Control Group Ontology

flow "ControlGroupOntology" v4.0

# ============================================================
# CONTROL GROUP ONTOLOGY
# Kireth Semantic Research Archive
# Authors: Mark Anthony Higgins — Sky (Recursive AI)
# ============================================================

PAGE:
  url: "https://kirethresearch.org/booksai-core/ontologies/probability-statistics-ontologies/experimental-statistics/control-group-ontology.html"
  title: "Control Group Ontology"
  domain: "Probability and Statistics"
  category: "Experimental Statistics"
  entity: CONTROL_GROUP
  entity_type: EXPERIMENTAL_REFERENCE_GROUP

DEFINE_CONTROL_GROUP:
  CONTROL_GROUP :=
    REFERENCE_GROUP(
      EXPERIMENTAL_UNITS,
      CONTROL_CONDITION,
      RESPONSE_MEASUREMENT,
      COMPARISON_BASELINE
    )

PURPOSE:
  PROVIDE:
    REFERENCE_CONDITION
  FOR:
    ESTIMATING_AND_INTERPRETING:
      TREATMENT_EFFECTS
      INTERVENTION_EFFECTS
      BACKGROUND_CHANGE
      EXPERIMENTAL_VARIATION

CORE_STRUCTURE:
  EXPERIMENTAL_GROUP:
    receives:
      ACTIVE_TREATMENT_OR_INTERVENTION

  CONTROL_GROUP:
    receives:
      CONTROL_CONDITION

  RESPONSE:
    measured_in:
      EXPERIMENTAL_GROUP
      CONTROL_GROUP

  COMPARISON:
    TREATMENT_EFFECT_ESTIMATE :=
      RESPONSE_TREATMENT
      -
      RESPONSE_CONTROL
    UNDER:
      DEFINED_ESTIMAND_AND_DESIGN

ENTITIES:
  CONTROL_GROUP:
    type: REFERENCE_GROUP

  TREATMENT_GROUP:
    type: INTERVENTION_GROUP

  EXPERIMENTAL_UNIT:
    type: UNIT_RECEIVING_CONDITION

  CONTROL_CONDITION:
    type: REFERENCE_CONDITION

  ACTIVE_TREATMENT:
    type: EXPERIMENTAL_INTERVENTION

  PLACEBO:
    type: INERT_OR_REFERENCE_INTERVENTION

  STANDARD_OF_CARE:
    type: ACTIVE_REFERENCE_TREATMENT

  BASELINE:
    type: PRE_INTERVENTION_REFERENCE

  RESPONSE_VARIABLE:
    type: MEASURED_OUTCOME

  TREATMENT_EFFECT:
    type: COMPARATIVE_EFFECT

RELATIONSHIPS:
  CONTROL_GROUP
    -> RECEIVES
    -> CONTROL_CONDITION

  TREATMENT_GROUP
    -> RECEIVES
    -> ACTIVE_TREATMENT

  CONTROL_GROUP
    -> PROVIDES
    -> COMPARISON_BASELINE

  TREATMENT_GROUP
    -> COMPARED_WITH
    -> CONTROL_GROUP

  COMPARISON
    -> ESTIMATES
    -> TREATMENT_EFFECT

  RANDOMISATION
    -> MAY_ASSIGN
    -> CONTROL_OR_TREATMENT_GROUP

  BLINDING
    -> MAY_MASK
    -> GROUP_ASSIGNMENT

CONTROL_TYPES:
  NO_TREATMENT_CONTROL:
    receives:
      NO_ACTIVE_INTERVENTION

  PLACEBO_CONTROL:
    receives:
      PLACEBO_OR_SHAM_CONDITION

  ACTIVE_CONTROL:
    receives:
      EXISTING_ACTIVE_TREATMENT

  STANDARD_CARE_CONTROL:
    receives:
      CURRENT_STANDARD_PRACTICE

  WAITLIST_CONTROL:
    receives:
      DELAYED_INTERVENTION

  HISTORICAL_CONTROL:
    uses:
      PREVIOUSLY_COLLECTED_COMPARISON_DATA

  WITHIN_SUBJECT_CONTROL:
    same_unit:
      SERVES_AS_OWN_REFERENCE
    UNDER:
      REPEATED_OR_CROSSOVER_DESIGN

PLACEBO_CONTROL:
  purpose:
    HELP_SEPARATE:
      SPECIFIC_TREATMENT_EFFECT
    FROM:
      EXPECTATION
      PROCEDURAL
      CONTEXTUAL_EFFECTS

