ESOMAR 37, question by question
Thirty-seven questions, and about eight of them decide whether the answers to the other twenty-nine mean anything.
Outline. Each heading below carries one line saying what the section has to establish. Expand each into two to four paragraphs, put at least one number or worked example in the piece that nobody else could publish, then delete this note.
The question set is good. The answers are usually written to pass rather than to be used, and the difference is visible question by question once you know where to look.
The eight that carry the weight
List them with the reason each one is load-bearing. Everything else is context.
Sources and recruitment: the answer that describes every supplier alive
Contrast a generic "diverse mix" with a percentage split by channel.
Deduplication: the honest answer admits a limit
Cross-panel dedup mostly does not work. A supplier claiming otherwise is describing something that does not exist.
Rejection and reversal: the pair, not the single number
A catch rate proves activity. A reversal rate proves accuracy. Almost nobody reports the second.
The questions the standard does not ask
Screen-out payment, participation caps, own-panel share per project. Say why each predicts data quality.
Publishing answers as a web page rather than a PDF
Why a gated PDF answers the question in form only, and what changes when it is a linkable page.
-
What a Partner Audit Changed
A supply partner walked the participant entry end to end before opening volume. Three of the things they asked for were things we had been meaning to do.
4 min read -
Reading a sample supplier’s quality report
Most quality reports are written to survive being read, not to be used. A handful of questions separate the ones with measurement behind them.
4 min read -
Three hops and the loss of provenance
Sample rarely travels in a straight line. Each time it changes hands something specific is lost, and by the third hand nobody can tell you where a respondent came from.
4 min read
