"Big data can contribute to the evidence base in development sectors where evaluations are often infeasible due to data issues. Given the rapidly increasing availability of big data and improving computation capacity, there is a great potential for using big data in future impact evaluations. Big da ... ta can also contribute to evaluations through providing new ways to identify control groups and establish counterfactuals, and can strengthen the analysis with data on pre-programme trends, covariates, and sub-groups, as well as enabling better robustness analyses. However, there are several analytical, ethical, and logistical challenges that may hinder the use of big data in impact evaluations. Standards should be set for the reporting of data quality issues, data representativeness, and data transparency. More interaction is needed between big data analysts, remote sensing scientists, and evaluators." (Conclusion) more

"Many systematic reviews are solely concerned with effectiveness or impact. While a review which tells you what works can help you decide what to do, it is of less use in telling you how to do it. Causal chain analysis-based systematic reviews, which analyse the working of a logic model or a theory ... of change for an intervention, can give useful information on programme design and implementation. Causal chain analysis is not yet common in systematic reviews. This brief lays out what causal chain analysis is, the benefits of using it, and how to do so. The causal chain analysis approach is based on specifying the logic model for an intervention. A logic model provides the basis for the questions to be answered in the systematic review, the types of studies to be reviewed, coding forms, and analysis. The causal chain analysis identifies weak and missing links in the causal chain, and thus which assumptions in the logic model may not hold. Programme designers and implementers can learn from these lessons to achieve better development outcomes." (Box 1: Highlights, page 2) more

"The framework for selecting appropriate methods of stakeholder engagement is presented here as five steps that can help people who are commissioning or conducting research or evaluations to orient themselves to their context, research purposes and, ultimately, options for stakeholder engagement. Th ... e first steps include understanding the value of generalisable evidence (Step 1) and localised evidence (Step 2) for informing decisions. Step 3 recognises the socio-political implications of these different ways of thinking. Step 4 helps researchers identify a starting point by utilising a matrix that signposts various tools and methods. Choosing a starting point depends largely on the following: whether the research findings are for local or general application: Does the knowledge need to be generalisable to many different settings, or is knowledge to suit the local setting sufficient? And: How much clarity and consensus is assumed about what is known when starting out: Is prior knowledge that a study will build on clear and widely agreed before the work begins? Step 5 considers which stakeholders to engage with, and how, when planning and conducting research, depending on the circumstances." (Page 3) more

"Middle-level theory (MLT) has several uses in development planning and evaluation. It helps predict whether a programme can be expected to work in a new setting. It offers insights into what design features are needed for success. IIt provides invaluable information for monitoring to see if the pro ... gramme is on track and to fix problems that arise. It reveals the causal processes and related assumptions to be tested in an evaluation and helps identify evaluation questions. It helps in interpreting evaluation findings, assessing their relevance and locating a description of them that is useful for programme design and evaluation in other settings. This CEDIL methods brief outlines ten steps for building and using a middle-level ‘causal-processtracing’ theory of change to serve these purposes." (Page 1) more