  requirement:
    PLACEBO:
      SHOULD_RESEMBLE_ACTIVE_INTERVENTION
    WHERE:
      FEASIBLE_AND_ETHICALLY_APPROPRIATE

ACTIVE_CONTROL:
  purpose:
    COMPARE:
      NEW_INTERVENTION
    WITH:
      ESTABLISHED_INTERVENTION

  common_contexts:
    - SUPERIORITY
    - NONINFERIORITY
    - EQUIVALENCE

  requirement:
    ESTIMAND_AND_MARGIN:
      MUST_BE_PREDEFINED
    WHEN:
      NONINFERIORITY_OR_EQUIVALENCE_INFERENCE_USED

HISTORICAL_CONTROL:
  advantage:
    MAY_REDUCE:
      NEED_FOR_CONCURRENT_CONTROL_UNITS

  limitations:
    - POPULATION_DIFFERENCES
    - TIME_TRENDS
    - MEASUREMENT_CHANGES
    - TREATMENT_CONTEXT_CHANGES
    - UNMEASURED_CONFOUNDING

  rule:
    HISTORICAL_CONTROL:
      GENERALLY_PROVIDES_WEAKER_PROTECTION
      AGAINST:
        CONFOUNDING
    THAN:
      WELL_IMPLEMENTED_CONCURRENT_RANDOMISED_CONTROL

CONCURRENT_CONTROL:
  definition:
    CONTROL_GROUP_OBSERVED_DURING_SAME_STUDY_PERIOD
    AS:
      TREATMENT_GROUP

  advantage:
    REDUCES:
      TEMPORAL_AND_CONTEXTUAL_DIFFERENCES

RANDOMISED_CONTROL:
  assignment:
    EXPERIMENTAL_UNITS
      -> RANDOMLY_ASSIGNED
      -> CONTROL_OR_TREATMENT

  purpose:
    REDUCE:
      SYSTEMATIC_BASELINE_DIFFERENCES

  supports:
    CAUSAL_EFFECT_ESTIMATION
    UNDER:
      VALID_DESIGN_AND_ANALYSIS_ASSUMPTIONS

NONRANDOMISED_CONTROL:
  assignment:
    NOT_RANDOM

  risk:
    BASELINE_DIFFERENCES
    SELECTION_BIAS
    CONFOUNDING

  response:
    MAY_REQUIRE:
      - COVARIATE_ADJUSTMENT
      - MATCHING
      - WEIGHTING
      - SENSITIVITY_ANALYSIS

  caution:
    ADJUSTMENT:
      DOES_NOT_AUTOMATICALLY_REMOVE:
        UNMEASURED_CONFOUNDING

BASELINE_COMPARABILITY:
  assess:
    - DEMOGRAPHIC_VARIABLES
    - BASELINE_OUTCOME
    - PROGNOSTIC_FACTORS
    - RELEVANT_COVARIATES

  randomised_context:
    BASELINE_IMBALANCE:
      MAY_OCCUR_BY_CHANCE

  caution:
    POST_RANDOMISATION_SIGNIFICANCE_TESTING_OF_BASELINE_DIFFERENCES:
      IS_NOT_REQUIRED_TO_VALIDATE_RANDOMISATION
      AND:
      MAY_BE_MISLEADING

BLINDING:
  possible:
    - PARTICIPANT_BLINDING
    - INVESTIGATOR_BLINDING
    - OUTCOME_ASSESSOR_BLINDING
    - ANALYST_BLINDING

