Get the most out of the Data
To ensure our clients get the most out of the data we offer experienced and professional facilitation on data analysis and use. This includes answers and guidance on how the information can be used, where the data will be housed, how the data can be presented, and how lessons and learning can be embedded into existing systems and operations.
Key components include:
1. Credible Impact Assessment
A COSA Impact Assessment offers an understanding of the intended and unintended longer-term effects (both positive and negative) that can be attributed to speciﬁc interventions or investments. It balances state-of-the-art scientific methods with business-like pragmatism.
A COSA impact assessment is distinguished by actively looking beyond single dimensions to include the environmental, social, and economic manifestations of change so as to better understand the dynamics of farming and supply chains.
Scientific credibility starts with gathering the right data. Only then can the data be mined with a broad array of thoughtful diagnostic and analytic tools. We go beyond simply knowing something to help identify attribution and the reasons for an outcome. Knowing how interventions such as training, certification, or credit affect an impact opens up practical solutions and more effective investments or policies.
Sustainability is intrinsically complex and so its analysis must successfully engage diverse credible approaches. Essentially this means having different tools for different situations. A sampling of the methods includes:
- Difference in Differences (DID) compares, using a simple linear model, the difference between the groups at baseline with the difference realized between the control and target after the intervention. Using this control group as a comparison at baseline helps control for differences between groups and helps mitigate the impact of how variability in conditions (independent of those caused by the intervention) may affect many of the observed changes. This is especially the case in agriculture, where yields (for example) can be significantly affected by local phenomena that can vary substantially from year to year.
- Propensity Score Matching (PSM) is a statistical matching technique used to more accurately compare groups by estimating the effect of a policy or intervention (treatment) by accounting for factors that may predict receiving it and could affect indicator performance. PSM helps address self-selection bias wherein producers choosing certification may be intrinsically different from producers who do not (e.g. they may be more entrepreneurial, higher-yielding, or have more access to credit).
- Instrumental Variables (IV) analysis is used to estimate a relationship when there is simultaneity (when the casual direction of a treatment is not immediately obvious rendering the results of simple regression analysis biased and inconsistent). This method “replaces” the endogenous variable (the treatment variable) for a highly correlated but exogenous proxy. In this case, the IV approach uses the exogenous variables (instruments) that predict treatment but do not predict the outcome variables (our indicators).
- Stochastic Frontier Analysis (SFA) uses the measured yields and inputs to estimate the highest level of yield that can be achieved for that sample of producers given the inputs utilized. It estimates the level of inefficiency for producers who did not reach that level and can estimate the components that might have contributed to this level of inefficiency. Because not all input data is available or relevant in each area of study, each SFA has a slightly different specification for the stochastic production function, though all the pertinent inputs are included. We follow the conventional specification for SFA and a simultaneous equation to explain the inefficiency term using components relevant to input use, such as producer demographics (sex, age, education), input technology (use of equipment), and locational fixed effects.
- Regression Discontinuity Design is a technique used when a cut-off point on a continuous variable (such as a poverty index or yield cut-off) is used to determine who receives a given project. The impact of the project can then be estimated by comparing outcomes for producers who just qualify for the project on this score, with outcomes for producers who just fail to qualify for the project given their score. While intuitive and straightforward, this technique only uses observations close to the cut-off point and has not yet been featured in any COSA projects.
- Sensitivity Analysis differs for each application but is ideally performed with each of the analytical approaches above to show that the approach was appropriate and to improve the rigor of our results. Furthermore, using an appropriate mix of these techniques when possible (e.g., PSM and DID or PSM and IV) is encouraged.
2. Presenting Information: Customized KPI and Management Dashboards
COSA offers real-time management dashboards that are practical and easy for managers to use. They can track standard or customized Key Performance Indicators (KPI) in real time and at the level of detail and nuance that makes sense for them.–
3. Knowledge Base and Benchmarking
COSA data grows with thousands of new surveys added every year and spans a number of countries. Our Knowledge base houses one of the largest selections of relatively comparable data on agricultural sustainability issues in the world. With forthcoming updates, it will even more easily support benchmarking or time-series analyses.
4. Better Informed Decisions and Policies
COSA’s assessments serve to present information in ways that facilitate better informed decisions and policies. Since sustainability is complex and affected by multiple factors, we go beyond providing just raw information and seek to distill the critical relationships and important issues with clear and concise determinations of the factors that have the most significant “effect” in a situation.