Application paper

Annual Review of Organizational Psychology and Organizational Behavior

The Science of Leadership: A Theoretical Model and Research Agenda Andrew M. Carton Management Department, The Wharton School, University of Pennsylvania, Philadelphia, Pennsylvania, USA; email: [email protected]

Annu. Rev. Organ. Psychol. Organ. Behav. 2022. 9:61–93

First published as a Review in Advance on November 10, 2021

The Annual Review of Organizational Psychology and Organizational Behavior is online at orgpsych.annualreviews.org

https://doi.org/10.1146/annurev-orgpsych-012420- 091227

Copyright © 2022 by Annual Reviews. All rights reserved

Keywords

leadership, social influence, motivation, vision, meaning, meaningfulness

Abstract

I review the empirical literature on leadership, focusing on papers published since 2010. To do so, I introduce a framework composed of two features: whether theories (a) involve the study of leaders or leading (i.e., the person versus the process) and (b) conceptualize leadership as a cause or a conse- quence (i.e., an independent versus dependent variable).This framework can enable future research to accumulate in a more programmatic fashion and help scholars determine where their own studies are located within the land- scape of leadership research. I end the review by critically evaluating existing work, arguing that the most popular subcategory of leadership research— lumped conceptualizations of leading, in which scholars examine multiple leader behaviors within a single construct—has significant limitations and may need to be replaced by a greater focus on split conceptualizations of leading, wherein scholars isolate single leader behaviors.

61

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

1. INTRODUCTION

All disciplines have quintessential topics that reside at the center of scholarly attention. At the core of physics is the study of quantummechanics, at the heart ofmodernmedicine is the germ theory of disease, and at the center of economics sits the Keynesian paradigm.Organizational psychology is no different. In addition to topics such asmotivation and teamwork, one subject that has captivated scholars since the field’s inception is leadership (Yukl et al. 2002). A comparison of the definitions of “organization” and “leadership” reveals the overlap between these two ideas: An organization is a collective in pursuit of a common purpose, and leadership is a process in which a person or set of people influence others to pursue a common purpose (Ouchi 1980). The parallels are self- evident and profound. Consequently, leadership remains one of the most well-studied topics in organizational psychology (Day 2000, Yukl et al. 2002).

In this article I review the empirical literature on leadership using a framework composed of two features: whether theories (a) involve the study of leaders or leading and (b) conceptualize leadership as a cause or a consequence (i.e., an independent versus dependent variable). I organize the literature within and between each of these categories, and then end the review by critically evaluating existing work, making several recommendations. Given the sheer scale of accumulated evidence, I focus primarily on research published in top management and psychology journals since 2010. For reviews that provide a relatively comprehensive synthesis of leadership theories in earlier periods, see Glynn & Raffaelli (2010) and Yukl et al. (2002). For prior reviews of topics relevant to leadership in this outlet, see Day & Dragoni (2015), Den Hartog (2015), Kozlowski et al. (2016), Liden et al. (2016), Antonakis et al. (2016), Tepper et al. (2017), Lyness & Grotto (2018), and Lord et al. (2020).

2. ORGANIZING THE LEADERSHIP LITERATURE

The teleological distinction between leaders and leading is rooted in how both concepts are de- fined.A leader is typically conceived as an individual who has a unique degree of status or authority with which to alter the behavior of others toward mutually agreed upon objectives (Pfeffer 1977). As I explain below, this status can be formal or informal, but most perspectives assume that both the leader and his/her/their follower(s) believe(s) that the leader has some legitimate claim to au- thority. As noted above, the act of leading is typically defined as a social influence process in which an individual or individuals attempt to influence others toward a common purpose (Yukl et al. 2002, Locke 1999).

Both theoretically and statistically, a key distinction between leaders and leading is the level of analysis. In multilevel theory, the individual resides at a higher level of analysis than any behav- ior that this individual may enact (Kozlowski & Klein 2000). Any individual leader may exhibit dozens or hundreds of distinct leader-relevant behaviors across the course of a single day, such as communicating vision, clarifying deliverables, andmaking resource allocation decisions. Although most individuals tend to act in consistent ways across time, each day they will enact some behav- iors but not others, and the manner and order in which they exhibit them may vary considerably. Indeed, behavior may vary just as much—and possibly more—within leaders versus between lead- ers ( Judge et al. 2002, Kenny & Zaccaro 1983, McClean et al. 2019; see also Hall et al. 2009 and Kenny & Livi 2009). Although their behavior varies each day, however, their personal attributes, including their demography and the degree to which their role as a leader is formally designated, remain relatively constant over long periods of time.1 Similarly, other peoples’ attributions of

1Some person-level attributes can certainly vary. For instance, leader identity can oscillate from situation to situation (Day & Sin 2011). Accordingly, temporal stability is a general mark of leader-level attributes rather than an invariant, definitional feature.

62 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

them—such as how capable they are to take on the role of a leader—tend to remain relatively stable (Epitropaki & Martin 2004, Dinh & Lord 2012; see also Hall et al. 2009 and Kenny & Livi 2009).2

The distinction between leader behaviors and leaders themselves is reflected by two traditions in the literature. The study of leaders is largely drawn from the social cognitive, identity, diversity, and personality traditions. It relates to questions such as which people emerge as leaders, how they are categorized, and how they are evaluated—each of which depends on viewing leaders as individuals who have stable attributes (Lord et al. 1984, Meindl 1995, van Knippenberg & Hogg 2003). By contrast, the study of leading (i.e., the process of leadership) primarily draws from the tradition that focuses on antecedents and consequences of discrete behaviors, such as giving feedback to employees. In this tradition, researchers focus on how behaviors influence a target’s psychological states (both cognitive and affective) and, in turn, the target’s behavior. For instance, the act of a leader admittingmistakes describes a bounded behavior that can be repeatedly observed in situ, even if it might be manifest in slightly different ways each time.

In short, a trait or psychological construct used to categorize a person’s inherent characteristics (e.g., a leader’s age) is more stable and thus fundamentally distinct from an externally observable behavior aimed at influencing others (e.g., a leader blaming a follower for a mistake) (Gravetter & Forzano 2011, p. 75). The study of leaders involves who leaders are and the positions they attain, whereas the study of leading involves how leaders exact influence as well as the processes they en- act. Given these related distinctions, I employ the study of leaders and the study of leading as the primary organizing principles of this article (see Figure 1). Toward the end of the article, I address the question of how leaders and leading can be studied together, given that they are inextricably intertwined, as it is leaders who do the leading. I also discuss when they can be understood sepa- rately, given that some individuals may be formally designated as leaders but fail to lead, whereas others who do not possess the label of leader can alter the behavior of others in ways that reflect the act of leading. In addition to the leaders/leading distinction, I employ a secondary organizing principle: leadership can be understood as a cause or a consequence. That is, both leaders and leading can be investigated as either an independent or a dependent variable (Glynn & Raffaelli 2010) (see Figure 1).

3. THE STUDY OF LEADERS

Scholars often examine leadership not as a process that is agnostic to the person who is doing the leading but with an eye toward understanding the leader as an individual.

2In reality, there are more than just two levels of analysis. For example, a leader’s one-time outburst may be nested within that person’s broader tendency to have frequent outbursts. And this tendency might be nested in a general disposition to engage in abusive leader behaviors. In turn, this disposition may be nested within even broader personality characteristics (e.g., narcissism), which in turn may be nested within demographic characteristics that are completely, or almost completely, immutable (perhaps people born during one gener- ation tend to be more narcissistic than others). Higher levels of analysis are more stable and enduring, and lower levels more granular and fluid. Critically, whether behavior is examined as a one-time episode or a more general tendency, it will be nested within numerous person-level attributes that do not change or change very infrequently. In this way, the hierarchical nesting of behavior within stable person-level factors remains rele- vant even if behaviors themselves can be conceptualized at different levels of granularity. For the remainder of the article, I thus treat behavior associated with leading as the lower level of analysis and person-level factors associated with leaders, such as traits, orientations, and social standing, as the higher level of analysis. Yet it is important to underline the reality that these hierarchical differences are not necessarily dichotomous or invariant.

www.annualreviews.org • The Science of Leadership 63

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

Leader attribute/

attribution of leader

Directing attention toward a common

purpose

Initiating action toward a common

purpose

Maintaining action toward a

common purpose

Causes Consequences

Leading as a dependent variable

Leading as an independent variable

The leader as an independent variable

The leader as a dependent variable

Leader

Behavior

Level of analysis

Figure 1

Two core attributes of leadership research: leaders versus leading and causes versus consequences. Extant research can be organized into four basic categories: leader behaviors (i.e., leading) as predictors or as outcomes and leader-level concepts (e.g., leader personality and perceptions of leaders) as predictors or as outcomes.

3.1. The Leader as an Independent Variable

Inspired by research in personality psychology, industrial/organizational psychology, human re- sources, and human factors, leadership scholars have long been interested in leader traits, disposi- tions, and other individual differences ( Judge et al. 2002). These variables reflect the leader’s gen- eral orientation across time and situations (DeRue et al. 2011). In this vein, scholars who examine the leader as an independent variable typically seek to understand how the leader’s stable personal- ity characteristics (leader traits) affect a range of outcomes, such as performance. This can include general habits and styles of behavior—but rather than considering behavior as episodic, contex- tualized, and “of the moment,” a leader-level perspective focuses on behavioral tendencies—that is, how leaders tend to act across time and situations.

One of the most dominant trends of the past decade involves the dark side of leader personal- ity. This research has unearthed a series of negative consequences of one dark trait in particular: narcissism. Narcissistic leaders engage in less task, relational, and change-oriented behaviors, are less responsive to indications that their risks are backfiring, and are more likely to be bolstered by social praise (Chatterjee & Hambrick 2011, Martin et al. 2016). Moreover, Gupta & Misangyi (2018) found that narcissism imbues leaders with an inflated sense of their own discretion, and Zhu & Chen (2015) found that narcissistic leaders often deliberately go against their directors’ suggestions in order to establish their superiority. What prompts narcissistic leaders to engage in the most extreme forms of destructive behavior? Possibly a mix of thin skin and vindictiveness: Liu et al. (2017) revealed that narcissistic leaders are especially prone to engage in self-interested behavior when they feel slighted by their organizations. Ultimately, however, the relationship be- tween narcissism and leadership might be weaker than often presumed: a meta-analysis of 92 samples found that psychopathic traits in leaders, including narcissism, are only weakly associated

64 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

with leader effectiveness, although the detrimental effect of these attributes is stronger for male leaders (Landay et al. 2019).

In addition to increasing their attention on narcissism, scholars have continued to focus on the Big Five personality factors: conscientiousness, agreeableness, neuroticism, openness, and ex- troversion. The general pattern that unifies these findings is the following: Longstanding beliefs about the advantages of some attributes, such as extroversion and conscientiousness, have come under scrutiny. Consider extroversion. Some findings reinforce received wisdom that extroverted leaders have an advantage. For example, Gupta & Misangyi (2018) found that extroverted CEOs are better at issue-selling because they aremore assertive andmore interested in building the social and political capital needed to facilitate change. However, Grant et al. (2011) drew on dominance complementarity theory to assert that extroverted leaders only yield strong performance when their followers are deferent. By contrast, proactive followers prefer more introverted leaders. In a similar vein, research over the past decade suggests that leader conscientiousness has surprisingly mixed effects (Badura et al. 2020, DeRue et al. 2012, Marinova et al. 2013). This may be because leaders who are conscientious have a tendency to get bogged down in detail and struggle to give followers a bigger picture view of how their work fits into the organization’s overall mission. In a similar way to how scholars have discovered that the advantages of extroversion and conscien- tiousness are bounded, there also may be a curvilinear effect for another trait that many would assume to be strictly positive: intelligence. Leaders who are too smart might have “too much of a good thing,” perhaps because they are more likely to engage in overanalysis at the expense of decisiveness (Antonakis et al. 2017).

In short, research on the traits that have received the most attention from leadership scholars over the past decade—especially narcissism, extroversion, conscientiousness, and intelligence— suggests that the relationships between leader traits and key outcomes have important boundary conditions and are not as straightforward as previous evidence and conventional wisdom might suggest.

3.2. The Leader as a Dependent Variable: The Role of Categorization, Identity, Networks, and Leader Well-Being

Scholars who investigate the leader as a dependent variable are interested in what happens to lead- ers themselves.This involves questions such as how leaders emerge and how the responsibilities of being a leader impact their well-being. It typically involves one of the following four approaches: implicit leadership theories, identity-based theories, network-based theories, and perspectives on leader well-being.

3.2.1. Implicit leadership theories. Rooted in research on social cognition, theories involv- ing attribution, categorization, and evaluation relate to how leaders are perceived as well as out- comes that hinge on these perceptions, such as leader selection and emergence (Foti et al. 2017, Shondrick & Lord 2010). The seminal work in this space has examined how people often jump to conclusions about a person’s leadership effectiveness based on little or no information about their behavior. Drawing from attribution theory, Meindl (1995) established that people often in- fer leader effectiveness not by observing a leader’s behavior but rather the outcomes that they presumed were caused by the leader: When organizations and groups perform well, leaders are presumed to be effective. A largely separate line of work is rooted in categorization theory. Lord et al. (1984) introduced leadership categorization theory (LCT) according to the premise that people have cognitive schemas about effective leaders based on leaders they have encountered elsewhere, such as personal experience and history textbooks.

www.annualreviews.org • The Science of Leadership 65

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

Whereas attribution-based theories have largely sputtered in the past 15 years, LCT has flour- ished. One of the main reasons is that LCT can account for an especially pernicious facet of cognition: instantaneous categorization. Even though leading is defined as a social influence pro- cess, and evaluators certainly do infer leadership potential from actual leader-relevant behavior [such as task-oriented communication patterns (Gerpott et al. 2019)], people jump to conclusions about a person’s leadership capability based on seemingly irrelevant information. This is because people engage in a largely subconscious process in which they evaluate leaders according to how well they match prevailing leader prototypes. This may include information such as how much a person talks (MacLaren et al. 2020), but, more disconcertingly, people often judge a person’s leadership potential without any information at all about that person’s behavior. This includes a target’s physical attributes or social category membership. This has made LCT an important vehicle for understanding why minority leaders experience discrimination. For example, the pro- totypical leader is assumed to be White (Rosette et al. 2008). As a result, the success of minority leaders is often attributed to the situation or to attributes that are irrelevant to leadership, whereas failure is attributed to their personhood (Carton & Rosette 2011, Park & Westphal 2013).

