Upload
toby-simpson
View
214
Download
2
Embed Size (px)
Citation preview
Designing Influential EvaluationsSession 4Approaches, Methods & Tools
Uganda Evaluation Week - Pre-Conference Workshop19th and 20th May 2014
2
Questions
Theory/Approach
Methods
Tools
Sequence of planning
Four contrasting key questions
To what extent can a specific (net) impact be attributed to the intervention?• Conditions that suit experiments and statistical models
Did the intervention make a difference?
• Contributory causes and causal packages
How has the intervention made a difference?
• Explanation and importance of theory
Will the intervention work elsewhere?
• External validity, transferability and generalisation
3
4
THEORIES & APPROACHES
5
Why are theories and approaches important? Different approaches to evaluation help to:
define the evaluation purpose and process determine different levels of participation draw boundaries for role of the evaluator
Theories and approaches: are linked to a specific philosophical orientation, design
and methodology stress the importance of socio-political and contextual
factors have considerable overlap have strengths and weaknesses
c. Helen Simons and Georgie Parry-Crooke
6
Evaluation theories give guidance
The nature of what we evaluate
How evaluators
should practice in the
real world
How to assign value to
programmes and their
performance
How to construct
knowledge
How to use knowledge
generated by evaluation
… “offer a set of rules,
prescriptions, prohibitions, and
guiding frameworks that specify what a good or proper evaluation is
and how evaluations
should be done”(Alkin
2004)
7
Theories and approaches: a personal selection
Utilization
focused
Results basedTheory
based
Democratically oriented
Realist
Empowerment
Participatory
Goal oriented
Case study
Responsive
Meta-evaluatio
n
Experiment-alism
Value for
moneyCost
benefit analysis
8
Results based evaluation Public management agenda Assessment against planned targets Helps establish key goals and
outcomes Permits managers to identify and take
action to correct weaknesses Supports a development agenda that
is shifting towards greater accountability
Participatory evaluation
Joint enterprise – engages participants at all stages of planning and implementation.
Akin to Empowerment Evaluation (different claims) Reduces threat of evaluation – transparent
process Recognizes value of all participants’ contributions Responsive to specific cultural/political context Shifts focus of who does evaluation and how - collaborative
roles/responsibilities Promotes understanding between partners Increases evaluation capacity and use - involving those
affected by outcome Close links to utilization-focused
9
10
Theory-based evaluationIncreasingly common for development evaluationClose links to managing for results agendaExamines how a programme is expected to workCentral to many donors’ evaluation policyBrings together elements of:
◦ social science theory the principles that shape social behaviour. e.g. social
cognitive learning theory; theory of health behaviour change; etc
◦ programme theory the assumptions that guide the way specific programmes,
treatments or interventions are implemented and expected to bring about change
Helps identify performance dimensions most critical to a programme’s success
11
Discussion Exercise
Consider two contrasting approaches: Participatory and Results-based Evaluation.
Working in small groups try to identify settings in which one or the other would be the preferable approach to use.
Set out your thinking with notes about strengths and weaknesses of each approach on a flipchart.
12
Discussion exercise
Work in groups of two or three and discuss the following question.
Why have theories; what benefits do they bring to evaluators?
List your ideas for discussion in plenary.
13
Discussion feedback - Why have theories?Common language among evaluatorsUnique knowledge baseDistinct identityReflects underlying concernsFacilitate communication among evaluatorsHelp understand and share good practicesProvide a rationale for evaluation proceduresHelp fit needs of stakeholdersChoice for appropriateness in different
settings
© Derek Poate, Helen Simons and Georgie Parry-Crooke
14
SELECTING METHODS
What is ‘Quality’ & ‘Rigour’?
15(c) Eliot Stern for UKES
Is it biased?
Is it precise?
Are methods sound?
Are they well-used?
Can I trust findings?
All evaluation and research faces quality challenges:
Why definitions matter:Common definitions of “impact”
Positive and negative, primary and secondary
long-term effects produced by a
development intervention, directly or indirectly,
intended or unintended (OECD/DAC)
The difference in the indicator of interest with the intervention and without the
intervention... tackling attribution by rigorously
identifying a counterfactual value (e.g. 3ie / J-PAL)
•Search for any effect•Longer-term change•Effects may be +ve and –ve•Effects are somehow ‘produced’ (a weaker notion of causality)
•Search for a given effect•Tends to be methods-led•Emphasis on attribution (direct link between cause and effect)•Reliance on counterfactual logic 16Sources: Stern et al (2012: 6); OECD-DAC (2002); White (2010).
