dot-detect_boundary: Detect Parameter Space Boundary Issues

In riskdiff: Risk Difference Estimation with Multiple Link Functions and Inverse Probability of Treatment Weighting

Detect Parameter Space Boundary Issues

Description

Detects when maximum likelihood estimates lie on or near the boundary of the parameter space for log-binomial and identity link models. Based on methods described in Donoghoe & Marschner (2018).

Usage

.detect_boundary(model, data, tolerance = 1e-06, verbose = FALSE)

Arguments

model	A fitted GLM object
data	The data used to fit the model
tolerance	Numeric tolerance for boundary detection (default: 1e-6)
verbose	Logical indicating whether to print diagnostic information

Value

A list containing:

boundary_detected

Logical indicating if boundary was detected

boundary_type

Character describing the type of boundary issue

boundary_parameters

Character vector of parameters on boundary

fitted_probabilities_range

Numeric vector with min/max fitted probabilities

separation_detected

Logical indicating complete/quasi-separation

References

Donoghoe MW, Marschner IC (2018). "logbin: An R Package for Relative Risk Regression Using the Log-Binomial Model." Journal of Statistical Software, 86(9), 1-22. doi:10.18637/jss.v086.i09

riskdiff documentation built on June 30, 2025, 9:07 a.m.