Peoples’ perception of a prototypical leader may shift depending upon the situation. For in- stance, evaluators prefer minorities more when their skillset matches the context, such as the pre- sumed fit of Asian Americans to engineering versus sales (Sy et al. 2010). Similarly, people may prefer Asian American leaders when organizations are failing, because they are presumed to be more communal in their orientation (Gündemir et al. 2019). Finally, the link between stereotyp- ing and leader outcomes can sometimes be circuitous. Zapata et al. (2016) found that minority leaders are more likely to experience stereotyping and undermining from their own subordinates when they treat their subordinates kindly. When subordinates are treated well, they do not pay close attention to their supervisors because they do not feel a sense of threat from them.However, this can create self-perpetuating spirals of discrimination because subordinates are more likely to stereotype leaders they do not pay attention to.

In addition to race, much of the work on implicit theories over the past decade has been de- voted to gender. Some of this research points to a disadvantage for women leaders. For example, McClean et al. (2018) found that men who speak up promotively were assumed to be promising leaders according to a rank ordering mechanism, whereas women who spoke up promotively were not. Similar to research showing that racial minorities are viewed as prototypical leaders only dur- ing exceptional periods, research on the glass cliff effect suggests that women may be preferred when firms are failing because they are assumed to be communal and self-sacrificing, and that this typecasting effect might reinforce the belief that women are not fit to lead when organizations are meeting or exceeding expectations (Ryan et al. 2011; see also Koenig et al. 2011).

Other trends have established that women leaders have a more general advantage. In a meta- analysis of 99 studies, Paustian-Underdahl et al. (2014) found that evaluators perceive women as more effective leaders than men. This may cause women to sometimes get more credit than men for enacting similar leader behaviors (Lanaj & Hollenbeck 2015). However, there are important boundary conditions on the advantage that women leaders have (Lanaj & Hollenbeck 2015). For instance, Rosette & Tost (2010) found that the preference for women leaders was limited to top- level positions (see also Leslie et al. 2017). Paustian-Underdahl et al.’s (2014) above-mentioned meta-analysis indicates that the most reliable advantage men have is in their own self-regard, as they tend to rate their own leadership more generously than women rate themselves.

Finally, some research has identified points of tension between different implicit leadership theories. One example is the connection found between LCT (Lord et al. 1984) and romance of leadership theory (Meindl 1995) in the effort to explain bias against Black leaders (Carton & Rosette 2011).

66 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

3.2.2. Identity-based theories. A related line of research involves identity-based theories of leadership. The logic is similar to that of categorization theory: Leaders are more likely to emerge when their attributes align with follower expectations.However, the basis for that alignment is not necessarily whether a person looks a certain way but whether their values and preferences align with those of followers (Hogg et al. 2012, Stam et al. 2014). The specific nature of identity will vary depending on what followers value. This can include relational identity (Chang & Johnson 2010), cultural identity (Lord & Brown 2001), and organizational identity (Sluss et al. 2012). For example, Lam et al. (2018) found that proactive employees identify more strongly with proactive leaders, especially when those leaders are more proactive than their predecessors, thereby creat- ing a contrast effect that amplifies the congruence between leader and follower identities. These identities are also central to the mutual sensemaking process between followers and leaders insofar as both parties engage in an iterative process in which some people claim the identity of a leader while others decide whether to bestow that identity onto them, thereby accepting the identity of a follower (DeRue & Ashford 2010).

Along these lines, in addition to the role of follower identity in predicting leader emergence, the leader’s own personal identity plays a key role. Simply put, leaders are unlikely to invest the time and energy required to claim the identity of being a leader and all that this entails (vying for leadership positions, undergoing leadership training, and accepting the risks of being a leader) unless they define themselves, at least in part, as a leader (DeRue & Ashford 2010, Kwok et al. 2018, Petriglieri & Stein 2012). In this vein, Guillén et al. (2015) found that people ascertain whether they are a leader by comparing their self-concept to well-known examples of leaders and their general understanding of what a leader is and should be.3

3.2.3. Network-based theories. Although theories centered on categorization and identity re- main the dominant paradigms for understanding leader evaluation, some research influenced by sociology takes a broader view by examining how the leader’s social network shapes when and why they emerge as leaders. For instance, Chiu et al. (2017) found that managers in more centralized networks are more likely to be endorsed by others as leaders. This echoes earlier findings show- ing the power of network centrality for driving perceptions of leader charisma (Balkundi et al. 2011). As another example, Brands et al. (2015) found that male leaders are seen as more charis- matic when they occupy central roles in their social networks, whereas female leaders are seen as more charismatic when their networks are cohesive. Other work has expanded the investigation beyond relational ties to include network structures based on perceptions. For instance, DeRue et al. (2015) found that network patterns of peoples’ perceptions of warmth and competence drove whether leaders emerged and whether their power was centralized or distributed.

3.2.4. Leader well-being. The body of leadership literature that is perhaps most exclusive to the past decade is that on leader well-being. Tracking the increased academic interest in men- tal health, scholars have begun to challenge the implicit assumption that only positive outcomes ensue when people occupy leadership positions. Although it is true that leader positions come with esteem, influence, and access to scarce resources, scholars have begun to uncover numerous downsides. In an experience sampling study, Foulk et al. (2018) found that leaders who experienced more psychological power perceived more incivility from others, harming their well-being. In a

3Although the extent to which people see themselves as a leader tends to be stable across long periods of time, it can (and often does) change within-person insofar as people who previously did not see themselves as leaders sometimes do at a later point, and vice versa (Miscenko et al. 2017).

www.annualreviews.org • The Science of Leadership 67

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

similar vein,Gabriel et al. (2020) found that becoming a leader made people more isolated. Recent research suggests that some employees are aware of many of these risks and thus avoid becoming leaders in the first place (Zhang et al. 2020). On the flip side, Li et al. (2020) found that becoming a leader can increase one’s conscientiousness but does not affect emotional stability, suggesting that the effects of assuming leadership roles are mixed and demand more attention.

4. THE STUDY OF LEADING

A behavior is an observable action that takes place in situ (e.g., providing task instructions). Au- thors sometimes take for granted that they are studying leading so long as the behavior under investigation is being enacted by a person who is formally designated as a leader. Although this approach is certainly an intuitive one, it culminates in two untenable conclusions. First, because almost any behavior can be enacted by most leaders, a working definition of leading as behaviors enacted by leaders is too inclusive to be helpful for organizing the literature. If leading can be everything, then it cannot be anything. Indeed, many behaviors exhibited by people in positions of formal authority will not necessarily reflect leading (Kniffin et al. 2020). Second, it would result in the circular logic that only people who are formally recognized as leaders can exhibit behaviors consistent with leadership, which is unfounded given that, at least from a logical standpoint, a per- son does not need to meet any formal preconditions in order to motivate other people toward a common purpose. To be sure, followers may need to perceive a person as having some legitimate status in order to permit that person to have influence over them (DeRue & Ashford 2010)—but it need not be formally designated (Lord et al. 2016). An alternative way to organize the literature is according to whether a behavior has a notable impact on an outcome that seems relevant to leading, such as follower performance. But behaviors centrally associated with leading cannot be identified or understood from a conceptual standpoint merely by examining their consequences; anointing a behavior as central to leadership merely because it reliably predicts key performance outcomes is inherently atheoretical and tautological.

To avoid these problems associated with circularity, a systematic indexing of the literature re- quires an identification of the behaviors most central to the root conceptualization of leading. Most definitions of leading converge on the following core tenets: Leading is an influence pro- cess that involves attempting to motivate followers toward a common purpose (see Vroom& Jago 2007 for an overview of relevant definitions). Given that motivation is the root action-oriented term in common conceptualizations of leadership, it would be informative to turn to theory on motivation to distill the principal attributes of leadership.Motivation involves three issues: direct- ing, initiating, and maintaining effort (Latham & Pinder 2005). A person must first decide what they want to achieve, and then take action and persist as they attempt to achieve it. Given that a common purpose is a part of most definitions of leading, it follows that leading involves directing people toward a common purpose and then influencing them to initiate and maintain effort in their pursuit of it. The term purpose can be understood as reflecting a specific target or general end-state. As such, this definition is consistent with the active pursuit of a specific target (e.g., becoming the most profitable company in the industry) (Locke 1999) as well as the pursuit of more diffuse end- states that reinforce the organization’s underlying value system (e.g., an individualistic culture) (Yukl 2008). In this way, any given study can be classified as an examination of leading rather than merely a study of actions performed by a leader to the extent that it meets these criteria. I use this conceptualization to organize the literature on leading, again focusing primarily on the past decade of empirical research.

68 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

4.1. Leading as an Independent Variable

Behavior can be perceived and understood at various levels of abstraction (Liberman & Trope 1998), ranging from a single behavioral episode that is defined richly and can be observed in situ (a person speaking about their company’s vision to a colleague on a Thursday afternoon) to the tendency for many different types of behaviors that are unlikely to be enacted at the same time to covary and thus potentially be united by some common theme (e.g., admitting mistakes and asking followers for feedback can perhaps be united by the theme “humility”). Additionally, a single behavior or a cluster of covarying behaviors can be enacted more or less consistently across time and different situations (e.g., a person’s tendency to communicate about long-term visions rather than short-term goals on various days and in different situations)—a point to which I return, below, when discussing the fluid boundaries between discrete leader behaviors and longstanding dispositions, styles, orientations, characteristics, and traits.

Given that scholars can study either single behaviors that tend to unfold in a single sequence of actions versus clusters of different behaviors that unfold in different settings, scholars have taken two basic approaches to understanding how the act of leading causes key organizational outcomes. The first involves investigating single, observable behaviors within single domains, such as com- municating vision or providing feedback. The second involves behavioral composites in which scholars aggregate multiple leadership behaviors into a single construct, such as assessing whether a leader articulates a clear purpose, makes ethical resource allocation decisions, and provides fol- lowers with autonomy. In principle, scholars can aggregate any number of discrete behaviors— from two to an indefinite number—into a single construct. Yet, as I review below, scholars who deploy this composite approach typically incorporate at least seven, and often many more, behav- iors into meta-constructs. In practice, this has resulted in bipolarization, such that scholars tend to fall into one of two extremes: either investigating one behavior at a time or lumping several (or sometimes several dozen) behaviors into a single meta-construct. This bifurcation reflects the longstanding distinction between such “splitters” and “lumpers” (Cronbach &Gleser 1957, Judge et al. 2002), wherein the splitter contingent investigates behaviors that can be observed as single episodes in situ, such as a leader who admits a mistake, whereas the lumper contingent synthesizes separate behaviors that covary yet do not unfold at the same time and in the same situation.

4.1.1. Splitters: research that isolates leadership behaviors. To examine the first dimension of leading—directing followers toward a common purpose—scholars continue to pay careful attention to the language of leadership (Steffens & Haslam 2013). The type of language most central to the establishment of a common purpose is that which refers to overarching goals, including vision, mission, aspirations, ultimate goals, and superordinate goals. Carton et al. (2014) argued that leaders establish a common purpose in part by communicating visions with more con- crete, image-laden rhetoric (e.g., “our vision is to make moviegoers laugh”) rather than abstract rhetoric (e.g., “our vision is to make people happy”). This not only gives each employee a clear mental picture of what the organization strives to accomplish but also gives different employees the same mental representation. Leaders can also make abstract values appear more proximal if they have an image-laden goal that can be used as a real-life symbol of the organization’s broadest aspirations. An example is when John F. Kennedy motivated NASA employees by remarking that “knowledge and peace are [on the moon],” thereby making the organization’s ultimate goals appear more closely connected to day-to-day work (Carton 2018, p. 336).

In addition to communicating with imagery, leaders may be able to further reinforce a shared sense of purpose by communicating a limited number of values (e.g., “our vision to make movie- goers laugh brings to life our core value of maximizing customer satisfaction”) rather than a long

www.annualreviews.org • The Science of Leadership 69

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

list of values that leads different employees to focus on different principles (Carton et al. 2014). This positions different employees to understand the broader meaning of the vision in the same way. A shared vision has many important consequences. For instance, Griffin et al. (2010) found that leader visions can spur both adaptive and proactive behaviors. It may also spur coordination and creativity (Carton et al. 2014). At the individual level, a common purpose spawned by a vivid vision increases employee persistence (Carton & Lucas 2018).

Although language is themost direct and pragmatic way to establish a common purpose, leaders can reinforce a shared sense of purpose through othermeans. For instance, organizations canmake use of ritualistic behavior in which employees attend events together and engage in ceremonial routines, creating emotions that reinforce the sense that they share the same values (Lepisto 2021). Another way to ensure that employees feel a common sense of direction during vision pursuit is by allowing and even encouraging their involvement in key decisions—that is, licensing participation. In a sample of 65 firms, Ashford et al. (2018) found that CEOs who pair vision articulation with feedback seeking improve firm performance. However, of the two variables, feedback seeking was more instrumental for improving the extent to which top management teams worked effectively together.

The second dimension of leading—helping followers initiate action toward a common purpose—has been understudied over the past decade; however, some important advancements have complicated received wisdom. For example, although it is well-known that one of the most effective ways to impel followers into action is through subgoals and related targets, such as dead- lines, it is important that these targets not be viewed as so challenging that employees do not initiate action in the first place. To the extent that employees are not intimidated by stretch goals, they can spur innovation because they require employees to think outside the box and develop new capabilities (Gary et al. 2017, Lemoine et al. 2016).

