BCa Bootstrap Validation for O-information

Evaluates the robustness of a candidate hyperedge using bias-corrected and accelerated (BCa) bootstrap confidence intervals.

Usage

bootstrap_Oinfo(
  data,
  vars,
  n_boot = 1000,
  cor_method = "pearson",
  alpha_boot = 0.05
)

Arguments

data

Preprocessed numeric data frame.

vars

Integer vector identifying variables in the candidate multiplet.

n_boot

Number of bootstrap samples. Must be an integer of at least 2.

cor_method

Correlation method passed to compute_correlation().

alpha_boot

Significance level for the BCa interval. The confidence level is 1 - alpha_boot.

Value

A list containing:

ci_lower

Lower BCa confidence limit.

ci_upper

Upper BCa confidence limit.

robust

Logical indicator whether the interval excludes zero.

Details

This procedure corresponds to the second stage of the hypergraph construction framework proposed by Marinazzo et al. (2024) and adopted by Possenti et al. (2026).

Bootstrap samples are generated by resampling observations with replacement. O-information is re-estimated for each bootstrap sample and a BCa confidence interval is computed.

A hyperedge is considered robust if its confidence interval does not include zero.

The BCa interval is preferred over standard percentile intervals because it adjusts for both bias and skewness in the bootstrap distribution. Failed or invalid intervals are returned as missing limits with robust = FALSE.

References

Marinazzo, D., Van Roozendaal, J., Rosas, F. E., Stella, M., Comolatti, R., Colenbier, N., Stramaglia, S., & Rosseel, Y. (2024). An information-theoretic approach to build hypergraphs in psychometrics. Behavior Research Methods, 56(7), 8057-8079.

Possenti, F., Girelli, L., Tieri, P., & Petti, M. (2026). Multiplex Hypergraph Modeling of Higher Order Structures in Psychometric Networks (Version 1). arXiv. https://doi.org/10.48550/ARXIV.2604.22744