Predicting polytomous career choices in healthcare using probabilistic expectations data

This paper compares career expectations and career outcomes of Swiss healthcare assistants (HCA), an occupation created to increase the supply of nurses. We investigate whether HCAs can predict their own professional careers two years ahead by eliciting their expectations for a range of career alternatives, including nursing and other studies. Polytomous choice situations have rarely been analyzed using numerical probabilities in the expectations literature. Our results show that almost all respondents give informative answers to the probabilistic online survey question. Individuals express considerable uncertainty about their future careers, with over 60% attaching positive probabilities to more than one career alternative. The analyses reveal that individuals' numerical expectations have substantial predictive value for their future careers, even after controlling for many variables. This finding confirms that individuals have private information not directly available to researchers, and that eliciting choice probabilities for polytomous choice situations is a viable approach in surveys. However, the mean shares for career alternatives implied by individual probabilities do not fully coincide with actual shares and are more accurate over 4 than over 2 years. The information conveyed in expectations and their deviations from outcomes enables us to derive policy recommendations to increase transitions to nursing.