Once leaders achieve the first two motivational steps—articulating a compelling vision and jumpstarting employees into action—they must focus on motivating employees to persist. In the past decade, scholars have moved beyond extrinsic incentives and rewards to a more cognitive un- derstanding of persistence, emphasizing the role of suborganizational visions and subgoals, per- haps because these cognitive markers are more likely than extrinsic rewards to lead to internalized and identified motivation, wherein employees see a personal connection to the organization’s vi- sion (Ryan & Deci 2000). One tactic is a suborganizational vision—a vision that a unit head crafts to lead the people they directly supervise. Suborganizational visions are an important instrument for allowing employees—especially those who work for peripheral units in an organization—to identify with an aspiration that is tailored to their own identity and interests while also linked to the organization’s broader vision (Lewis & Clark 2020). In a similar vein, leaders can encour- age followers to partake in the process of vision crafting. In a study of 197 leader-follower dyads, Kearney et al. (2019) found that communicating a common vision ensures that leaders control their organization’s direction as long as the leader simultaneously empowers followers such that they have the opportunity to shape and own the vision.

In addition to suborganizational visions, leaders can make use of subgoals, which provide a short-term referent that employees aspire to achieve and a clear demarcation between success and failure. In addition to helping employees regulate effort, subgoals can help employees see a connection between their day-to-day work and the organization’s vision. For example, John F. Kennedy helped employees see a connection between their day-to-day work and the objective of landing on the moon by articulating three subgoals: reaching Earth’s orbit, breaking Earth’s orbit, and reaching the moon’s orbit (Carton 2018). Subgoals can vary in terms of how they are supervised and how they are set. In terms of how they are supervised, leaders who are deeply involved in the day-in, day-out process of managing subgoals are more effective at promoting

70 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

persistence than those who involve themselves only as goal deadlines draw near (Chen&Nadkarni 2017). Regarding how they are set, there are still more questions than answers, suggesting an area that is ripe for future research. For instance, little is known about the ideal number, time range, and level of subgoals, as well as how these characteristics vary by context. For example, whereas a greater number of subgoals may help employees regulate their effort and stay on course, too many subgoals may cause employees to become so distracted by details that they lose sight of how their work connects to a shared vision (Carton 2018). Beyond subgoals, there are a number of other tactics through which leaders can motivate persistence, including their affective tone (Staw et al. 2019) and their efforts to shed light on the social impact of their employees’ day-to-day work (Grant 2012). However the role of leading in driving follower persistence remains a relatively neglected area of research and is in need of fresh insights.

4.1.2. Lumpers: research that synthesizes multiple leader behaviors into single constructs. Numerous theories agglomerate many distinct behaviors, usually ranging from a dozen to several dozen, into a single construct. Typically, these theories are labeled “[adjective] leadership” insofar as they reflect a bundle of behaviors that pertain to a common theme. This includes authentic leadership (Gardner et al. 2011), empowering leadership (Cheong et al. 2019), ethical leadership (Bedi et al. 2016), humble leadership (Feng et al. 2014), servant leadership (Van Dierendonck 2011), and transformational leadership (Wang et al. 2011). In some cases, scholars will emphasize one or a subset of these behaviors in their theorizing, yet still use the multi-behavior index in the methods. In these cases, the empirical test adds to the existing body of evidence for these lumped theories.

Most of these theories feature behaviors that fit into at least one of the three aforementioned tenets of leadership—directing action toward a common purpose, initiating action, and main- taining action—however, most of them include several behaviors that cut across these categories while also incorporating behaviors that are less essential to the definitional core of leading. For instance, transformational leadership features one behavior that is central to leading (vision com- munication) and others that are less central, such as tailoring messages toward each person as an individual. It is important to note that there are many different conceptualizations and measures within each leadership theory; therefore, in each case I introduce the construct with a conceptu- alization and measure that has been published in the past decade in a top outlet and epitomizes how scholars employ the construct.

A prototypical model of empowering leadership is composed of 24 distinct behaviors, including giving followers responsibility, encouraging initiative, showing concern, listening, and recogniz- ing followers’ strengths and weaknesses (Amundsen & Martinsen 2014). It has been linked to outcomes such as organizational citizenship behaviors (van Dijke et al. 2012), creativity (Zhang & Bartol 2010), team development (Lorinkova et al. 2013), goal setting (Schilpzand et al. 2018), and innovation (Chen et al. 2011).

A popular model of ethical leadership involves 38 behaviors, including holding followers ac- countable, giving subordinates influence, exhibiting concern for environmental sustainability, showing interest in followers, clarifying expectations, and keeping promises (Kalshoven et al. 2011). Schaubroeck et al. (2012) found that ethical leadership boosted unit ethical culture. Mayer et al. (2013) found that a leader’s ethical behavior is most likely to drive ethical actions (e.g., whistleblowing) when it is not only the leader who is ethical but also peers. Importantly, ethi- cal leadership may not always have expected effects. Leaders who act ethically one day may act more abusive the next day because they are depleted from regulating their own behavior (Lin et al. 2016).

www.annualreviews.org • The Science of Leadership 71

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

A popular approach to understanding humble leadership involves nine behaviors, including actively seeking feedback, acknowledging when others have superior skills, complimenting others on their strengths, and being willing to learn from others. Humble leadership has been linked to an empowering organizational climate (Ou et al. 2014), performance (Owens et al. 2013), and team creativity (Hu et al. 2018). It may curb deviance but also promote it by enhancing employees’ sense of entitlement (Qin et al. 2020).

A version of paradoxical leadership involves five dimensions, each of which is associated with multiple behaviors. This includes combining self-centeredness with other-centeredness, treating subordinates uniformly while allowing individualization, and maintaining decision control while allowing autonomy (Zhang et al. 2015). It increases collective self-efficacy (Shao et al. 2019), em- ployees’ ability to manage goals (Pearce et al. 2019), and proactivity (Zhang et al. 2015).

A prevalent model of servant leadership is composed of 28 behaviors, including recognizing when something is going wrong, emphasizing community service, and not compromising ethical principles (Liden et al. 2014). It has been linked to organizational identification (Liden et al. 2014), team effectiveness (Hu & Liden 2011), and performance (Chen et al. 2015, Liden et al. 2014).

A widely studied version of transformational leadership features more than 10 behaviors sorted into four classes, including inspiring followers through vision communication, considering them individually, and challenging them intellectually (Bass & Avolio 1994). It has been linked to per- formance (Grant 2012), regulatory focus ( Johnson et al. 2017), and helping behavior (Wang & Howell 2010). It also enhances job performance (Zhang et al. 2014).

In sum, the literature has devoted substantial attention to leading as an independent variable, with the primary emphasis on lumped (rather than split) frameworks.

4.2. Leading as a Dependent Variable: Development and Interventions

Leading is often driven by “upstream” antecedents. In particular, industrial/organizational psy- chologists and human resource scholars continue to expand our repository of knowledge on leader training and development.This work is predicated on the idea that leaders are not merely born but can be made—an assumption that has long been questioned but was recently given strong support in Lacerenza et al.’s (2017) meta-analysis. The authors aggregated 335 samples and discovered powerful effects for leader development programs on a variety of behaviors, including how lead- ers react to threats. They also examined a variety of moderators. Some are intuitive. For instance, on-site training worked better than virtual training. Yet others were surprising. For instance, 360 degree feedback may not provide an advantage over feedback from a single source, and longer training programs are more effective than shorter ones (debunking long-held assumptions that longer programs are depleting and may promote reactance). Other findings were perhaps intu- itive in terms of the direction of the effect but still surprising in terms of magnitude. To illustrate, high-level leaders are not merely less likely to acquire new leader behaviors than low-level leaders; they are four times less likely.

A related literature is that on leader interventions—frameworks or nudges that can cultivate desirable behaviors. One point of weakness among many leaders involves vision communication. Carton et al. (2014) found that approximately 90% of leaders communicate abstract visionary rhetoric (e.g., “make people happy”) even though concrete visions (“make moviegoers laugh”) are more likely to motivate and coordinate employees—findings consistent with prior research on leader communication (Emrich et al. 2001). To combat this so-called blurry vision bias, Carton & Lucas (2018) found that leaders craft more image-laden visions when instructed to mentally project themselves deep into the future and imagine witnessing a vivid, real-life moment that will take place once their organization’s vision is achieved. For example, the leader of a computer repair company may envision her employees feeling proud when a customer calls to express gratitude

72 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

for their efforts. This approach works better than explicitly instructing leaders to use image-based words, suggesting that some leader interventions should be indirect.

5. INTEGRATING THE STUDY OF LEADERS AND LEADING

A small number of articles over the past decade have investigated the distinction between lead- ers and leading. One of the key insights from this research is that it is possible for individuals to enact behaviors central to leadership (e.g., vision communication) without being formally des- ignated as leaders, and likewise people who are formally designated as leaders might not exhibit exclusively (or even primarily) behaviors that fall within the domain of leading (Ashford & Sitkin 2019, Bedeian & Hunt 2006). That is, leading does not require a formal leadership position, and a formal leadership position does not guarantee leading.

Another perspective relates not to drawing distinct boundaries between leaders and leading but to mapping the interplay between them. There are two basic ways this can be manifest: causal relationships and contingent relationships.

5.1. Causal Relationships

Consistent with the notion that leaders and leading can both be understood as independent or dependent variables, it is possible that one can cause the other. First, exhibiting actions associated with leadership, such as vision communication or providing feedback to others as they pursue long- term goals, can cause a person to attain a leadership position and/or affect them once they ascend to a leadership role. For example, Gerpott et al. (2019) found that individuals who communicated about change were more likely to emerge as leaders early in a team’s project, whereas those who focused on relationships were more likely to emerge as leaders later in the team’s project. This suggests that communication patterns determine which people emerge as leaders and when.

The second is when characteristics of the leader cause leading. The straightforward way this can be understood is with respect to traits, dispositions, and styles predicting leader behaviors, as these individual differences typically reflect general ways of behaving across time and situations. For instance, a person’s trait extroversion might predict behaviors such as information sharing and feedback seeking such that extroverted people are more likely to share information and seek out feedback. However, there are less direct (and less intuitive) channels. For instance, a leader’s religious identity can prompt a self-fulfilling prophecy, such that followers who have the same identity infuse the leader with more credibility, which then empowers the leader to exhibit more effective behaviors ( Jackson & Johnson 2012). In a similar vein, Johnson et al. (2012) found that individualistic and collective identities combine to predict which leaders are likely to enact trans- formational behaviors. Accordingly, a leader’s identity can be a distal cause of their own subsequent leadership behavior (Day & Sin 2011; see also DeRue&Ashford 2010).This is a process involving social perception insofar as targets see how other people treat them—such as deferring to them or deriding them—and they interpret this treatment in a way that increases (or decreases) their own self-identification as a leader, which, in turn, shapes how they behave. The first row of Figure 2 provides a visual depiction of the two basic causal relationships between the study of leaders and leading.

5.2. Contingent Relationships

Most contingent approaches examine how the consequences of leading are moderated by at- tributes of (or attributions employees make of ) the leader—a type of person-times-behavior

www.annualreviews.org • The Science of Leadership 73

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

The leader

Leading

The leader

Leading

The leader

Leading

The leader

Leading

The leader

Leading

The leader

Leading

The leader

Leading

The leader

Leading

The leader

Leading

The leader

Leading

The leader

The leader

Leading Leading

Outcome Cause

Outcome Cause

Causal relationships

Contingent relationships

Integration of causal and contingent relationships

Figure 2

Integrating theory on leaders and leading: causal versus contingent approaches. The figure outlines a number of ways that leader behaviors and leader-level concepts can be connected. For example, the effectiveness of vision communication (a leader behavior) might depend on the characteristics of the leader who is doing the communicating (a leader-level perception).

interaction. For example, Eisenberger et al. (2010) found that the association between leader– member exchange and affective organizational commitment was greater when employees believed their supervisor’s personal qualities reflected the central identity of the organization (e.g., a highly compassionate supervisor belonging to an organization that prides itself on compassionate rela- tions with the surrounding community). As another example,Wang et al. (2013) found that leader gender (a leader-level attribute) amplifies the negative link between authoritarian behaviors and subordinate performance, such that the relationship is stronger for women than men. They also found that the positive association between benevolent behaviors and subordinate performance is stronger for men.Wang et al.’s (2013) findings exemplify how examining the contingent relation- ship between leaders and leading is central to understanding the role of diversity in the leadership process. This is because diversity usually involves relatively immutable person-level attributes (as- pects of individual leaders). The second row of Figure 2 provides a visual depiction of the various ways that leading can be contingent on the relatively immutable features of leaders and also how outcomes associated with leaders can be contingent on specific behaviors associated with leading.

5.3. Approaches that Combine Causal and Contingent Relationships

An even more nuanced—and especially thought-provoking—possibility involves integrating causal and contingent approaches (see Figure 2, third row). For example, the relationship between a leader attribute and leading can be moderated by a separate leader attribute. Along these lines, Tost et al. (2013) found that a leader’s power increases their propensity to act in a

74 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

dominant manner, and, in turn, harms team communication and performance. However, this pattern only sustains when the person holds a formally recognized position as a leader, which influences whether and when other team members believe the leader’s authority is legitimate. Another possibility is that a separate leader attribute moderates the relationship between leading and a leader attribute. To illustrate, Wang et al. (2018) found that a leader action (expressing anger) influenced whether employees evaluated the leader positively; however, whether the leader was categorized as abusive moderated this effect.

