Skip to content
No Vape

A vaping model’s pseudo R-squared is not a quit percentage

Keep the named fit statistic beside its model instead of converting it into explained health, prediction accuracy or your probability of quitting.

No Vape · Updated

Open book on a cafe table
Illustrative image · Lucille Emi Oh / Unsplash

Keep the adjective and the method name

A table may place a statistic under a regression result rather than beside an individual predictor. Before translating it, identify whether the outcome is categorical and whether the label names Nagelkerke, McFadden or another measure. Dropping the word pseudo can change what a summary appears to claim.

The original Korean Journal of Health Promotion report Factors Associated with Current E-Cigarettes Use of Female Workers uses a complex-sample logistic model and names Nagelkerke pseudo R-squared in its table note. Its data concern the 2023 Community Health Survey. This example establishes that the named statistic appears in actual vaping research; no use estimate, health association or quitting result is being transferred to readers.

Korean Journal of Health Promotion: Factors Associated with Current E-Cigarettes Use of Female Workers: Using the 2023 Community Health Survey

Do not borrow the ordinary variance interpretation

UCLA’s statistical resource distinguishes several pseudo R-squared definitions and explains why their values can differ. Logistic models are not fitted using the same squared-error objective as ordinary least-squares regression. Consequently, a familiar-looking number does not automatically carry the ordinary variance-explained interpretation.

The resource describes Nagelkerke as a rescaling of Cox–Snell’s measure and McFadden’s as a different likelihood-based construction. You do not need to reproduce either formula to retain their names. The important reading decision is to avoid treating two differently labeled fit values as measurements on one universal quality ruler.

UCLA Office of Advanced Research Computing: FAQ: What are pseudo R-squareds?

Correct a fictional caption without calculating a replacement

Imagine a fictional logistic table lists a named pseudo R-squared, while a caption calls it “the percentage of adults who will quit.” Your correction removes that outcome claim and keeps the actual statistic beside the actual model. It does not produce a new quit percentage from the same number. No published value or observed quitting result is supplied in this illustration.

A likelihood-based value such as the named Nagelkerke statistic here is not the percentage of classifications that were correct, and it does not show performance on new participants. Those questions need their own reported assessments. Keep uncertainty and validation separate instead of letting a single fit statistic answer every question about the model.

Leave a specific question where information is missing

A useful note contains the model, the named pseudo R-squared, what the authors say it assesses and any unresolved interpretation. If the report only says R-squared for a categorical model, ask which definition was used. Do not supply the most familiar one or impose a universal good-versus-bad cutoff.

This introductory distinction cannot certify the analysis, resolve confounding or establish a clinical benefit. Consequential interpretation requires the full report and suitable expertise. A personal vaping record does not need a pseudo R-squared to be useful: clear observations can guide an ordinary next action while an unclear research statistic remains an open question.

Sources and scope

Sources inform the guidance. The worksheets and examples are No Vape editorial suggestions, not tested treatments.

  1. Korean Journal of Health Promotion — Factors Associated with Current E-Cigarettes Use of Female Workers: Using the 2023 Community Health Survey
  2. UCLA Office of Advanced Research Computing — FAQ: What are pseudo R-squareds?
  3. No Vape — No Vape — app features and store links

Written with AI assistance and checked against the cited sources. No clinical review is claimed. This guide covers practical planning, not individual medical treatment. How we research and write