Abstract

Interviewer falsification can substantially bias survey estimates. To avoid such biases, most studies implement statistical methods to identify falsifications. However, these are often time and labor intensive. In this project, we will test novel methods of dimensionality reduction to identify interviewer falsification, with a focus on uniform manifold approximation and projection (UMAP). We will test the approach using data containing verified falsifications. THe approach will be implemented with further survey data to detect potential problems with implementing the new methods.