NCES seeks to improve standards and models for Non-Response Bias Analysis (NRBA) to be used by contractors who conduct data collections including both surveys and assessments. Increasing levels of non-response to NCES data collections (a widely prevalent problem for federal statistical data collections) have increased the need for understanding and adjusting for the bias introduced.
Statements of best practices need to be augmented by fully worked and annotated examples that can act as guidance when data collections are implemented. Therefore NCES has proposed that an exemplar be constructed for each of four different types of data collections.
The University of Michigan team will serve as the statistical resource and will develop these exemplars using (actual) data from four data collections that will be specified by NCES.