References for the logistic-regression project
What the diagnostics draw on, and what has actually been checked
None of the three sources has been checked. All three claims are transcribed from Andy’s earlier notes rather than a fresh reading.
The working documents are Picking a Method and Diagnostics; the folder’s other documents are on the project index.
Status markers
- ❌ Not checked. No claim in the working documents has been verified against this source.
- ⚠️ Transcribed, unverified. A number or claim came from this source but has not been checked against it.
- ✅ Checked. Verified against the source.
Sources
❌ Harrell FE. Regression Modeling Strategies, 2nd ed. Springer, 2015. Free online at hbiostat.org/rmsc.
Section 10.2.2 is cited in Picking a Method for the claim that estimating a baseline rate to within ±10% needs at least 96 patients. The same book is cited more extensively in the sizing-studies project, which carries the fuller record.
❌ Harrell FE. \(R^2\) measures for models with binary and ordinal outcomes. Blog post. hbiostat.org/bib/r2.html
Cited in Diagnostics, Section 6, as the discussion of which \(R^2\) candidate to use, and as the source for the rms::R2Measures function. Not read against the specific claim.
❌ Harrell FE. Lecture notes, Binary Logistic Regression, Section 8-40. Not linked in the source material and not located here. Cited in Diagnostics, Section 6, as the source of “other metrics” not yet implemented. Finding a public copy of these notes is the first step in checking this one.