IATI Consultations Archive

Live discussions and consultations can be found at discuss.iatistandard.org.

Support disaggregation of performance data

Problem: There is currently no way to disaggregate results data -- by gender, by age, by geography, or any other way. 

Proposed solution: Allow more than one <target> and <actual> element under <period>. Support <location> and/or <dimension> elements under the <target> and <actual> elements. 

Details:

The <location> element would contain a reference to a geography already defined for the activity. 

<actual value="110">
  <location ref="Clinic-001" />
</actual>

 

The <dimension> element would have two attributes: @name and @value. For example:

<period>
  <actual value="100">
    <dimension name="sex" value="male" />
  </actual>
  <actual value="110">
    <dimension name="sex" value="female" />
  </actual>
</period>

More than one dimension can be included in an actual or target element:

<actual value="110">
  <dimension name="sex" value="female" />
  <dimension name="age value="adult" />
</actual>

Location and other dimensions can be combined:

<actual value="110">
  <location ref="Clinic-001" />
  <dimension name="age value="adult" />
</actual>

Targets can be disaggregated in the same way:

<period>
  <target value="100">
    <dimension name="sex" value="male" />
  </target>
  <target value="110">
    <dimension name="sex" value="female" />
  </target>
  <actual>
    ...
  </actual>
  ...
</period>

None of these elements is required, so existing documents would still be valid. 

 

Have more questions? Submit a request

16 Comments

  • 0
    Avatar
    Catherine Marschner

    Regarding this schema proposal, I agree this would cover the need with one point - it should retain all the fields currently in the schema now, allowing publishers to continue publishing as they have been, and just add an extension that allows disaggregations if available.

    MCC is very keen to see a way to report disaggregated indicators as well, and I'm hoping this can be discussed at the upcoming TAG.  I am particularly interested in disaggregating by gender and would love to pilot publishing some disaggregated results data if we can agree on an extension.

     

  • 0
    Avatar
    IATI Tech Team

    This item has been moved to the '2.02 Decimal Upgrade Proposals' forum for inclusion the upcoming decimal upgrade.

  • 0
    Avatar
    Bill Anderson

    I am fully behind getting a disaggregation solution into 2.02.

    Two questions:

    1. Could we not represent location, dimension-name and dimension-value as attributes of value rather than sub-elements?
    2. Do we need vocabularies and codes for dimensions (or are we happy doing arithmetic using free-text filters?)

    Location should be represented as value/@location-ref which should be described in the same activity by an instance of location/@ref.

  • 0
    Avatar
    Herb Caudill

    Thanks, Bill. 

    1. I've proposed having dimensions as sub-elements so that you can have more than one. See the third example above, where a datum applies to `age: adult` and `sex: female`. 
    2. The only logical place to define dimension vocabularies would be as part of the indicator repository schema. There are very, very few useful external vocabularies. See http://blog.devresults.com/proposed-indicator-definition-schema/#disaggregation for more thoughts on this.
    3. Location & activity values: agreed.
  • 0
    Avatar
    Bill Anderson

    Attributes vs. sub-elements

    Could we not have specific attributes for age (which needs a vocab anyway), gender and location plus an additional variable dimension?

  • 0
    Avatar
    Herb Caudill

    We can't come up with a universal age vocab. In the real world there are infinite ways to put ages into buckets, and legitimately so. Depending on what you're measuring you may be interested on whether someone is a newborn vs an older infant; or an adolescent vs a young adult; or of child-bearing age vs not; etc. 

    You could treat sex as a "special" disaggregation, but I wouldn't. For one thing, even on something as simple as gender you're going to have different ways of labeling male vs female, not to mention vocabularies with three or more options. So I think it's best to leave this to each indicator repository to define. 

    Location and activities disaggregations are "special", however, and we already know how to refer to these characteristics unambiguously. Agreed on that point. 

    In practice we very often see data being disaggregated by more than 2 dimensions. So I would stick with the original proposal of supporting n generic dimensions, defined globally for each repository. 

     

     

     

     

  • 0
    Avatar
    IATI Tech Team

    This proposal has been discussed amongst the IATI Technical Team following the end of the initial suggestion phase of the v2.02 upgrade.

    The IATI Technical Team is happy progress this proposal however more discussion and consensus on its practical implementation is needed before a final decision to proceed can be made.  We invite further contributions before 7 October.

