Skip to content

Class: Effect

The inferential quantity tested about a map: what was compared, whether the comparison is a mediated path, and what statistic tested it. Estimation facts -- model family, estimator, stage, software -- belong to the ModelEstimation that produced the map.

The comparison is cells: one entry per level of a model term that entered the contrast, each with the side it sat on. Three things a reader might expect to find here are read off that instead:

- what the contrast was adjusted for -- the terms of the ModelEstimation, plus those

_ of the stages it reaches through inputs_from, minus the terms these cells name._

- what the map was tested against -- a modelled reference is that condition's level

_ with a negative cell; the unmodelled implicit baseline, or zero, is the absence of a_

_ second cell; a decoding metric's reference is PerformanceMetric.reference_value._

URI: neurostudy:Effect

 classDiagram
    class Effect
    click Effect href "../Effect/"
      Effect : cells





        Effect --> "1..*" Cell : cells
        click Cell href "../Cell/"



      Effect : kind





        Effect --> "0..1" EffectKind : kind
        click EffectKind href "../EffectKind/"



      Effect : mediation





        Effect --> "0..1" Mediation : mediation
        click Mediation href "../Mediation/"



      Effect : statistic





        Effect --> "1" Statistic : statistic
        click Statistic href "../Statistic/"

Slots

Name Cardinality and Range Description Inheritance
cells 1..*
Cell		 What this contrast compared: one entry per level of a model term that entered... direct
kind 0..1
EffectKind		 What kind of test these cells describe, stated as well as encoded direct
mediation 0..1
Mediation		 The mediation model this effect's quantity is a path through direct
statistic 1
Statistic		 The statistic this effect was tested with, and its degrees of freedom direct

Usages

used by used in type used
Analysis effect range Effect

Rules

Rule Applied Preconditions Postconditions Elseconditions
slot_conditions {'cells': {'value_presence': 'PRESENT'}}

Identifier and Mapping Information

Schema Source

  • from schema: https://neurostuff.org/schemas/neuroimaging-study-storage

Mappings

Mapping Type Mapped Value
self neurostudy:Effect
native neurostudy:Effect

LinkML Source

Direct

name: Effect
description: "The inferential quantity tested about a map: what was compared, whether\
  \ the comparison is a mediated path, and what statistic tested it. Estimation facts\
  \ -- model family, estimator, stage, software -- belong to the ModelEstimation that\
  \ produced the map.\nThe comparison is `cells`: one entry per level of a model term\
  \ that entered the contrast, each with the side it sat on. Three things a reader\
  \ might expect to find here are read off that instead:\n- **what the contrast was\
  \ adjusted for** -- the terms of the `ModelEstimation`, plus those\n  of the stages\
  \ it reaches through `inputs_from`, minus the terms these cells name.\n- **what\
  \ the map was tested against** -- a modelled reference is that condition's level\n\
  \  with a `negative` cell; the unmodelled implicit baseline, or zero, is the absence\
  \ of a\n  second cell; a decoding metric's reference is `PerformanceMetric.reference_value`."
from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage
rank: 1000
attributes:
  cells:
    name: cells
    description: 'What this contrast compared: one entry per level of a model term
      that entered it, each with the side it sat on. The only place direction lives.

      Every cell names its term, so the axis of a comparison is explicit. Two cells
      naming one term compare levels within that factor; cells naming different terms
      compare across them. A condition contrast, a cohort comparison, a pre-post change,
      a crossover comparison of arms and a region-by-condition dissociation all take
      this one form, and what distinguishes them is what the named term''s levels
      range over, which `FactorLevel` states on the model.

      Required and non-empty. A one-sample test is one signed cell: a lone `positive`
      cell tests that coefficient against zero, which is what the unmodelled implicit
      baseline is, and the sign distinguishes activation from deactivation. A map
      with no inferential test at all is not an Analysis.'
    in_subset:
    - mvp
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - Effect
    range: Cell
    required: true
    multivalued: true
    inlined_as_list: true
  kind:
    name: kind
    description: 'What kind of test these cells describe, stated as well as encoded.

      Redundant on purpose: the kind follows entirely from the cells by the ordered
      steps of extraction-readme.md §3, so a `kind` that disagrees with the cells
      beside it is a record contradicting itself. Neither half is authoritative --
      a stated `contrast` over cells that derive `simple_effect` may be cells missing
      their second level, or a kind read off the analysis''s title rather than its
      test. The pair says a reader should look, which is more than either half says
      alone.