6. ANALYSIS: MOVING FORWARD BY TAKING A CRITICAL LOOK BACKWARD

Substantial amounts of time, attention, and funding are dedicated every year to the study of lead- ership. As such, the stakes are not trivial for getting the science right. The two-part framework presented in this article—distinguishing between (a) the study of leaders and leading and (b) lead- ership as an independent versus dependent variable—may be helpful for not only synthesizing and reviewing existing research but also critically analyzing the state of the science in an effort to guide future research. I now pivot to this critical analysis. It leads to three basic conclusions. First, the most promising area of future research involves explicitly disentangling the role of leaders and leading within the same study. Second, critical areas of inquiry have gone almost entirely unexam- ined. Third, an extensive amount of research on leading has been dedicated to theories that may be misspecified. I suggest that these theories should be respecified, repurposed, or possibly even abandoned. I now expand on these three major issues while paying special attention to “bright spots”—existing studies of leadership that are exemplary and worth emulating (see Table 1 for a summary of the key recommendations from my critical analysis of the literature).

6.1. Doing More of What Works: Decoupling and Synthesizing Leaders and Leading

The main organizing principle of this article is the distinction between leaders and leading. Al- though it is true that leaders and leading should be theoretically and empirically separated, I do not suggest that they should be studied separately. After all, it is leaders who do the leading. Thus, it is inevitable that researchers will strive to theorize and test both leaders and leading simultane- ously.When this happens, it is easy to confuse and confound the two approaches from a theoretical and statistical point of view. As such, the question is not whether the study of leaders and leading should be integrated, but how this integration can be achieved.

Rather than studying leaders and leading in separate articles, I argue that the science of lead- ership will advance more systematically if scholars account for leaders and leading at the same time, but with separate variables. I covered examples of this approach in Section 5, Integrating the Study of Leaders and Leading. This includes research on how a person who exhibits certain be- haviors is more likely to be viewed as a legitimate leader [i.e., a behavior, such as a focus on change, causes someone to be classified as a leader (Gerpott et al. 2019)], leaders who are viewed as more prototypical have the confidence to lead more effectively ( Jackson & Johnson 2012), and certain forms of leading have effects only if they are enacted by people categorized as formal rather than informal leaders (Tost et al. 2013).

Figure 2 maps the ways that the study of leaders and leading can be integrated. Of special note are integrative approaches that have been largely overlooked. For instance, scholars over the past decade have largely neglected occasions when the effect of a leader attribute is moderated by leading (e.g., perhaps the effect of leader conscientiousness on group productivity is moderated by

www.annualreviews.org • The Science of Leadership 75

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

Table 1 Recommendations for future research

Focus area Dominant patterns in existing research Recommendations for future research Position of leadership within

causal models Positioning leading as an independent variable More frequently positioning leading as a

dependent variable Leading as an independent

variable Lumping multiple behaviors within a single

construct Focusing on features of single behaviors that occur in discrete episodes

Leading as a dependent variable

Examining attributes of training programs (e.g., program length)

Examining how effectively leaders acquire specific behaviors

Interplay between leaders and leading

Examining concepts associated with leaders (e.g., leader traits, how leaders are evaluated) separately from leading (e.g., the process in which people attempt to motivate others toward a common goal)

Focusing on the interplay between leaders and leading by modeling them as separate factors within the same theory

Behavioral core of leading (direction, initiation, and persistence)

Understanding how leaders provide direction Understanding how leaders boost action initiation and persistence

The leader as an independent variable

Identifying linear effects and main effects Identifying curvilinear effects and boundary conditions, especially reversals

Desirability of leadership Assuming that attaining leadership positions is desirable

Identifying costs of attaining and maintaining a leadership position, and taking self-selection more seriously (many people do not want to become the CEO)

Leader emergence and leader effectiveness

Focusing on separate causes of leader emergence and leader effectiveness

Determining whether the factors that predict leader emergence might detract from leader effectiveness, and vice versa

When an attribute of leading morphs into an attribute of a leader

Taking for granted the distinctions between ephemeral behaviors that are not tied to a person’s stable personality (e.g., one-off actions, behavioral episodes) and stable behavioral patterns (e.g., habits, behavioral orientations, traits)

Carefully delineating the relationship between short-term and long-term patterns of behavior as well as situational features that make people act in stable ways, even if they do not have an underlying disposition to do so

the leader’s clarity of communication) or when a predictor of leading is moderated by an attribute of the leader (e.g., perhaps the impact of a leader development program on vision communication depends on leader narcissism). To gain a well-rounded understanding of the points of overlap between leaders and leading, scholars should account for these possibilities.

Of course, scholars will often want to emphasize either leaders or leading in their theorizing and methods, depending on the research question. But even when scholars do not explicitly model both leaders and leading, I suggest that leadership research will advancemore systematically if they acknowledge that the effects of one likely depend on the other in their introduction and discussion sections. A norm of acknowledging the interplay between leaders and leading can guide future research while also preventing confusion about which level of analysis is being examined.

In short, distinguishing variables at the level of leaders (e.g., demographic characteristics) from variables at the level of leading (e.g., motivating followers to exert more effort) and then examin- ing their interplay will enable proper theoretical and empirical specification and, in turn, a more precise and complete picture of leadership in any given setting. Furthermore, by bringing them into the fold in the same studies while keeping them specified as separate constructs, the distinc- tions between leaders and leading may ultimately become clearer, potentially helping to prevent future instances of misspecification.

76 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

When researchers are interested in examining multiple behaviors in the same study, it is advis- able in most situations to separate them into different constructs in both the theory and methods. This can be done via variance partitioning models, in which researchers allow for the likelihood that each behavior has a unique set of antecedents and consequences, and that the best way to un- derstand how multiple behaviors influence the same outcome is by analyzing each one separately while controlling for the others. This can also be achieved via process models, in which scholars examine how leaders enact behaviors one after the other, such that each behavior can be enacted more effectively if it was preceded by a specific behavior. For example, the effectiveness of subgoals might depend on leaders first constricting attention to a single common goal (Carton 2018).

6.2. Improving How Scholars Prioritize Among Topics

With respect to the study of leaders, research on categorization, emergence, and evaluation contin- ues to be healthy. However, there are important opportunities for growth. One involves boundary conditions and second-order effects of leader traits, including work on curvilinear effects of Big Five factors, negative effects of ostensibly “good” qualities (e.g., extroversion and conscientious- ness), and conditions that complicate the negative effects of dark qualities such as narcissism. Ad- ditionally, research on the downsides of attaining leadership positions for the leaders themselves is an especially intriguing new area worthy of greater attention (Gabriel et al. 2020). Another area that has been overlooked has been the role of culture. Little is currently known about how the cultural background of leaders attenuates or amplifies certain effects between leading and key out- comes. Additionally, little is known about how culture influences how leaders are appraised and which leaders are selected (compare with Gündemir et al. 2019).

In regards to the study of leading, scholars can employ a dimensional scaling approach to es- tablish the extent to which different behaviors are more or less indicative of the three core, defini- tional criteria of leading: directing, initiating, and maintaining action toward a common purpose (see Figure 1). In the spirit of Hollenbeck et al.’s (2012) dimensional scaling approach, scholars can open their construct definition sections by locating the behavior(s) under investigation on all three dimensions as if they were coordinates on a Leadership GPS Map. Such an approach can enable a more systematic and linear accumulation of research. It can also help identify topics that are deeply relevant to some or all of the three aspects of leadership but have not received commen- surate attention. Whereas some concepts, including vision communication, are being examined with increasing frequency, other concepts are relatively neglected. Of particular note are behav- iors that involve helping people initiate action. Motivating employees to take action with tactics such as goal setting and removing obstacles was a key consideration in past decades, including in work that focused on initiating structure ( Judge et al. 2004), leadership and goal setting (Locke 1999), and path-goal theory (House 1996). In recent years, however, these topics have fallen out of favor.

A dimensional scaling approach can also help identify false negatives—topics that technically fall within the domain of leading but were not labeled as such by the study authors. This may in- volve forms of goal setting andmeaning-making that involvemanagers helping people understand, and better regulate effort in service of, the purpose that unifies their efforts (Kim 2018). Addition- ally, it can help scholars identify false positives—concepts that are labeled leadership even though they only meet the criteria of leading to a negligible degree. For instance, laissez-faire leadership involves apathy and neglect and thus should probably not bear the label of leadership, even if it is performed by a person who has the formal title of a leader (Breevaart & Zacher 2019).

Finally, whereas a wealth of research has been invested in leader-level attributes as independent and dependent variables and leading as an independent variable, much less attention has been

www.annualreviews.org • The Science of Leadership 77

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

dedicated to the study of leading as a dependent variable (compare with Lacerenza et al. 2017). Yet this area may be the most practically relevant of all areas of leadership research, because it pertains to how people can acquire behaviors that are central to leadership and known to predict valued outcomes. Toward this end, it is disconcerting that research with potentially the greatest practical import has received such little attention. As scholars focus more attention on leader interventions, I suggest that they focus their attention on the acquisition of specific behaviors, such as providing constructive feedback, rather than the attributes of the training program, such as whether feedback is given to leaders from multiple sources versus single sources or on-site versus off-site (compare with Lacerenza et al. 2017). When scholars and practitioners conceive of leadership training as the acquisition of specific behaviors (e.g., vision communication), it is likely to spur them to articulate the conditions (e.g., organizational culture) and interventions (e.g., mental time travel) that explicitly spur those behaviors.

6.3. (Mis)understanding the Multilevel Nature of Leadership

Following from the distinction between leaders (individuals who occupy positions) and leading (behaviors that influence employees), leadership is inherently a multilevel construct that can most precisely and comprehensively be investigated at two levels of analysis. According to the precepts of multilevel theory, behaviors are nested within leaders, not vice versa (Kozlowski & Klein 2000). That is, individual leaders enact many different behaviors (some but not all of which are associated with leading) across the span of each work day and work week. As they do so, they carry with them certain stable, if not immutable, characteristics—such as their personality traits and demographic characteristics. Said another way, behaviors might best be understood at the within-person level, whereas leader-level effects may best be understood at the between-person level. Such a concep- tion of leadership is consistent with both multilevel theory as well as empirical manifestations of multilevel theory, such as repeated measures, clustering, fixed effects, and hierarchical linear modeling. For instance, leader-level effects can be examined as fixed effects, whereas each leader’s behaviors (and their unique consequences) can be modeled to vary week to week, day to day, or even moment to moment (Kozlowski & Klein 2000, Dinh & Lord 2012).

Scholars have previously recognized similar dichotomies, such as the distinction between leader traits and leader behaviors (DeRue et al. 2011), how leaders are trained versus how leaders behave in medias res (Day 2000), or how leaders are perceived versus how they exact influence (Pfeffer 1977).However, the reality that these dichotomies reflect two different levels of analysis has largely been overlooked (compare with Yammarino et al. 2005,McClean et al. 2019). This point is critical because it determines a key form of theoretical and empirical specification: whether a theory’s assumptions, arguments, hypotheses, hypothesis tests, and conclusions consistently focus on the same level of analysis (leaders or leading) or focus on both levels of analysis in a consistent way.

The most important implication of this distinction is the following: Theories of leaders should be consistently specified at the leader level, and theories of leading should be consistently spec- ified at the level of behaviors. In a figural representation, this means that leader-level attributes would reside in a separate box that represents a higher level of analysis than any single, observable behavior and the causal pathway in which it is embedded (see Figure 1). As such, a precise investi- gation of leadership—such as identifying causes and effects or detailing process models—typically requires that scholars separate leaders from leading when developing theory and measures, even when (as recommended above) they investigate both levels of analysis within a single theory, hy- pothesis, and/or study.This has especially important implications for one class of theories covered in this review: those that lump together many discrete leader behaviors within single constructs. I turn to this topic now.

78 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

6.3.1. The status quo: What would happen if lumped theories of leading were left un- changed? In the effort to achieve parsimonious theories, scholars often aggregate many discrete leader behaviors within single constructs. This can be well-justified, especially when scholars are interested in prediction rather than explanation. Indeed, in some cases it may be more useful, both theoretically and practically, to lump these behaviors under a single meta-construct than to split them. For example, in leadership development it can be extremely powerful to identify a bun- dle of conceptually similar behaviors that empirically covary and strongly predict either negative or positive outcomes—especially if leaders who acquire one behavior within that broader bundle automatically acquire most or all of the other behaviors in that bundle.

There are other advantages to lumped theories of leading beyond leader development.Lumped theories can be useful for leader selection because the covariation that they capture often reflects dispositions, orientations, styles, and traits that some people have and others do not. If some of these orientations are more predictive than are others, or more compatible with certain organi- zational cultures than are others, then they can benefit leader selection theories and programs. Similar to personality traits—which involve the study of dispositions that cut across time and contexts—scholars may be interested in determining whether an individual who consistently ex- hibitsmany related behaviors (e.g., those related to humility) will be poised to consistently improve effectiveness in a single unit across time (temporal stability) or in many different units (cross- situational stability).

Additionally, lumped constructs can help organize the literature by simplifying it and enable likeminded scholars to more easily identify collaborators and audiences who share a conceptual interest that is much broader than any single leader behavior. It is likely to be easier to identify an audience interested in humble leadership than one interested in leaders who admit mistakes. Indeed, one of the hallmarks of academic expertise is the ability to identify themes that unite underlying indicators rather than to push for more granular parsing and proliferation (Heath & Heath 2007).

The advantages of lumped theories are largely rooted in the reality that they are highly pre- dictive and have “broad bandwidth” insofar as they incorporate a good deal of information about leading within a single construct. However, there is a well-known trade-off between bandwidth and fidelity, and a corresponding trade-off between predictive power and explanatory power. Be- cause lumped theories are thematic abstractions that are removed from the dozens of individual behavioral episodes that characterize them, they may be highly predictive but will often obscure attempts by researchers to explain why, how, and when leading impacts key outcomes ( Judge et al. 2002). This is especially problematic given that most modern leadership theories prioritize not prediction but explanation: identifying antecedents, mediators, and moderators that illuminate why, how, and when leader behaviors yield specific consequences. This is likely because leading is a process that inherently involves the interplay between one person’s behavior and another person’s psychological states and reactions. That is, the notion of antecedents and explanatory mechanisms is baked into the concept itself. Furthermore, leadership is an applied concept in- sofar as theorists and practitioners are interested in how it can be improved, and to understand how something can be improved, it is critical to first understand how it works. The upshot is that scholars are especially interested in the conditions that shape how leadership processes unfold as well as why and when functional processes can be magnified and dysfunctional processes stymied. In other words, explanation reigns supreme.

