Add predictor selection chapter (RMB Ch10) with HERS and WCGS DAG examples

[Copilot]

• Fix CI failure: corrected include path from _sec_hers_intro.qmd to _subfiles/shared/_sec_hers_intro.qmd in _sec-pred-sel-dag.qmd

[d-morrison]

@copilot fix https://github.com/d-morrison/rme/actions/runs/24222907017/job/70718039145?pr=400

[Copilot]

@copilot fix https://github.com/d-morrison/rme/actions/runs/24222907017/job/70718039145?pr=400

Fixed in commit 1cf4786. The CI failure was caused by renv::snapshot() accidentally removing broom.helpers (required by gtsummary >= 2.5.0) along with several other packages from renv.lock. The fix restores the original renv.lock state (including broom.helpers 1.22.0) while adding the new dagitty 0.3-4 entry.

[Copilot]

Pull request overview

Adds a new “Predictor Selection” chapter (RMB Ch10) to the Quarto book/website, including an HERS DAG example and accompanying sections.

Changes:

• Adds dagitty to project dependencies (DESCRIPTION Suggests + renv.lock).
• Introduces new chapter/section QMD files for predictor selection overview and HERS DAG-based discussion.
• Wires the new chapter into the Quarto book + website nav, and updates the spelling WORDLIST.

Reviewed changes

Copilot reviewed 7 out of 8 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
DESCRIPTION	Adds dagitty to Suggests so renders can use it.
renv.lock	Locks dagitty version for reproducible installs.
predictor-selection.qmd	New top-level chapter file that includes the new sections and examples.
_sec-pred-sel-overview.qmd	New introductory overview of predictor selection strategies.
_sec-pred-sel-dag.qmd	New DAG theory + HERS DAG section and LDL modeling workflow.
_quarto-book.yml	Adds the new chapter to the book build.
_quarto-website.yml	Adds the new chapter to website render list + navbar.
inst/WORDLIST	Adds new technical terms for spellchecking.

[d-morrison]

@copilot apply changes based on the comments in this thread

[Copilot]

@copilot apply changes based on the comments in this thread

Applied all review changes in 2f09646:

• Grammar: "describe" → "describes" in _sec-pred-sel-overview.qmd
• DAG: removed age -> HT edge (HERS is an RCT); updated the age bullet to say it's prognostic but not a confounder; updated the adjustmentSets() notes to reflect the empty adjustment set
• WORDLIST: removed "Github" (wrong capitalization, "GitHub" already present) and duplicate "Hua"

[Copilot]

@copilot resolve the merge conflicts in this pull request

Resolved in 0c60d75. Two conflicts were present:

• .github/copilot-instructions.md: kept both the "Figures and Tables: Use Div Format" section (from this branch) and the new "Landscape Tables in PDF" section (from main)
• inst/WORDLIST: merged the union of both word lists, also removing nested conflict markers that had been left in the file from a previous incomplete merge

[Copilot]

add more details on the HERS study, drawing from https://demorrison.ucdavis.edu/files/Hulley1998HERS.pdf

is there another example dataset we can use as with an observational study design, so that con...

Added in 9cd2203:

• HERS study details (in _sec-pred-sel-dag.qmd): expanded with enrollment criteria, follow-up duration, LDL/HDL changes (−11%/+10%), null CHD result, time trend, and adverse events (thromboembolic, gallbladder), all with attribution to @HulleyStephen1998RToE.

• WCGS as observational study example: added a new ## An observational study example: WCGS section with a DAG showing age → TypeA and smoking → TypeA arrows (making them true confounders), adjustmentSets() output, and explanation that confounding is plausible since Type A behavior was not randomly assigned — contrasting with HERS.

@copilot elaborate on 10% rule (change of estimate rule), also adapting content from https://demorrison.ucdavis.edu/files/KleinbaumLogistic.pdf (with attribution, as always - make a note about always ...

Added in 9cd2203:

• Elaborated 10% rule in _sec-pred-sel-primary.qmd: formal definition with the percent-change formula, a note distinguishing bias vs. significance, and attribution to @kleinbaum2010logistic [Chapter 8].
• Numerical WCGS example (#sec-pred-sel-10pct-wcgs): fits a full logistic model for CHD in WCGS, drops each confounder one at a time, and shows the percentage change in the Type A coefficient in a table.
• Attribution guidelines added to copilot-instructions.md under ## Attribution for Adapted Content.

Also fixed the callout heading levels (# → ##) across predictor-selection.qmd, _sec-pred-sel-primary.qmd, and _sec-pred-sel-details.qmd as flagged in the automated review.

[d-morrison]

@copilot fix https://github.com/d-morrison/rme/actions/runs/24529862797/job/71709947757?pr=400

[Copilot]

@copilot fix https://github.com/d-morrison/rme/actions/runs/24529862797/job/71709947757?pr=400

Fixed in 3ff2d2e. The include path for _sec_hers_intro.qmd in _sec-pred-sel-dag.qmd was pointing to the root (_sec_hers_intro.qmd) instead of the correct subfile location (_subfiles/shared/_sec_hers_intro.qmd).

[github-actions]

PR Preview Action v1.8.1
Preview removed because the pull request was closed.

_{2026-04-16 22:03 UTC}

[d-morrison]

@copilot

correct the statement of the main goal of the HERS study: it was estimate the causal effect of HT on CHD event risk.

define a confounder as a common cause of an exposure and outcome.