Attachment and emotion regulation: Compensatory interactions and leader–member exchange

The Leadership Quarterly 23 (2012) 686–701

Contents lists available at SciVerse ScienceDirect

The Leadership Quarterly journal homepage: www.elsevier.com/locate/leaqua

Attachment and emotion regulation: Compensatory interactions and leader–member exchange☆ David A. Richards a,⁎, Rick D. Hackett b a b

Faculty of Business, Lakehead University, Canada DeGroote School of Business, McMaster University, Canada

a r t i c l e

i n f o

Available online 13 April 2012 Keywords: Attachment theory Emotion regulation Leader–member exchange (LMX)

a b s t r a c t The current study draws on attachment theory (Bowlby, 1982) to examine how attachment (a relationship-based trait disposition), and the interaction between attachment and emotion regulation, relate to LMX quality. Data were collected from subordinates and supervisors in a variety of work settings. Attachment anxiety and attachment avoidance negatively predicted LMX quality. Moderator effects were found between attachment and emotion regulation. © 2012 Elsevier Inc. All rights reserved.

1. Introduction There is a voluminous and compelling literature underscoring the importance of reliable and supportive parental care during early childhood to the development of “relationship scripts”. Relationship scripts underlie the propensity to initiate and build trusting, enduring and enriching relationships (Kahn & Kram, 1994; Keller, 2003). Much of this research has evolved from Bowlby's (1982) attachment theory, which has been the basis for several studies of workplace relationships (Harms, 2011). Given the parallels in the power differential in parent–child and leader–follower relationships, attachment theory may offer insights into the processes underlying leader–member exchange (LMX) (Harms, 2011; Martin, Epitropaki, Thomas, & Topaka, 2010). Surprisingly then, there are no studies of LMX that have drawn from attachment theory. Attachment theory posits that people are born with an innate tendency to seek proximity to others (Bowlby, 1973, 1980, 1982). Individuals who, in early childhood, received consistent parental support develop secure attachment styles (and a positive view of self and others); while those who received inconsistent support develop anxious attachment styles (and a negative view of self); and persons who consistently received a lack of parental support develop avoidant attachment styles (and a negative view of others). Individuals with higher attachment anxiety tend to be especially anxious to retain the support and acceptance of their relationship partner, expressed as insecurity, lack of trust, and high dependency. People high in avoidant attachment typically suppress desire for affiliation and avoid close relationships altogether. As these styles are fairly stable throughout adulthood (Mikulincer & Shaver, 2005) they could adversely affect the propensity to initiate, develop and sustain high quality LMX. This study contributes to the LMX literature in several ways. First, using the dyad as the unit of analysis, we offer the first assessment of the relationship between attachment style and LMX quality. Second, our measure of attachment aligns with the original dimensional conceptualization of attachment style (Richards & Schat, 2011), as opposed to typographical approaches to its measurement (e.g. Hazan & Shaver, 1987). Third, we test for the moderating influence of emotion regulation (ER). ER refers to the process by which individuals attempt to influence their own emotions; when they experience them, and how they express

☆ An earlier version of this research was presented at the annual meeting of the Academy of Management, Montreal, Quebec, Canada, in August 2010. This research was supported in part by funding from Lakehead University to David A. Richards. We thank Peter Bycio for his helpful comments on the various drafts of this manuscript, and the reviewers and editor, David Day, for their direction throughout. ⁎ Corresponding author at: 955 Oliver Road, Lakehead University, Thunder Bay, Ontario, Canada, P7B 5E1. E-mail address: [email protected] (D.A. Richards). 1048-9843/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.leaqua.2012.03.005

