Temporal Stability of Stated Preferences: The Case of Junior Nursing Jobs

Abstract

Summary With the growing use of discrete choice experiments (DCEs) in health workforce research, the reliability of elicited job preferences is a growing concern. We provide the first empirical evidence on the temporal stability of such preferences using a unique longitudinal survey of Australian nursing students and graduate nurses. The respondents completed DCEs on nursing positions in two survey waves. Each position is described by salary and 11 non‐salary attributes, and the two waves are spaced 15months apart on average. Between the waves, most final‐year students finished their degrees and started out as graduate nurses. Thus, the survey covers a long timespan that includes an important period of career transition. The relative importance of different job attributes appears stable enough to support the use of DCEs to identify key areas of policy intervention. There is virtually no change in the groupings of influential job characteristics. Conclusions regarding the stability of willingness‐to‐pay, however, are different because of unstable preferences for salary. The instability of preferences for salary was also found previously in the context of comparing alternative elicitation methods. This prompts us to push for further work on the reliability of stated preferences over monetary attributes. Copyright © 2016 John Wiley & Sons, Ltd.

Stated preferences over job characteristics: A panel study

“…Our previous findings raise several questions for the current analysis of temporal stability by method of elicitation. As summarized above, we found that preferences for salary were unstable between the two cases of the DCE within the same wave (Yoo and Doiron 2013) and between the two waves of the same task (Doiron and Yoo 2017). In wave 1, salary was undervalued in the single-profile case task relative to the multi-profile case task.…”

“…The temporal stability of preference estimates in the traditional multi‐profile case has been discussed more extensively in our previous work (Doiron and Yoo ). The results presented in table are consistent with our previous conclusions that, although preferences exhibit a high degree of stability across waves, salary stands out with the largest drop in its weight over time .…”

“…For this reason, our previous studies based on the first‐wave of the multi‐profile case BWS (Doiron et al. , Doiron and Yoo ) specified the random utility function in terms of ln( salary ) to estimate the marginal utility of money, instead of including dummies for three salary levels ($950, $1000 and $1250) to estimate the discrete utility difference between each level and the base level of $800…”

“…Each identified utility weight in this component then measures extra utility that a level of some attribute offers over the worst level of the same attribute. For hospital type and clinical rotations, we choose "private hospital" and "no rotation" as the omitted levels; our previous studies (Yoo and Doiron 2013, Doiron et al 2014, Doiron and Yoo 2017 found that these levels gave less utility than "public hospital" and "three rotations. "…”

“…While there is evidence validating stated preferences using revealed preference estimates (Brownstone et al 2000, Small et al 2005 are still many open questions regarding appropriate methods to use in various contexts and characteristics that may be specific to stated preferences. As an example, as far as we are aware, the only direct evidence on the stability of stated job preferences comes from our own studies, Yoo and Doiron (2013) and Doiron and Yoo (2017) that we will summarize shortly. In this paper, we contribute to this literature by analyzing the stability of stated preferences over job characteristics in nursing.…”