Real Men Don’t Make Mistakes: Investigating the Effects of Leader Gender, Error Type, and the Occupational Context on Leader Error Perceptions
Despite the fact that leaders make mistakes, little attention has been paid to the effects of errors on subordinate perceptions. This study investigated the influence of errors on perceptions of leader competence, effectiveness, and desire to work for the leader. It also examined the effects of gendered expectations on perceptions of male and female leader errors by investigating the interactions that occur between the leader’s gender, the type of error, and the occupational context.
A sample of 284 undergraduates read a series of fictional employee emails describing a leader’s behavior and responded to several measures while envisioning themselves as subordinates of the leader.
Results suggested task and relationship errors exert damaging and differential effects on perceptions of leader task and relationship competence, respectively, and equally damage desire to work for the leader. Male leaders were perceived as less task and relationship competent, desirable to work for, and effective than female leaders for committing errors in a masculinized domain.
This study suggests leader errors matter, and that current leadership models ought to be expanded to account more clearly for them. Moreover, it offers insight into the role of gendered expectations in determining perceptions of male and female leader errors.
This study is one of the first to empirically examine leader error perceptions and the effects of gender stereotypes on these perceptions. It represents a step toward understanding evaluations of male and female leaders, not when they succeed, but when they make mistakes.
KeywordsLeadershipLeader errorsGenderFollower attributions
Implicit leadership theories suggest a tendency for people to perceive leaders as heroic figures capable of single handedly shaping the destiny of their organizations (Meindl et al. 1985; Meindl and Ehrlich 1987). However, a fledgling body of work on leader error (e.g., Eubanks and Mumford 2010; Hunter et al. 2011) underscores the fact that leaders operate in complex environments marked by high levels of ambiguity, and thus often make decisions that unintentionally harm their constituencies. Simply put, leaders make mistakes. Moreover, these mistakes can have far reaching negative implications, as evidenced in well-known historical examples such as the nuclear meltdown at Chernobyl and the Challenger space shuttle disaster. More recently, high-profile errors on the part of leaders, including Netflix CEO Reed Hastings’ controversial decision to spin off the company’s mail order DVD service into a new company called Qwikster, equally highlight the profound consequences of such mistakes.
Despite growing recognition of the importance in studying leader errors, little attention has been paid toward understanding perceptions of those who make them. Moreover, the role of gender in shaping such perceptions remains largely neglected. That is, while we know a good deal about the effects of gendered expectations on perceptions of successful male and female leaders (e.g., Heilman et al. 1995, 1989; Heilman and Okimoto 2007), it is unclear how these expectations influence evaluations of male and female leaders who make mistakes. Such an understanding appears timely, given the greater parity among men and women in supervisory and middle management positions (Eagly et al. 2003).
The present effort thus has several objectives. First, we seek to determine the impact of leader errors on subordinate perceptions, as well as the potentially unique effects of task and relationship errors on different outcomes. Second, we examine whether male and female leaders are perceived differently after making errors. We hypothesize that it is the interaction between a leader’s gender and both the type of error and the occupational context that partially determine such perceptions. With these goals in mind, we first provide a background on leader error, then discuss Eagly and Karau’s (2002) role congruity theory to provide a theoretical foundation for our hypotheses. Finally, we discuss our findings, their implications, and avenues for future research.
Defining Leader Error
The study of human errors is an active area of investigative inquiry, with a rich research tradition across the academic disciplines of cognitive psychology, human factors, organizational behavior, and military and medical research. Indeed, definitions of error abound (e.g., Reason 1990; Zapf and Reason 1994; Zhao and Olivera 2006). Drawing on these definitions, Hunter et al. (2011) identified four themes among them. First, errors must have been avoidable by the actor and not simply the result of situational factors (Reason 1990; Senders and Moray 1991). Second, errors may stem from either an action or inaction. Third, errors yield unintended outcomes that were not part of one’s original goal or plan of action (Zapf et al. 1992). These outcomes tend to be undesired, but at times may be positive or beneficial (Sitkin 1996; Van Dyck et al. 2005).
Finally, errors may be domain specific, making it necessary for error taxonomies unique to different positions, professions, or contexts (Senders and Moray 1991). Indeed, while there is likely conceptual overlap among errors made in different roles and situational contexts, Senders and Moray (1991) argued for the need to specify error types within different research domains. Thus, applying these themes to leadership, Hunter et al. (2011) defined leader error as occurring when: An avoidable action (or inaction) is chosen by a leader that results in an initial outcome outside of the leader’s original intent, goal, or prediction (p. 240). Bearing in mind this definition, we now discuss several distinctions between leader errors and related constructs, and then turn our attention to Hunter et al.’s (2011) error taxonomy to introduce task and relationship errors as error types that may be differentially perceived when committed by male and female leaders.
Leader Errors Versus Ineffectiveness
Though it may seem plausible that leader errors are conceptually indistinct from general ineffectiveness, Hunter et al. (2011) argued that one of the defining features of an error is that it is inherently goal-directed in nature. That is, the term “leader error” is applicable only to leaders working toward goals or outcomes associated with their leader role. Thus, leaders who behave in a laissez-faire fashion would not be committing errors of inaction because they lack intended end states or goals associated with leadership. The implication is that errors are not due to low motivation (Campbell et al. 1993), but rather consist of valid attempts at effectiveness that result in outcomes outside the leader’s plans. Thus, in our view, errors reflect a subset of behaviors that may eventually result in low ratings of performance, but are not one in the same. While errors often fall under the umbrella of leader ineffectiveness with laissez-faire leadership and management by exception (Kelloway et al. 2005), they are a unique form of ineffectiveness, marked by conscious, goal-directed behaviors that yield unintended outcomes.
