Metrics in the CGIAR – reviewing Panel Report and next steps - Jeffrey Sayer
1.
2. Definitions
• Data are observations, such as weight, height, plot size.
• Metrics are computed by aggregating and combining
raw data, for example, yield, or height for age. They often
represent the values on which indicators are built.
• Indicators are summary measures that reflect system
properties. Examples include infant mortality rates and
the prevalence of acute malnutrition or changes in these
values.
3. Methods
• Diverse panel – 2 face to face meetings
• Questionnaire – workshop
• Review of CRPs
• Contact persons in CRPs
• Review of what others are doing
• Study of some examples of metrics use
4. Concerns
• CGIAR has made valuable contributions – Gene
banks – commodity crops – IFPRI models
• Leading specialists on DMI
• But overall patchy – no strategy, poor archiving
and curation – low accessibility – lack of skills
• Proliferation of recent initiatives – some naïve
5. ISSUES
Purpose – accountability vs learning
Research resource - understanding changes
Data management – ground rules in the open access
policy
Impact assessment
Global assessment and monitoring – SDGs etc.
Alignment – with others
Resources for basic metrics – measuring a few things
well
Rationalising CGIAR efforts
Key scientific issues
6. Action for CO
• Normative role for Consortium
• Peer review of data – quality control
• Ontology – Consistency in use of terms – i.e.
“sentinel site”
• Community of practice
• Fund allocation
7. ISPC Commentary
• Provision of comprehensive, accessible high-quality data
and metrics on agricultural systems should a major public
goods product of the CGIAR;
• Need for both a learning focus and accountability in a
data, metrics and indicators system.
• Indicator targets for CRPs must use consistent vocabulary
to maximise comparability beyond Open Access Policy
• ISPC advises caution against establishing a centralized system
8. • special attention to aggregation/disaggregation of data
collected at different spatial and temporal scales, to allow
analysis of trade-offs and interactions across system
components
• More is not better – get the basics right