Missing-Data Sensitivity

Scientific meaning

What it measures

Assesses missing-data sensitivity using evidence appropriate to statistical validity and inference, distinguishing random uncertainty from systematic error.

Why it matters

Material weakness in missing-data sensitivity can change the direction, magnitude, certainty, or biological interpretation of the research conclusion.

Measurement criteria

Prespecified threshold; valid reference standard; complete 2×2 classification; confidence intervals; spectrum and prevalence assessment.

Typical methods of assessment

Structured critical appraisal; quantitative estimation with uncertainty; prespecified thresholds; independent replication; sensitivity and subgroup analyses.

Computation profile

Computation mode

QuantitativeProportionComputation

Readiness

TemplateComputable

Formula status

GenericTemplateDefined

Method family

proportion estimation

Output datatype

xsd:decimal

Unit or scale

Proportion or percentage (0–1 or 0–100%)

Numeric range

0 to 1

Directionality

ContextDependentDirection

Required inputs

• numerator
• denominator
• operational_definition
• assessment_context

Optional inputs

• weight
• stratum
• confidence_level

Machine-readable expression

{
  "language": "BEMO-Expression-JSON",
  "operator": "divide",
  "arguments": [
    "numerator",
    "denominator"
  ],
  "constraints": [
    "denominator > 0"
  ]
}

Computation policies

Aggregation

Not specified in source; must be defined and versioned before composite use.

Normalization

None by default; any normalization must be justified, versioned, and validated.

Missing data

Must be declared before computation; report missingness; no silent imputation; perform sensitivity analysis when material.

Uncertainty required

Yes

Uncertainty method

Confidence interval, credible interval, bootstrap distribution, inter-rater reliability, or sensitivity analysis as scientifically applicable.

Confidence interval required

Yes

Threshold policy

Prespecify, justify, version, and sensitivity-test thresholds; do not derive and evaluate on the same data without correction.

Decision thresholds

No universal threshold declared

Quality control

Source provenance; input validation; duplicate control; uncertainty reporting; independent or orthogonal validation where applicable.

Calibration

As applicable; required when interpreted probabilistically.

Applicability and validation

Evidence object

Result / experiment / study / body of evidence, as applicable

Evidence level

Result / experiment / study / body of evidence, as applicable

Study types

All quantitative biomedical studies

Domain

Statistical Validity and Inference

Required data sources

Context-specific; use source references, registered studies, primary data, and authoritative biomedical resources as applicable.

Validation status

Source maturity: Mature; BEMO computation profile requires independent validation.

Benchmark

A representative, versioned benchmark set is required before production use.

Reference standard

Use an independent reference standard when one exists; document expert-adjudicated alternatives.

External validation required

Yes

Interpretation safeguards

Common misinterpretations

Treating missing-data sensitivity as interchangeable with overall study quality, or interpreting a favorable value as proof that all other bias and validity domains are satisfactory.

Limitations

Depends on context, operational definition, thresholds, data quality, and evaluator judgment; it should not be used as a stand-alone summary score without domain-level evidence.

Dependency status

Not yet curated. Closely related metrics are represented; causal or computational dependencies require expert curation.

Source provenance

Source workbook

Pure_Biomedical_Evidence_Research_Metrics_Part_2(3).xlsx

Source worksheet

Biomedical Evidence Metrics

Source row

455

Source record SHA-256

6708217d617153345c000ff0b34cc96bd34facf9514e5d1a84b3e56623ffb545

Definition source

Pure_Biomedical_Evidence_Research_Metrics_Part_2(3).xlsx row 455; see source_references

Ontology IRI

https://w3id.org/bemo/BEMO_2000448

Profile IRI

https://w3id.org/bemo/profile/BEMO_2000448

SHACL shape

https://w3id.org/bemo/shapes/BEMO_2000448_AssessmentShape

Scientific references (5)

1. https://www.consort-spirit.org/Source framework, standard, guideline, or originSource-preserved; URL reachability not asserted by this package
2. https://www.equator-network.org/reporting-guidelines/strobe/Source framework, standard, guideline, or originSource-preserved; URL reachability not asserted by this package
3. https://www.tripod-statement.org/Source framework, standard, guideline, or originSource-preserved; URL reachability not asserted by this package
4. https://www.equator-network.org/reporting-guidelines/reporting-recommendations-for-tumor-marker-prognostic-studies-remark/Source framework, standard, guideline, or originSource-preserved; URL reachability not asserted by this package
5. https://database.ich.org/sites/default/files/E9_Guideline.pdfSource framework, standard, guideline, or originSource-preserved; URL reachability not asserted by this package

Governance and FAIR status

Lifecycle

Candidate

Approval

Draft

Owner

BEMO Project

Curator

Unassigned

Reviewer

Unassigned

License

Pending owner approval; CC BY 4.0 recommended for OBO compatibility.

Findable

Provisional: stable local ID assigned; public namespace registration pending.

Accessible

Provisional: package is distributable; permanent public release location pending.

Interoperable

Yes: OWL 2, RDF, SKOS, SHACL, JSON-LD, CSV, and OBO-style identifiers.

Reusable

Provisional: rich metadata supplied; open-license owner approval pending.

Governance note

Candidate term pending scientific, ontology-engineering, and computation-method review.

Editor note

Source content preserved verbatim; computation metadata is a generated default and must be curated before normative use.