Further, while ineffectiveness implies negative outcomes, this is not necessarily the case for errors. Indeed, although errors often result in undesirable outcomes, especially in the short run, they may actually have positive effects, particularly over the long term (Hunter et al. 2011). For example, Coca-Cola’s decision in 1985 to pursue an alternative product labeled “New Coke,” an event often cited as a classic marketing error (Oliver 1986; Rickard 1995), led to widespread public anger and losses in sales. Yet, in the end, Coke’s decision to reinstitute “Coke Classic,” led to renewed public support and, ultimately, increased financial success (Hunter et al. 2011).
Finally, while poor performance may reflect instances of deviance, abuse, coercion, and other forms of purposely self-serving and destructive leader behavior (Padilla et al. 2007), Hunter et al.’s (2011) definition of leader error applies to those whose goals are typically formed for the good of the group or organization, and thus tend to be socialized in nature. For example, though leader errors may overlap conceptually with aspects of leader derailment and incompetence, particularly with respect to relationship problems (McCall and Lombardo 1983; Morrison et al. 1987), the latter often includes persistent displays of voluntarily destructive behavior (e.g., bullying, manipulation, deception, absenteeism, fraud, theft), which harm the leader’s constituencies and “derail” his or her career over time (Einarsen et al. 2007).
Task and Relationship Errors
In deriving a taxonomy of leader errors, Hunter et al. (2011) emphasized the importance of considering both task errors, including those associated with gathering information, problem-solving, planning and organizing, and decision-making, and relationship errors, including those related to supporting, recognizing and rewarding, and mentoring and developing. Indeed, the famous two-factor model of leadership behavior emerging out of the Ohio State and Michigan studies (Likert 1961, 1967; Stogdill 1974) underscores the importance of leaders engaging in initiating structure (task-oriented) and consideration (relationship-oriented) behaviors. Yukl et al.’s (2002) revised three-factor model further includes these meta-categories, with a third group of behaviors focused on initiating and implementing change in organizations.
While these models possess empirical and theoretical merit, they have been criticized for focusing on behaviors that are most prominent to subordinates as opposed to behaviors that are critical to performance (Hunter et al. 2007). In an attempt to provide a more comprehensive taxonomy of leadership behavior, Fleishman et al. (1991) proposed a four-factor model based on a review of over 65 differing classification schemes. It encompasses the task and relationship-oriented behaviors identified by other models and incorporates additional behaviors less frequently witnessed by subordinates (e.g., planning, gathering information). As such, drawing on such work, Hunter et al. (2011) suggested leader errors fall into four broad categories, including those related to (1) information search and structuring, (2) information use in problem-solving, (3) managing material resources, and (4) managing personnel resources.
While the first three categories consist of errors that are primarily task-related in nature (e.g., disregarding information when developing plans, improperly managing supply usage), the latter category comprises behaviors that are relationship-oriented (e.g., overlooking employee concerns, being overly hurtful during performance appraisals). Drawing on Hunter et al.’s (2011) taxonomy, we suggest task and relationship errors may elicit different gendered expectations of male and female leaders who make them. Before arguing for the potential interactions between the leader’s gender and the type of error, however, we present our main effect hypotheses.
The Impact of Leader Errors on Follower Perceptions
For leaders to be effective, followers must trust their ability to make difficult decisions, execute their initiatives and act as positive organizational figureheads (Yukl 2007). Simply put, it is critical that followers perceive their leaders as competent. Such attributions are central to the leader–follower influence process, as shown in theories of charismatic leadership (Conger 1999; Conger and Kanungo 1987). For instance, the extraordinary effects of charismatic leaders are only possible when followers attribute them charisma based on observable aspects of their behavior. Indeed, leaders who display confidence, one means of obtaining attributions of charisma, are perceived as more competent and influential (Shamir 1995). Thus, when leaders commit errors, followers may be less likely to associate them with competence and less willing to follow them.
Research on implicit leadership theories further suggests people hold cognitive schemas specifying the traits and behaviors expected of leaders based on past socialization and personal experiences with leadership (Epitropaki and Martin 2004). When activated, these schemas aid in understanding and responding to managerial behavior (Poole et al. 1989; Weick 1995), thereby influencing the ways in which leaders are perceived and evaluated by social observers (Jelinek et al. 1983). Research suggests leaders who do not fit stereotypical views of an ideal, prototypical leader are less likely to emerge and be seen as effective by subordinates (Eden and Leviatan 1975; Lord et al. 1984; Rush et al. 1977). While there is some evidence of context-specific categorization (Maurer and Lord 1991; Lord et al. 1982), a few common elements emerge among prototypical leaders. In particular, most implicit theories do not include leaders who make mistakes; it is much more common for followers to view leaders as infallible (Hunter et al. 2007; Schyns and Hansbrough 2008).
Indeed, Meindl et al. (1985) and Meindl and Ehrlich (1987) research on the “Romance of Leadership” indicates that people tend to possess heroic conceptualizations of leadership, which mistakenly frame leaders as all powerful individuals capable of controlling the performance of their respective organizations. Specifically, studies suggest leaders are accorded greater influence on and blame for various organizational and group failures, despite the many potential causal factors outside their control (Meindl et al. 1985; Giessner and van Knippenberg 2008; Lord et al. 1978; Rush et al. 1977). Moreover, even as researchers, we have not been immune to these heroic stereotypes, as evidenced by the heroic leadership bias plaguing much, albeit not all, of the leadership literature (Yukl 2007; Hunter et al. 2007).
In sum, when leaders commit errors and break with the lofty performance expectations and infallible prototypes associated with implicit leadership theories, they are likely perceived as less competent and desirable to work for by followers. While this hypothesis speaks to errors more generally, it is useful to examine the differential effects of different error types on criteria as well. Because subordinates are likely to draw conclusions about a leader’s abilities based on observable aspects of their behavior, it seems reasonable to suggest that task errors are likely to exert more harmful effects on perceptions of the leader’s task competence, while relationship errors are likely to be more damaging to perceptions of the leader’s relationship competence.
