Skip to content
No Vape

Normal residuals in a vaping study do not mean normal health

Locate the observed-minus-fitted object and the actual diagnostic before treating statistical normality as a statement about people, raw counts or clinical results.

No Vape · Updated

Disposable vape resting on concrete
Illustrative image · Markus Winkler / Unsplash

Ask what was plotted or tested

The word normal has an everyday health meaning and a statistical distribution meaning. A paper can test a raw variable’s distribution, examine residuals from a regression or compare a clinical measurement with a reference range. These are different tasks. Read the exact object named instead of shortening all three to “the results were normal.”

The original Medical Sciences report titled Reduced Expiratory Flow in Young Adult Vapers and Smokers describes checking model residual normality using Q–Q plots and a Shapiro–Wilk test, alongside other diagnostics. This guide uses that methods wording only. It supplies no interpretation of lung-function values, symptom explanation or claim that a diagnostic establishes safety.

Medical Sciences / MDPI: Reduced Expiratory Flow in Young Adult Vapers and Smokers: A Cross-Sectional Study

Keep a fitted value separate from an observed one

In a fictional arithmetic example, an observation has value ten and a model fits eight, leaving a residual of two under observed-minus-fitted notation. Another row can have a different fitted value and the same residual. The residual records a discrepancy relative to that fit; it is not another participant or a directly measured new outcome.

NIST’s modeling handbook describes histograms and normal probability plots of residuals as checks on assumptions about random errors. It also warns that other model problems affect interpretation. A graph that appears close to its expected shape therefore needs the actual model and its other conditions, rather than being treated as a stand-alone quality certificate.

NIST Engineering Statistics Handbook: How can I test whether or not the random errors are distributed normally?

Read the diagnostic without choosing a repair

Find which model generated the residuals, which plot or test was used and what the authors say about any departure. A residual plot and a raw-data histogram can look different without contradicting each other. The difference follows from looking at different quantities. Do not replace one with the other while summarizing the methods.

If the paper reports additional checks for changing spread, influential observations or the form of a relationship, keep those separate too. This reading exercise does not tell you to delete observations, transform outcomes or select a preferred test. Such decisions require the analysis purpose and methodological judgment, not an isolated screenshot or a convenient p-value.

Use statistical language without adding a health verdict

Your final note can state that the investigators assessed a specified distribution assumption for residuals from a named model. If the source only says assumptions were checked, retain that limited wording and ask what was actually assessed. A nonsignificant test is not proof that every assumption is true, and a departure is not a diagnosis of the people studied.

Good numerical behavior does not establish representative recruitment, correct causal interpretation or personal treatment benefit. Seek qualified interpretation for consequential health questions with the complete report. For your own routine, preserve actual observations in ordinary words rather than trying to make a small personal log look statistically normal. The useful result here is identifying what the technical check actually checked.

Sources and scope

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

  1. Medical Sciences / MDPI — Reduced Expiratory Flow in Young Adult Vapers and Smokers: A Cross-Sectional Study
  2. NIST Engineering Statistics Handbook — How can I test whether or not the random errors are distributed normally?
  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