This leads to the Achilles heel of lumped theories: It becomes increasingly difficult to explain the causes and consequences of leader behaviors (a) as the number of leader behaviors within a lumped construct increases and (b) as the antecedents, mediators, and moderators of the different

www.annualreviews.org • The Science of Leadership 79

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

leader behaviors within a lumped construct become increasingly varied in quantity and in kind. This is in large part because the antecedents, moderators, and mediators are likely to influence, and be influenced by, each constituent element of the lumped construct in different ways. Imag- ine a hypothetical theory that includes two behaviors: how effectively a leader provides feedback and how fairly a leader makes resource allocation decisions. These two behaviors may empirically covary, but suppose researchers want to examine what predicts these behaviors. It is highly likely that the predictors of one behavior are distinct from the other, and these differences may be over- looked, obscured, or warped in order to conjoin them at the same level of analysis as a two-factor lumped construct of leading.

Now suppose researchers also want to specify a serial mediation chain that explains the im- pact of these two behaviors on employee commitment. The chances are now even higher that feedback and fairness will have distinct causal pathways. Not only are they likely to have distinct antecedents, but they are also likely to impact commitment through distinct mediation chains. Similar to the lumped leadership construct, these mediation chains can be understood via lumped or emergent constructs, but it would be difficult or impossible for these constructs to reflect the influence process as it happens in a work context, insofar as specific leader actions alter specific psychological states of employees, and, in turn, how they behave. Perhaps leader feedback impacts commitment primarily by increasing follower self-efficacy, whereas leader fairness impacts com- mitment primarily through a follower’s belief in a just world. Even if leader fairness and feedback covary, it would not adequately reflect the influence process associated with this two-factor form of leading to omit one of these two mediators or to lump both of them together into a meta- construct that is so general that it ceases to be meaningful. Either approach would obscure how this form of leading exacts its impact in the work context.

Now suppose the researchers want to specify moderators—that is, which conditions attenuate the causes or effects of feedback and fairness. It is likely that feedback and fairness have distinct moderators, even in the very unlikely case that the above-mentioned antecedents and mediators are the same. Furthermore, the moderators may be different for feedback versus fairness on dif- ferent mediated pathways (Dinh & Lord 2012).

If a researcher investigates the impact of this lumped meta-construct on employee commit- ment, then it becomes difficult or impossible to answer many important questions. Is one behav- ior more important for explaining commitment? Does each behavior influence commitment via unique mediators? Is the effect of one of these behaviors, but not the other, contingent on a factor such as resource allocation? Is one behavior easier to influence via training than the other? Are there interdependencies between both behaviors, such as interactions (e.g., perhaps specific versus vague feedback only impacts commitment when the leaders who provide this feedback also make ethical decisions)?

Now suppose scholars examine not two correlated leader behaviors but five, ten, or fifteen.The possibility of misspecification in the process of explanation rises from probable to almost certain, until it becomes almost impossible to isolate the nomological network of discrete leader behaviors from one another, both theoretically and empirically. Even scholars who comprehensively account for a host of antecedents,moderators, and mediators of varied leader behaviors will have a difficult time identifying (and explaining coherently) which of them gowith which leader behaviors.This is a problem from a conceptual level, as a given theory can become unwieldy and haphazard quickly. But then it becomes an even bigger problem empirically, because lumped theories are almost always tested as unitary factors that are holistically predicted by a set of antecedents, holistically predict mediators and consequences, and are holistically conditioned by moderators.

Taken together, the chance of misspecification, imprecision, error, and confusion rise consid- erably, perhaps exponentially, as scholars expand their focus on a broader nomological network

80 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

in an effort to explain why, how, and when various leader behaviors impact specific outcomes. Given that leading, by definition, is a subtype of social influence, it requires the examination of why, how, and when one person influences other people via discrete behaviors and the psycho- logical states that each of those behaviors uniquely shapes. Aggregating behaviors at a higher level of analysis ceases to reflect social influence in this way. This is a particularly acute prob- lem for the study of social influence in organizational psychology and organizational behavior. Understanding a social influence process in organizational psychology typically requires (at a minimum) a leader behavior and a follower psychological state—and in organizational behav- ior it typically requires (at a minimum) a leader behavior and a follower behavior—and the de rigueur approach these days is to identify a leader or follower psychological state that mediates this effect. Furthermore, many theories currently include at least one moderator. As such, the- ories of leading can only meaningfully explain a social influence process when several mecha- nisms of explanation are included, increasing the likelihood that lumped theories of leading will be misspecified.

In sum, the distinction between explanation and prediction puts a damper on efforts to in- crease the parsimony of theories of leading by lumping together “like with like.” The band- width gained by including more leader behaviors within a single meta-construct is likely offset by the inability to accurately and precisely understand why, how, and when different forms of lead- ing impact consequences such as employee commitment, burnout, engagement, well-being, and performance.

A related issue is that covarying is distinct from co-occurring. At any given point during the work day, leaders exhibit a limited set of their behavioral repertoire, and each individual behavior likely triggers a unique set of consequences via a unique causal pathway. That is, theoretical and analytical approaches to studying leading must account for how leading is a process that occurs in situ. This is especially imperative given that day-to-day behavior may vary more within leaders than between leaders (McClean et al. 2019).

In this vein, constructs that lump many behaviors into a single construct are more challeng- ing to study because the odds of identifying each of these aspects of the nomological network are lower. As indicated in my review above, these constructs incorporate several to several dozen behaviors—and often the measures also lump in non-behaviors, including assessments of leader traits (e.g., blanket assessments of charisma, ethicality, and humility) and evaluators’ beliefs and attitudes about the leader (e.g., trust in the leader). As a result, it is almost impossible to tease apart the within-person, within-dyad, and within-situation causes and effects of many of these constructs. Often scholars can only speculate about these aspects of the influence process.

There are still other limitations that further exacerbate concerns with lumping.One of them is that whereas lumping is popular among psychologists who study traits and dispositions, scholars who examine other influence processes tend to focus on discrete behaviors that unfold within a single episode of time, such as what is prevalent in research on voice or behavioral decision theory. To attract scholars whose backgrounds and identities reside outside of leadership, it is important to recognize that theymay bemore familiar with the custom of differentiating among thematically similar constructs (e.g., studying only one form of justice at a time, such as procedural justice) than aggregating them.

A final concern with lumped theories of leading is that the methods used—often questionnaires that rely on individuals to rate leaders on multiple dimensions—are deeply susceptible to halo effects such that it is not clear whether these behaviors actually do covary, wherein leaders who practice one behavior actually engage in a behavior found to covary with it, or whether evaluators merely perceive an illusory covariation.

www.annualreviews.org • The Science of Leadership 81

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

6.3.2. The way forward: Should lumped theories of leading be respecified, repur- posed, or abandoned? Given the challenges associated with lumped constructs of lead- ing, I summarize three possible alternative paths forward: respecification, repurposing, and abandonment.

6.3.2.1. The case for respecifying lumped conceptualizations of leading. Respecification would involve shifting lumped theories from conceptualizations of leading to conceptualizations of lead- ers. This would fundamentally change the assumptions and arguments used to build and analyze each theory. One option is to respecify them as implicit theories. In survey research, peoples’ im- plicit theories of leadership explain as much variation in their perceptions of leader behavior as leader behavior itself (Wang et al. 2019). For example, when a leader looks a certain way (e.g., possesses dominant facial features), belongs to a high performing unit, or enacts just one behavior associated with good leadership (e.g., respectful treatment), various perceptual biases can come into play, including halo effects and associative memory (the tendency for the perception of one behavior to activate perceptions of other behaviors in a semantic network of receptors inmemory). As a result, perceivers will be inclined to “see” in that leader other behaviors associated with good leadership that might not objectively be there. Ultimately, constructs such as authentic leadership and servant leadership might indicate peoples’ implicit theories about their leaders better than their leaders’ actual behavior. Given that surveys are such a prevalent method for assessing these forms of leadership, it may be the case that existing research is already tapping into followers’ im- plicit leadership theories more than actual leader behaviors, suggesting that it may be possible for scholars to pivot how they conceptualize these theories more so than change themethods they use.

A second option is to respecify them as trait-based theories. This may be especially useful for theories of leadership that directly correspond to personality traits that reflect a person’s overrid- ing behavioral dispositions. For example, servant leadership may reflect a trait similar to the duty subdimension of conscientiousness (Colbert & Witt 2009), and humble leadership may be inte- grated more directly with work on the trait of humility ( Johnson et al. 2011).This could be a boon for research on leader selection systems. Research from the Ohio State Leadership Studies in the middle of the twentieth century is an example of how a leadership model can evolve from strictly a behavioral theory of leading to a description of leader styles. This research program initially began as a battery of scales intended to distinguish between dozens of possible leader behaviors, but factor analyses unearthed two stable, basic dimensions (consideration and initiating structure) that explained a substantial degree of variance in key leadership outcomes ( Judge et al. 2004). By the end of the twentieth century, these two dimensions came to be viewed as behavioral tendencies (markers of individual leaders) rather than behavioral episodes (markers of leading).

To determine whether to respecify theories as implicit theories or traits, scholars can conduct further methodological studies to parse the extent to which survey responses represent disposi- tional tendencies (humble leaders act in consistent ways onmultiple dimensions) versus perceptual biases (people see covariation among similar behaviors even if this covariation does not exist in reality) via a variety of nonsurvey methods, including observational and archival studies.

6.3.2.2. The case for repurposing lumped conceptualizations of leading. Repurposing is an alternative to respecification, as it does not involve changing the assumptions or tenets of a theory but rather how a theory is employed. One possibility involves more frequently employing lumped theories as dependent variables in main effects models. This would involve a substantial pivot, given that the vast majority of research on lumped theories of leading features them as predictors of downstream outcomes. But it would yield numerous advantages. It would reduce the need for

82 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

lumped theories to focus on explaining a complex, multistaged, iterative social influence process. Rather, the focus can be more exclusively placed on leader behavior. This can be especially useful for research on leadership development and training. For example, in leadership development it may be extremely powerful to identify a bundle of conceptually similar behaviors that empirically covary—especially if leaders who acquire one behavior within that broader bundle automatically acquire most or all of the other behaviors in that bundle.

As a hypothetical example, consider humble leadership. Imagine that the process of learning and practicing one key behavior associated with humble leadership—admitting mistakes—spurs a process of generalized habit formation and associative learning that automatically triggers the other behaviors associated with this form of leading, such as spotlighting follower strengths,mod- eling teachability, granting voice, legitimizing follower development, and reversing roles with fol- lowers. In this example, a leader needs to acquire only one orienting behavior to establish a more general disposition of humility that spreads to dozens of inter-related behaviors that the leader did not consciously acquire but still reliably exhibits. This example assumes that the covariation among the distinct behaviors associated with humble leadership is both formative and reflective (MacCallum & Browne 1993). For the formative path, a single indicator (in this case, admitting mistakes) partly or wholly causes a broader latent variable (in this case, humble leadership). For the reflective path, the increase in humble leadership boosts the remaining indicators associated with that form of leadership (e.g., spotlighting follower strengths, giving voice, and modeling teach- ability). This approach essentially involves exploiting the covariation between behaviors lumped within meta-constructs.

Along these lines, lumped theories can help leaders better comprehend and remember leader behaviors when participating in leadership development programs. In this way, a feature of these theories that is often problematic (that they activate and influence perceivers’ implicit theories of leadership) can be used for good. Research on cognitive psychology suggests that people are better able to remember ideas when they are chunked within a higher-order label (Gobet 2005). As such, even if servant leadership—as a theory of leading—cannot tease apart how, when, and why each of the 36 items impacts downstream outcomes, the reality that the range of behaviors and perceptions these items capture are packaged underneath a single umbrella term highlight- ing their commonalities can make training and development more efficient and effective. This is theoretically meaningful because it identifies where leader behaviors come from and practically useful because it enables practitioners to focus on a common underlying theme that unifies highly predictive behaviors rather than a multitude of disaggregated behaviors.

A final option involves treating lumped theories of leading as parent literatures that can engage and orient readers, given that readers are familiar with ideas such as transformational leadership, but then focus both hypothesis development as well as each study’s empirical investigation on a specific behavior or set of disentangled behaviors that fall within the parent literature. For ex- ample, Ashford et al. (2018) opened their paper by discussing humble leadership; however, they then immediately focused their analysis on two individual behaviors, feedback seeking and vision articulation, which is more precise from the standpoint of theoretical specification. They then manipulated both behaviors with two different treatment conditions in a factorial design, which is more precise from the standpoint of empirical specification.

6.3.2.3. The case for abandoning lumped conceptualizations of leading. Despite considerable upside, both respecification and repurposing are laden with risks. First, it would not be clear how to reconcile traits such as servant leadership with other traits (especially the Big Five). Second, scholars would have to tease apart the distinctions between traits versus styles, orientations, and foci—labels that have all been used to capture similar ideas. Third, existing methodologies often

www.annualreviews.org • The Science of Leadership 83

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

cannot tease apart the extent to which a construct such as servant leadership is an implicit theory such that respondents to surveys (a) subconsciously categorize leaders as more or less servant- like and then rate them according to a series of behaviors that merely reflect an implicit theory, or (b) record the leader’s behaviors in a fashion that reflects the leader’s actual behavior. In the former case, covariation among behaviors is a halo effect. In the latter case, the behaviors actually do covary in real life. Fourth, some of the lumped conceptualizations of leading mix behaviors (e.g., the leader asking for input) with traits, orientations, tendencies, characteristics, and styles (e.g., “the leader is caring” or “the leader helps develop my career”). This would make it difficult to respecify a lumped theory’s entire set of subdimensions and items as behaviors.

