Consumer samples that still balance at the quota cell

A consumer sample can hit its national age and gender targets and still be wrong, because the cells inside those targets filled unevenly. We interlock quotas by default and report how each one filled, so you can see where the sample is thin before you write the conclusions.

Nested quotas are the common failure. You ask for national representation on age, gender and region, and you get it — on each dimension separately. Then you cut the data by women aged 18 to 24 in the north and find nineteen people, half of whom came in on the last afternoon of field from one recruitment source.

Terms we use in the field report

Interlocked quota

A bucket defined by more than one variable at once — age and gender and region — rather than each variable balanced independently. It costs more to fill and it is the only version that survives a crosstab.

Screen-out rate

The share of entrants removed by your screener before the questionnaire proper. Reported per bucket, because an average hides the one that is starving.

Source concentration

How much of a bucket came from a single recruitment channel. High concentration in a thin one is the usual explanation for a result nobody can replicate.

Fill curve

When each cell reached quota across the field period. Those that fill only in the final hours are the ones to treat carefully.

Category behaviour as a profiling layer

Household attributes get you a representative shape. They do not get you buyers of a category. Where a study needs category incidence — people who bought pet food in the last month, switched broadband in the last year, hold a specific policy type — we profile that separately on entry and screen against the stored answer rather than asking it cold in the screener, which is where over-claiming happens.

The stored answer is still re-asked in field. If the two disagree, the respondent is removed and the contradiction is recorded against their quality score. That removal appears in your rejection log with the reason attached.