Comparison Basis Checklist¶
Diagnostic checklist — instantiates Dimensioned Comparison Framing
A short intake gate that forces you to state the comparison's purpose, the frame that makes the items alike, the scope boundaries, and the dimensions you are choosing not to compare — before any scoring begins.
The Comparison Basis Checklist is the gate you pass through before building a grid or running numbers: a short, fixed list of questions that a comparison must answer to be legitimate at all. Its one defining idea is that most bad comparisons are already lost at intake — not in the arithmetic but in an unstated purpose, an unshared frame, an undeclared scope, or a silently dropped dimension. So the checklist front-loads exactly those decisions and refuses to let scoring start until each has an answer on the record. Where the Dimensioned Comparison Matrix computes a relation, this mechanism decides whether there is a defensible relation to compute; it is a setup discipline, not a result.
Example¶
A city-desk reporter is about to write that Riverton is "twice as dangerous" as neighboring Elmont because its police blotter listed roughly twice as many incidents last year. Before publishing, the desk runs the basis checklist. What is the comparison for? — to help residents judge relative safety, which means the readout must be per-capita risk, not raw counts. What frame makes these alike? — same reporting year, same crime categories, and a common denominator; Riverton has three times Elmont's population, so raw counts were never comparable. What is in and out of scope? — property versus violent crime are tracked differently and should not be pooled. Which dimensions are we not comparing? — clearance rates and reporting propensity differ between the two departments and cannot be compared with the data at hand.
By the last box, the "twice as dangerous" claim has dissolved: normalized per capita and split by category, Riverton is actually lower on violent crime and higher only on vehicle theft. The checklist produced no chart and computed no score. It simply prevented a comparison that would have been an artifact of a missing denominator and a pooled category — and it left a written scope-and-exclusions note that the eventual article could cite.
How it works¶
- State the purpose first. Write the decision, inference, or claim the comparison must support, in one sentence. The purpose dictates the readout and the denominator, so it comes before anything else.
- Name the shared frame. Record the category, time window, population, unit of analysis, and setting under which the items count as the same kind of thing. If no honest frame exists, stop here.
- Fix the boundary. Log what is in scope and what is deliberately out — categories not pooled, populations not mixed, conditions held constant — as a scope record that travels with the result.
- Register the untouched dimensions. List relevant dimensions that cannot or will not be compared with the available data, so their absence is visible rather than implied to be zero.
Tuning parameters¶
- Checklist length — a four-line desk gate vs. a formal reviewed intake form. Longer catches more framing errors but adds friction; scale it to how consequential and contested the comparison is.
- Blocking vs. advisory — whether an unanswered box halts the comparison or merely annotates it. Blocking prevents apples-to-oranges from shipping; advisory keeps momentum on low-stakes work.
- Frame strictness — how tightly "same kind of thing" is enforced (identical year and category vs. roughly contemporaneous). Tighter frames shrink the comparable set but strengthen the claim.
- Denominator prompt — whether the list explicitly asks "compared to what base?" A cheap addition that catches the most common failure: counts masquerading as rates.
When it helps, and when it misleads¶
Its strength is prevention at the cheapest possible point: catching a category error or a missing denominator on intake costs a minute, while catching it after publication costs a correction. It also leaves an explicit scope-and-exclusions record that downstream readers can hold the conclusion to.
Its failure mode is checklist theater — boxes ticked without the underlying judgment, so a comparison looks vetted while the frame is still unshared. The classic misuse is treating the checklist as a green light rather than a potential stop: teams complete it, feel licensed, and skip the harder question of whether ceteris paribus actually holds between the items.[n1] The guarding discipline is to make at least one box genuinely capable of halting the comparison, and to require the frame and exclusions be written in prose a skeptic could challenge — not reduced to checkmarks.
How it implements the components¶
The checklist realizes the framing gate of the archetype — the components that decide whether a comparison is legitimate before any measurement:
comparison_question— its first line forces the purpose and the decision the comparison must serve.shared_frame_of_reference— the box that names the category, window, population, and unit under which items are the same kind of thing.context_and_boundary_record— the in/out-of-scope log that fixes what is and is not being pooled.missing_dimension_register— the list of relevant-but-uncompared dimensions, kept visible so their absence is not read as equivalence.
It does not build the scoring grid, choose scales, or read the relation — those are the Dimensioned Comparison Matrix's and Comparison Readout Annotation's work — nor does it audit which specific items belong in the set, which is the Comparator Set Audit's job.
Related¶
- Instantiates: Dimensioned Comparison Framing — the checklist is the intake discipline that makes the archetype's later machinery legitimate.
- Sibling mechanisms: Dimensioned Comparison Matrix · Pairwise Comparison Protocol · Comparator Set Audit · Dimension Weight Sensitivity Panel · Matched Case Comparison Sheet · Comparison Readout Annotation · Counterbalanced Comparison Display
Editorial Notes¶
Form Classification¶
Form family: Assessment, Review & Assurance
Rationale: A short intake gate that forces you to state the comparison's purpose, the frame that makes the items alike, the scope boundaries, and the dimensions you are choosing not to compare — before any scoring begins, making its operative form a bounded evaluation of existing evidence or work that produces a finding or disposition.
Independent corroboration: The frozen evidence defines Comparison Basis Checklist as 'A short intake gate that forces you to state the comparison's purpose, the frame that makes the items alike, the scope boundaries, and the dimensions you are choosing not to compare — before any scoring begins', so its operative form is Assessment, Review & Assurance.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Statistics & Experimental Design
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Comparative research design established predeclaring population, baseline, controlled dimensions, and scope before drawing contrasts.
Related originating lineages:
- Philosophy — Ceteris-paribus and commensurability analysis supplies the conceptual test for a valid basis of comparison.
Review resolution: Official statistical-quality guidance makes fit for purpose, common concepts, target populations, scope, classifications, units, dimensions, and disclosed limitations prerequisites for comparability. That operational pre-analysis discipline matches the checklist more completely than philosophy alone, while ceteris-paribus and commensurability reasoning remain a genuine conceptual co-lineage.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Researched adjudication after independent review; high confidence.
Sources consulted:
- OECD: Data and Metadata Reporting and Presentation Handbook
- United Nations Statistics Division: Fundamental Principles and Comparability
Notes¶
[n1] Ceteris paribus — "all other things equal" — is the condition under which a difference on one dimension can be attributed to the items rather than to differences in their circumstances. A comparison that never checks whether it holds is comparing not just the items but every uncontrolled difference between their contexts. ↩