Ethical leadership is arguably-one of the most critical forms of leadership for individual stakeholders as well as society at large, as has been demonstrated by recent events, such as the COVID-19 pandemic and the social justice movement. Other events highlight the need for ethical leadership, such as corporate and political scandals as well as recognition that businesses should promote the welfare of stakeholders beyond shareholders (Freeman, 2019). Ethical leadership characterizes the extent to which leaders engage in normatively appropriate conduct and promote the wellbeing of stakeholders (Brown et al., 2005). Yet, while a great deal of conceptual (Kleshinski et al., 2021) and empirical work has been conducted on ethical leadership (for meta-analytic reviews see Banks et al., 2018, Hoch et al., 2018), there are several critical gaps that provide unique opportunities for the advancement of theory and practice in this area.
One opportunity is with respect to focusing specifically on ethical leader behaviors. Broadly speaking, the field of leadership is plagued by theoretical discussion that conflates, for example, leader behaviors with follower subjective evaluations (Antonakis et al., 2016, Day, 2014, Fischer and Sitkin, 2022, Fischer et al., in press). Follower behavior as an outcome is also measured via supervisor, subjective evaluations or follower self-evaluations of, for instance, counterproductive work behaviors and organizational citizenship behaviors (Fischer, Hambrick, Sajons, & Van Quaquebeke, 2020). A recent review of the organizational behavior literature showed that only 3 % of the variables used measure actual behaviors, yet theoretical discussions often refer to behaviors (only 19 % of studies included one behavioral variable; Banks et al., in press) meta-analytic reviews of the ethical leadership literature further demonstrated these concerns in that behaviors are theoretically conflated with subjective evaluations (Banks et al., 2018, Hoch et al., 2018). The ethical leadership literature is similarly affected by the conflation of leader behaviors (x1) with follower evaluations (y1) which are conceptually and empirically distinct (x1 ≠ y1). Both ethical leader behaviors and subjective evaluations of those behaviors are theoretically relevant, yet they are not conceptually interchangeable as we later demonstrate (Fischer & Sitkin, in press). Thus, we aim to contribute to the ethical leadership literature by focusing specifically on the ethical behaviors exhibited by leaders.
A second opportunity presents itself in terms of demonstrating clear causal linkages between ethical leader behaviors and outcomes. The ethical leadership literature, like other areas of leadership, suffers from endogeneity bias, which occurs when “the effect of x on y cannot be interpreted because it includes omitted causes” (Antonakis et al., 2010, p. 1087). Endogeneity bias can result from a host of factors including omitted variables, omitted selection, simultaneity, common-method variance, as well as measurement error, and leads to correlation and regression analyses being uninterpretable (Hill et al., 2021). The inability to draw causal inferences limits the advancement of theory of the antecedents that cause ethical leader behavior as well as the extent to which (and circumstances under which) ethical leader behavior causes attitudinal and behavioral outcomes. Past evidence indicates the majority of leadership research suffers from this type of bias (Antonakis et al., 2010), and the issue persists in the ethical leadership literature, as a majority of the ethical leadership literature is correlational (Banks et al., 2021). We intend to fill this gap via experimental methods that allow one to rule out endogeneity bias as an alternative explanation.
We first begin by presenting a conceptualization of ethical leadership. Next, we argue for signaling theory as a parsimonious, foundational, and meta-theoretical framework that allows for greater theoretical specificity between ethical leader behaviors and evaluations of those behaviors. Using a constant comparative method, Study 1 was conducted by coding Chief Executive Officer (CEO) letters to shareholders (n = 10,919 sentences). The eight verbal ethical leader signals (ELSs) which emerged were then experimentally manipulated in Study 2 to demonstrate that they lead to evaluations of ethical leadership (n = 264). While evaluations of ethical leadership are important, it is also necessary to investigate the extent to which ELSs can result in changes to others’ behaviors. Hence, in Study 3 we illustrated that ELSs reduce financial theft as a form of counterproductive behaviors (n = 434). Study 4 experimentally investigated the extent to which ELSs increase task performance and extra role behavior (n = 434). Finally, in Study 5, a machine learning algorithm, DeepEthics, was created to automatically score text for ELSs on a large scale for future research. We conclude this work with recommendations for theory and practice.