D.A. Richards, R.D. Hackett / The Leadership Quarterly 23 (2012) 686–701

687

them behaviorally (Gross, 1998a). We present theory supportive of the notion that ER strategies (i.e., reappraisal and/or suppression) interact with attachment style (of leader and subordinate) to influence LMX quality. 2. Theory and hypotheses 2.1. LMX Grounded in role theory (Merton, 1968), social exchange theory (Blau, 1964), and attribution theory (Heider, 1958; Kelley, 1967), LMX focuses on the dyadic exchange between a leader and a subordinate, as well as the process through which the relationship develops. Graen and Uhl-Bien (1995) noted that “LMX clearly incorporates an operationalization of a relation-based approach to leadership” (p. 109), one founded on social exchanges that are mutually beneficial to both parties (Graen & Uhl-Bien, 1995). Quality LMX relationships require high levels of mutual trust and are composed of mutual respect, positive affect and loyalty. High quality LMX is also characterized by perceived equity in the contributions of each party (Liden & Maslyn, 1998) and positively predicts a variety of work related outcomes, including satisfaction with supervision, affective organizational commitment, and objective measures of performance (r = .71, .31 and .11, respectively; Gerstner & Day, 1997). LMX incorporates a social exchange perspective (Liden, Sparrowe, & Wayne, 1997; Sparrowe & Liden, 1997) wherein tangible and intangible currencies are exchanged (Blau, 1964; Cropanzano & Mitchell, 2005). Social exchanges can be differentiated from other forms of exchange in that they are voluntary actions that the agent expects to be reciprocated (Blau, 1964). Within the leader–subordinate dyad a series of reciprocal exchanges come to define increasingly stable role expectations. The quality of the relationship that ultimately develops is thought to be determined by the degree to which each party perceives the exchanges to be fair and equitable (Wayne, Shore, & Linden, 1997), underscoring the importance of interpersonal and contextual factors (Sparrowe & Liden, 1997). 2.2. Attachment theory A key premise of the current study is that an attachment style involving a negative view of the self or of others (Brennan, Clark, & Shaver, 1998 adversely affects the relationships an employee has at work. Attachment theory (cf. Bowlby, 1973, 1980, 1982) posits that individuals have an innate tendency to seek proximity to others in times of need. Specifically, innate attachment behaviors are intended to attract and maintain proximity to attachment figures (i.e., supportive others) to defend against psychological or physical threats (Mikulincer & Shaver, 2005). Bowlby (1973) proposed that patterns of attachment arise from early experiences with supportive others (primarily parents or caregivers) including the extent to which they are available and responsive. From early experiences, people are thought to develop relatively stable and lasting scripts or mental models of attachment that influence their subsequent interpersonal experiences (Mikulincer & Shaver, 2007). Later, Harms (2011) conceptually broadened attachment style to include a composite of motives (e.g. desire for emotionally close relationships); abilities (e.g. ability to develop close relationships) and perceptions (e.g. the degree to which others wish to develop close relationships). Also moving beyond Bowlby, Harms (2011) proposed that one's generalized attachment orientation reflects a history of prior attachment relationships but it may be modified based on experiences and expectations specific to new relationships. Others have posited that the attachment style enacted in any particular relationship may reflect not only one's generalized attachment style but also one's experiences with the particular other (cf. Keller, 2003), an idea awaiting empirical investigation. As per trait activation theory (Tett & Burnett, 2003), perhaps one's core attachment style is activated to lesser or greater degrees by the actions of others. For example, attachment systems developed early in life tend to be activated by distress or fear (Ainsworth, Blehar, Waters, & Wall, 1978; Mikulincer & Shaver, 2005). Still, psychological processes sustain attachment patterns (e.g. seeking feedback that supports view of self and others; Mikulincer & Shaver, 2007); and a change in attachment style becomes difficult with repeated, uncorrected use of habitual models and schemas (Keller & Cacioppe, 2001). The conceptualization of adult attachment has evolved from a categorical typology to a dimensional conceptualization consisting of attachment anxiety and attachment avoidance (Ross, McKim, & DiTommaso, 2006). Attachment security (low levels of attachment anxiety and avoidance) relates to optimism, positive views of the self and others, and confidence that help will be available in times of distress, thereby enabling optimal functioning (Mikulincer, 1995). Attachment anxiety, on the other hand, is characterized by a negative view of self, an overdependence on relationships (Mikulincer & Shaver, 2005), and a tendency to be hyper-vigilant to social and emotional cues from others (Fraley, Niedenthal, Marks, Brumbaugh, & Vicary, 2006). Highly anxious people also experience distress associated with separation from attachment figures and fear that they will be rejected or abandoned (Bowlby, 1973). Attachment avoidance entails viewing others as unavailable or untrustworthy in times of need (Bowlby, 1973; Mikulincer & Shaver, 2005) leading to the “deactivation of proximity seeking, inhibition of the quest for support, and active attempts to handle distress alone” (Mikulincer, Shaver, & Pereg, 2003, p.85). Deactivation of attachment systems is undertaken to avoid the anticipated additional frustration associated with the unavailability of a trustworthy attachment figure (Cassidy & Kobak, 1988). This results in the denial and suppression of attachment needs, the dismissal of threat-related signals, the denial of the importance of relationships, and the avoidance of emotional involvement or intimacy (Mikulincer & Shaver, 2005). Avoidant individuals broaden deactivation to include a generalized distancing from distress even in situations that are not attachment-related (Mikulincer et al., 2003). Finally, attachment styles are differentiated both theoretically and empirically from personality (Harms, 2011).