Leaders who commit errors will be perceived as less (a) task competent, (b) relationship competent, and (c) desirable to work for than leaders who do not.
Task errors will result in lower perceived task competence than relationship errors.
Relationship errors will result in lower perceived relationship competence than task errors.
Role Congruity Theory
An extension of social-role theory (Eagly 1987), role congruity theory states that gender discrimination toward female leaders stems from societal gender roles, which shape normative expectations of the qualities and behaviors believed to be desirable for each sex (Eagly and Karau 2002). Gender roles consist of descriptive and prescriptive norms (Cialdini and Trost 1998; Eagly and Karau 2002). Descriptive norms, or stereotypes, link men with agentic qualities (e.g., dominant, assertive, confident) and women with communal attributes (e.g., nurturing, perceptive, service-oriented) (Eagly 1987; Heilman 2001). Injunctive norms prescribe which behaviors are gender-appropriate, thereby dictating how men and women ought to behave in various social settings.
The theory suggests prejudice toward female leaders results from perceived incongruity between stereotypes of women as communal and leaders as agentic. By assuming the masculine role of leader, female leaders are perceived as violating expectations of their feminine gender role, thereby provoking disdain from social observers (Heilman et al. 2004). While leadership is generally ascribed masculine qualities (Heilman 1983; Ragins and Sundstrom 1989; Schein 2001), the theory suggests that less incongruity should result when the position in question is defined in feminine terms; i.e., in more communal ways (Eagly and Karau 2002).
In sum, while role congruity theory suggests gender roles create implicit expectations of women as generally incompetent and ill suited for leadership, it also suggests they may be seen as more competent and better suited than men to succeed in leadership roles where the task is communal in nature (e.g., nursing). Thus, in contrast to masculine job settings, male and female leaders may be held to similar performance expectations in feminine jobs, with women holding perhaps a slight advantage over men in such contexts (Eagly and Karau 2002). We now integrate our discussion of role congruity theory and leader error and present our interaction hypotheses.
The Interaction Between Gender and Error Type
Given societal gender roles that associate women with communality and interpersonal competence, female leaders may be held to higher standards of social conduct than their male counterparts. People likely expect female leaders to be socially astute and to display concern for followers. When they make relationship errors, they may elicit more negative reactions than male leaders—who are, at least partially, excused for errors that are less incongruent with their gender role. For instance, research indicates female supervisors who display anger, which may reflect a relationship error, are conferred lower status and performance assessments relative to male supervisors who express identical levels of anger (Brescoll and Uhlmann 2008; Glomb and Hulin 1997). Research also suggests that when people are only informed of a female manager’s success, they assume she lacks the nurturing, interpersonal qualities related to communality—triggering attributions of her as selfish, cold, and manipulative (Heilman et al. 1989, 1995, 2004).
Thus, when female leaders are directly observed making relationship errors, which run in stark contrast to their gender role, this incongruity may result in more damaged perceptions of their relationship competence relative to male leaders. In turn, male leaders may experience higher performance expectations with respect to task errors. Studies suggest people set lower acceptable standards of task competence for women, resulting in them having to outperform men to be perceived as capable leaders (Biernat and Kobrynowicz 1997; Foschi 1996; Foschi et al. 1994). Thus, while task errors committed by a female leader may serve as implicit confirmation of her gender role and assumed lack of task ability (“See, a woman can’t do this job!”), a male leader may elicit higher levels of perceived incongruity and greater losses in perceptions of his task competence for similar mistakes (“Wait, shouldn’t he be good at this?”).
Leader gender and the type of error will interact such that male leaders who commit task errors will suffer greater losses in perceptions of their task competence than female leaders.
Leader gender and the type of error will interact such that female leaders who commit relationship errors will suffer greater losses in perceptions of their relationship competence than male leaders.
The Interaction Between Gender and the Occupational Context
Although role congruity theory focuses on explaining prejudice toward female leaders who succeed in masculine domains, its underlying logic may be extended to understanding how male and female leaders are viewed after making errors in gendered work situations. The theory suggests gendered expectations vary depending on the nature of the work task. Though female leaders may be perceived as possessing less leader-like qualities (Eagly and Karau 2002), they may be concurrently expected to outperform male leaders in feminine work roles that require more communal qualities associated with women. As such, while men may possess the advantage of being seen as more leader-like (Massengill and di Marco 1979; Schein 1973), women may be perceived as more competent leaders in feminine jobs (Eagly and Karau 2002; Mueller 1986; Sapiro 1983).
Thus, we argue male and female leaders may be held to similar performance standards in traditionally feminine work settings. However, male leaders may be held to higher standards in male-typed jobs given the masculinized nature of the work as well as the masculine qualities associated with the leader role more broadly. Thus, we hypothesize that when male and female leaders commit errors in a feminine work domain, such leaders will not be perceived differently in terms of their competence and desirability as leaders. Yet, in a masculinized work context, we predict that male leaders will experience greater losses in perceptions of their competence and desirability as leaders, given men are generally expected to outperform women as leaders in such settings and are likely to violate their gender role for committing errors in these contexts.
Leader gender and the occupational context will interact such that male and female leaders who make errors in a feminized job will not be perceived differently in terms of their (a) task competence, (b) relationship competence, and (c) desirability as leaders. Yet, male leaders who make errors in a masculinized job will be perceived as less task competent, relationship competent, and desirable to work for than similar female leaders in such contexts.
The Interaction Between Gender, Error Type, and the Occupational Context
Extending these arguments a step further, while male leaders who make task errors in a masculinized work setting may be more likely to elicit greater perceived incongruity from social observers than female leaders who commit task errors in these domains, female leaders who commit relationship errors in feminine work contexts may arouse greater perceived incongruity than male leaders who commit relationship errors in these settings. As such, we also predict a three-way interaction between leader gender, the type of error, and the occupational context.
