A skipped vaping-survey question is different from a no answer

Look for the question’s entry condition

A survey may ask follow-up questions only after a particular earlier response. Find the if instruction and the question it refers to. The eligible base for the follow-up can therefore be smaller than the overall sample even when everyone completed the survey. That routing issue is different from participants disappearing at a later follow-up wave.

The questionnaire appendix to Health Canada’s 2023 vaping-panel report contains conditional item routing and separate do-not-know options. We use it as a primary example that those states can be distinguished. This guide does not transfer its code values to other datasets or reproduce the report’s behavior results. Every dataset needs its own instructions and definitions.

Health Canada / Environics Research: Vaping Panel Surveys and Interviews 2022-2023 — Final Report

Draw a fictional route

Imagine an invented survey first asking whether a respondent used a specified product during a defined period. Only those answering yes receive a question about a purchase for that product. A person answering no to the first item may never see the purchase item. A blank there means unasked under the route, not necessarily a recorded zero-dollar purchase answer.

Another person can be asked the purchase question and answer do not know. A third can be asked and leave it unanswered. Make separate boxes for the three situations. The example is hypothetical and does not establish how any named survey computes a total. The useful output is a route map showing which information was collected and which was not requested.

Check how the report handles each state

Locate the questionnaire, codebook, and table footnote if available. A report may derive a variable from several answers or deliberately exclude some states from a percentage denominator. Read the actual rule. Do not treat a special numeric code as a large spending amount or assume that a familiar code in one file carries the same meaning in another.

Some questionnaires combine do not know and prefer not to say. If the source combines them, retain the combined label instead of separating motives you cannot observe. Likewise, a derived value is not necessarily a direct answer to the displayed question. Keep any documented derivation with the result and leave undocumented handling unresolved. This is a reading exercise, not instructions to edit a public dataset.

Phrase the denominator precisely

Instead of saying everyone in the survey answered this question, state that the quoted result concerns the defined people routed to it, with the stated handling of unknown answers. If you cannot verify that group, do not calculate a replacement percentage from the total sample. A missing routing explanation deserves a focused request to the responsible methods source.

This distinction prevents a blank cell from becoming an invented quitting outcome, purchase, or opinion. It cannot recover facts about people who were not asked and does not establish representative sampling or clinical applicability. Keep survey-reading notes separate from your personal record: in your own log, lack of an entry likewise needs its actual explanation rather than a confident behavior claim.