New methods for measuring key variables used in impact assessment - for example, DNA fingerprinting and soil hyperspectrometry - have come on-stream in the past decade. SPIA, and a number of researchers in the CGIAR centers, have used these new methods on a pilot scale, finding that measurement error is a significant problem in a lot of agricultural survey data typically used in impact assessments. This workshop brought together 30 researchers who are contributing to this literature - both internal to CGIAR and external academic researchers - to look ahead at the next phase of research on these issues.