Leader gender, the type of error, and the occupational context will interact such that (a) male leaders who make task errors in a masculine job will be perceived as less task competent than female leaders who make task errors in such jobs, while (b) female leaders who commit relationship errors in a feminine job will be seen as less relationship competent than male leaders who make relationship errors in such jobs.
Design and Participants
This study used a 2 (leader gender: male vs. female) × 2 (occupational context: nursing vs. construction) × 3 (type of error: task, relationship, no errors) between-subjects design. The sample consisted of 301 undergraduate students from a large northeastern university, including 222 females and 79 males between the ages of 18 and 36 (M = 19.32, SD = 1.76). Participants, on average, possessed 2.71 (SD = 1.81) years of work experience and reported working currently in a wide range of jobs, including customer service representative at a major retailer, supervisor at a large restaurant chain, manager at a local boutique, and trainer in the United States Air Force. Participants logged into an online survey, which randomly assigned them to one of twelve study conditions. They were then asked to provide demographic information and to complete several covariate measures. Subsequently, they read a series of fictional employee emails describing an organization, a group of its employees, and the group’s leader. Next, they completed several measures and open-ended questions while envisioning themselves as followers of the leader.
Twelve sets of employee emails, each representing one of the study’s conditions, were created, consisting of communications between subordinates of the leader and announcements from HR representatives. The emails provided information about the organization, the leader’s behavior and performance, and only differed with respect to the three independent variables. Stereotypically Caucasian names were selected for employees to control for any confounding effects of ethnicity. “Appendix” presents sample emails from one of the study’s conditions.1
In the female condition, the leader was given the name “Barbara Smith,” while in the male condition the leader was named “Bill Smith.” These names were chosen given their distinction as stereotypically masculine and feminine, Caucasian sounding names. The last name of “Smith” was held constant to create as much equivalence on descriptors as possible.
Nursing and construction scenarios were selected to represent stereotypically feminine and masculine work domains, respectively. Nursing and construction were chosen because of the feminine (Bush 1976) and masculine (Iacuone 2007) stereotypes associated with each, again, respectively. In the construction scenario, the leader was described as the “foreman,” while in the nursing scenario the leader was referred to as the “head nurse.” Both task and relationship errors were placed in the context of these two settings. For example, while one task error in the nursing scenario consisted of the head nurse mismanaging an order of drugs, solutions, and other medical equipment, the same error in the construction scenario entailed the foreman mismanaging the order of wall tiles, aluminum pipes, and other materials.
Three conditions, two of which manipulated the type of error, were created using Hunter et al.’s (2011) taxonomy of leader errors. No errors were committed in the third condition. In the task error condition, the leader committed three task errors, including overlooking important information while making planning decisions, developing weak plans, and improperly managing resources. In the relationship error condition, the leader failed to take into account employee concerns regarding increased support for innovation, lost his or her temper, as well as overlooked his or her subordinates’ concerns regarding changing healthcare plans.
Hostile sexism entails particularly unwelcoming and aggressive attitudes toward women in non-traditional gender roles (Glick and Fiske 1996). It is associated with negative evaluations of the non-traditional “career woman” (Glick et al. 1997), negative assessments of female candidates for managerial positions (Masser and Abrams 2004), and less favorable attitudes toward female leaders (Sakalli-Ugurlu and Beydogan 2002). As such, we included Glick and Fiske’s (1996) 11-item hostile sexism scale as a potential covariate (α = .88) to control for those individuals who might perceive female leaders more negatively, on average, as a result of their high levels of hostile sexism rather than the independent variables of interest. Sample items included: “Women are too easily offended” and “Women seek to gain power by getting control over men.” Items were rated on a “1” (Disagree) to “5” (Agree) scale.
A case analysis revealed that males represented, on average, 26 % of individuals per condition. Thus, we included participant gender as a potential covariate to partial out any variance due to the disproportionate influence of female participants, who might view female leaders less negatively, on average, following an error compared to male participants (Glick and Fiske 1996). Further, consistent with role congruity theory, we predicted that while males might view female leaders more negatively for making relationship errors in a feminized work setting, females might perceive male leaders more negatively for task errors committed in a masculine work context. As a result, we included the four-way interaction between participants’ gender, the leader’s gender, the type of error, and the occupational context as a potential covariate.
Leader Task Competence
Cushenbery et al.’s (2009) nine-item scale (α = .92), based on task-related leader behaviors summarized by Yukl (2007), was used to measure perceived leader task competence. Sample items include: “I believe this leader would do well at….” “… organizing work activities to improve efficiency” and “… planning short-term operations.” Responses were measured on a “1” (Disagree) to “5” (Agree) scale.
Leader Relationship Competence
Cushenbery et al.’s (2009) 11-item scale, based on relationship-oriented behaviors discussed by Yukl (2007), was used to measure perceived leader relationship competence (α = .97). Sample items include: “I believe this leader would do well at…” “providing support and encouragement to someone with a difficult task” and “recognizing contributions and achievements.” Items were measured on a “1” (Disagree) to “5” (Agree) scale.
Desire to Work for the Leader
Cushenbery et al.’s (2009) six-item scale, developed from work by Burke et al. (2007), was used to measure desire to work for the leader (α = .96). Items include: “I would be willing to serve under this leader” and “I would like to work with this leader on future projects.” Items were measured on a “1” (Disagree) to “5” (Agree) scale.