688

D.A. Richards, R.D. Hackett / The Leadership Quarterly 23 (2012) 686–701

For example, they typically yield significant prediction beyond the “Big Five” personality factors, especially for relationship related outcomes (e.g. Noftle & Shaver, 2006). 2.3. Attachment theory and leadership Attachment theory has been applied to enhance the understanding of leadership behavior and leader–follower relationships. For example, self-verification theory (SVT; Swann, 2012) provides a basis to relate the influence of attachment styles to leader– follower relationships (Keller & Cacioppe, 2001). SVT suggests that individuals solicit behavior from others that maintains the coherency and consistency in their mental models of self-concept (Swann, Rentfrow, & Guinn, 2003), even if the self concept is a negative one. As explained below, SVT in combination with attachment theory can been used to help account for the selffulfilling nature of individuals' leadership expectations (Keller & Cacioppe, 2001). Individuals with mental models/schemas grounded in having received consistent caregiver responsiveness early in life should develop a secure sense of self. They seek self concept verification through interactions with others, guided by the expectation that the relationship partner is trustworthy and will be responsive to their needs. Thus, leader (follower) behavior should be oriented toward building trust and garnering the responses that will verify the self-concept. SVT holds that analogous processes play out for those with a negative view of self as reflected in insecure attachment styles. Keller (2003) describes how attachment style can help account for individual differences in implicit leadership theories. Keller and Cacioppe (2001) suggest that avoidant leaders (followers) are not likely to be motivated to seek resolution of issues that obstruct relationship development; and that anxious leaders (followers) will leave conflict unaddressed out of fear that the existing relationship will be harmed. Furthermore, a secure leader (follower) attempting to build a strong relationship with an anxious follower (leader) will likely eventually become discouraged with the follower's (leader's) incessant need for affirmation of selfworth resulting in the tendency to avoid the relationship partner altogether. Similarly, secure leaders (followers) may be discouraged by attachment-avoidant followers (leaders) who resist relationship building efforts (Keller & Cacioppe, 2001). Importantly, the relationship enhancing or diminishing behaviors thought by Keller and Cacioppe (2001) and Keller (2003) to be associated with the various combinations of leader–follower attachment styles has yet to be tested empirically. Since Harms (2011) thoroughly reviews the empirical literature linking adult attachment style to leadership, it will only be briefly summarized here. For example, Popper, Amit, Gal, Mishkal-Sinai, and Lisak (2004) found that relative to non-leaders, leaders are more likely to possess secure attachment, and less likely to possess attachment anxiety or avoidance. Secure attachment was also positively associated with transformational leadership (Popper, Mayseless, & Castelnovo, 2000) and socialized charismatic leadership (Popper, 2002). With regard to leader attachment anxiety, it relates positively with self-serving leadership motives and negatively with follower performance (Davidovitz, Mikulincer, Shaver, Izsak, & Popper, 2007). Moreover, leader attachment avoidance negatively predicted: (a) levels of prosocial motives, (b) follower's felt security, and (c) followers' socioemotional functioning and mental health (Davidovitz et al., 2007). Moreover, attachment secure leaders tend to adopt a relational oriented leadership approach; while attachment avoidant leaders rely more on task oriented leadership (Doverspike, Hollis, Justice, & Polomsky, 1997). Finally, Berson, Dan, and Yammarino (2006) showed that attachment secure individuals were the most likely to emerge as leaders in experimental groups. 2.4. Emotion regulation As noted earlier, ER refers to the process by which individuals attempt to influence their own emotions, including when and how they experience them, and how they are expressed (Gross, 1998a). While emotions obviously influence one's own thoughts and behaviors, they can also “shape the emotions, thoughts, and behaviors of others” (Hareli & Rafaeli, 2008, p. 36). Barsade (2002) found that individual affective states are contagious and affect both individual and group processes, with positive mood resulting in greater cooperation, lower conflict, and higher subjective performance ratings. Also, leaders' positive and negative moods influence the affective states of subordinates (Johnson, 2008; Sy, Côté, & Saavedra, 2005). Further, Dasborough and Ashkanasy (2002) propose that affective states influence the attributions followers make about their leader's intentions, which in turn, contribute to the quality of leader–follower relationships. Followers' positive affect relates positively to attributing charisma to their leader (Johnson, 2008). Newcombe and Ashkanasy (2002) suggest that emotions may play an even larger role in relationship development than does the content of leader–follower exchanges. In any case, ER is likely to influence how dispositional tendencies (e.g. attachment style) are “manifested” within leader–subordinate relationships. Awareness and regulation of emotions help individuals achieve situation-appropriate responses at work, interacting with an individual's attachment style to influence one's behavior. Importantly, the emotional responses associated with one's attachment style may be tempered by ER. Emotions affect performance (Côté & Miners, 2006) in part because negative emotions interfere with positive interpersonal relationships. Also, Wong and Law (2002) showed that the awareness and ER are important in leader–subordinate relationships in that emotional intelligence (which includes ER as a component) was positively associated with improved outcomes such as job satisfaction and citizenship behavior. There are thought to be two main types of ER (Gross, 1998a; 1998b): antecedent-focused and response-focused. The former involves strategies that alter the emotional impact of a situation. Cognitive reappraisal is an antecedent-focused ER strategy, in that it entails changing the way one thinks about a situation so as to control the emotional response that is experienced (Gross & John, 2003). In contrast, response-focused ER, such as suppression, entails using strategies that alter one's behavioral response