It would also be challenging to repurpose theories of leading from an independent to a depen- dent variable. If scholars take this approach, they would still run into problems with specification, because it might be difficult to learn if, when, and why leaders learn one behavior of, say, servant leadership better than another. Additionally, scholars would not be able to study leading simul- taneously as an independent and dependent variable, because of the aforementioned problems associated with specification as an independent variable. Furthermore, if the subdimensions of a construct such as servant leadership, empowering leadership, or humble leadership do covary in real life (i.e., their covariation is not a statistical artifact), the fused nature of all of the individual behaviors makes it challenging to determine whether the behaviors all have a common cause that is omitted or if one behavior is more fundamental insofar as it activates the others.

Finally, there are pragmatic considerations that reflect how scholars do their work and how knowledge tends to accumulate. Respecifying/repurposing lumped constructs can create confu- sion for scholars who want to synthesize research before and after the shift. It also may create issues when some researchers shift specification but others do not or when journal referees use their prior understanding of the theory’s specification to evaluate a theory’s updated specification.

In short, the prospect of respecifying/repurposing lumped theories is laden with coordination costs that further imperil the accumulation of knowledge. When these concerns are considered in tandem with other known limitations of lumped theories, including the tendency for surveys associated with these theories to make use of double-barreled items and for some subdimensions to be tautological (van Knippenberg & Sitkin 2013), it leaves me torn about whether to recom- mend a pivot toward respecification/repurposing of lumped constructs or a “clean break” such that these constructs are abandoned. The former step might preserve the value and evocativeness of longstanding leadership theories, maintain continuity in the labels used to understand leader- ship, and respect the decades of research and smart minds that have been dedicated to their study. At a minimum, these theories not only have established predictive validity but also have served as shared forms of meaning that galvanize communities of leadership scholars.

But my own view is that the latter approach—a wholesale break from the past—might be most generative. It would require leadership scholars to return to first principles without being con- strained by the vestiges of constructs that are, in many ways, beyond repair. Fresh starts allow for clear thinking and stem escalation of commitment (Alvesson & Einola 2019, Gottfredson et al. 2020, van Knippenberg & Sitkin 2013). I encourage other scholars to articulate points of agree- ment and disagreement with my position here and, ultimately, to wager whether lumped concep- tualizations of leading should be respecified, repurposed, abandoned, or (if I am misguided in my criticisms) left as is.

6.3.3. Avoiding future instances of misspecification. The abundance of lumped theories raises the question of why they are so popular, despite their inherent limitations. In an effort to guide future research, I speculate about some reasons here.

The first involves etymology—theories that emerged for one purpose and then drifted to another without clear justification. Many of the most influential lumped conceptualizations of

84 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

leadership, including transformational leadership and charismatic leadership, had their origins in social science disciplines other than organizational behavior. For example, the idea of a transforming leader arose in history to explain the arc of specific leaders, such as Gandhi, across time and place (Burns 1978). Accordingly, these perspectives were not intended to facilitate the study of the causes and effects of individual behaviors in situ.

The second is intuition. The study of leaders is intuitive, whereas the study of leading is not. It is easier to think about the phenomenon of leadership at the level of the leader than the level of behaviors (Schlenker & Weigold 1989). People tend to classify and simplify the world at the level of the individual, such that it is easier to make assessments about individual traits or general states of being than to make assessments about individual behaviors that are not explicitly tied to identifiable individuals (Macrae & Bodenhausen 2000). People are easily distinguishable from one another, whereas behaviors are fuzzy and sometimes spill into each other. A person is perceived as a complete whole; a behavior, less so. In this way, it may appear to make more sense to think that the “atom”—the most fundamental unit of analysis—in any given situation involving leadership is the leader rather than a behavior associated with leading. Moreover, people do not merely find a focus on the individual leader as more intuitive but also preferable. Hunt (1999) noted that transformational and charismatic leadership rejuvenated the study of leadership in part because of their focus on extraordinary individuals (e.g., Nelson Mandela and Margaret Thatcher) rather than stand-alone behaviors. But what is intuitive and easily understandable does not necessarily most closely reflect the phenomenon of interest.

The third reason is convention. There are several common practices that reinforce the tendency to lump multiple behaviors into single constructs. One popular approach is to create a construct and label it “leadership” with a modifier before the word leadership (e.g., authentic leadership, charismatic leadership, empowering leadership, ethical leadership, humble leadership, servant leadership, or transformational leadership). This convention implies that a set of leader behaviors can best be understood as a single theoretical construct that can be modeled in an undifferentiated manner. Instead, it may be more helpful to consider a label such as “empowering leadership” not as a construct that can be hypothesized or modeled as a single variable, but as an area of research or a subdiscipline that captures a range of related behaviors that need to be teased apart in separate programs of research.

There are also methodological conventions that encourage misspecification (van Knippenberg & Sitkin 2013). Perceivers are unlikely to detect an accurate amount of variation across discrete behaviors practiced by a given leader when asked about these discrete behaviors all at once, even via validated scales. These devices are susceptible to halo effects, such that leaders who are su- perlative in one sense are viewed as superlative in others. And even when scholars correlate survey measures with objective measures of performance in order to reduce common method variance (Podsakoff et al. 2003), the surveys may still be biased by the tendency for perceivers to attribute good things to leaders who lead in the context of success. Thus, although it is typically appro- priate for social cognitive theories of leaders to use methods that reflect perception, cognition, and categorization (e.g., surveys with self-report questionnaire items or Q sorting), these meth- ods may be less appropriate for theories of leading because of the tendency for halo effects. These shortcomings highlight a danger of statistical machinery. When scales with excellent psychomet- ric properties reinforce a theory’s flaws (specification at the level of the leader rather than the level of the behavior), they may create illusory empirical support for those flaws. The methodology al- most perfectly obscures misspecification, leading to a cycle whereby theories that do not provide explanatory power at the level that the phenomenon takes place (as an influence process between two or more individuals) appear strongly supported, and, in turn, beget more scholarly attention (van Knippenberg & Sitkin 2013).

www.annualreviews.org • The Science of Leadership 85

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

By contrast, unobtrusive methods such as content analysis (e.g., video, audio, or text coding) can be particularly effective for assessing leading as long as the observable manifestations of the behaviors are well-defined. Content analysis allows raters to account for discrete leader behaviors before performance outcomes are measured. Additionally, controlled experiments certainly can (and should) be used to capture the consequences of single behavioral episodes in an internally valid way; however, it is better to examine multiple leadership behaviors in a factorial design than to force-fit them into a single treatment condition.

The final consideration is practicality. Whereas scholars can examine entire traits of leaders in one study, it is impossible to capture the construct of leading in one study; rather, it requires a program of research. Consistent with the issue of bandwidth fidelity and the trade-off between comprehensiveness and precision (Ones & Viswesvaran 1996), although it may appear that more variance in outcomes can be explained by specifying behavior at the level of the leader, it is much more difficult to precisely specify causes and effects at this level of analysis.

Given that the limitations of lumped theories are so firmly rooted in biases that all researchers (that is, all humans) have, including the desire to follow convention, intuition, and practicality, it is essential that the community of leadership scholars generate norms to guard against gratu- itous lumping. I suggest that the best ways to do this are for more researchers to ask whether the antecedents, mediators, and moderators that are most frequently studied in tandem with lumped theories can explain all of the behaviors that are subsumed within a lumped theory in the same way. Even though a lumped theory may appear to have substantial reach, that reach may come at the cost of so much precision that it renders the science untenable. As a result, it may be best for scholars to respecify, repurpose, or abandon lumped theories and direct more of their attention to isolating single leader behaviors.

7. CONCLUSION

The distinctions between (a) leaders and leading and (b) leadership as a cause versus a consequence can provide scholars with a framework for organizing existing research, identifying areas in need of greater scrutiny, and directing attention toward the most promising theories and paradigms. Overall, the study of leaders stands in great health. By contrast, the study of leading is haphaz- ard, especially when it is featured as an independent variable. To continue to advance the science of leading, scholars should pay particular attention to the attributes that make lumped theories seductive despite their serious limitations and ask whether the upside to these theories is worth the considerable cost. Although focusing on the causes and consequences of individual leader be- haviors may result in leadership research plodding along in a highly piecemeal and fragmented fashion, such an approach may be the best way to do justice to the complexity of the phenomenon and slowly but surely accumulate evidence in the most rigorous fashion possible.

DISCLOSURE STATEMENT

The author is not aware of any affiliations,memberships, funding, or financial holdings that might be perceived as affecting the objectivity of this review.

ACKNOWLEDGMENTS

I’m indebted to Annual Review of Organizational Psychology and Organizational Behavior Associate Editor Sue Ashford and Editor Frederick Morgeson for their detailed feedback throughout the review process. I also thank Samir Nurmohamed, Michael Parke, and Ned Wellman for helpful conversations.

86 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

LITERATURE CITED

AlvessonM,EinolaK.2019.Warning for excessive positivity: authentic leadership and other traps in leadership studies. Leadersh. Q. 30(4):383–95

Amundsen S,Martinsen ØL. 2014. Empowering leadership: construct clarification conceptualization and val- idation of a new scale. Leadersh. Q. 25(3):487–511

Antonakis J, Bastardoz N, Jacquart P, Shamir B. 2016. Charisma: an ill-defined and ill-measured gift. Annu. Rev. Organ. Psychol. Organ. Behav. 3:292–319

Antonakis J, House RJ, Simonton DK. 2017. Can super smart leaders suffer from too much of a good thing? The curvilinear effect of intelligence on perceived leadership behavior. J. Appl. Psychol. 102(7):1003–21

Ashford SJ, Sitkin SB. 2019. From problems to progress: a dialogue on prevailing issues in leadership research. Leadersh. Q. 30(4):454–60

Ashford SJ, Wellman N, Sully de Luque M, De Stobbeleir KE, Wollan M. 2018. Two roads to effectiveness: CEO feedback seeking vision articulation and firm performance. J. Organ. Behav. 39(1):82–95

Badura KL, Grijalva E, Galvin BM, Owens BP, Joseph DL. 2020. Motivation to lead: a meta-analysis and distal-proximal model of motivation and leadership. J. Appl. Psychol. 105(4):331–54

Balkundi P, Kilduff M, Harrison DA. 2011. Centrality and charisma: comparing how leader networks and attributions affect team performance. J. Appl. Psychol. 96(6):1209–22

Bass BM,Avolio BJ. 1994. Improving Organizational Effectiveness Through Transformational Leadership.Thousand Oaks, CA: Sage

Bedeian AG, Hunt JG. 2006. Academic amnesia and vestigial assumptions of our forefathers. Leadersh. Q. 17(2):190–205

Bedi A, Alpaslan CM, Green S. 2016. A meta-analytic review of ethical leadership outcomes and moderators. J. Bus. Ethics 139(3):517–36

Brands RA,Menges JI, Kilduff M. 2015. The leader-in-social-network schema: perceptions of network struc- ture affect gendered attributions of charisma. Organ. Sci. 26(4):1210–25

Breevaart K, Zacher H. 2019. Main and interactive effects of weekly transformational and laissez-faire lead- ership on followers’ trust in the leader and leader effectiveness. J. Occup. Organ. Psychol. 92(2):384–409

Burns JM. 1978. Leadership. New York: Harper and Row Carton AM. 2018. “I’m not mopping the floors, I’m putting a man on themoon”: howNASA leaders enhanced

the meaningfulness of work by changing the meaning of work. Adm. Sci. Q. 63(2):323–69 Carton AM, Lucas BJ. 2018. How can leaders overcome the blurry vision bias? Identifying an antidote to the

paradox of vision communication. Acad. Manag. J. 61(6):2106–29 Carton AM, Murphy C, Clark JR. 2014. A (blurry) vision of the future: how leader rhetoric about ultimate

goals influences performance. Acad. Manag. J. 57(6):1544–70 Carton AM, Rosette AS. 2011. Explaining bias against black leaders: integrating theory on information pro-

cessing and goal-based stereotyping. Acad. Manag. J. 54(6):1141–58 Chatterjee A, Hambrick DC. 2011. Executive personality, capability cues, and risk taking: how narcissistic

CEOs react to their successes and stumbles. Admin. Sci. Q. 56(2):202–37 Chang D, Johnson RE. 2010. Not all leader–member exchanges are created equal: importance of leader rela-

tional identity. Leadersh. Q. 21(5):796–808 Chen G, Sharma PN, Edinger SK, Shapiro DL, Farh JL. 2011. Motivating and demotivating forces in teams:

cross-level influences of empowering leadership and relationship conflict. J. Appl. Psychol. 96(3):541–57 Chen J, Nadkarni S. 2017. It’s about time! CEOs’ temporal dispositions temporal leadership and corporate

entrepreneurship. Adm. Sci. Q. 62(1):31–66 Chen Z, Zhu J, Zhou M. 2015. How does a servant leader fuel the service fire? A multilevel model of servant

leadership, individual self identity, group competition climate, and customer service performance. J. Appl. Psychol. 100(2):511–21

CheongM,Yammarino FJ,Dionne SD, Spain SM,Tsai CY. 2019. A review of the effectiveness of empowering leadership. Leadersh. Q. 30(1):34–58

Chiu CYC, Balkundi P, Weinberg FJ. 2017. When managers become leaders: the role of manager network centralities, social power, and followers’ perception of leadership. Leadersh. Q. 28(2):334–48

www.annualreviews.org • The Science of Leadership 87

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

Colbert AE,Witt LA. 2009. The role of goal-focused leadership in enabling the expression of conscientious- ness. J. Appl. Psychol. 94(3):790–96

