Constant Comparison Matrix¶
Analytic matrix — instantiates Theory-Responsive Case Sampling Design
Compares each new case against prior cases and the current categories, forcing every difference into a model revision.
A Constant Comparison Matrix is the analytic engine of the loop: it takes each newly collected case or incident and compares it, cell by cell, against the cases already collected and against the current categories, so that every similarity tightens a category and every difference forces either a new property or an outright revision. It works retrospectively, on data already in hand, and its output is model change — not a case selection and not a stopping decision. Its defining move is refusing to let a case pass without an explicit comparison: nothing is filed as "more of the same" until it has been checked against what "the same" is supposed to mean.
Example¶
A nurse-researcher is building an account of how patients on long-term dialysis cope with the rigidity of their treatment schedule. Early interviews yield a category, rigid adherence — patients who organize their whole week around treatment and resist any deviation. A new interview then describes something that looks similar but isn't: a patient who negotiates with the clinic to shift sessions around a grandchild's visits, treating the schedule as bargainable. Fed into the matrix — rows are incidents, columns are categories and their properties — this incident is compared against the prior rigid adherence incidents and refuses to sit in the same cell.
The comparison forces a split: what looked like one category is two, distinguished by a newly named property, schedule as fixed versus negotiable. That property, in turn, reorganizes several earlier incidents that had been filed too coarsely. The matrix does not decide which patient to interview next; it changes what the model is, so that whoever does choose the next case is choosing against a sharper account than existed an hour ago.
How it works¶
- Code in units. Break each case into incidents small enough to compare, not whole-case impressions.
- Compare incident to incident. Test each new incident against prior ones: same category, or a difference that needs a name?
- Compare incident to category. Ask whether the current category can hold the incident, or must gain a property, split, or merge.
- Fold the divergence into the model. Every difference that survives comparison becomes a logged revision to the categories and their properties.
Tuning parameters¶
- Comparison unit — fine incidents versus whole-case gestalts. Finer units catch subtler distinctions but multiply the comparisons to run.
- Split-versus-collapse bias — how readily a difference spawns a new category. Splitting eagerly yields resolution but risks a shattered, over-specified model; collapsing eagerly risks papering over real distinctions.
- Negative-evidence weight — how much a single dissenting incident is allowed to move a category. High weight keeps the model honest but jittery; low weight is stable but complacent.
- Representation — a literal grid versus running comparative memos. The grid surfaces gaps at a glance; memos capture nuance a cell cannot.
When it helps, and when it misleads¶
Its strength is discipline against narrative fluency — the pull to see confirmation everywhere. Under the constant comparative method, every case must earn its place by changing a category or being explicitly checked and found not to,[1] which is what keeps an emerging account anchored to data rather than to the analyst's growing conviction.
Its failure modes are the matrix's own excesses. Fine-grained comparison invites combinatorial explosion and over-splitting, a model so finely diced that its categories no longer generalize. The opposite misuse is coding new incidents to fit a preferred category — quietly immunizing the model against the very differences the matrix exists to surface. The guarding discipline is to periodically check, informally, that each category earned its properties from the data and not from the frame the analyst brought in.
How it implements the components¶
A Constant Comparison Matrix fills the analysis half of the loop, not its selection or stopping halves:
emerging_model_frame— the matrix's columns are the current categories and properties; running comparisons is what keeps that frame current.analysis_sampling_loop— it is the analysis step of the loop, working each collected case against the model before the next is chosen.model_revision_register— every difference that survives comparison is logged as a concrete revision.
It does not register the model's open gaps or map them to candidate cases — that prospective grid is Theoretical Gap Matrix, which owns analytic_gap_register and case_contrast_palette.
Related¶
- Instantiates: Theory-Responsive Case Sampling Design — supplies the analytic comparison that turns collected cases into a revised model.
- Sibling mechanisms: Boundary Case Probe · Case Selection Audit Trail · Grounded Theory Sampling Memo · Maximum Variation Case Round · Negative Case Sampling Pass · Rival Explanation Discriminator · Saturation Review Memo · Theoretical Gap Matrix · Transferability Claim Check
Editorial Notes¶
Form Classification¶
Form family: Analysis, Modeling & Optimization
Rationale: Compares each new case against prior cases and the current categories, forcing every difference into a model revision, making its operative form a computation, comparison, model, or analytic representation used to infer, estimate, or choose.
Independent corroboration: The frozen evidence defines Constant Comparison Matrix as 'Compares each new case against prior cases and the current categories, forcing every difference into a model revision', so its operative form is Analysis, Modeling & Optimization.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Ethnography & Qualitative Methods
Origin pattern: Single lineage
Present-day reach: Specialized
Rationale: Grounded-theory methodology cohered constant comparison of each new case with prior cases and provisional categories so analysis continually revises the emerging theory.
Related originating lineages:
- Sociology & Anthropology — The method arose within sociological qualitative research before becoming a general analytic practice.
Review resolution: Both reviewers agree on ethnography_qualitative_methods as primary. Constant comparison arose within grounded qualitative analysis, with sociology_anthropology materially forming that lineage; the matrix format is an encyclopedia synthesis serving a specialized analytic task.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; high confidence.
References¶
[1] The constant comparative method — from Glaser and Strauss's The Discovery of Grounded Theory (1967) — the practice of continuously comparing each new datum against prior data and against emerging categories, so that categories and their properties are generated from the comparisons rather than imposed in advance. withdrawn registry ↩