D.A. Richards, R.D. Hackett / The Leadership Quarterly 23 (2012) 686–701

689

to an already felt emotion as elicited by some event (Gross & John, 2003). While ER is not actually quite so compartmentalized, reappraisal and suppression are the two key ER strategies (Gross, 2010). 2.5. Attachment theory and LMX Attachment theory is likely to be helpful in enhancing our understanding of the dyadic exchange within LMX, given the interpersonal nature of the leader–subordinate relationship and that there are “more similarities than differences between LMX and close personal relationships” (Martin et al., 2010, p. 59). As applied to non-work contexts, attachment anxiety and avoidance are both negatively associated with relationship satisfaction (Mikulincer & Shaver, 2007). LMX is thought to develop over three stages of role taking, role making, and role routinization (Graen & Uhl-Bien, 1995). In the role-taking (Stage 1), interactions between leader and follower (typically “strangers”) are formal; the exchange is purely contractual and immediate. In the role making (Stage 2), dyad members become “acquaintances”, and the relationship becomes less formal, with resources exchanged at both a personal and professional level, with each party expecting equitable returns over short time spans. Nonetheless, the discretionary boundaries of the exchange relationship are still being tested. Finally, in the role routinization (Stage 3), exchanges are less transactional and more social (i.e. “in-kind” and allowing for broader time frames for reciprocation). It is in this final stage that high quality LMX is achieved, reflecting mutual respect, perceived equity in contribution, loyalty and positive affect (Liden & Maslyn, 1998). 2.5.1. Attachment anxiety and LMX Mutual trust is seen as being of increasing importance to the development of LMX from stages 1 to 3 (Brower, Schoorman, & Tan, 2000; Ferris et al., 2009; Hackett, Farh, Song, & Lapierre, 2003; Sue-Chan, Au, & Hackett, 2012). Consistent with the notion that the lower propensity to trust characteristic of high attachment insecurity may adversely impact the LMX trust building process, Yakovleva, Reilly, and Werko (2010) found a reciprocal relationship between propensity to trust and trust within co-worker dyads. As for LMX relationships, the theoretical foundation for the influence of attachment style is clear given Harms (2011) conceptualization of attachment styles “as an integrated system of motives, abilities and perceptions which interact with one another to produce patterns of behavior” (p. 288). For example, in line with SVT, if both parties are high in attachment anxiety, they will enter Stage 1 of LMX not trusting one another to provide consistent support (or equitable exchange) and will be motivated to confirm a negative self-concept of being unworthy of, and incapable of earning, such support. More generally, it is likely that the motivational, cognitive and perceptual dynamics exemplified above will be expressed behaviorally in ways that hamper LMX quality, interfering in particular with the role taking and role making stages of development (Graen & Scandura, 1987). Harms (2011) noted that followers who are high in attachment anxiety will tend to “cling” to their leader; resisting offers of greater autonomy and independence that accompanies delegation of increasingly challenging tasks that is centrally instrumental to LMX development (Dienesch & Liden, 1986). In such instances leaders with secure attachment will likely seek to distance themselves, and withdraw their support, attention and delegation, thereby adversely impacting LMX. In this way, followers inadvertently prompt their leader to meet their expectation of intermittent support (Harms, 2011). Moreover, as with anxious attachment parents who promote dependency among their children (George & Solomon, 1999), anxious leaders will tend to delegate less and discourage expressions of initiative and independence from their subordinates. This will undermine followers' perceived ability to fulfill their work role, which will build dependence and thwart LMX development. Hypothesis 1. Attachment anxiety will be negatively associated with the quality of LMX in leader–subordinate dyads.

2.5.2. Attachment avoidance and LMX As implied earlier, attachment avoidance is related to a preference for working alone, dissatisfaction with co-workers, the use of work to avoid socializing, and avoiding conflict with co-workers (Hardy & Barkham, 1994; Hazan & Shaver, 1990). Deactivation strategies involve disengagement from interaction with others and efforts to manage difficulties alone (Cassidy & Kobak, 1988; Mikulincer et al., 2003). Consistent with this disposition, Doverspike et al. (1997) found that leaders high in attachment avoidance employ more of a task (rather than relational) focused leadership style. Also, drawing from SVT, leaders and/or followers high in attachment avoidance will behave in ways aimed at verifying their self-concept of a socially distant “lone wolf”. Taken together, both leader and follower are likely to remain at the purely transactional stage (Stage 1) of LMX development. Further, Keller and Cacioppe (2001) note that attachment avoidant individuals tend not to trust others, consistent with their earlier experiences of care-giver non-responsiveness (Shaver, Collins, & Clark, 1996). This is expressed in nonreceptiveness and inattentive behavior toward their relationship partner and is likely to be viewed by a securely attached leader (follower) as distant and unworthy of respect, loyalty, affect and relationship building contributions that strengthen LMX (Dienesch & Liden, 1986). Hypothesis 2. Attachment avoidance will be negatively associated with the quality of LMX in leader–subordinate dyads.

2.5.3. Interaction effects In addition to the direct effects of attachment insecurity on LMX, there is also a theoretical basis to expect that the interaction of leader attachment and subordinate attachment would have an effect on LMX quality. Since attachment anxiety involves an