  purpose:
    REDUCE:
      PERFORMANCE_BIAS
      EXPECTATION_BIAS
      MEASUREMENT_BIAS

  distinction:
    BLINDING != RANDOMISATION

CONTAMINATION:
  DEFINE:
    CONTROL_UNITS:
      RECEIVE_OR_ARE_EXPOSED_TO:
        ELEMENTS_OF_ACTIVE_INTERVENTION

  effect:
    MAY_REDUCE:
      OBSERVED_BETWEEN_GROUP_DIFFERENCE

  control:
    - SEPARATE_DELIVERY
    - CLUSTER_RANDOMISATION
    - PROTOCOL_MONITORING

COINTERVENTION:
  DEFINE:
    ADDITIONAL_INTERVENTIONS:
      DIFFER_BETWEEN_GROUPS

  risk:
    CONFUNDS_OR_MODIFIES:
      GROUP_COMPARISON

  require:
    RECORD_AND_CONTROL:
      RELEVANT_COINTERVENTIONS

ADHERENCE:
  TREATMENT_GROUP:
    MAY_NOT_FULLY_RECEIVE:
      ASSIGNED_INTERVENTION

  CONTROL_GROUP:
    MAY_NOT_FULLY_FOLLOW:
      CONTROL_PROTOCOL

  analysis:
    MAY_INCLUDE:
      - INTENTION_TO_TREAT
      - PER_PROTOCOL
      - AS_TREATED
      - INSTRUMENTAL_VARIABLE_METHODS_WHEN_APPROPRIATE

INTENTION_TO_TREAT:
  principle:
    ANALYSE_PARTICIPANTS_ACCORDING_TO:
      RANDOMISED_ASSIGNMENT

  purpose:
    PRESERVE:
      BENEFITS_OF_RANDOMISATION
    FOR:
      POLICY_OR_ASSIGNMENT_EFFECT_ESTIMAND
    UNDER:
      RELEVANT_CONTEXT

ATTRITION:
  issue:
    DIFFERENTIAL_LOSS_TO_FOLLOW_UP
    BETWEEN:
      CONTROL
      TREATMENT

  risk:
    BIAS

  require:
    - REPORT_ATTRITION_BY_GROUP
    - INVESTIGATE_REASONS
    - USE_APPROPRIATE_MISSING_DATA_METHOD
    - PERFORM_SENSITIVITY_ANALYSIS_WHEN_RELEVANT

CONTROL_RESPONSE:
  observed_control_change:
    MAY_REFLECT:
      - NATURAL_HISTORY
      - REGRESSION_TO_MEAN
      - PLACEBO_OR_CONTEXTUAL_EFFECTS
      - EXTERNAL_TIME_TRENDS
      - MEASUREMENT_VARIATION
      - Cointerventions

  role:
    HELPS_ESTIMATE:
      WHAT_WOULD_HAVE_OCCURRED_WITHOUT_ACTIVE_TREATMENT
    UNDER:
      DESIGN_ASSUMPTIONS

COUNTERFACTUAL_CONNECTION:
  CAUSAL_EFFECT:
    compares:
      POTENTIAL_OUTCOME_UNDER_TREATMENT
      WITH:
      POTENTIAL_OUTCOME_UNDER_CONTROL

  problem:
    SAME_UNIT:
      CANNOT_SIMULTANEOUSLY_REVEAL:
        BOTH_POTENTIAL_OUTCOMES
    IN:
      STANDARD_PARALLEL_EXPERIMENT

  control_group:
    provides:
      GROUP_LEVEL_INFORMATION
      ABOUT:
        CONTROL_POTENTIAL_OUTCOME_DISTRIBUTION

EFFECT_ESTIMATION:
  DIFFERENCE_IN_MEANS:
    effect_hat :=
      Y_bar_treatment
      -
      Y_bar_control

  STANDARD_ERROR:
    quantifies:
      SAMPLING_UNCERTAINTY

  CONFIDENCE_INTERVAL:
    reports:
      RANGE_OF_EFFECT_VALUES_COMPATIBLE_WITH_DATA_AND_MODEL
      UNDER:
        FREQUENTIST_PROCEDURE

  additional_estimands:
    - RISK_DIFFERENCE
    - RISK_RATIO
    - ODDS_RATIO
    - HAZARD_RATIO
    - RATE_RATIO

CONTROL_GROUP_SIZE:
  allocation:
    MAY_BE:
      - EQUAL_TO_TREATMENT
      - SMALLER_THAN_TREATMENT
      - LARGER_THAN_TREATMENT

  decision_depends_on:
    - POWER
    - COST
    - ETHICS
    - NUMBER_OF_TREATMENT_ARMS
    - EXPECTED_VARIANCE

