---
title: "Measurement-quality stress tests and negative controls"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Measurement-quality stress tests and negative controls}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

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

Milestone #2 turns measurement-quality perturbations into predeclared software-validation scenarios.

```{r}
sp <- eyeprocess_stress_evidence_plan()
head(expand_eyeprocess_stress_evidence_plan(sp))
eyeprocess_reliability_evidence_plan()
eyeprocess_negative_control_evidence_plan()
```

Stress dimensions include gaze missingness, pupil dropout, calibration offsets, sampling jitter, AOI-label noise, device shifts, and trial imbalance. Reliability evidence includes split-half, ICC, temporal stability, and Bland-Altman-style agreement. Negative controls include permutations, temporal shifts, placebo windows, and intentionally known leakage.

Thresholds and perturbations are study-specific reporting conventions. Reliability is not validity, and negative-control/leakage findings are not misconduct labels.