690

D.A. Richards, R.D. Hackett / The Leadership Quarterly 23 (2012) 686–701

activation of proximity seeking, this behavior by each member of the dyad toward the other could perhaps provide a conduit for the mutual satisfaction of attachment needs. Moreover, Keller (2003) proposed that leader and follower will experience consistently positive relationships only where their attachment styles match. Specifically, she argued that congruence in attachment styles give rise to shared implicit leadership theories, which enhance perceived similarity, common expectations, and mutual identification, liking, and ultimately higher LMX. Hypothesis 3. The interaction between leader and subordinate attachment anxiety will be positively related to LMX quality; such that LMX quality will be higher when both LMX dyad members are high in attachment anxiety than when only one member is high in attachment anxiety. In contrast, the interaction effect of leader and subordinate attachment avoidance is likely to have the opposite effect, contrary to Keller's (2003) congruency model. Specifically, attachment avoidance will further limit LMX quality when both members of the dyad have higher levels of avoidance because the deactivation of proximity seeking and the lack of trust in others will further inhibit opportunities for the relationship to develop. Hypothesis 4. The interaction between leader and subordinate attachment avoidance will be negatively related to LMX quality; such that LMX quality will be lower when both LMX dyad members are high in attachment avoidance than when only one member is high in attachment avoidance. 2.6. Attachment theory, LMX and the influence of emotion regulation Day and Crain (1992) showed that negative affectivity of subordinates moderated the association between subordinates' ability and LMX quality, in that the association was weaker for subordinates high in negative affectivity. Fisk and Friesen (2012) identified the importance of leader emotion regulation in the development of LMX and on subordinate attitudes and behavior. This is consistent with the notion that ER can influence the nature of the leader–follower dyad. In relation to attachment theory, secure individuals are better able to regulate emotions than are either anxiously and avoidantly attached individuals (Cooper, Shaver, & Collins, 1998). As described earlier, activation and deactivation of the attachment system are strategies used to regulate emotion (Mikulincer et al., 2003). As explained below, the use of ER strategies such as cognitive reappraisal and suppression may compensate for the tendency of attachment insecure individuals to behave in ways that undermine relationship building. Anxious attachment individuals are inclined to ruminate on negative thoughts and feelings, and they have difficulty distancing themselves from distressing experiences (Mikulincer & Florian, 1998). This difficulty may reflect the failure to adequately employ reappraisal and/or suppression. On the other hand, attachment avoidant individuals may be more likely to use suppression (Gross & John, 2003) as they are prone to distancing themselves from experiencing emotions (Mikulincer et al., 2003). In any case, the notion that attachment anxiety and avoidance may predispose individuals to use and/or refrain from a given ER strategy is important since the degree to which a strategy is used may mitigate the influence of attachment style on relationship development. Therefore, we examine the possibility that reappraisal and suppression interactively influence the relationship between attachment style and LMX. 2.6.1. ER using reappraisal Emotional regulation may be beneficial in facilitating social exchange by making interactions between leaders and subordinates more predictable (Glasø & Einarsen, 2008). For example, reappraisal may help leaders and followers with higher attachment anxiety build higher quality LMX by managing the affective reactions to environmental cues that would otherwise trigger their attachment systems and the negative emotion arising from unmet attachment needs. Specifically, as noted earlier, high attachment anxiety individuals tend to experience greater negative emotions because of their hyper-vigilance and their inability to effectively manage distress through internal resilience or proximity seeking (Mikulincer et al., 2003). Effectively implemented, reappraisal has the potential to reduce the experience of negative emotions, possibly allowing for more constructive social interactions (Dasborough & Ashkanasy, 2002) and higher LMX. Hypothesis 5. Reappraisal will moderate the negative association between attachment anxiety and LMX quality, such that LMX will be higher among attachment anxious dyadic partners when they use reappraisal. Reappraisal should also help manage attachment avoidance, as it entails reframing events in ways that generate fewer of the negative emotions that trigger deactivating strategies that result in interpersonal disengagement. As attachment avoidant individuals are especially inclined to deny (leaving unacknowledged) potential threats to their physical and psychological well being, they may be presented with fewer opportunities to use reappraisal (Mikulincer et al., 2003). Stated alternatively, without acknowledgement of these threats, there is less likely to be a priming of adaptive ER strategies (Mikulincer et al., 2003). However, when denial becomes more difficult (i.e. the threats are especially salient), reappraisal will help lower the likelihood of experiencing the negative emotions that undermine relationship development. Hypothesis 6. Reappraisal will moderate the negative association between attachment avoidance and LMX quality, such that LMX will be higher among attachment avoidant dyadic partners when they use reappraisal.

D.A. Richards, R.D. Hackett / The Leadership Quarterly 23 (2012) 686–701

691

2.6.2. ER using suppression While reappraisal helps prevent the experience of negative emotions, suppression alters their behavioral expression. Suppressing felt negative emotions can result in greater levels of stress and emotional labor (Grandey, 2003), negative cognitive effects (Goldin, McRae, Ramel, & Gross, 2008; Richards & Gross, 2000), and impairment of interpersonal communications and relationship development (Butler et al., 2003). Since, suppression masks negative emotions; it potentially allows individuals high in attachment anxiety and attachment avoidance to engage more constructively in a manner that promotes LMX. Such exchanges could also help satisfy attachment needs, particularly those of the relationship partner (Popper & Amit, 2009), resulting in enhanced LMX quality. However, suppression is likely to benefit the relationship partner most because the person suppressing emotions endures the stress associated with using this strategy. Hypothesis 7. Suppression will moderate the association between attachment anxiety and LMX quality, such that LMX will be higher among attachment anxious dyad partners when they use suppression. Individuals with higher levels of attachment avoidance are unlikely to seek proximity, but are likely to benefit from using suppression. Specifically, suppression enables the avoidant individual to contain negative emotions that could potentially trigger the attachment system. Attachment avoidant individuals desire to deactivate proximity seeking because a negative view of others results in expectations that proximity seeking will result in disappointment or even punishment. Suppression happens for these individuals as a way of avoiding negative emotional experiences. Therefore avoidant individuals who suppress negative emotions are able to function better in a general way, and in interpersonal contexts especially (Cassidy & Kobak, 1988; Mikulincer et al., 2003). Again this effect will be stronger for the other member of the dyad because the person using suppression experiences the distress that comes with use of suppression. Hypothesis 8. Suppression will moderate the association between attachment avoidance and LMX quality, such that LMX will be higher when attachment avoidant dyadic partners use suppression.

