Case Selection Audit Trail¶
Provenance log — instantiates Theory-Responsive Case Sampling Design
Preserves the versioned, time-ordered record of the sampling path — memos, access constraints, and each case's model effect — so the sequence can be reconstructed and defended.
A Case Selection Audit Trail is the retrospective, append-only record of how the sampling sequence actually unfolded: which cases were taken in what order, which versions of the sampling memos were live at each step, what access constraints shaped what was reachable, and what each analyzed case did to the model. It authors none of that reasoning — it preserves it, time-stamped and immutable, so that a later reader (an auditor, a reviewer, or oneself six months on) can reconstruct why the case order came out as it did and rule out the suspicion that rationales were retrofitted to findings. Its defining property is chronology under lock: the value is not in the reasoning it holds but in proving when that reasoning existed.
Example¶
A national transport safety board is investigating a cluster of signal-passed-at-danger incidents on a commuter rail network. The investigation is adaptive: after the first three cases point at fatigue, the team adds a near-miss from a neighboring network, then deliberately adds a case where crew fatigue was demonstrably absent. Every one of those moves is logged in the audit trail — case order and timestamps, the memo version that justified each pick, a note that two depot cases were unreachable because the operator withheld rosters (an access constraint, not an analytic choice), and a before/after snapshot of what each case did to the working causal model.
Two years later an independent oversight panel re-opens the file after a similar accident. The politically loaded question is whether the board cherry-picked cases to reach a predetermined "fatigue" conclusion. The audit trail settles it: the record shows the fatigue-absent case was selected and analyzed before the fatigue finding was finalized, and that the two missing depot cases were excluded for access, not because they were inconvenient. The trail does not prove the analysis was correct — but it proves the sequence was honest, which is exactly what the panel needed to see.
How it works¶
- Append-only logging. Each selection appends a dated entry; earlier entries are never silently edited.
- Version-stamp the memos. Every sampling memo is captured in the version that was live when the case was chosen, so later revisions cannot rewrite the recorded intent.
- Record access constraints. Note which cases were unreachable or substituted, and why, so exclusions read as access limits rather than analytic verdicts.
- Snapshot model effects. Store a before/after of the model state each case produced, linking the sequence to its consequences.
Tuning parameters¶
- Granularity — every micro-decision versus one entry per case. Finer granularity buys reconstructability at the cost of upkeep and noise.
- Immutability strictness — hard append-only versus editable-with-history. Stricter immutability defeats retrofitting but is heavier to maintain.
- Snapshot depth — full model state versus a delta per case. Full snapshots reconstruct any past state; deltas are lighter but chainable only in order.
- Access-constraint detail — a bare "unreachable" flag versus a documented contingency. More detail defends exclusions better but exposes sensitive field relationships.
When it helps, and when it misleads¶
Its strength is defeating post-hoc rationalization: a locked chronology is what lets an inquiry claim its selection was reasoned in advance rather than fitted to results, and it is the operational backbone of confirmability.[1] It is what turns "trust us" into "check the record."
Its failure mode is that a trail can be complete but performative — every step faithfully logged while the reasoning at each step was still biased. The trail proves order and timing, not soundness; treating a thorough audit trail as though it validated the selection logic is the classic misuse, lending procedural polish to a flawed inquiry. The guarding discipline is to pair the trail with an independent read of the rationales' content, not just their timestamps — the log shows you can reconstruct the path, not that the path was wise.
How it implements the components¶
A Case Selection Audit Trail fills the preservation slice of the archetype, not the reasoning:
memo_version_history— its spine: every sampling memo kept in versioned, time-stamped form, so intent cannot be rewritten after the fact.field_access_contingency_plan— it records the access constraints and workarounds that shaped which cases were reachable, so exclusions are legible as access limits.model_revision_register— it snapshots each case's before/after model effect, tying the sequence to its consequences.
It does not author the selection reasoning it stores: selection_rationale_record, analytic_gap_register, and case_learning_question are written by Grounded Theory Sampling Memo; the trail only versions and preserves them.
Related¶
- Instantiates: Theory-Responsive Case Sampling Design — supplies the auditability half of the archetype's adaptivity-versus-auditability tradeoff.
- Consumes: Grounded Theory Sampling Memo — the memos whose versions and ordering the trail preserves.
- Sibling mechanisms: Boundary Case Probe · Constant Comparison Matrix · 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: Record, Log & Register
Rationale: Preserves the versioned, time-ordered record of the sampling path — memos, access constraints, and each case's model effect — so the sequence can be reconstructed and defended, making its operative form a durable record, ledger, register, or trace whose value depends on preserving actual state or history.
Independent corroboration: The frozen evidence defines Case Selection Audit Trail as 'Preserves the versioned, time-ordered record of the sampling path — memos, access constraints, and each case's model effect — so the sequence can be reconstructed and defended', so its operative form is Record, Log & Register.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Ethnography & Qualitative Methods
Origin pattern: Single lineage
Present-day reach: Specialized
Rationale: Qualitative research established chronological audit trails for sampling memos, access constraints, analytic effects, and confirmability.
Review resolution: Ethnography and qualitative research are primary because an explicit case-selection trail is a method of reflexivity, confirmability, and transparent sampling. Historical and accounting records are useful analogies but do not constitute separate origins for this research-sampling mechanism.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
The trail records that you can see why the sequence unfolded, not that it was right. Keeping this distinction sharp is what stops an audit trail from being mistaken for validation — reconstruction and soundness are separate goods, and only the first is the trail's job.
References¶
[1] Confirmability — in Lincoln and Guba's trustworthiness framework (Naturalistic Inquiry, 1985), the degree to which findings are shaped by the data and participants rather than the researcher's bias. An audit trail is the canonical means of demonstrating it. registry ↩