      The disagreement it most often catches is a decoding result encoded as a comparison:
      accuracy above chance is one signed cell and a `simple_effect` (§5.9), while
      two directionless cells derive `contrast` with its direction lost.'
    in_subset:
    - mvp
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - Effect
    - Correction
    range: EffectKind
  mediation:
    name: mediation
    description: The mediation model this effect's quantity is a path through. Present
      only for a mediation analysis and absent otherwise, so an effect carries either
      a complete Mediation or none.
    in_subset:
    - mvp
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - Effect
    range: Mediation
    inlined: true
  statistic:
    name: statistic
    description: The statistic this effect was tested with, and its degrees of freedom.
      Not the value of the statistic at any location -- peak magnitudes belong to
      the reported coordinates.
    in_subset:
    - mvp
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - Effect
    range: Statistic
    required: true
    inlined: true
rules:
- postconditions:
    slot_conditions:
      cells:
        name: cells
        value_presence: PRESENT
  description: An effect that compared nothing tested nothing; cells cannot be empty.

Induced

name: Effect
description: "The inferential quantity tested about a map: what was compared, whether\
  \ the comparison is a mediated path, and what statistic tested it. Estimation facts\
  \ -- model family, estimator, stage, software -- belong to the ModelEstimation that\
  \ produced the map.\nThe comparison is `cells`: one entry per level of a model term\
  \ that entered the contrast, each with the side it sat on. Three things a reader\
  \ might expect to find here are read off that instead:\n- **what the contrast was\
  \ adjusted for** -- the terms of the `ModelEstimation`, plus those\n  of the stages\
  \ it reaches through `inputs_from`, minus the terms these cells name.\n- **what\
  \ the map was tested against** -- a modelled reference is that condition's level\n\
  \  with a `negative` cell; the unmodelled implicit baseline, or zero, is the absence\
  \ of a\n  second cell; a decoding metric's reference is `PerformanceMetric.reference_value`."
from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage
rank: 1000
attributes:
  cells:
    name: cells
    description: 'What this contrast compared: one entry per level of a model term
      that entered it, each with the side it sat on. The only place direction lives.

      Every cell names its term, so the axis of a comparison is explicit. Two cells
      naming one term compare levels within that factor; cells naming different terms
      compare across them. A condition contrast, a cohort comparison, a pre-post change,
      a crossover comparison of arms and a region-by-condition dissociation all take
      this one form, and what distinguishes them is what the named term''s levels
      range over, which `FactorLevel` states on the model.

      Required and non-empty. A one-sample test is one signed cell: a lone `positive`
      cell tests that coefficient against zero, which is what the unmodelled implicit
      baseline is, and the sign distinguishes activation from deactivation. A map
      with no inferential test at all is not an Analysis.'
    in_subset:
    - mvp
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: Effect
    domain_of:
    - Effect
    range: Cell
    required: true
    multivalued: true
    inlined: true
    inlined_as_list: true
  kind:
    name: kind
    description: 'What kind of test these cells describe, stated as well as encoded.

      Redundant on purpose: the kind follows entirely from the cells by the ordered
      steps of extraction-readme.md §3, so a `kind` that disagrees with the cells
      beside it is a record contradicting itself. Neither half is authoritative --
      a stated `contrast` over cells that derive `simple_effect` may be cells missing
      their second level, or a kind read off the analysis''s title rather than its
      test. The pair says a reader should look, which is more than either half says
      alone.

      The disagreement it most often catches is a decoding result encoded as a comparison:
      accuracy above chance is one signed cell and a `simple_effect` (§5.9), while
      two directionless cells derive `contrast` with its direction lost.'
    in_subset:
    - mvp
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: Effect
    domain_of:
    - Effect
    - Correction
    range: EffectKind
  mediation:
    name: mediation
    description: The mediation model this effect's quantity is a path through. Present
      only for a mediation analysis and absent otherwise, so an effect carries either
      a complete Mediation or none.
    in_subset:
    - mvp
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: Effect
    domain_of:
    - Effect
    range: Mediation
    inlined: true
  statistic:
    name: statistic
    description: The statistic this effect was tested with, and its degrees of freedom.
      Not the value of the statistic at any location -- peak magnitudes belong to
      the reported coordinates.
    in_subset:
    - mvp
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: Effect
    domain_of:
    - Effect
    range: Statistic
    required: true
    inlined: true
rules:
- postconditions:
    slot_conditions:
      cells:
        name: cells
        value_presence: PRESENT
  description: An effect that compared nothing tested nothing; cells cannot be empty.