3. Method 3.1. Sample Participants were recruited through the StudyResponse project (Stanton & Weiss, 2002), a nonprofit service that facilitates online research by enlisting willing panelists. Participants had to be currently employed and have a supervisor who was willing to enroll in StudyResponse. Participants received a link to their respective surveys and used an ID code provided by StudyResponse to maintain anonymity. We excluded any unmatched leaders and subordinates or cases that had missing focal variables, resulting in a final sample of 150 matched dyads. Subordinates averaged 37.1 years old (ranging from 21 to 65) while the gender distribution was approximately equal (50.7% male). Thirteen per cent had a high school education, another 15% were attending college or university, 43% had graduated from college or undergraduate university programs, and 27% had graduate degrees or were attending graduate school (2 participants did not answer the question). The gender distribution among managers was also approximately equal (50.7% were male), with an average age of 41.2 years. They were more likely to be attending or to have completed graduate school (40%). Various employment sectors were represented, including manufacturing (28.7%), technology sector (10.0%), banking (7.3%), retail (8.7%), healthcare (8.0%), government and public service (6.0%), construction (5.3%), travel and transportation (4.7%), education (4.0%), hospitality and food services (3.3%), and other (14%). Subordinates reported a slightly shorter work week (41 h) than did the managers (44 h).

3.2. Measures All of the following measures were completed by both the subordinates and supervisors, except for dyadic tenure, which was measured from the subordinate's perspective. 1

3.2.1. LMX The 12-item multidimensional LMX-MDM (Liden & Maslyn, 1998) was used with 5-point scales ranging from 1 = strongly disagree to 5 = strongly agree. The subordinate version referred to “my supervisor” while the leader version employed “this employee”. Sample items include “I like my supervisor/this employee very much as a person” and “I am impressed with my supervisor's/this employee's knowledge of his/her job.” 2 1 Dyadic tenure was also measured from the supervisor's perspective. There was a low level of agreement in some cases, so we opted for the subordinate rating, which was likely more salient, and therefore more accurate. 2 The LMX-MDM can be used as a one-dimensional scale with the four sub-factors contained under a higher order factor (Erdogan, Kraimer, & Liden, 2004; Pellegrini & Scandura, 2006). The dimensions of LMX were highly correlated (r = .69 to .84; r = .82 to .98 when corrected for attenuation) and the dimensional analysis did not result in meaningful differences. Accordingly, we treated LMX as a unidimensional scale.

692

D.A. Richards, R.D. Hackett / The Leadership Quarterly 23 (2012) 686–701

3.2.2. Attachment Supervisor and subordinate attachment was measured using Richards and Schat's (2011) adaptation of Brennan et al.'s (1998) Experience in Close Relationships scale. The 36-item instrument consists of two subscales — attachment anxiety and attachment avoidance. The anxiety scale has 18 items (e.g., “I need a lot of reassurance that I am liked and appreciated by other people.” and “If I can't get others to show interest in me, I get upset or angry”). The avoidance scale also consists of 18 items (e.g., “I usually discuss my problems and concerns with other people” and “I am very comfortable being close to others”; both negatively keyed). The response scale ranges from 1 = strongly disagree to 7 = strongly agree. 3.2.3. ER Reappraisal and suppression by both leaders and subordinates was assessed using the ER Questionnaire (Gross & John, 2003). The reappraisal items include “I control my emotions by changing the way I think about the situation I'm in” and “When I want to feel less negative emotion, I change the way I'm thinking about the situation”. Suppression items include “When I am feeling negative emotions, I make sure not to express them” and “I keep my emotions to myself”. A 7-point Likert scale ranging from 1 = strongly disagree to 7 = strongly agree was used. 3.2.4. Control variables Regression analyses controlled for age, gender, and education of the leader and subordinate as well as dyadic tenure (the length of time that the leader and subordinate worked together). 3.3. Data analysis The Actor Partner Interdependence Model (APIM) was the primary analysis used to test our hypothesis. It retains the individual unit measures, nested within the dyads. The APIM is especially useful in examining the effects of relationship partner characteristics (e.g. attachment style) on the quality of the relationship (Cook & Kenny, 2005; Kashy & Kenny, 2000). Use of the APIM allows for the estimation of separate actor and partner effects associated with relationship style. For example, in the present case, the actor effect reflects the relationship between a leader's (or follower's) attachment style and his/her own reported LMX quality (i.e. within-person); whereas the partner effect reflects the relationship between the leader's (follower's) attachment style and the follower's (leaders') reported LMX quality (i.e., cross-person). The APIM also allows for reciprocal influence between leader and follower attachment styles, and for correlated error — the residual nonindependence in leader and follower ratings of LMX quality. As shown in Fig. 1 there are two actor effects and two partner effects (cf. Kenny, Kashy, & Cook, 2006). Our data included independent dyads consisting of a unique leader and subordinate pair such that each could be identified by their role. To determine whether our analysis should be conducted   at the individual or dyadic level, we examined the correlation pffiffi between leader and subordinate LMX. A t-test t ðn−2Þ ¼ rpffiffin−22 (Kenny et al., 2006) revealed that the correlation was significant 1−r