  • 0
    Avatar
    Sarah Johns

    Agree with this proposal - this should be viewed as a 'live' element and developed in tandem with discussions around vocab development and disaggregation of data used in monitoring the SDGs. As this is likely to require this element to flexible, I'd agree with Herb's idea of a dimension sub-element as it gives the most flexibility. The objective is NOT to aggregate up but to surface hitherto 'hidden' communities - women, children, people with disabilities etc. 

  • 0
    Avatar
    Rolf Kleef

    I assume multiple dimensions and locations should be combined as "AND"?

    <actual value="110">
      <dimension name="sex" value="female" />
      <dimension name="age value="adult" />
    </actual>

    means: "110 adult females"?

    (The dimension element also offers a better wat to indicate whether data can be aggregated, to replace the current attribute of the result element...)

  • 0
    Avatar
    Herb Caudill

    Rolf - yes, that's correct. 

  • 0
    Avatar
    Catherine Marschner

    My understanding is that MCC could use this extension, even without including it in an upgrade, since it doesn't change the original schema, it just extends it.  I had a number of conversations in Ottawa on this topic and had left with the impression that IATI was moving towards allowing donors to pilot these sorts of extensions, and after they had been in use for awhile, consider them for inclusion in a subsequent update.  Based on this understanding, I was planning on piloting the proposed extension this fall to add gender disaggregations to a number of our indicators.  Let me know if this is still a workable path forward!

  • 0
    Avatar
    IATI Tech Team

    Please see below for the technical implementation that is suggested in the Version 2.02 - Formal Proposal:

    New element: result/indicator/target/location

    • Add new element to the activity standard: location

    • Example

      • Proposed:
        <target value="110">

         <location ref="AF-KAN" />

         <location ref="KH-ABC" />

        </target>

    • Add definition
      Location: A location already defined and described in the iati-activity/location element
      Ref: A cross-reference to the internal reference assigned to a defined location: iati-activity/location/@ref

     


    New element: result/indicator/target/dimension

    • Add new element to the activity standard: dimension

    • Example

      • Proposed
        <target value="110">
         <dimension name="sex" value="female" />
         <dimension name="age value="adult" />
        </target>

    • Add definition
      Dimension: ‘A category used for disaggregating the result by gender, age, etc’
      Name: ‘Freetext description of a category being disaggregation’
      Value: ‘Description of the value being disaggregated.’

     

    New element: result/indicator/actual/location

    • Add new element to the activity standard: location

    • Example

      • Proposed:
        <actual value="110">
         <location ref="AF-KAN" />
         <location ref="KH-ABC" />
        </actual>

    • Add definition
      Location: A location already defined and described in the iati-activity/location element
      Ref: A cross-reference to the internal reference assigned to a defined location: iati-activity/location/@ref

     


    New element: result/indicator/actual/dimension

    • Add new element to the activity standard: dimension

    • Example

      • Proposed
        <actual value="110">
         <dimension name="sex" value="female" />
         <dimension name="age value="adult" />
        </actual>

    • Add definition
      Dimension: ‘A category used for disaggregating the result by gender, age, etc’
      Name: ‘Freetext description of a category being disaggregation’
      Value: ‘Description of the value being disaggregated.’

  • 0
    Avatar
    Herb Caudill

    Looks good!

    One small note: The same changes should apply to the <target> element as well as the <actual> element. So for example

    <target value="100">
        <dimension name="sex" value="female" />
        <dimension name="age value="adult" />
    </target>

     

     

  • 0
    Avatar
    IATI Tech Team

    In light of the above comment, the following elements have been added to the planned technical implementation, as detailed above (post created on 'October 12, 2015, 15:40'):

    • result/indicator/target/location

    • result/indicator/target/dimension

    The formal proposal and proposed v2.02 summary table spreadsheet have also both been updated in this regard.

  • 0
    Avatar
    IATI Tech Team

    This proposal has been added as a GitHub issue, for inclusion in the version 2.02 development branch of the IATI-Schemas: https://github.com/IATI/IATI-Schemas/issues/282

  • 0
    Avatar
    IATI Tech Team

    This proposal has now been incorporated into a version 2.02 code repositories (see the above GitHub issue link for technical details). The relevant pages on a development version of the 2.02 reference and documentation site is available to view here:

    We welcome scrutiny on the implementation of this proposal and encourage the community to feedback and suggest solutions for any errors, omissions and problems. This should be done before Monday 7 December, when the process will commence to release version 2.02 as a live version of the IATI Standard. More information on the inspection phase is available here.

Please sign in to leave a comment.