Scale and benchmarks for leader effectiveness
1. Low rating
Either explicitly states or suggests that the leader is ineffective
States or implies that the leader is not doing what is necessary to perform well in his or her current leadership position
Describes the leader’s behavior as harmful to subordinates and/or the organization and its goals
Describes the leader and/or their performance in highly negative terms
Ex: She is ineffective because she is irresponsible and relies on her junior nurses to figure problems out. A hospital should NEVER be short of supplies
2. Low to average rating
3. Average rating
Does not explicitly state or suggest whether the leader is effective or not
Wavers between whether the leader is effective or not; does not make a definitive evaluation one way or the other
Either states or suggests that he or she is not in any position to offer an opinion on the leader and his or her effectiveness
Ex: This leader shows strong leadership skills and knows what she wants, but may have some problems communicating what she really means
Ex: Personally from only a few e-mails, none of which are from her, all I can tell is there might be some issues at the work site with her but I’m unsure how to base this question on a few e-mails
4. Average to high rating
5. High rating
Either explicitly states or suggests that the leader is effective
Describes the leader’s behavior as beneficial to subordinates and/or the organization and its goals
States or implies that the leader is doing what is necessary to perform well in his or her current leadership position
Describes the leader and their performance in highly positive terms
Ex: He is extremely effective in my opinion. He takes action where action is needed to effectively and swiftly get the job done. Not to mention on top of getting the job done, he does it well
To ensure participants interpreted the study’s manipulations in the intended ways, they were asked to indicate the gender of the leader (male or female) and the occupational context (nursing or construction) of their assigned conditions. Participants in the error conditions were also asked to report the type of error (task or relationship) committed by the leader. In terms of the gender and occupational context manipulations, 96.7 and 98.3 % of participants reported the correct condition, respectively. Further, for the task error conditions, 94.9 % of participants identified the correct condition, while 95.5 % of participants in the relationship error conditions reported that the leader committed a relationship error. Results generally indicated that the experimental manipulations functioned as intended. However, to cleanse the data of those who may have paid inadequate attention during the study, we removed 17 participants who failed any or all the manipulation checks from our analysis—resulting in a final sample size of 284. In addition, a post hoc study (N = 29 per condition) was conducted to determine whether the target individual (i.e., Barbara/Bill) was recognized as the leader, as opposed to other characters in the scenarios. Across conditions, 100 % of respondents correctly reported Bill/Barbara as the leader.
Confirmatory Factor Analysis
Model fit statistics from the CFA of the three dependent variables
A chi-square difference test showed that the three-factor model displayed a significantly better fit to the data compared to the one-factor model [Δχ2 (3, N = 284) = 2587.67, p < .001] and the three alternative two-factor models described above: [Δχ2 (2, N = 284) = 3217.02, p < .001], [Δχ2 (2, N = 284) = 1657.07, p < .001] and [Δχ2 (2, N = 284) = 926.24, p < .001], respectively. Further, goodness-of-fit statistics [χ2(272) = 478.36; RMSEA = .06; CFI = .98; NFI = .95, SRMR = .05, GFI = .83] and factor loadings (>.77) were all consistent with recommended guidelines (e.g., Bagozzi and Yi 1988; Jöreskog and Sörbom 1993), providing adequate empirical support for the three-factor solution.
Preliminary Analysis of Covariates
The threshold for retaining covariates was set at the p ≤ .10 level. Covariates not meeting this threshold were identified through stepwise deletion and removed from the final analysis to maximize degrees of freedom. A multivariate analysis of covariance (MANCOVA) indicated no significant effects for any of the covariates on the dependent variables. Results of the univariate analysis of covariance (ANCOVA) for leader task competence were as follows: participant gender [F(1, 264) = 1.45, η2 = .01, p > .10], hostile sexism [F(1, 264) = .33, η2 = .00, p > .10], as well as the interaction between gender and the independent variables [F(1, 264) = 2.49, η2 = .01, p > .10].
With respect to leader relationship competence, neither participant gender [F(1, 264) = .88, η2 = .00, p > .10], hostile sexism [F(1, 264) = 1.19, η2 = .00, p > .10], nor the four-way interaction term [F(1, 264) = .86, η2 = .00, p > .10] were significant. Moreover, results indicated no significant effects for desire to work for the leader gender [F(1, 264) = 1.23, η2 = .01, p > .10]; hostile sexism [F(1, 264) = .01, η2 = .00, p > .10]; four-way interaction [F(1, 264) = 1.46, η2 = .01, p > .10], nor for the content-coded leader effectiveness variable gender [F(1, 264) = .20, η2 = .00, p > .10]; hostile sexism [F(1, 264) = 2.45, η2 = .01, p > .10]; four-way interaction, [F(1, 264) = .03, η2 = .00, p > .10]. As such, we removed all three control variables from our final MANOVA analysis.
Means, standard deviations, and intercorrelations among study variables
Leader task competence
Leader relationship competence
Desire to work for the leader
Summary of univariate analyses of covariance for the study’s dependent variables
Leader task competence
Leader relationship competence
Desire to work for the leader
Type of error
Leader gender × type of Error
Leader gender × occupational context
Type of error × occupational context
Leader gender × type of error × occupational context
Tests of Main Effect Hypotheses
With respect to H1a, our results indicated a significant main effect for the type of error on perceived leader task competence, F(2, 272) = 208.50, η2 = .59, p < .01. A pairwise comparison test revealed that participants in the error conditions perceived leaders as less task-competent (M = 2.36, SD = .68) compared to those in the non-error conditions (M = 3.93, SD = .54) (Mdiff = 1.57, SD = .61, p < .01), supporting H1a. Regarding effect size, the η2 value, which estimates the proportion of variability in the dependent variable(s) explained by membership in the groups that define the independent variable(s) and is equivalent to the R2 of a dummy-coded variable in regression analysis (Pedhazur 1982), indicated that 59 % of the observed differences, or variability, in perceived task competence was explained by leaders committing errors or not.
