Maximum Variation Case Round¶
Sampling round — instantiates Theory-Responsive Case Sampling Design
Samples deliberately across the widest range of cases to see which findings survive maximum difference.
A Maximum Variation Case Round selects a batch of cases spread as far apart as possible across the dimensions that might matter, in a single deliberate round, to learn which findings survive maximum difference. The logic is simple: a pattern that persists across cases chosen precisely because they are unlike each other is a robust one, and a pattern that appears in some corners of the spread but not others is dimension-specific rather than general. It samples across the breadth of the case universe — not toward a single edge, and not toward a disconfirmation — and its output is a partition of findings into the invariant and the variable.
Example¶
A public-health team is trying to understand why a school hand-washing program reduces absenteeism in some districts and barely moves it in others. Rather than study more of the districts they can reach easily, they design a maximum-variation round: in one batch they include a large urban district and a tiny rural one, one with reliable piped water and one that trucks water in, one with low staff turnover and one that churns teachers every year, and a mix of budgets. The point is not to represent the state's districts in proportion — it is to stretch the sample across every dimension a skeptic might invoke.
When the analysis comes back, one finding survives the whole spread: wherever a single "hygiene-champion" teacher takes ownership, uptake climbs, regardless of size, water source, or budget. That pattern's persistence across maximum difference is what makes it credible. A second finding — that automated soap dispensers help — appears only in the piped-water districts, marking it as infrastructure-dependent rather than general. The round has separated the robust core from the context-bound trimming, without ever claiming the batch mirrors the population.
How it works¶
- Enumerate the dimensions that might matter. List the axes a critic could say the finding depends on — size, resources, culture, maturity.
- Pick cases that maximize spread. Choose the batch to stretch each dimension to its extremes rather than cluster near the middle.
- Hold the analysis framework constant. Apply the same coding and comparison to every case so difference in findings reflects the cases, not the method.
- Partition the results. Separate findings that survive the whole spread (candidate invariants) from those tied to particular dimensions (context-bound), flagging the latter for follow-up.
Tuning parameters¶
- Dimensions spanned — how many axes the round stretches at once. More axes test robustness harder but blur which dimension drives any variation observed.
- Cases per round — a lean spread versus a wide batch. More cases sharpen the partition but multiply cost and burden.
- Spread-versus-comparability — how extreme the variation is allowed to get. Push too far and the cases stop being comparable, so nothing can be concluded from their agreement or disagreement.
- Burden ceiling — the access and consent load the round is permitted to impose across its varied populations.
When it helps, and when it misleads¶
Its strength is turning "we studied whoever we could reach" into a deliberate robustness test: maximum-variation sampling makes any pattern that survives wide difference genuinely credible, and exposes the ones that were quietly context-bound.[n1] It is the loop's breadth instrument.
Its failure modes come from the spread itself. Stretch the cases so far that they are no longer comparable and the round yields nothing — agreement across incommensurable cases is not robustness, it is coincidence. The signature misuse is treating a maximum-variation batch as if it were a representative sample and reporting proportions from it; the round is built for range, not for standing in for a population. The guarding discipline is to hold one analysis framework across the whole spread and to report invariants and their exceptions, never frequencies.
How it implements the components¶
A Maximum Variation Case Round fills the breadth slice of the archetype, not its edge-finding or disconfirming slices:
case_universe_and_access_boundary— it maps the case universe and its access limits, then samples deliberately across that whole span.case_contrast_palette— it draws on the palette's contrast and variation roles to maximize difference across the batch.ethics_and_burden_gate— casting a wide net across many different populations multiplies access and consent burden, so the round is run through the burden gate.
It does not author scope conditions from a single edge case — that is Boundary Case Probe via scope_and_transferability_note; nor does it hunt a case chosen to disconfirm the account, which is Negative Case Sampling Pass via negative_or_deviant_case_trigger.
Related¶
- Instantiates: Theory-Responsive Case Sampling Design — supplies the breadth test that separates robust findings from context-bound ones.
- Sibling mechanisms: Boundary Case Probe · Case Selection Audit Trail · Constant Comparison Matrix · Grounded Theory Sampling Memo · Negative Case Sampling Pass · Rival Explanation Discriminator · Saturation Review Memo · Theoretical Gap Matrix · Transferability Claim Check
Editorial Notes¶
Form Classification¶
Form family: Experiment, Test & Rehearsal
Rationale: The round deliberately exposes a proposed finding to maximally different cases under a held-constant analysis to learn which claims survive the variation.
Nearest alternative: Analysis, Modeling & Optimization — Common coding supports comparison, but the defining operation is the deliberate variation and challenge rather than computation alone.
Review outcome: Adjudicated after independent review; high confidence.
Origin Attribution¶
Primary origin: Ethnography & Qualitative Methods
Origin pattern: Single lineage
Present-day reach: Multi-domain
Rationale: Maximum-variation sampling is an established purposive case-selection strategy in qualitative research.
Related originating lineages:
- Sociology & Anthropology — For Maximum Variation Case Round, institutions, membership, social scale, norms, and collective meaning materially shaped the mechanism's characteristic form.
Review resolution: Both independent reviews place the primary provenance in ethnography_qualitative_methods. The queued differences (alternate_origin_disagreement) concern secondary metadata, not primary lineage. The final retains sociology_anthropology only where a reviewer supplied a formative-lineage rationale; downstream use or broad applicability by itself is not treated as origin. origin_mode=single_lineage because one disciplinary lineage remains dominant and application breadth alone does not create another origin. domain_reach=multi_domain records established application breadth separately from provenance. confidence=high preserves the more cautious evidence assessment. encyclopedia_synthesis=false records whether either reviewer identified deliberate corpus-level composition.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
A maximum-variation round is a robustness test, never a representativeness claim. The two are easy to conflate because both involve "a spread of cases," but only representative sampling licenses statements about how common something is — this round licenses only statements about what survives difference.
[n1] Maximum variation sampling — one of the purposeful-sampling strategies catalogued by Michael Quinn Patton: deliberately choosing cases that differ widely so that any shared pattern is a strong finding and the variation itself documents the range to which the account must answer. ↩