Cronbach LJ, Gleser GC. 1957. Psychological Tests and Personnel Decisions. Champaign: Univ. Ill. Press Day DV. 2000. Leadership development: a review in context. Leadersh. Q. 11(4):581–613 Day DV, Dragoni L. 2015. Leadership development: an outcome-oriented review based on time and levels of

analyses. Annu. Rev. Organ. Psychol. Organ. Behav. 2:133–56 Day DV, Sin HP. 2011. Longitudinal tests of an integrative model of leader development: charting and un-

derstanding developmental trajectories. Leadersh. Q. 22(3):545–60 Den Hartog DN. 2015. Ethical leadership. Annu. Rev. Organ. Psychol. Organ. Behav. 2:409–34 DeRue DS, Ashford SJ. 2010. Who will lead and who will follow? A social process of leadership identity

construction in organizations. Acad. Manag. Rev. 35(4):627–47 DeRue DS,Nahrgang JD, Ashford SJ. 2015. Interpersonal perceptions and the emergence of leadership struc-

tures in groups: a network perspective.Organ. Sci. 26(4):1192–1209 DeRue DS, Nahrgang JD, Hollenbeck JR, Workman K. 2012. A quasi-experimental study of after-event re-

views and leadership development. J. Appl. Psychol. 97(5):997–1015 DeRue DS, Nahrgang JD, Wellman E, Humphrey SE. 2011. Trait and behavioral theories of leadership: an

integration and meta-analytic test of their relative validity. Pers. Psychol. 64(1):7–52 Dinh JE, Lord RG. 2012. Implications of dispositional and process views of traits for individual difference

research in leadership. Leadersh. Q. 23(4):651–69 Eisenberger R, Karagonlar G, Stinglhamber F, Neves P, Becker TE, et al. 2010. Leader–member exchange

and affective organizational commitment: the contribution of supervisor’s organizational embodiment. J. Appl. Psychol. 95(6):1085–1103

Emrich CG, Brower HH, Feldman JM, Garland H. 2001. Images in words: presidential rhetoric, charisma, and greatness. Adm. Sci. Q. 46(3):527–57

Epitropaki O,Martin R. 2004. Implicit leadership theories in applied settings: factor structure, generalizability, and stability over time. J. Appl. Psychol. 89(2):293–310

Feng JM, Liu SS, Wu KJ, Wang HC. 2014. A review of the literature of humble leadership. Foreign Econ. Manag. 36:38–48

Foti RJ, Hansbrough TK, Epitropaki O, Coyle PT. 2017. Dynamic viewpoints on implicit leadership and followership theories: approaches, findings, and future directions. Leadersh. Q. 28(2):261–67

Foulk TA, Lanaj K, Tu MH, Erez A, Archambeau L. 2018. Heavy is the head that wears the crown: an actor- centric approach to daily psychological power, abusive leader behavior, and perceived incivility. Acad. Manag. J. 61(2):661–84

Gabriel AS, Lanaj K, Jennings RE. 2020. Is one the loneliest number? A within-person examination of the adaptive and maladaptive consequences of leader loneliness at work. J. Appl. Psychol. https://doi.org/10. 1037/apl0000838. In press

GardnerWL,Cogliser CC,Davis KM,Dickens MP. 2011. Authentic leadership: a review of the literature and research agenda. Leadersh. Q. 22(6):1120–45

Gary MS, Yang MM, Yetton PW, Sterman JD. 2017. Stretch goals and the distribution of organizational performance. Organ. Sci. 28(3):395–410

Gerpott FH, Lehmann-Willenbrock N, Voelpel SC, Van Vugt M. 2019. It’s not just what is said but when it’s said: a temporal account of verbal behaviors and emergent leadership in self-managed teams. Acad. Manag. J. 62(3):717–38

Glynn MA, Raffaelli R. 2010. Uncovering mechanisms of theory development in an academic field: lessons from leadership research. Acad. Manag. Ann. 4(1):359–401

Gobet F. 2005. Chunking models of expertise: implications for education. Appl. Cogn. Psychol. 19(2):183–204 Gottfredson RK,Wright SL, Heaphy ED. 2020. A critique of the Leader-Member Exchange construct: Back

to square one. Leadersh. Q. 31(6):101385 Grant AM. 2012. Leading with meaning: beneficiary contact, prosocial impact, and the performance effects

of transformational leadership. Acad. Manag. J. 55(2):458–76 Grant AM,Gino F,Hofmann DA. 2011. Reversing the extroverted leadership advantage: the role of employee

proactivity. Acad. Manag. J. 54(3):528–50

88 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

Gravetter FJ, Forzano LAB. 2011. Research Methods for the Behavioral Sciences. Belmont, CA: Wadsworth. 4th ed.

Griffin MA, Parker SK, Mason CM. 2010. Leader vision and the development of adaptive and proactive per- formance: a longitudinal study. J. Appl. Psychol. 95(1):174–82

Guillén L,Mayo M,Korotov K. 2015. Is leadership a part of me? A leader identity approach to understanding the motivation to lead. Leadersh. Q. 26(5):802–20

Gündemir S, Carton AM, Homan AC. 2019. The impact of organizational performance on the emergence of Asian American leaders. J. Appl. Psychol. 104(1):107–22

Gupta A, Misangyi VF. 2018. Follow the leader (or not): the influence of peer CEOs’ characteristics on in- terorganizational imitation. Strateg. Manag. J. 39(5):1437–72

Hall RJ, Lord RG, Foster KE. 2009. Considerations in applying the social relations model to the study of leadership emergence in groups: a leadership categorization perspective. InMulti-Level Issues in Organi- zational Behavior and Leadership. Bingley, UK: Emerald Group

Heath C, Heath D. 2007.Made to Stick: Why Some Ideas Survive and Others Die. New York: Random House HoggMA, van KnippenbergD,Rast DE III. 2012.The social identity theory of leadership: theoretical origins,

research findings, and conceptual developments. Eur. Rev. Soc. Psychol. 23(1):258–304 Hollenbeck JR, Beersma B, Schouten ME. 2012. Beyond team types and taxonomies: a dimensional scaling

conceptualization for team description. Acad. Manag. Rev. 37(1):82–106 House RJ. 1996. Path-goal theory of leadership: lessons legacy and a reformulated theory. Leadersh. Q.

7(3):323–52 Hu J, Erdogan B, Jiang K, Bauer TN, Liu S. 2018. Leader humility and team creativity: the role of team

information sharing, psychological safety, and power distance. J. Appl. Psychol. 103(3):313–23 Hu J,Liden RC. 2011.Antecedents of team potency and team effectiveness: an examination of goal and process

clarity and servant leadership. J. Appl. Psychol. 96(4):851–62 Hunt JG. 1999. Transformational/charismatic leadership’s transformation of the field: an historical essay.

Leadersh. Q. 10(2):129–44 Jackson EM, Johnson RE. 2012. When opposites do (and do not) attract: interplay of leader and follower

self-identities and its consequences for leader–member exchange. Leadersh. Q. 23(3):488–501 JohnsonMK,RowattWC,Petrini L. 2011.A new trait on themarket: Honesty–Humility as a unique predictor

of job performance ratings. Pers. Individ. Differ. 50(6):857–62 Johnson RE, King DD, Lin SHJ, Scott BA,Walker EMJ,Wang M. 2017. Regulatory focus trickle-down: how

leader regulatory focus and behavior shape follower regulatory focus. Organ. Behav. Hum. Decis. Process. 140:29–45

Johnson RE, Venus M, Lanaj K, Mao C, Chang CH. 2012. Leader identity as an antecedent of the fre- quency and consistency of transformational consideration and abusive leadership behaviors. J. Appl. Psychol. 97(6):1262–72

Judge TA, Bono JE, Ilies R, Gerhardt MW. 2002. Personality and leadership: a qualitative and quantitative review. J. Appl. Psychol. 87(4):765–80

Judge TA, Piccolo RF, Ilies R. 2004.The forgotten ones? The validity of consideration and initiating structure in leadership research. J. Appl. Psychol. 89(1):36–51

Kalshoven K, Den Hartog DN, De Hoogh AH. 2011. Ethical leadership at work questionnaire (ELW): de- velopment and validation of a multidimensional measure. Leadersh. Q. 22(1):51–69

Kearney E, Shemla M, van Knippenberg D, Scholz FA. 2019. A paradox perspective on the interactive effects of visionary and empowering leadership.Organ. Behav. Hum. Decis. Process. 155:20–30

Kenny DA, Livi S. 2009. A componential analysis of leadership using the social relations model. InMulti-Level Issues in Organizational Behavior and Leadership. Bingley, UK: Emerald Group

Kenny DA, Zaccaro SJ. 1983. An estimate of variance due to traits in leadership. J. Appl. Psychol. 68(4):678–85 Kim Y. 2018. Enhancing employee communication behaviors for sensemaking and sensegiving in crisis situ-

ations: strategic management approach for effective internal crisis communication. J. Commun. Manag. 22(4):451–75

Kniffin KM, Detert JR, Leroy HL. 2020. On leading and managing: synonyms or separate (and unequal)? Acad. Manag. Discov. 6(4):544–71

www.annualreviews.org • The Science of Leadership 89

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

Koenig AM, Eagly AH, Mitchell AA, Ristikari T. 2011. Are leader stereotypes masculine? A meta-analysis of three research paradigms. Psychol. Bull. 137(4):616–42

Kozlowski SWJ, Klein KJ. 2000. A multilevel approach to theory and research in organizations: contextual, temporal, and emergent processes. In Multilevel Theory, Research, and Methods in Organizations: Founda- tions, Extensions, and New Directions, ed. KJ Klein, SWJ Kozlowski, pp. 3–90. San Francisco: Jossey-Bass

Kozlowski SWJ, Mak S, Chao GT. 2016. Team-centric leadership: an integrative review. Annu. Rev. Organ. Psychol. Organ. Behav. 3:21–54

Kwok N, Hanig S, Brown DJ, Shen W. 2018. How leader role identity influences the process of leader emer- gence: a social network analysis. Leadersh. Q. 29(6):648–62

Lacerenza CN, Reyes DL, Marlow SL, Joseph DL, Salas E. 2017. Leadership training design delivery and implementation: a meta-analysis. J. Appl. Psychol. 102(12):1686–1718

LamW, Lee C, Taylor MS, Zhao HH. 2018. Does proactive personality matter in leadership transitions? Ef- fects of proactive personality on new leader identification and responses to new leaders and their change agendas. Acad. Manag. J. 61(1):245–63

Lanaj K, Hollenbeck JR. 2015. Leadership over-emergence in self-managing teams: the role of gender and countervailing biases. Acad. Manag. J. 58(5):1476–94

Landay K, Harms PD, Credé M. 2019. Shall we serve the dark lords? A meta-analytic review of psychopathy and leadership. J. Appl. Psychol. 104(1):183–96

Latham GP, Pinder CC. 2005.Work motivation theory and research at the dawn of the twenty-first century. Annu. Rev. Psychol. 56:485–516

Lemoine GJ, Blum TC, Roman PM. 2016. Detrimental effects of “stretch” goals in specialty substance use disorder treatment organizations. J. Subst. Abuse Treat. 64:13–23

Lepisto D. 2021. Ritual work and the formation of a shared sense of meaningfulness. Acad. Manag. J. https:// doi.org/10.5465/amj.2018.0854. In press

Leslie LM, Manchester CF, Dahm PC. 2017. Why and when does the gender gap reverse? Diversity goals and the pay premium for high potential women. Acad. Manag. J. 60(2):402–32

Lewis A, Clark J. 2020. Dreams within a dream: multiple visions and organizational structure. J. Organ. Behav. 41(1):50–76

Li WD, Li S, Feng JJ, Wang M, Zhang H, Frese M, Wu CH. 2020. Can becoming a leader change your personality? An investigation with two longitudinal studies from a role-based perspective. J. Appl. Psychol. 106(6):882–902

Liberman N, Trope Y. 1998. The role of feasibility and desirability considerations in near and distant future decisions: a test of temporal construal theory. J. Pers. Soc. Psychol. 75(1):5–18

Liden RC, Anand S, Vidyarthi P. 2016.Dyadic relationships.Annu. Rev. Organ. Psychol. Organ. Behav. 3:139–66 Liden RC,Wayne SJ, Liao C,Meuser JD. 2014. Servant leadership and serving culture: influence on individual

and unit performance. Acad. Manag. J. 57(5):1434–52 Lin SHJ, Ma J, Johnson RE. 2016.When ethical leader behavior breaks bad: how ethical leader behavior can

turn abusive via ego depletion and moral licensing. J. Appl. Psychol. 101(6):815–30 Liu H, Chiang JTJ, Fehr R, Xu M, Wang S. 2017. How do leaders react when treated unfairly? Leader nar-

cissism and self-interested behavior in response to unfair treatment. J. Appl. Psychol. 102(11):1590–99 Locke EA. 1999. The Essence of Leadership: The Four Keys to Leading Successfully. Washington, DC: Lexington

Books Lord RG, Brown DJ. 2001. Leadership values and subordinate self-concepts. Leadersh. Q. 12(2):133–52 Lord RG, Epitropaki O, Foti RJ, Hansbrough TK. 2020. Implicit leadership theories, implicit followership

theories, and dynamic processing of leadership information. Annu. Rev. Organ. Psychol. Organ. Behav. 7:49–74

Lord RG, Foti RJ, De Vader CL. 1984. A test of leadership categorization theory: internal structure informa- tion processing and leadership perceptions.Organ. Behav. Hum. Perform. 34(3):343–78

Lord RG, Gatti P, Chui SL. 2016. Social-cognitive relational and identity-based approaches to leadership. Organ. Behav. Hum. Decis. Process. 136:119–34

