Skip to content
No Vape

A vaping study removed outliers: read the reason before accepting the result

Separate verified errors from unusual valid observations and look for the reported rule, scope and consequence of exclusions.

No Vape · Updated

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

Find what was unusual and at which level

The word outlier could describe a laboratory measurement, a participant value or an entire study estimate. Identify the unit first. Removing one assay measurement is different from excluding every observation from one person. A headline that says a few outliers were removed may leave that distinction unstated.

NIST explains that unusual observations can be errors, random variation or scientifically interesting information. Its guidance distinguishes investigation from automatic deletion. That limited concept supports this reading exercise; we do not supply a numerical cutoff, recommend an analysis method or decide whether a real vaping dataset should exclude anything.

US National Institute of Standards and Technology: NIST Engineering Statistics Handbook: Detection of Outliers

Separate a verified error from a statistical flag

A documented instrument failure or coding mistake gives a different basis from a value lying far from the others. Look for what the authors actually established. A statistical flag can initiate investigation without proving why the value occurred. Do not fill an absent explanation with an accusation or with automatic reassurance.

In an invented example, one recorded duration contains a known unit-entry error, while another duration is unusual but supported by the source record. Both stand out numerically, yet their information status differs. The example offers no rule for correcting either and no inference about someone’s health or nicotine exposure.

Read when and how the rule was applied

Locate the stated exclusion criterion, whether it was planned and whether it was applied consistently to the relevant data. Ask how many observations or participants remained for the particular reported result. Do not assume that a general methods sentence explains every different analysis in the paper.

The article or supplementary material may show an alternative analysis retaining values or using another approach. If so, keep that comparison attached to its specific purpose. Repeating an analysis with a different rule does not create an independent replication or remove every other source of uncertainty.

Keep the consequence visible without inventing a corrected estimate

Ask whether the reported conclusion depends materially on the exclusions and whether the authors discuss that dependence. If the information is unavailable, say you could not verify the consequence. Do not calculate a replacement effect from a partial table or average results that used different populations.

Exclusions may alter more than a headline number: the population described can also change. Conversely, the presence of an exclusion does not automatically invalidate the report. A proportionate reading retains the rationale, remaining scope and unresolved questions together.

Do not use the research label to tidy your own record

A personal day with more recorded use is not an error merely because it looks different. Correct a known recording mistake with its basis, and retain uncertainty where you do not know the right value. A research analyst’s stated exclusion process is not permission to erase inconvenient personal observations.

This guide is an editorial methods-reading aid, not statistical or clinical review. Consequential interpretation belongs with a qualified professional. The useful result is a precise account of what was removed and why, without turning clean-looking data into a certificate of trustworthy quitting evidence.

Sources and scope

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

  1. US National Institute of Standards and Technology — NIST Engineering Statistics Handbook: Detection of Outliers
  2. 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