Skip to contents

Unified Regression Calibration Wrapper (External Reliability Study)

RC_ExReliab.Rd

A single formula interface for regression calibration in external reliability studies. The user simply specifies `link = "linear"`, `"logistic"`, or `"log"`, and the wrapper selects the appropriate model: * `"linear"` → Gaussian (identity link) * `"logistic"` → Binomial (logit link) * `"log"` → Poisson (log link)

Usage

RC_ExReliab(
  formula,
  main_data,
  rep_data,
  link = c("linear", "logistic", "log"),
  return_details = FALSE
)

Arguments

formula

A formula or character string such as "Y ~ sbp(sbp2, sbp3) + chol(chol2, chol3) + age + weight". Terms of the form `var(rep1, rep2, ...)` are treated as error-prone exposures with replicates in `rep_data`; other terms are treated as covariates W.

main_data

Data frame holding the outcome, error-prone exposures, and covariates.

rep_data

Data frame holding replicate columns referenced in `formula`.

Character; one of `"linear"`, `"logistic"`, or `"log"`.

return_details

Logical; if TRUE, return parsed, prepared, and RC internals.

Value

A list with * `uncorrected`: naive regression estimates * `corrected` : sandwich-corrected regression calibration estimates plus optional `details` if `return_details = TRUE`.