)
#>           metric       mean         sd      lower      upper
#> 1   mean_abs_dev 0.07808955 0.02457258 0.03846060 0.10963050
#> 2 median_abs_dev 0.06588545 0.02220565 0.02991325 0.09716984
#> 3    correlation 0.97498013 0.01592382 0.94570362 0.99516211
#> 4    max_abs_dev 0.19047875 0.07735301 0.07527392 0.32270329