(r = .53, p b .001; t(148) = 7.50) which suggested a lack of independence, and warranted analysis at the dyadic level. We confirmed the level of analysis using hierarchical linear modeling (HLM) (Bryk & Raudenbush, 1992) by calculating the intraclass correlation 2 τ ICC ¼ ðτþσ 2 Þ (ICC = .51, p b .001, χ (148) = 459.24) prior to testing the hypotheses at the dyadic level. HLM7 was used to estimate actor and partner effects as recommended by Kenny et al. (2006). The computational specifics of the APIM using HLM are detailed elsewhere (e.g. Campbell & Kashy, 2002; Kenny et al., 2006) and are described only briefly here. Our analyses included both actor and partner variables in each of the models tested, which allowed for simultaneous assessment of their effects on leader-member exchange quality. HLM requires that the data be organized in two separate files. The level-1 file contains within-dyad variables and mixed variables. The data include mixed variables that vary both between and within dyads (Kashy & Kenny, 2000; Yakovleva et al., 2010). Specifically, attachment style, ER and LMX are mixed variables because there is an assessment specific to each dyad member, as well as variation among the dyads on each of the variables. As noted by Campbell and Kashy (2002) “(a)ttachment avoidance is an example of a mixed variable because some people are more avoidant than others, and the average level of avoidance within a couple differs across different couples…Actor and partner effects can be directly estimated for mixed predictor variables only”

X1 Subordinate’s IV

a

Y1 Subordinate’s DV

E1

Y2 Leader’s DV

E2

p

p X2 Leader’s IV

a

Fig. 1. APIM schemata. Adapted from Fig. 7.1 in Kenny et al. (2006, p. 145).(a = actor effect, p = partner effect, IV = independent variable, DB = dependent variable).

1. Sub. age 2. Sub. gender 3. Sub. educ. 4. Ldr. age 5. Ldr. gender 6. Ldr. educ 7. Dyad tenure 8. Sub. anx 9. Sub. avoid 10. Sub. reapp 11. Sub. suppr 12. Ldr. anx 13. Ldr. avoid 14. Ldr. reapp 15. Ldr. suppr. 16. Sub-LMXa 17. Ldr-LMXb

M

SD

37.07 .49 4.03 41.18 .49 4.43 5.18 3.49 3.68 4.95 4.32 3.44 3.53 5.02 4.08 3.93 4.05

9.15 .50 1.34 9.86 .50 1.28 4.36 1.25 .87 1.11 1.15 1.23 .87 1.03 1.17 .72 .64

1 .06 −.10 .50⁎⁎ .07 −.06 .35⁎⁎ −.20⁎ .10 .05 .02 −.10 .17⁎ −.14 −.02 .11 .08

2

−.32⁎⁎ .17⁎ .05 −.05 −.06 −.19⁎ .04 .08 −.28⁎⁎ −.15 −.06 −.02 −.21⁎⁎ .02 .15

3

−.21⁎ −.06 .46⁎⁎

4

−.07 .14 −.05 .05 .06 .21⁎⁎

.03 −.16⁎ .11 .30⁎⁎ −.07 −.05 −.08 −.27⁎⁎

.05 .06 .17⁎ −.03 −.08

.01 −.14 −.17⁎ .16⁎ .14

5

6

7

.09 −.11 .02.01 .03 −.01 .07 .07 −.05 .06 .11 .21⁎⁎

−.13 −.06 −.06 .10 −.05 .07 .04 .06 .19⁎ −.07 −.05

.02 −.02 −.07 −.02 .06 .07 −.01 .06 .03 −.06

8

9

(.95) .27⁎⁎ .00 .29⁎⁎ .61⁎⁎ .26⁎⁎

(.87) −.23⁎⁎ .33⁎⁎ .18⁎ .40⁎⁎

.01 .25⁎⁎ −.28⁎⁎ −.14

−.07 .15 −.29⁎⁎ −.15

10

11

(.87) .12 .12 −.14 .43⁎⁎

(.78) .33⁎⁎ .22⁎⁎

−.01 .22⁎⁎ .22⁎⁎

.11 .42⁎⁎ .05 .11

12

13

14

15

16

17

(.95) .41⁎⁎ −.03 .45⁎⁎ −.15 −.11

(.88) −.28⁎⁎ .43⁎⁎ −.14 −.13

(.88) .21⁎⁎ .20⁎ .14

(.74) −.05 −.02

(.94) .53⁎⁎

(.93 )

Sub. = Subordinate; Ldr. = Leader; Educ. = Education; Anx. = Attachment Anxiety; Avoid = Attachment Avoidance; Reapp = Reappraisal; Suppr = Suppression. aSub-LMX = LMX rated by the subordinate. bLdr-LMX = LMX rated by the leader. Pairwise deletion N = 148–150. ⁎p b .05. ⁎⁎p b .01.

