Exploratory and confirmatory sampling optimize different goals
Aliases: exploratory sampling · confirmatory sample · maximum variation
What it is
Exploratory sampling often maximizes information diversity, using fewer contrasting cases to expose unknown mechanisms. Confirmatory sampling makes the target population, conditions, and measurement process explicit so a bounded claim can confront new evidence. “Small and varied” versus “controlled and repeatable” names different objectives; exploration is not arbitrary, and confirmation does not require demographic homogeneity.
Why it happens
Exploration gains value from encountering a context not yet seen, so maximum variation, critical cases, and theoretical sampling increase contrast. Confirmation must distinguish the focal effect from composition, implementation, and measurement drift, requiring stable inclusion, stratification, procedure, and analysis. If sample change is confounded with condition, an apparent result cannot be assigned to mechanism or population.
Studying it
Exploratory plans state variation axes, case rationale, and an information-adequacy rule while preserving case differences. Confirmatory plans define population, frame or recruitment boundary, inclusion, quotas or randomization, measurement version, and repeatable scripts. Replication need not recruit identical people; it admits a new eligible sample under the same rule. Population generalization additionally requires a selection account.
Where it stops holding
Small exploratory samples do not estimate prevalence, and highly varied cases can confound several differences. Overrestriction in confirmation improves internal comparability while weakening external validity. Rare populations and field settings may not reproduce identical conditions; transparent context, mechanism, and deviations can support theoretical rather than laboratory-style replication.
Related
- Same group: Q1.09.1 Method–phase alignment is common, not definitional · Q1.09.3 Reporting exploration as validation hides untested hypotheses · Q1.09.4 Mixed purposes can interfere within one guide
- Adjacent: Q1.04 Sampling and representativeness · Q1.08 Sample size
- Search terms:
maximum variation sampling·sampling frame·theoretical replication