Presentation to "Rethinking Impact: Understanding theComplexity of Poverty and Change", 26 - 28 March 2008Cali, Colombia. It oulined four key tasks for impact evaluation and sets out some methods that can be used for each of these.
1. Four key tasks in impact assessment of complex interventions Rethinking Impact: Understanding the Complexity of Poverty and Change 26 - 28 March 2008 Cali, Colombia Professor Patricia Rogers Royal Melbourne Institute of Technology, Australia
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13. What works here and now? What do we mean by ‘works’? What works in what contexts? (implementation environments and participant characteristics) What works? COMPLEX COMPLICATED SIMPLE
14. Types of interventions Simple intervention Complicated or complex intervention Single causal strand. Intervention is sufficient to produce the impacts Multiple simultaneous causal strands required to produce the impacts Universal mechanism. Intervention is necessary to produce the impacts Different causal mechanisms operating in different contexts Linear causality, proportional impact Recursive, with feedback loops, leading to disproportionate impact at critical levels Agreement and certainty about pre-identified outcomes Disagreement and uncertainty about outcomes and emergent outcomes
15. Simple causal attribution intervention is both necessary and sufficient to produce the impact Impact Intervention
16. Causality in an INUS world Intervention is an Insufficient but Necessary part of a causal package which is itself Unnecessary (not the only way) but Sufficient to produce the impact Impact Intervention Other factor – context or contribution of another intervention Alternative intervention Mackie, 1974, Mark, 2001
17. Causal analysis of problems Predisposition Triggering events or conditions Impacts
18. Causal analysis of interventions Intervention Favourable context Impacts
19. Eg of developing testable hypotheses not involving a comparison group Change in rate of all road fatalities
21. Uptake requires further ongoing adaptation Complicated message Clear messages Reporting and supporting use Unique, highly contingent causality Causal packages and non-linearity Clear counter-factual likely Analysing cause Harder to plan for given emergence Evidence needed about multiple components More likely to have standardised measures developed Describing impacts May be emergent Likely to differ, reflecting different agendas Likely to be agreed Deciding impacts Complex Complicated Simple