Regarding H1b, results further indicated a significant main effect for the type of error on perceived leader relationship competence F(2, 272) = 128.38, η2 = .47, p < .01. A comparison test, again, revealed that participants in the error conditions perceived the leader as significantly less relationship competent (M = 2.52, SD = .63) compared to those in the non-error conditions (M = 3.66, SD = .63) (Mdiff = 1.14, SD = .63, p < .01). Thus, H1b was supported. Results showed that 47 % of the variability in perceived relationship competence was attributable to whether participants were in the error or non-error conditions. In terms of H1c, a main effect was also found for the type of error on desire to work for the leader F(2, 272) = 381.01, η2 = .73, p < .01. A comparison test revealed that those in the error conditions displayed a significantly lower desire to work for the leader (M = 2.01, SD = .60) than those in the non-error conditions (M = 4.01, SD = .61) (Mdiff = 2.00, SD = .61, p < .01). With respect to effect size, the η2 estimate indicated that 73 % of the variance in desire to work for the leader was due to whether leaders made errors or not.
In addition, and in support of H2, a follow-up pairwise comparison test further indicated that task errors resulted in significantly lower perceptions of the leader’s task competence (M = 2.20, SD = .66) compared to relationship errors (M = 2.53, SD = .66) (Mdiff = .33, SD = .66, p < .05). In contrast, a separate post hoc comparison test further revealed that relationship errors lead to significantly lower perceptions of the leader’s relationship competence (M = 2.27, SD = .62) relative to task errors (M = 2.76, SD = .64) (Mdiff = .49, SD = .63, p < .01). Thus, H3 was supported.
Tests of Interaction Hypotheses
Results failed to support H4’s predicted interaction between the leader’s gender and the type of error on perceived leader task competence (H4a) [F(2, 272) = .71, η2 = .01, p > .05] and relationship competence (H4b) [F(2, 272) = 2.00, η2 = .01, p > .05]. Yet, in support of H5a, a two-way interaction was found between the leader’s gender and the occupational context, F(1, 272) = 5.89, η2 = .02, p < .05. When female leaders made mistakes in the nursing condition, they were viewed similarly (M = 2.32, SD = .64) to male leaders (M = 2.47, SD = .63) (Mdiff = .15, SD = .64, p > .05). However, in the construction scenario, male leaders who committed errors were seen as significantly less task-competent (M = 2.17, SD = .75) than similar female leaders (M = 2.48, SD = .62) (Mdiff = .31, SD = .70, p < .05). Two-percent of the variability in perceived task competence was attributable to the leader’s gender in conjunction with the context he or she made errors in.
Supporting H5b, a significant two-way interaction was also found between the leader’s gender and the occupational context on perceived leader relationship competence, F(1, 272) = 4.15, η2 = .01, p < .05. In the nursing scenario, female leaders guilty of errors were perceived similarly (M = 2.63, SD = .54) to male leaders who made mistakes (M = 2.53, SD = .63) (Mdiff = .10, SD = .59, p > .05). However, in the construction setting, male leaders who made errors were viewed as less relationship competent (M = 2.24, SD = .72) than female leaders who did so (M = 2.65, SD = .54) (Mdiff = .41, SD = .63, p < .05). With respect to effect size, the η2 estimate showed that 1 % of the variability in perceived leader relationship competence was attributable to the leader’s gender in relation to the occupational context where he or she committed errors.
Summary of hypotheses and results
Hypotheses and predicted mean trends
Degree of support
Main effect hypotheses
Leaders who commit errors will be perceived as less (a) task competent, (b) relationship competent, and (c) desirable to work for than leaders who do not
Task errors will result in lower perceived leader task competence than relationship errors
Relationship errors will result in lower perceived leader relationship competence than task errors
Leader gender and the type of error will interact such that (a) male leaders who commit task errors will suffer greater losses in perceptions of their task competence than female leaders, while (b) female leaders who commit relationship errors will suffer greater losses in perceptions of their relationship competence than male leaders
Leader gender and the occupational context will interact such that male and female leaders who make errors in a feminized job will not be perceived differently in terms of their (a) task competence, (b) relationship competence, and (c) desirability as leaders. Yet, male leaders who make errors in a masculinized job will be perceived as less task competent, relationship competent, and desirable to work for than female leaders who commit errors in such contexts
Leader gender, the type of error, and the occupational context will interact such that (a) male leaders who commit task errors in a masculinized job will be perceived as less task competent than female leaders who commit task errors in such contexts, while (b) female leaders who commit relationship errors in a feminine job will be seen as less relationship competent than male leaders who make relationship errors in such jobs
Sample qualitative responses of leader effectiveness perceptions
Feminine task (nursing)
Masculine task (construction)
Feminine task (nursing)
Masculine task (construction)
“She’s ineffective because she’s not competent in terms of the human aspect of leadership. This leader does not seem to have the faith of her workers.”
“She’s ineffective, irresponsible and relies on her junior nurses to figure out all the problems. A hospital should NEVER be short of supplies!”
“The leader is ineffective because she messed up the supply of drugs. This is dangerous because what if someone’s life depended on a drug they didn’t have? She’s incompetent because she’s messed up the work schedule, causing people to come in on their day off, which lowers morale and work ethic.”
“Although she was unorganized and demanding, she seemed to care about the organization and wanting to improve.”
“She wants the business to do well by recruiting members and accepting them into the company. The leader may be ineffective because she overlooked a simple decision that threw off the whole company.”
“She was welcoming and warm, but her employees seem to distrust her. If her employees seem to notice a pattern of unfair and unorganized treatment she must not be a very effective leader.”
“He is ineffective because of his inability to make important changes, organize his supplies and his staff, and build a healthy working relationship with the nurses.”
“He cannot organize his materials and workers; this makes him ineffective because it does not allow he or his workers to do their jobs to their full potential.”
“He is ineffective because he doesn’t talk to his staff about their schedules and he doesn’t notice when there is an increase in patients or how to respond to that change.”
“A lack of communication skills makes this leader ineffective!”
“He’s simply irresponsible. To be a leader, manager, supervisor, etc., you must know your responsibilities, keep track of the responsibilities of those under you, and be very responsible. It seems like he lacks these qualities, or isn’t doing his job well as the leader of the junior construction workers.”