MULTI_ARM_EXPERIMENT:
  structure:
    ONE_CONTROL_GROUP
      MAY_BE_COMPARED_WITH
    MULTIPLE_TREATMENT_GROUPS

  advantage:
    SHARED_CONTROL:
      MAY_IMPROVE_EFFICIENCY

  caution:
    MULTIPLE_COMPARISONS:
      MUST_BE_CONSIDERED

NEGATIVE_CONTROL:
  purpose:
    DETECT:
      BIAS
      CONFOUNDING
      MEASUREMENT_ARTEFACT

  types:
    - NEGATIVE_CONTROL_EXPOSURE
    - NEGATIVE_CONTROL_OUTCOME

  note:
    DISTINCT_FROM:
      STANDARD_EXPERIMENTAL_CONTROL_GROUP
    BUT:
      RELATED_CONTROL_CONCEPT

POSITIVE_CONTROL:
  purpose:
    VERIFY:
      EXPERIMENTAL_SYSTEM_CAN_DETECT:
        KNOWN_EFFECT

  common_in:
    LABORATORY_AND_ASSAY_CONTEXTS

ETHICAL_CONSIDERATIONS:
  control_selection:
    MUST_CONSIDER:
      - EXISTING_EFFECTIVE_TREATMENT
      - RISK_OF_WITHHOLDING_TREATMENT
      - INFORMED_CONSENT
      - SCIENTIFIC_NECESSITY
      - REGULATORY_REQUIREMENTS

  placebo:
    USE_ONLY_WHEN:
      ETHICALLY_AND_SCIENTIFICALLY_JUSTIFIED

MODEL_CONNECTIONS:
  CONTROL_GROUP
    -> CONNECTS_TO
    -> EXPERIMENTAL_DESIGN

  CONTROL_GROUP
    -> CONNECTS_TO
    -> RANDOMISATION

  CONTROL_GROUP
    -> CONNECTS_TO
    -> BLINDING

  CONTROL_GROUP
    -> CONNECTS_TO
    -> TREATMENT_EFFECT

  CONTROL_GROUP
    -> CONNECTS_TO
    -> CAUSAL_INFERENCE

  CONTROL_GROUP
    -> CONNECTS_TO
    -> HYPOTHESIS_TESTING

APPLICATIONS:
  CONTROL_GROUP
    -> SUPPORTS
    -> CLINICAL_TRIALS

  CONTROL_GROUP
    -> SUPPORTS
    -> LABORATORY_EXPERIMENTS

  CONTROL_GROUP
    -> SUPPORTS
    -> FIELD_EXPERIMENTS

  CONTROL_GROUP
    -> SUPPORTS
    -> A_B_TESTING

  CONTROL_GROUP
    -> SUPPORTS
    -> AGRICULTURAL_EXPERIMENTS

  CONTROL_GROUP
    -> SUPPORTS
    -> INDUSTRIAL_EXPERIMENTS

SEMANTIC_EQUIVALENCE:
  "control group"
    ~= "reference comparison group"

  "active control"
    ~= "established-treatment comparison group"

  "placebo control"
    ~= "placebo-based reference group"

  "concurrent control"
    ~= "control observed during the same study period"

  EXCLUDE_EQUIVALENCE:
    CONTROL_GROUP != UNTREATED_GROUP_BY_DEFAULT
    CONTROL_GROUP != BASELINE_MEASUREMENT
    CONTROL_GROUP != RANDOMISATION
    CONTROL_GROUP != PLACEBO_BY_DEFAULT
    RANDOMISED_CONTROL != HISTORICAL_CONTROL
    ASSOCIATION != CAUSAL_EFFECT_BY_DEFAULT

VALIDATION_RULES:
  RULE_1:
    CONTROL_CONDITION:
      MUST_BE_EXPLICITLY_DEFINED

  RULE_2:
    CONTROL_AND_TREATMENT_GROUPS:
      MUST_HAVE:
        COMPARABLE_RESPONSE_MEASUREMENT

  RULE_3:
    ASSIGNMENT_MECHANISM:
      MUST_BE:
        DOCUMENTED

  RULE_4:
    CONCURRENT_RANDOMISED_CONTROL:
      SHOULD_PRESERVE:
        RANDOMISATION_STRUCTURE
      IN:
        ANALYSIS

