Match the headline to the people counted
A programme may introduce a survey of thousands and then discuss only respondents in one age group or region. The overall sample size is not the number answering for that subgroup. Full Fact’s polling guide identifies this as a common interpretation problem. Find the base for the actual statement, rather than borrowing the largest number in the report.
Read weighted and unweighted bases separately
The British Polling Council distinguishes the responses collected from the weighted data used to better match the intended population. A weighted base is not a fresh set of additional people interviewed. Check the table headings and methodology so you know whether the displayed group size is an original respondent count or part of the weighting presentation.
Use a cautious comparison
In a fictional survey of 2,000 people, a subgroup might contain only 80 respondents. A percentage for that group should not be described as though all 2,000 answered on its behalf. The example illustrates the base problem, not a rule that every group of 80 is useless. Precision depends on the design and weighting as well as the raw count.
Keep uncertainty visible in the summary
If the pollster warns against interpreting small cells, retain that warning. Avoid making a dramatic claim about a narrow audience from a rounded percentage alone. A useful account identifies the group, relevant base and limitation, then distinguishes it from the headline result for the whole sample. That is more informative than treating every table column as an equally strong national finding.
Questions, answered
Does the full sample size apply to every subgroup?
No. Check the base for the group actually discussed.
Does weighting mean more people were interviewed?
No. It changes how collected responses contribute to the result.