“He can’t even order the supplies correctly. Not only that, he didn’t even inform his employees of the mishap!”
“The fact that he said he would carry out the tasks, but then threw them on someone else is not someone who I would like to work for. I would like to work for a leader who at the beginning would tell me what my responsibilities were and let me work towards them.”
“I think what makes this leader effective is that she does not make mistakes and takes care of everything so precisely.”
“She cares about her employees by not scheduling them on their days off or making them work late. She is good at carrying out tasks and planning in advance because she ordered all the medicines on time and for the whole week.”
“What makes this leader effective was how organized she was. She coordinated plans that made everyone else’s schedules a little bit easier, and she also made sure to have the available resources.”
“This leader is effective because she gets ahead of the game by planning and thinking ahead, thereby benefiting everyone in the company.”
“She is effective because she got together materials and organized the schedules of her workers.”
“She is an effective leader because she took the time to make sure that all the materials were necessary for the job and took the time to think about the laborers.”
“She is effective because she was able to foresee a possible shortage in supplies and was able to restock them in a timely manner.”
“What makes this leader effective is that he is willing to take the time to organize schedules, medications, and things like that in order to benefit the patients and staff. He is willing to listen and help with problems that arise.”
“This leader is effective because he is always prepared and doing more than is required. He is organized and restocks the supplies in the hospital. He is well prepared.”
“He plans in advance by ordering supplies before they are needed. Also, he tries to make his employees happy by scheduling them when they want and giving them days off. If leaders can make their employees happy, there is better chance of them working harder.”
“He is extremely effective in my opinion. All the workers had nothing but good things to say about him and how much he is getting done. He is always up to par with what needs to be done and doesn’t have the workers taking time out of their personal lives to do his work. Instead he takes action where action is needed to effectively and swiftly get the job done. Not to mention on top of getting the job done, he does it well!”
“I feel this leader is effective because he is familiar with the status of his work’s needs. He evaluates a situation and proceeds to adjust for the needs of the company. He doesn’t direct blindly; he takes current situations into account.”
Consistent with implicit leadership theories, our findings suggest errors exert damaging effects on perceptions of leaders who commit them. Leaders who made mistakes were viewed as less task and relationship competent, desirable to work for, and effective than leaders who did not. Furthermore, while task errors resulted in significantly lower perceptions of the leader’s task competence compared to relationship errors, relationship errors yielded significantly lower perceptions of the leader’s relationship competence relative to task errors. Overall, these results suggest that task and relationship errors exert differential effects depending on the criterion.
The interactions found between the leader’s gender and the occupational context on perceived task and relationship competence and desire to work for the leader further suggest male leaders are evaluated more negatively than female leaders for errors made in masculinized work domains. These findings were corroborated by the interaction found for the content-coded leader effectiveness variable. Consistent with role congruity theory, male leaders who commit errors in masculinized work contexts may be seen as violating expectations of male performance in gender congruent occupations, resulting in more negative evaluations of them compared to their female counterparts—who are presumably expected to fail in masculine work settings.
Finally, although results failed to support H4’s predicted interaction between the type of error and the leader’s gender and H6’s three-way interaction, one methodological issue may be responsible. While we carefully adhered to Hunter et al.’s (2011) error taxonomy in deriving our operational definitions of task and relationship errors, these definitions consisted of behaviors men and women might be expected to perform equally well on, thereby diluting our predicted interactions. For instance, there is some evidence people tend to associate female managers with being highly organized and attentive to detail (Muller and Rowell 1997), perhaps balancing performance expectations among male and female leaders on certain task-related leadership behaviors (e.g., planning, organizing). As such, when leaders made planning errors in the emails, participants might have held male and female leaders to similar performance expectations, thus resulting in similar evaluations of both genders (Eagly and Karau 2002; Heilman et al. 2004).
Further, while gender stereotypes associate female leaders with being socioemotional experts (Forsyth et al. 1997; Forsyth et al. 1985), some of the relationship errors depicted in the emails might have been interpreted as less gender typed than previously thought. For example, although failing to take into account employee concerns regarding changing healthcare plans or the need for increased innovation may damage working relationships between leaders and followers, these errors might have also been construed as having task-related elements. In contrast, an operational definition focusing on more specific socioemotional errors, such as failing to express warmth toward followers or failing to facilitate group cohesiveness, might have resulted in clearer differential performance expectancies, and thereby more discrepant evaluations of male and female leaders responsible for these errors.
It is important to bear in mind that our findings were based on an experimental design consisting of undergraduates reading fictional employee emails. Experiments that utilize “paper people” scenarios have been criticized by some, who suggest these methods fail to produce the “experimental realism” needed to create the same psychological impact one would expect in an actual organization (Landy 2008; Murphy et al. 1986). Thus, caution should be given to the generalizability of our findings to cases of leader error in real-world organizations. It should also be noted that the use of the term “foreman” (which includes the word “man”) in the experimental emails may have been perceived as less gender neutral relative to the term “head nurse.” As such, future studies should attempt to use job titles that avoid gendered terminology.
Broadly speaking, however, in contrast to field studies, this study offers the advantage of equating the objective characteristics and behavior of male and female leaders, which might function as confounds by accounting for apparently prejudicial reactions (Eagly and Karau 2002; Kanter 1977). As such, it provides causal support for the influence of gender on perceptions of male and female leaders who commit errors. Nonetheless, future studies should investigate the consequences of leader errors in the field. Morgeson (2005), for instance, asked external leaders of self-managing teams to generate examples of problems or events, which affected the team’s effectiveness and how they intervened to get them back on track. These critical incidents were placed into surveys and administered to team members, who rated the leader’s effectiveness in managing each event and their satisfaction with such leadership. As such, this approach might be employed as a way of determining the impact of real, rather than hypothetical, leader errors.
