No. “Excellent” and “right?” encourage agreement.

Answer

No. “Excellent” and “right?” encourage agreement.

Voluntary-response bias: strong opinions are more likely to appear.

Answer

Voluntary-response bias: strong opinions are more likely to appear.

Use balanced alternatives without praise or criticism.

Example: “Do you support, oppose or have no opinion about a new sports session?”

Answer

Example: “Do you support, oppose or have no opinion about a new sports session?”

Divide replies by questionnaires sent.

Answer

Randomly vary which drink is tasted first and include “no preference”. The labels should not reveal the recipe.

Use anonymous A/B labels, identical servings and balanced tasting order; ask the same neutral preference question.

Answer

Use anonymous A/B labels, identical servings and balanced tasting order; ask the same neutral preference question.

Record non-response and qualify any town-wide inference.

Current users may have different views from non-users. Randomly sample households from a complete town address list across neighbourhoods, then use a consistent adult-selection rule.

Answer

Current users may have different views from non-users. Randomly sample households from a complete town address list across neighbourhoods, then use a consistent adult-selection rule.

Ask separate neutral questions about cost, access and overall opinion, each with balanced options.

It assumes affordability/convenience and combines different qualities.

Answer

It assumes affordability/convenience and combines different qualities.

Report both denominators and explain possible non-response bias.

18 of the 20 respondents support it; only 10% of invitees replied. This does not establish 90% support among all invitees.

Answer

18 of the 20 respondents support it; only 10% of invitees replied. This does not establish 90% support among all invitees.

Offer the same private response method and follow-up to everyone selected; include absent pupils later.

Use support / oppose / no opinion; randomly sample from the full school register within year groups in proportion to their sizes.

Answer

Use support / oppose / no opinion; randomly sample from the full school register within year groups in proportion to their sizes.

Neither result automatically represents the whole population. Larger size cannot repair A’s systematic selection and wording problems.

A has self-selection and wording bias. B has a better starting design but a 40% response rate and possible non-response bias. Follow up all selected non-respondents consistently, examine response across known groups, and report uncertainty.

Answer

A has self-selection and wording bias. B has a better starting design but a 40% response rate and possible non-response bias. Follow up all selected non-respondents consistently, examine response across known groups, and report uncertainty.