Broadening Impact Evaluation
“Impact evaluations are evaluations that assess the contribution of an intervention towards some outcome or
goal.
The contribution may be intended or unintended, positive or negative, long-term or short-term.
Impact evaluations attempt to identify a clear link
between causes and effects, and explain how the intervention worked and for whom”
Different approaches to causal inference
Power relations – who counts?
Explanations – mixed methods 17
Appropriate DesignFactors to consider:Evaluation purpose : Who will use the
evaluation, and what for? Evaluation Questions: What needs to be
measured? Is it ‘net effect’? Is it ‘how the impact occurred’?
Attributes of the intervention (+context): Small-N interventions/ long term and intangible effects/etc
‘Right Rigour’: how much certainty is needed?
Resources available, and Time constraints
18Sources: Stern et al (2012); Clemens & Demombynes (2013); White & Philips (2012).
Impact Evaluation Design Options
19Sources: Mayne (2013); Stern et al (2012).
Counterfactual frameworks that depend on the difference between two otherwise identical cases(Experimental and Quasi-
experimental designs)
Regularity frameworks that depend on the frequency of association between cause
and effect(Statistical approaches)
Comparative frameworks that depend on combinations
of causes that lead to an effect
(Case-based’ approaches, simulations and network
analysis)
Generative frameworks that depend on identifying
the causal links and mechanisms that explain
effects(Theory-based approaches)
Design Options for assessing causationNon-experimental
designs
Two defining features of ‘Impact Evaluation’Impact Evaluation =1. The need to demonstrate some causal effect
(i.e. that the intervention caused an effect or impact to happen).
2. The need to provide explanatory analysis by answering ‘how’ and ‘why’ this effect came about – as well as for whom.
Impact Evaluation is NOT the same as Process Evaluation – which considers a broader range of questions on programme relevance, the attainment of objectives, plus the quality and efficiency of implementation.
20
DATA COLLECTION
22
Case study
RCT/ quasi
experiment
Choices in data collection
Purposive or
random samplin
g Economi
c modellin
g
Document
reviewParticipator
y/ qualitative/
FGD
Direct measure-
ment
Individual/ group key informant interviews
Questionnaire survey
QUANT vs. QUAL designs
Strengths of QUANT designs Strengths of QUAL designs
• Statistical control of selection bias
• Statistical confidence when generalising from a sample to pop.
• Quantification (numbers) for outcomes and impacts
• Ability to replicate data collection and analysis
• Accepted standards for sampling, data collection and analysis.
But common weaknesses:• Difficulty capturing sensitive
inform-ation, and hard-to-reach groups
• Data reduction losses information
• No analysis of context
• Examines the broader context within which a programme operates
• Flexibility to evolve, and provide a more holistic view
• Multiple sources provide complex understanding and interpretation
• Ability to capture views on sensitive issues, and from the marginalised
• Accessible to non-specialistsBut common weaknesses:• Lack of generalizability• Hard to reach consensus with
multiple views• Subjective, and prone to bias
23Sources: Bamberger (2013).
Design optionsFramework Inference Methods
Counterfactual Difference between identical or matched cases
RCTDifference in differencePropensity score matching
Generative (theory-based)
Identify and confirm causal processes
Theory of changeContribution analysisRealist evaluationCongruence analysis
Comparative (case based)
Comparison and across and within cases of causal factors
Grounded theoryEthnographyQCA
Participatory Validation by participants of effect caused by interventions
Participatory or democratic evaluationLearning by doingMost significant change
Evening exerciseRead the handout texts dealing
with arguments for and against the use of RCTs and quasi-experimental designs.
Prepare for a short debate to argue the case for and against use of experimental approaches in evaluation. You will be asked to argue for a specific point of view for the debate.
SummaryTheories and approaches: guide generic
orientation that influences the choice of methods and the spirit in which methods and tools and are selected and used.
Methods: a way of carrying out an evaluation or a type of evaluation (using one or more tools) in line with an evaluation approach (research design).
Tools: describe standard techniques for collecting and analysing data.
Evaluation design needs to deal explicitly with how questions of attribution will be dealt with and the nature of causal inference.
END