Lorinkova NM, Pearsall MJ, Sims HP Jr. 2013. Examining the differential longitudinal performance of direc- tive versus empowering leadership in teams. Acad. Manag. J. 56(2):573–96

90 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

Lyness KS, Grotto AR. 2018. Women and leadership in the United States: Are we closing the gender gap? Annu. Rev. Organ. Psychol. Organ. Behav. 5:227–65

MacCallum R, BrowneMW. 1993. The use of causal indicators in covariance structure models: some practical issues. Psychol. Bull. 114:533–41

MacLaren NG, Yammarino FJ,Dionne SD, Sayama H,MumfordMD, et al. 2020. Testing the babble hypoth- esis: Speaking time predicts leader emergence in small groups. Leadersh. Q. 31(5):101409

Macrae CN, Bodenhausen GV. 2000. Social cognition: thinking categorically about others.Annu. Rev. Psychol. 51:93–120

Marinova SV, Moon H, Kamdar D. 2013. Getting ahead or getting along? The two-facet conceptualization of conscientiousness and leadership emergence. Organ. Sci. 24(4):1257–76

Martin SR, Côté S,Woodruff T. 2016. Echoes of our upbringing: how growing up wealthy or poor relates to narcissism leader behavior and leader effectiveness. Acad. Manag. J. 59(6):2157–77

Mayer DM,Nurmohamed S, Treviño LK, Shapiro DL, Schminke M. 2013. Encouraging employees to report unethical conduct internally: It takes a village. Organ. Behav. Hum. Decis. Process. 121(1):89–103

McClean EJ, Martin SR, Emich KJ, Woodruff CT. 2018. The social consequences of voice: an examination of voice type and gender on status and subsequent leader emergence. Acad. Manag. J. 61(5):1869–91

McClean ST,Barnes CM,Courtright SH, JohnsonRE.2019.Resetting the clock on dynamic leader behaviors: a conceptual integration and agenda for future research. Acad. Manag. Ann. 13(2):479–508

Meindl JR. 1995. The romance of leadership as a follower-centric theory: a social constructionist approach. Leadersh. Q. 6(3):329–41

Miscenko D, Guenter H, Day DV. 2017. Am I a leader? Examining leader identity development over time. Leadersh. Q. 28(5):605–20

Ones DS, Viswesvaran C. 1996. Bandwidth–fidelity dilemma in personality measurement for personnel selec- tion. J. Organ. Behav. 17(6):609–26

OuAY,Tsui AS,Kinicki AJ,WaldmanDA,XiaoZ,SongLJ. 2014.Humble chief executive officers’ connections to top management team integration and middle managers’ responses. Adm. Sci. Q. 59(1):34–72

Ouchi WG. 1980. Markets, bureaucracies, and clans. Adm. Sci. Q. 25(1):129–41 Owens BP, Johnson MD, Mitchell TR. 2013. Expressed humility in organizations: implications for perfor-

mance teams and leadership.Organ. Sci. 24(5):1517–38 Park SH,Westphal JD. 2013. Social discrimination in the corporate elite: how status affects the propensity for

minority CEOs to receive blame for low firm performance. Adm. Sci. Q. 58(4):542–86 Paustian-Underdahl SC, Walker LS, Woehr DJ. 2014. Gender and perceptions of leadership effectiveness: a

meta-analysis of contextual moderators. J. Appl. Psychol. 99(6):1129–45 Pearce CL, Wassenaar CL, Berson Y, Tuval-Mashiach R. 2019. Toward a theory of meta-paradoxical leader-

ship.Organ. Behav. Hum. Decis. Process. 155:31–41 Petriglieri G, Stein M. 2012. The unwanted self: projective identification in leaders’ identity work. Organ.

Stud. 33(9):1217–35 Pfeffer J. 1977. The ambiguity of leadership. Acad. Manag. Rev. 2(1):104–12 Podsakoff PM, MacKenzie SB, Lee JY, Podsakoff NP. 2003. Common method biases in behavioral research:

a critical review of the literature and recommended remedies. J. Appl. Psychol. 88(5):879–903 Qin X, Chen C, Yam KC, Huang M, Ju D. 2020. The double-edged sword of leader humility: investigating

when and why leader humility promotes versus inhibits subordinate deviance. J.Appl. Psychol.105(7):693– 712

Rosette AS, Leonardelli GJ, Phillips KW. 2008. The White standard: racial bias in leader categorization. J. Appl. Psychol. 93(4):758–77

Rosette AS,Tost LP. 2010.Agentic women and communal leadership: how role prescriptions confer advantage to top women leaders. J. Appl. Psychol. 95(2):221–35

Ryan MK,Haslam SA, Hersby MD, Bongiorno R. 2011. Think crisis–think female: the glass cliff and contex- tual variation in the think manager–think male stereotype. J. Appl. Psychol. 96(3):470–84

Ryan RM, Deci EL. 2000. Intrinsic and extrinsic motivations: classic definitions and new directions. Contemp. Educ. Psychol. 25(1):54–67

Schaubroeck JM,Hannah ST, Avolio BJ, Kozlowski SWJ, Lord RG, et al. 2012. Embedding ethical leadership within and across organization levels. Acad. Manag. J. 55(5):1053–78

www.annualreviews.org • The Science of Leadership 91

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

Schilpzand P,Houston L, Cho J. 2018.Not too tired to be proactive: daily empowering leadership spurs next- morning employee proactivity as moderated by nightly sleep quality. Acad. Manag. J. 61(6):2367–87

Schlenker BR, Weigold MF. 1989. Goals and the self-identification process: constructing desired identities. In Goal Concepts in Personality and Social Psychology, ed. LA Pervin, pp. 243–90. Mahwah, NJ: Lawrence Erlbaum

Shao Y, Nijstad BA, Täuber S. 2019. Creativity under workload pressure and integrative complexity: the double-edged sword of paradoxical leadership. Organ. Behav. Hum. Decis. Process. 155:7–19

Shondrick SJ,Lord RG.2010. Implicit leadership and followership theories: Dynamic structures for leadership perceptions, memory, and leader-follower processes. Int. Rev. Ind. Organ. Psychol. 25:1–33

Sluss D, Ployhart RE, Cobb MG, Ashforth BE. 2012. Generalizing newcomers’ relational and organizational identifications: processes and prototypicality. Acad. Manag. J. 55(4):949–75

Stam D, Lord RG, Knippenberg DV,Wisse B. 2014. An image of who we might become: vision communica- tion, possible selves, and vision pursuit.Organ. Sci. 25(4):1172–94

Staw BM, DeCelles KA, de Goey P. 2019. Leadership in the locker room: how the intensity of leaders’ un- pleasant affective displays shapes team performance. J. Appl. Psychol. 104(12):1547–57

Steffens NK, Haslam SA. 2013. Power through ‘us’: leaders’ use of we-referencing language predicts election victory. PLOS ONE 8(10):e77952

Sy T, Shore LM, Strauss J, Shore TH, Tram S, et al. 2010. Leadership perceptions as a function of race– occupation fit: the case of Asian Americans. J. Appl. Psychol. 95(5):902–19

Tepper BJ, Simon L, Park HM. 2017. Abusive supervision. Annu. Rev. Organ. Psychol. Organ. Behav. 4:123–52 Tost LP,Gino F,Larrick RP. 2013.When power makes others speechless: the negative impact of leader power

on team performance. Acad. Manag. J. 56(5):1465–86 Van Dierendonck D. 2011. Servant leadership: a review and synthesis. J. Manag. 37(4):1228–61 van Dijke M, De Cremer D, Mayer DM, Van Quaquebeke N. 2012. When does procedural fairness pro-

mote organizational citizenship behavior? Integrating empowering leadership types in relational justice models.Organ. Behav. Hum. Decis. Process. 117(2):235–48

van Knippenberg D,Hogg MA. 2003. A social identity model of leadership effectiveness in organizations. Res. Organ. Behav. 25:243–95

van Knippenberg D, Sitkin SB. 2013. A critical assessment of charismatic—transformational leadership re- search: Back to the drawing board? Acad. Manag. Ann. 7(1):1–60

Vroom VH, Jago AG. 2007. The role of the situation in leadership. Am. Psychol. 62(1):17–24 Wang AC, Chiang JTJ, Tsai CY, Lin TT, Cheng BS. 2013. Gender makes the difference: the moderating

role of leader gender on the relationship between leadership styles and subordinate performance.Organ. Behav. Hum. Decis. Process. 122(2):101–13

Wang G, Oh IS, Courtright SH, Colbert AE. 2011. Transformational leadership and performance across cri- teria and levels: a meta-analytic review of 25 years of research.Group Organ. Manag. 36(2):223–70

Wang G, Van Iddekinge CH, Zhang L, Bishoff J. 2019. Meta-analytic and primary investigations of the role of followers in ratings of leadership behavior in organizations. J. Appl. Psychol. 104(1):70–106

Wang L, Restubog S, Shao B, Lu V, Van Kleef GA. 2018. Does anger expression help or harm leader effec- tiveness? The role of competence-based versus integrity-based violations and abusive supervision. Acad. Manag. J. 61(3):1050–72

Wang XHF, Howell JM. 2010. Exploring the dual-level effects of transformational leadership on followers. J. Appl. Psychol. 95(6):1134–44

Yammarino FJ, Dionne SD, Chun JU, Dansereau F. 2005. Leadership and levels of analysis: a state-of-the- science review. Leadersh. Q. 16(6):879–919

Yukl G. 2008. How leaders influence organizational effectiveness. Leadersh. Q. 19(6):708–22 Yukl G, Gordon A, Taber T. 2002. A hierarchical taxonomy of leadership behavior: integrating a half century

of behavior research. J. Leadersh. Organ. Stud. 9(1):15–32 Zapata CP, Carton AM, Liu JT. 2016.When justice promotes injustice: why minority leaders experience bias

when they adhere to interpersonal justice rules. Acad. Manag. J. 59(4):1150–73 Zhang C, Nahrgang JD, Ashford SJ, DeRue DS. 2020. The risky side of leadership: conceptualizing risk

perceptions in informal leadership and investigating the effects of their over-time changes in teams. Organ. Sci. 31(5):1138–58

92 Carton

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

Zhang X, Bartol KM. 2010. Linking empowering leadership and employee creativity: the influence of psycho- logical empowerment, intrinsic motivation, and creative process engagement.Acad.Manag. J. 53(1):107– 28

Zhang Y, LePine JA, Buckman BR, Wei F. 2014. It’s not fair. . . or is it? The role of justice and leadership in explaining work stressor–job performance relationships. Acad. Manag. J. 57(3):675–97

Zhang Y, Waldman DA, Han YL, Li XB. 2015. Paradoxical leader behaviors in people management: An- tecedents and consequences. Acad. Manag. J. 58(2):538–66

Zhu DH, Chen G. 2015. CEO narcissism and the impact of prior board experience on corporate strategy. Adm. Sci. Q. 60(1):31–65

www.annualreviews.org • The Science of Leadership 93

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

OP09_TOC ARjats.cls December 8, 2021 14:13

Annual Review of Organizational Psychology and Organizational Behavior

Volume 9, 2022Contents

From Traditional Research to Responsible Research: The Necessity of Scientific Freedom and Scientific Responsibility for Better Societies Anne S. Tsui � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 1

Recovery from Work: Advancing the Field Toward the Future Sabine Sonnentag, Bonnie Hayden Cheng, and Stacey L. Parker � � � � � � � � � � � � � � � � � � � � � � � � � � �33

The Science of Leadership: A Theoretical Model and Research Agenda Andrew M. Carton � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �61

Stigmatized Work and Stigmatized Workers Glen Kreiner, Christine A. Mihelcic, and Sven Mikolon � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �95

The Power of Listening at Work Avraham N. Kluger and Guy Itzchakov � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 121

Compensation, Benefits, and Total Rewards: A Bird’s-Eye (Re)View Ingrid Smithey Fulmer and Junting Li � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 147

Smart Heuristics for Individuals, Teams, and Organizations Gerd Gigerenzer, Jochen Reb, and Shenghua Luan � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 171

When Gender Matters in Organizational Negotiations Hannah Riley Bowles, Bobbi Thomason, and Inmaculada Macias-Alonso � � � � � � � � � � � � � � � � 199

New Developments in Social Network Analysis Daniel J. Brass � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 225

Trust Within the Workplace: A Review of Two Waves of Research and a Glimpse of the Third Kurt T. Dirks and Bart de Jong � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 247

Cross-Cultural Innovation and Entrepreneurship Ute Stephan � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 277

Relational Dynamics of Leadership: Problems and Prospects Terri A. Scandura and Jeremy D. Meuser � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 309

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.

OP09_TOC ARjats.cls December 8, 2021 14:13

The Structure of Intrinsic Motivation Ayelet Fishbach and Kaitlin Woolley � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 339

Revisiting Behavioral Integrity: Progress and New Directions After 20 Years Tony Simons, Hannes Leroy, and Lisa Nishii � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 365

Informal (Field-Based) Learning Scott I. Tannenbaum and Mikhail A. Wolfson � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 391

Assessing Interests in the Twenty-First-Century Workforce: Building on a Century of Interest Measurement Christopher D. Nye � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 415

Accumulating Knowledge in the Organizational Sciences Frank A. Bosco � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 441

Errata

An online log of corrections to Annual Review of Organizational Psychology and Organizational Behavior articles may be found at http://www.annualreviews.org/ errata/orgpsych

A nn

u. R

ev . O

rg an

. P sy

ch ol

. O rg

an . B

eh av

. 2 02

2. 9:

61 -9

3. D

ow nl

oa de

d fr

om w

w w

.a nn

ua lr

ev ie

w s.

or g

A cc

es s

pr ov

id ed

b y

U ni

ve rs

ity o

f C

al if

or ni

a -

R iv

er si

de o

n 08

/1 2/

22 . F

or p

er so

na l u

se o

nl y.