Presentation on theme: "Mywish K. Maredia Michigan State University"— Presentation transcript:
1Designing Impact Evaluations: What are the Appropriate Research Questions and Methods? Mywish K. MarediaMichigan State UniversityWorkshop for Managers of Impact EvaluationMay 13, 2013InterAction, th St. NW, Suite 210Washington, DC
3Your Role in Explaining the ‘Miracle’ As Impact Evaluation Managers, your role is to ensure that a plan is in place to bridge the knowledge gap between how project outputs result in impacts; andDoing so based on evidence generated using credible methodology
4Focus of this presentation Discuss the development of appropriate impact evaluation questionsWhat types of impact evaluation designs are most appropriate in different contexts and given the evaluation questions?Purpose:To share some preliminary thoughts; present a framework; andTo facilitate discussion and exchange of ideas / experience
5Clarifying the term: Impact evaluation What it is:It is concerned with establishing a causal link between realized impacts (the effect) and an intervention (the ‘cause’) which could be a program, activity, policy change, etc.The goal of the analysis is to ‘rule out’ other possibilities / explanations for the observed effects
6What do we mean by evaluation design? Every evaluation is essentially a research or a discovery project/activityIf your results are to be reliable, you have to give the evaluation a structure that will tell you what you want to knowThat structure – the arrangement of discovery- is the evaluation’s designThe design depends on what kinds of questions your evaluation is meant to answer
7Development of Impact Evaluation Questions Characteristics of ‘appropriate’ IE questions:They should be narrow/specificFocus on small number of questions (~5)Focused on ‘summative’ evaluation of a project/interventionThey should reflect the input of program staff and sponsors
8Examples of common impact evaluation (research) questions Overall impact (effectiveness)Did it work? Did the intervention produce the intended impacts in the short, medium and long term?For whom, in what ways and in what circumstances did the intervention work?What unintended impacts (positive and negative) did the intervention produce?Source: Rogers (2012) Introduction to Impact Evaluation. Impact Evaluation Notes No.1
9Examples of common impact evaluation (research) questions (cont’d) Nature of impacts and their distributionAre impacts likely to be sustainable?Did these impacts reach all intended beneficiaries?Influence of other factors on the impactsHow did the intervention work in conjunction with other interventions, programs or services to achieve outcomes?What helped or hindered the intervention to achieve these impacts?
10Examples of common impact evaluation (research) questions (cont’d) How it worksHow did the intervention contribute to intended impacts?What were the particular features of the intervention that made a difference?To what extent are differences in impact explained by variations in implementation?Matching intended impacts to needsTo what extent did the impacts match the needs of the intended beneficiaries?
11Common impact evaluation designs (focused on causal analysis) Methods for examining the factual. For e.g.:Comparative case studiesBeneficiary/expert attributionMethods for creating counterfactualExperimental designs or RCTs (based on the principle of random assignment)Pipeline comparisonsOther methods/approaches (using statistical techniques to form credible comparison groups). For e.g.,:Propensity score matching (PSM)Instrumental variables (IV)Regression discontinuity (RD)Difference in difference (DD)
12IE MethodsIn theory, there are multiplicity of methods and approaches that can be used to assess impactsEach have problems and limitationsThere is no ‘one size fits all’ method/approach
13Choosing an appropriate impact evaluation design Depends on… 1. The nature of the research questionsResearch QuestionsAppropriate MethodsWhat is the effectiveness of a programObservational and correlational methodsWhether observed effects can reasonably be attributed to the intervention and not to other sourcesExperimental and quasi-experimental methodsWhat is the net impact of the programCost effectiveness; cost-benefit analysis with qualitative methods to summarize the full range of impacts
14Choosing an appropriate impact evaluation design (cont’d) Depends on… 2. The nature of your programNature of your programMethods to considerWill you roll out your program over time?Pipeline designWhat is the unit of intervention--individuals, groups, communities?Those that give enough statistical power based on the number of ‘units of observations’Is program assigned to participants or do they self-select?RCT, quasi-experimentalThere is no credible reason for other influencing factors (e.g., water pump)Before/after comparison
15Choosing an appropriate impact evaluation design (cont’d) Depends on… 3. what participants / stakeholders will consent to? 4. What are your resources and time constraints?
16Practical Considerations in Designing Impact Evaluation Establishing the program theory (the logic behind a program and the causal chain from inputs to outcome to impact)Understanding the program settingHow participants are selected (to mitigate selection bias)Decision tree on which method is applicable and should be explored
17Practical Considerations in Designing Impact Evaluation (Cont’d) Sample sizePower calculation – Does the setting allow for enough numbers of units of intervention and units of observation for a robust design?Tradeoff between power and costTime frameIs there enough time to observe the impact?FlexibilityStrive for rigor, but be flexible…