Additionally, it may be argued that the experimental emails utilized in the present study reflect a simplistic treatment of errors in organizations. That is, they explicitly suggested errors were solely the fault of the target leader, rather than attributable to circumstances inherent to the environmental context. However, as Hunter et al. (2011) noted, leader errors are not simply a result of personality traits or flaws in ego, but rather may stem from a complex amalgamation of factors from multiple levels of analysis. Given such complexity, it is critical to recognize that errors, especially leader errors, tend to be in the “eyes of the beholder”; that is, perceptions of errors and their associated effects often reflect a complex interaction between leader behaviors and attributions made by social observers regarding the causal factors related to such behaviors.
Indeed, a rich body of work on the fundamental attribution error (Jones 1979; Nisbett and Ross 1980; Ross 1977) suggests that individuals are quick to draw conclusions about others’ dispositions and personalities based on visible aspects of their behavior, even when plausible situational explanations for the behavior exist (Tetlock 1985). Moreover, romanticized, heroic conceptions of leadership seem to further exacerbate misattributions of blame for various group and organizational failures to a target’s leadership ability, while undervaluing the influence of extraneous, situational forces (Meindl et al. 1985; Weber et al. 2001). As such, the present study’s approach partly illuminates the snap judgments made about leaders when presented with error information, even information which is indirect in nature. A next step might be to examine how varying situational explanations for errors impact follower judgments, or explicitly measuring these attributions as potential moderators and mediators.
We would also like to emphasize that while this study, like others, has primarily couched errors as constituting inherently negative events, there are potential long-term positive effects associated with errors (e.g., learning, innovation, resilience) (Sitkin 1996; van Dyck et al. 2005). For example, van Dyck et al. (2005) found that organizations with an error management culture, stressing open discussion of errors, sharing error knowledge, helping in error situations, timely error detection and analysis, effective error recovery, and coordinated efforts at handling errors, were characterized by higher levels of firm performance and survivability. These findings were consistent with other studies by Edmondson (1996), Rochlin (1999), and Helmreich and Merritt (2000), who emphasized the importance of open communication, support for error reporting, as well as continuous learning as a means of increasing organizational performance and safety.
With respect to leader errors, Hunter et al. (2011) noted that while the proximal effects of errors most often are undesirable, their distal outcomes may, in fact, be quite beneficial. They suggested leaders who reside in organizations marked by an effective error management culture are more likely to learn from their mistakes, thus reducing the chances of such errors occurring in the future. Moreover, they proposed that leaders in these cultures would be more willing and able to innovate in such contexts. Indeed, it is suggested increased innovation is tied to a belief that one will not be blamed or chastised for making errors (Edmondson 1996). Thus, when errors are accepted as a natural part of work and are freely discussed, both lower-level employees and leaders are more likely to engage in the processes of exploration and experimentation necessary for creative performance (Dorman and Frese 1994; van Dyck et al. 2005; Zhou and Shalley 2003).
Theoretical and Practical Implications
This study makes several theoretical contributions to the leadership literature. First, our results suggest leader errors matter; errors damage perceptions of a leader’s competence and followers’ desire to work for them. Though some theoretical models of leadership, such as Bass (1985) and Bass and Avolio (1990, 1994) full-range model, partially acknowledge certain behaviors that might be categorized as errors (e.g., laissez-faire leadership and passive management by exception), current models should be expanded to consider both passive errors and more active mistakes (e.g., losing one’s temper on followers). Our findings also provide support for the differential effects of different error types on criteria, suggesting future studies and theoretical models should consider how different errors uniquely impact various outcomes.
This study also reflects one of the first to incorporate gender into current discussions of leader error. Mistakes are inevitable outcomes of occupying positions of authority (Hunter et al. 2011), positions that are increasingly, at least in middle management, being filled by men and women (Eagly and Carli 2003; Heilman 2001). Thus, it is critical we address the question of how male and female leaders who commit errors are viewed differently by subordinates. We believe this study makes a substantive contribution to our understanding of leadership phenomena by providing preliminary evidence of gendered perceptions of male and female leader errors.
From a practical standpoint, while it is impractical to suggest leaders should attempt to avoid errors altogether given the complex nature of their positions (Finkelstein 2003), leaders should recognize the different types of errors they make and consider how these errors impact their followers in different ways. For instance, a leader who commits a relationship error in the form of losing his or her temper or disregarding a subordinate’s concerns will likely fail to mend damaged perceptions of his or her relationship competence with more effective task behaviors. Instead, the leader should demonstrate increased support and consideration for subordinates.
Future Research Directions
The present study also provides several avenues for future research. First, more specific operationalizations of task and relationship errors, which consist of behaviors that hold distinct gendered expectancies for men and women, are needed to investigate the possible differential evaluations of male and female leaders who commit such errors. Second, future research might examine how other demographic characteristics of the leader, such as race and ethnicity, affect such evaluations. Although the experience of racial minorities has garnered some attention from leadership researchers, it has been more as a side note than as a core feature of understanding leadership (Murtadha and Watts 2005; Ospina and Su 2009). Third, it is possible that leader errors are not perceived as distinct from errors that any employee might make. For this reason, future studies might investigate whether leader errors are conceptually distinct from errors in general.
Fourth, the United States has become progressively more gender egalitarian over time (Twenge 1997). Thus, future research might address how differences in ratings of competence across genders might vary in comparison to cultures that are less egalitarian than the US. Fifth, research might examine how gendered expectations of specific leadership roles within the same organization shape perceptions of errors made by male and female leaders within these roles. For instance, while line management positions, which include direct responsibility for profit and loss, are often associated with men, staff management roles, which focus on supporting roles, are stereotypically perceived as more feminine in nature (Cabrera et al. 2009; Lyness and Heilman 2006). In conclusion, we hope this study spurs others to investigate these avenues of inquiry and believe there is much to be gained from pursuing this line of research.
All stimulus materials are available upon request from the first author.