---
title: "Hierarchical Effect and Variance Atlases"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Hierarchical Effect and Variance Atlases}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

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

The expanded group-level layer can inspect raw posterior deviations, rank
uncertainty, and baseline latent variance partitioning.

```{r, eval=FALSE}
draws <- group_effect_draws_table(
  fit,
  groups = "participant_id",
  coefficients = "Intercept",
  ndraws = 1000
)

ranks <- group_effect_rank_probability_table(
  fit,
  group = "participant_id",
  coefficient = "Intercept"
)

partition <- random_intercept_variance_partition(fit)

plot_group_effect_distribution(draws)
plot_group_effect_rank_probability(ranks)
plot_random_intercept_variance_partition(partition)
```

Rank probabilities and variance fractions remain descriptive posterior
quantities; they do not establish substantive group importance or causal
variance attribution.