3D impact analysis A new tool to approach impact evaluations April 23, 2015 Rob D. van den Berg Visiting Fellow, IDS CDI is a joint initiative between: and and For more information: www.ids.ac/cdi or email: cdi@ids.ac.uk 1 Overview • • • • • • • What is impact? What is evidence? What is causality? What is attribution/contribution? Time Space Scale 2 Impact • Impact is an ordinary word in the English language – “the effective action of one thing or person on another; the effect of such action; influence; impression” • Its meaning cannot be scientifically claimed • Demand for impact evidence can refer to a wide variety of effects, influences and impressions 3 Evidence • Evidence is an ordinary word in the English language – “the quality or condition of being evident; clearness; evidentness” • Its meaning cannot be scientifically claimed • Demand for impact evidence can refer to a wide variety of qualities or conditions 4 Causality • The word “cause” is an ordinary word in the English language – “A person or thing that gives rise to an action, phenomenon, or condition” • Its meaning cannot be scientifically claimed • Demand for evidence of cause and effect can refer to a wide variety of actions, phenomena and conditions 5 Attribution / Contribution • Both words are ordinary words in the English language, with great variety in meaning – Attribution: in copyright law, requiring an author to be credited; in marketing, assigning a value to a marketing activity based on desired outcome; journalism, practice of attributing information to its source – Contribution: donation, sharing, payment, publication, a song by Mica Paris 6 Impact Evaluation • Focus of Impact Evaluations: – Impact = evidence of causality between an intervention and the desired effect by establishing a counterfactual through controlled experimentation, which attributes part of the effect to the intervention • This partially meets the demand for impact evidence in politics, the media and society • So what to do with other demands? 7 Meeting impact demand • Broaden the concepts of impact and causality • Broaden the range of scientific methods and tools • Develop a framework for understanding demand for impact evidence • Incorporate issues of time, space and scale • This is urgent, given the adoption of the Sustainable Development Goals in September 2015 8 Understanding causality • Schaffer (2013) proposes two kinds of causality: “difference” and “production” – Difference: with/without (counterfactual) analysis – Production: A “produces” B (natural systems, physics & technology) • Concepts that include causality: – Catalytic roles (the change agent speeds up change but is not involved in the change itself) – Dynamic and chaotic systems (feedback loops, iterative processes, Fibonacci sequences) • Terry Pratchett: “hardly anything important has a single cause” 9 Issues of time, space and scale • Some changes can be observed immediately – others take decades – Short-term results: vaccinations, technology transfer, new livelihood approaches etc. – Either short- or long-term: market transformations, societal change – Long-term results: health trends, ecosystem services, ozone layer • Some changes are local, other regional, national or even global • Some changes concern one actor, intervention or institution, others involve multiple actors or institutions, and multiple sectors • Sustainable development involves longer time horizons, overlapping locations and many scales Matrix of evaluable impact • Impact can be evaluated at different moments in time: ex ante, in real time and ex post – These can be refined: ex ante tends to be done once, but real time and ex post have many possibilities – Different processes tend to have different time horizons • Geographical space runs from local to global – These can also be refined: the boundaries of societies, economies and natural systems are different from each other and may overlap • Scales of involvement can go from one actor to a multiplicity, from one market to a full economic system, from one governance level (or sector) to many – Actors, markets and governance may not fully overlap Appraisal Inception Implementation End-of-project 2 years ex-post 5-8 years ex-post 12 Local Regional National Global Multisector Multiactor One 13 Matrix dimensions space and time Ex ante Local Inception Mid-term End of project Ex post < 2 years Ex post 58 years Experimentation National Regional Global Ecosystem Monitoring and data analysis (including “big data”) Mixed methods / theory of change approaches (overlap with other rows) 14 Matrix dimensions space and scale One inter- Multiple vention interventions Enabling environment Market change Market transform -ation Climate change Local RCTs National Regional Global Mixed methods / theory of change Monitoring Data analysis Ecosystem (overlap with other rows) Double evaluand evaluations 15 Matrix dimensions scale and time One inter- Multiple vention interventions Enabling environment Market change Market transform -ation Climate change Ex-ante Inception Real-time End-ofproject RCTs and quasiexperimental Monitoring Data analysis Ex-post Mixed methods / theory of change 16 Counterfactual analysis One inter- Multiple vention interventions Local National Regional Global RCTs Enabling environment Quasiexperimental and QCA Social Network analysis Market change Market transform -ation Climate change Delphi Modelling of data and experimentation (quasi- and natural) Ecosystem (overlap with other rows) 17 Production causality One inter- Multiple vention interventions Enabling environment Market change Market transform -ation Climate change Local National Regional Inspection, validation before/after data Systems evaluation Global Ecosystem (overlap with other rows) Verification of data, trend analysis 18 Thank you! rdwinterberg@gmail.com For more information: www.ids.ac/cdi or email: cdi@ids.ac.uk 19