Qing Yang, Shuang Li, Kai Wang, Xin Chang, Shengya Cao, Guihua Zhang
In Study 1, high appraisal intensity reduced perceived professional autonomy (t(237.92) = 6.66, p < 0.001, Cohen's d = 0.86). In Study 2, professional autonomy positively predicted innovative work behavior (b = 0.192, p < 0.001), but appraisal intensity did not significantly predict autonomy (b = -0.133, p = 0.055), and the indirect effect was not significant. In Study 3, appraisal intensity was negatively associated with autonomy (b = -0.189, p = 0.002), autonomy positively predicted innovative work behavior (b = 0.126, p = 0.028), and transparency was positively associated with autonomy (b = 0.150, p = 0.011). However, neither the interaction between appraisal intensity and transparency (b = 0.046, p = 0.300) nor the index of moderated mediation was significant.
INTRODUCTION: Hospitals increasingly use algorithmic systems to monitor, compare, score, and appraise employees. Although such systems may improve consistency and efficiency, intensive use may constrain medical professionals' autonomy and their innovative behavior. This study examined the relationships among algorithmic performance appraisal intensity, perceived professional autonomy, and innovative work behavior, and tested whether algorithmic transparency moderates these relationships.
METHODS: Three studies were conducted. Study 1 was a between-subjects vignette experiment (N = 240) comparing high- and low-intensity appraisal. Study 2 used a two-wave survey of medical professionals (N = 287) to test mediation. Study 3 used a three-wave, multisource design (N = 273), with supervisors rating innovative work behavior, to test moderation and moderated mediation.
RESULTS: In Study 1, high appraisal intensity reduced perceived professional autonomy (t(237.92) = 6.66, p < 0.001, Cohen's d = 0.86). In Study 2, professional autonomy positively predicted innovative work behavior (b = 0.192, p < 0.001), but appraisal intensity did not significantly predict autonomy (b = -0.133, p = 0.055), and the indirect effect was not significant. In Study 3, appraisal intensity was negatively associated with autonomy (b = -0.189, p = 0.002), autonomy positively predicted innovative work behavior (b = 0.126, p = 0.028), and transparency was positively associated with autonomy (b = 0.150, p = 0.011). However, neither the interaction between appraisal intensity and transparency (b = 0.046, p = 0.300) nor the index of moderated mediation was significant.
DISCUSSION: Across the three studies, intensive algorithmic appraisal showed inconsistent but generally autonomy-constraining relationships, whereas professional autonomy consistently supported innovative work behavior. Transparency was positively associated with autonomy but did not reliably buffer the negative effects of appraisal intensity. Hospitals should retain contextual human judgment and treat transparency as a supportive design feature rather than assume that it offsets intensive algorithmic control.