---
title: "Empirical validation programmes in eyeprocess"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Empirical validation programmes in eyeprocess}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(collapse = TRUE, comment = "#>")
```

eyeprocess 0.9 treats validation as an explicit research object rather than an implicit property of an estimator. `process_validation_design()` declares the measurement regimes to study; `run_process_validation()` records recovery, interval coverage, convergence, warnings, and failures; and `freeze_validation_reference()` supports regression-style evidence freezing.

```{r, eval=FALSE}
d <- process_validation_design(n_persons=c(50,150), n_trials=c(10,30), missingness=c(0,.15), sampling_rate_hz=c(60,120), replications=100)
x <- run_process_validation(d)
validation_recovery_table(x)
validation_coverage_table(x)
validation_failure_profile(x)
plot(x, type="recovery")
```

The built-in simulator is a software-validation fixture with known truth. It is not a substantive psychological theory. Recovery under a supplied data-generating process does not establish external validity.