Ask who had a chance to be included before asking how many people answered.
Begin with selection
Statistics Canada distinguishes probability sampling from non-probability selection. In non-probability research, selection does not follow a random process that assigns each population member a known selection probability. This affects what can be concluded from the resulting sample.
An illustrative example
Imagine a survey of 10,000 people who volunteer through one brand’s social feed. The response count is substantial, but it does not automatically make the sample representative of everyone who might buy the product. This is a hypothetical example, not a Northbound study or a claim about any particular panel.
Make the quality discussion broader
A useful research brief considers coverage, recruitment, questionnaire quality, weighting and the decision the evidence must support. Adding interviews can improve precision within a design, but cannot by itself repair every source of bias. Our preference is to explain the trade-offs in plain language and to avoid presenting the sample count as a stand-alone quality certificate.
Evidence note. This is an editorial research guide or perspective, first published on 25 October 2026. It does not report original Northbound polling. Read our publication policy.