Critical appraisal without a quality score

Ask the methodological questions that fit the source. Do not compress different weaknesses, strengths and evidence roles into one badge or number.

This framework asks questions; it does not answer them automatically. The individual evidence record states the extracted design and known limits. The original source remains necessary for independent appraisal.

Questions for every source

Identity and provenance: How were the organism, population, stock and life stage identified and sourced?: A published name or seller label can hide a taxonomic, population or life-stage transfer.
Independent unit: What unit was independently assigned or sampled, and how many units support each comparison?: Animal counts do not create independent replication when animals share a vessel, site, family or treatment history.
Comparison and context: What was the actual comparator, allocation process, environment, duration and material condition?: A result has meaning only against the comparison and conditions that produced it.
Endpoint and measurement: Which endpoint was measured, with what method, unit, timing, resolution and decision rule?: One endpoint cannot silently become survival, welfare, diagnosis, reproduction or long-term population performance.
Denominators and missingness: Are starting counts, exclusions, losses, missing observations and analysis denominators reconciled?: Unreported or changing denominators can alter the apparent direction, precision and applicability of a result.
Transfer boundary: What is the nearest tempting aquarium claim that this source design cannot establish?: Direct evidence can still be narrow, and adjacent evidence can be useful only while the inference remains visible.

Questions that depend on evidence class

Controlled experiment

Treatment assignment: Were independent units assigned to treatments, and were baseline conditions comparable?: Unclear allocation or baseline imbalance can confound the treatment comparison.
Replication: Were treatment vessels or other exposure units independently replicated rather than subsampled?: Repeated animals or assays inside one exposure unit do not replicate the environmental treatment.
Exposure verification: Was the treatment, dose, feed intake or environmental exposure measured and maintained as described?: Nominal treatment labels may not equal the exposure animals actually received.
Analysis: Did the statistical model match the unit, repeated measures, multiplicity and missing outcomes?: A precise p-value cannot repair a mismatched unit or unaccounted comparison structure.

Field observation

Sampling frame: How were sites, dates, habitats and animals selected, and what population can that sampling represent?: Convenient or uneven sampling limits prevalence, abundance and seasonal inference.
Sampling effort: Was effort standardized or recorded across sites, times and environmental conditions?: Observed counts can change because detectability or effort changed rather than the population.
Association: Which co-varying environmental or biological factors could explain the observed association?: Field correlation does not by itself identify cause or a husbandry intervention.
Detection and absence: Could non-detection reflect gear, visibility, season, behavior or identification limits?: Not observed is not equivalent to absent.

Morphology or taxonomy

Examined material: Which specimens, localities, sexes, stages, types or sequence accessions were actually examined?: A diagnosis applies to examined material and methods, not automatically to a trade-line photograph.
Diagnostic characters: Which characters or loci distinguish the taxon, and were alternatives explicitly compared?: Color alone may be variable, convergent or unrelated to the diagnostic character.
Name history: Were synonyms, historical combinations, type material and later revisions checked?: An older title can describe a population that later work placed in another taxon.
Consumer identification limit: Can the required character or method be observed reliably in an ordinary live-animal image?: A valid taxonomic method may still be unavailable to a keeper examining one photograph.

Review or synthesis

Search and selection: Were search sources, dates, queries, eligibility decisions and excluded material reported?: An unsystematic source set can omit contrary or newer evidence.
Source roles: Are primary experiments, field studies, reviews and expert assertions kept distinguishable?: Repeated citation through reviews does not create new independent evidence.
Included-study appraisal: Were the units, methods, limitations and conflicts of included sources critically appraised?: A synthesis inherits limitations when it only counts or repeats conclusions.
Currency: What was the search end date, and could later taxonomy, corrections or studies change the synthesis?: A well-conducted review can become stale without becoming fraudulent.

System observation or method study

System relationship: Was the measured unit an aquarium system, analytical method, another taxon or another biological level?: A system or method result is not automatically a Neocaridina animal response.
Inference mechanism: What explicit mechanism or shared property makes the adjacent evidence relevant?: Similarity must be stated and tested rather than assumed from a broad label.
Target validation: Has the method or inference been validated in Neocaridina, the intended water, life stage and decision range?: Analytical validity in one matrix or biological validity in another species may not transfer.
Permitted use: Is the source being used only for its method or system role rather than promoted to animal evidence?: Useful adjacent evidence becomes misleading when its role is silently upgraded.

How to use the framework

Open the canonical evidence record and the original source.
Start with the six questions shared by every source.
Apply the four questions for the record's evidence class.
Write the narrow result and consequential limitation before considering aquarium action.
Use the evidence challenge when the record omits or misstates a method, result, conflict or boundary.

A limitation does not automatically invalidate a study, and a long checklist does not make one source more important than another. Appraisal decides what a source can support, not whether it deserves a decorative score.

Framework version 1.0, reviewed 2026-08-12.