How bad is occupational coding error? A task-based approach
Abstract
Studies of occupational choice and mobility are often plagued by rampant occupational coding error. Use of task-based occupation measures, such as O*Net, may mitigate the bias caused by coding error if the occupation is misclassified as an occupation similar to the true occupation. Measuring occupational changes in "task space", I find that task-based measures reduce the problems of coding error, but only slightly. If one does not correct for coding error, one overestimates traditional occupational mobility rates by about 90%; using task-based measures, the overestimate of mobility is still 75%. I also show that when tasks are used as regressors and coding error is not corrected, estimates will be attenuated by 15%-20%. Task-based measures are a slight improvement over census occupation codes but are no panacea for dealing with coding error.
Publication Title
Economics Letters
Recommended Citation
Speer, J. (2016). How bad is occupational coding error? A task-based approach. Economics Letters, 141, 166-168. https://doi.org/10.1016/j.econlet.2016.02.025
