Analyzing further training participation rates across waves in the NEPS data
(October 2016)Abstract
The spell-based nature of the National Educational Panel Study poses some challenges for analyzing training participation rates across waves. Raw training participation rates of each wave using courses compiled in the SpCourses data set and in the SpEmp data set differ by up to 75 percent across waves. Such differences do not prevail in other data sets for the same time period. In this technical paper we argue that the differences arise due to different time spans for which individuals indicate their training participation. Controlling for each individual's reference period reduces the differences in training participation rates across waves to plausible levels.