Skip to content
No Vape

Inverse probability weights in vaping research: which population is being compared?

Map exposure, censoring and survey weights to their stated purposes before calling a weighted observational result representative or randomized.

No Vape · Updated

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

Read the purpose beside each weight

A paper may contain several kinds of weights in one analysis. Write their names on separate lines before summarizing the result. Treatment or exposure weights concern the observed exposure process. Censoring weights concern remaining observed during follow-up. Survey weights concern a stated population and sampling process. Find the actual explanation rather than treating every weighted estimate as nationally representative.

An original US adult smoking-and-vaping cohort analysis explicitly separates its inverse probability treatment and censoring weights from PATH survey weights. Its primary and secondary analyses target different population interpretations. That makes the distinction inspectable in real research; we import no outcome estimate or claim that vaping helps a reader quit.

Addiction (original research): Prospective association between e-cigarette use frequency patterns and cigarette smoking abstinence among adult cigarette smokers in the United States

Identify the constructed comparison target

Austin and Stuart’s methodological paper explains weighting by the probability of the exposure actually received and distinguishes weights targeting an average population effect from weights targeting the treated population. Its artificial population is a statistical construction. It is not a new set of enrolled people, and the word treatment in the method does not make an observed vaping pattern a prescribed therapy.

Your reading note can ask: is the reported estimate meant for the study population, the exposed group or a separately specified population? Keep the authors’ answer attached to the result. Do not infer the target from the largest enrollment count or assume it stayed unchanged when another set of weights was added.

Austin and Stuart / Statistics in Medicine (2015): Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies

Try a purpose-and-target map

Imagine a fictional report with one analysis using exposure-related weights and a second adding survey weights. A post labels both results national quitting probabilities. Your map separates the analytic comparison from the population extension and asks which statement each analysis supports. It does not calculate new weights or turn either result into an individual forecast.

Use four fields: observed participants, purpose of each weight, stated effect target and documented limitations. This differs from asking who was retained in matched pairs: weighting changes contributions without necessarily constructing those pairs. It also goes beyond distinguishing raw from weighted respondent counts; the question here is why this comparison was constructed.

Keep the assumptions and diagnostics visible

Austin and Stuart discuss measured-covariate balance, exposure probabilities and assumptions including no unmeasured confounding. A balanced weighted sample does not establish that every relevant difference was recorded. Unusual weights can require methodological attention; this guide supplies no universal trimming rule or threshold for accepting a study.

If the paper does not identify the target or explain a weight, retain that unresolved question. Do not replace the uncertainty with either automatically causal or automatically invalid. Seek suitable expertise for consequential interpretation of the full analysis. Personal tracking can describe your own routine without pretending to create a pseudopopulation or recover an unobserved alternative outcome.

Austin and Stuart / Statistics in Medicine (2015): Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies

Sources and scope

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

  1. Addiction (original research) — Prospective association between e-cigarette use frequency patterns and cigarette smoking abstinence among adult cigarette smokers in the United States
  2. Austin and Stuart / Statistics in Medicine (2015) — Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies
  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