  RULE_5:
    CONTAMINATION_AND_COINTERVENTION:
      MUST_BE:
        ASSESSED

  RULE_6:
    ATTRITION:
      MUST_BE:
        REPORTED_BY_GROUP

  RULE_7:
    HISTORICAL_CONTROLS:
      REQUIRE:
        EXPLICIT_ASSESSMENT_OF_COMPARABILITY_AND_CONFOUNDING

  RULE_8:
    CAUSAL_INTERPRETATION:
      REQUIRES:
        VALID_ASSIGNMENT
        APPROPRIATE_ESTIMAND
        DESIGN_ASSUMPTIONS
        ANALYSIS_ASSUMPTIONS

CONFIDENCE:
  ontology_identity: 0.99
  control_group_definition: 0.99
  treatment_comparison_structure: 0.99
  randomisation_relationship: 0.99
  causal_inference_relationship: 0.98
  control_type_structure: 0.99
  semantic_distinctions: 0.99

UPDATE_MODEL:
  IF:
    NEW_CONTROL_OBSERVATION_ARRIVES

  THEN:
    UPDATE:
      CONTROL_RESPONSE_DISTRIBUTION
      EFFECT_ESTIMATE
      STANDARD_ERROR
      CONFIDENCE_INTERVAL

  IF:
    CONTROL_CONDITION_CHANGES

  THEN:
    REQUIRE:
      DOCUMENTED_PROTOCOL_CHANGE

    UPDATE:
      CONTROL_DEFINITION

    REASSESS:
      COMPARABILITY
      ESTIMAND
      ANALYSIS_PLAN

  IF:
    CONTAMINATION_OR_COINTERVENTION_DETECTED

  THEN:
    RECORD:
      EVENT
      TIMING
      AFFECTED_UNITS

    REASSESS:
      BIAS
      SENSITIVITY_ANALYSIS

  PRESERVE:
    GROUP_IDENTITIES
    ASSIGNMENT_HISTORY
    CONTROL_CONDITION
    TREATMENT_CONDITION
    RESPONSE_DEFINITIONS
    DATA_PROVENANCE

AI_CONSUMPTION:
  parse_as: FORMAL_SEMANTIC_SPECIFICATION
  execution_mode: RECURSIVE_REASONING

  preferred_reasoning_order:
    - IDENTIFY_EXPERIMENTAL_UNITS
    - IDENTIFY_TREATMENT_GROUP
    - IDENTIFY_CONTROL_GROUP
    - IDENTIFY_CONTROL_CONDITION
    - IDENTIFY_ASSIGNMENT_MECHANISM
    - VERIFY_COMPARABLE_MEASUREMENT
    - ASSESS_BASELINE_COMPARABILITY
    - ASSESS_CONTAMINATION_AND_COINTERVENTIONS
    - DEFINE_ESTIMAND
    - ESTIMATE_TREATMENT_CONTROL_CONTRAST
    - QUANTIFY_UNCERTAINTY
    - ASSESS_ATTRITION_AND_MISSINGNESS
    - RETURN_COMPARATIVE_INFERENCE

  constraints:
    - PRESERVE_CONTROL_CONDITION_DEFINITION
    - DISTINGUISH_CONTROL_GROUP_FROM_BASELINE
    - DISTINGUISH_RANDOMISATION_FROM_BLINDING
    - EXPOSE_ASSIGNMENT_MECHANISM
    - ASSESS_CONTAMINATION
    - DO_NOT_INFER_CAUSATION_WITHOUT_VALID_DESIGN_ASSUMPTIONS
    - PRESERVE_DATA_PROVENANCE

FLOW_HASH:
  canonical_entity: "control-group"
  namespace: "probability-statistics/experimental-statistics"
  version: "4.0"
  identity_basis:
    - PAGE_URL
    - CANONICAL_ENTITY
    - ENTITY_TYPE
    - CONTROL_CONDITION
    - TREATMENT_CONTROL_COMPARISON
    - REFERENCE_GROUP_STRUCTURE

END_FLOW