D.A. Richards, R.D. Hackett / The Leadership Quarterly 23 (2012) 686–701

Table 1 Means, standard deviations, and correlations.

693

694

D.A. Richards, R.D. Hackett / The Leadership Quarterly 23 (2012) 686–701

Table 2 APIM analysis results: Direct effects predicting LMX. Actor

Anxiety Avoidance Reappraisal Suppression

Partner

Coefficient

SE

t

df

Coefficient

SE

t

df

− 0.09 − 0.17 0.08 0.01

0.04 0.05 0.04 0.04

− 2.28⁎ − 3.46⁎⁎⁎ 2.01⁎ 0.27

288 288 288 288

0.00 − 0.05 0.11 0.02

0.04 0.05 0.04 0.04

− 0.08 − 1.11 2.69⁎⁎ 0.53

288 288 288 288

⁎p b .05. ⁎⁎p b .01. ⁎⁎⁎p b .001.

(p. 329). The within-person interactions (e.g. attachment anxiety and reappraisal) and age, gender, and education are also mixed variables and therefore part of the level-1 dataset. The level two file contains between-dyad variables, which are variables that only differ between dyads (Kashy & Kenny, 2000). In this case dyadic tenure (which is the same within the dyad but differs between dyads) was included as was the interaction term of leader and subordinate attachment dimensions. Within the dyad there are only two cases (leader and subordinate), and the random component must therefore be removed from the slopes of the level-2 variables (Campbell & Kashy, 2002). 4. Results Descriptive statistics, including alpha reliability coefficients for the study variables are presented in Table 1. 4.1. Direct effects: Attachment and LMX APIM findings related to the first two hypotheses are presented in Table 2. We controlled for subordinate and leader demographic variables (age, gender, and education), and dyadic tenure, and for dyad role (leader vs. subordinate). As noted earlier, although the members of the dyad are distinguishable, the dyad role variable in the hierarchical linear models was not significant — meaning that the effects were the same for both dyadic members. For this reason, our results do not differentiate the leader and subordinate effects, but rather the actor and partner effects. Hypotheses 1 and 2 were partially supported, in that there was a significant and negative actor effect for attachment anxiety (t(288) = −2.28, p b .05) and attachment avoidance (t(288) = −3.46, p b .001), but no significant partner effects for either dimension of attachment. Hence, leader's attachment insecurity (anxiety and avoidance) associates negatively with their own perceptions of LMX quality; and similarly for subordinates. Further, the leader's (follower's) attachment insecurity was unrelated to their subordinates' (leader's) reported LMX (i.e. no cross-member or partner effects). Though not hypothesized, for reappraisal there was a significant actor effect (t(288) = 2.01, p b .05) and a stronger significant partner effect (t(288) = 2.69, p b .01). This means that the use of reappraisal by the leader (subordinate) was positively associated with their own and the subordinate's (leader's) evaluation of LMX quality (i.e. cross-member effects). Together these findings suggest that attachment operates at the individual level, while reappraisal has both an individual and a dyadic level influence. 4.2. Interaction effects of subordinate and leader attachment Consistent with Baron and Kenny (1986), to test for interaction effects, the dimensions of attachment and ER were centered and the interaction terms were calculated in SPSS prior to including them in the hierarchical linear models (Kenny et al., 2006). Centering involves subtracting the sample mean of the variables from individual scores prior to calculating the product of those variables (the interaction term) to improve interpretability of the results (Aiken & West, 1991). As with the direct effects, we constructed hierarchical linear models to determine the actor and partner effects. The control variables were also included in the analysis. The results of this analysis are provided in Table 3. We analyzed the interaction of subordinate and leader attachment dispositions for both anxiety and avoidance. Support for Hypothesis 3 was found in a significant positive interaction effect for attachment anxiety (t(287) = 4.19, p b .001). As shown in Fig. 2, low levels of attachment anxiety for both actors and partners relate to higher levels of LMX. In contrast, LMX quality suffers when leaders and subordinates have larger differences in their attachment anxiety (primarily expressed as an actor effect). The interaction effect for avoidance was not significant, so Hypothesis 4 was not supported. Table 3 APIM analysis results: Interaction effects of subordinate and leader attachment in predicating LMX.

Subordinate anxiety × leader anxiety Subordinate avoidance × leader avoidance ⁎p b .05. ⁎⁎p b .01. ⁎⁎⁎p b .001.

Coefficient

SE

t

df

0.09 0.00

0.02 0.04

4.19⁎⁎⁎ .10

287 287

D.A. Richards, R.D. Hackett / The Leadership Quarterly 23 (2012) 686–701

695

Fig. 2. The interaction between actor and partner attachment anxiety (leader–subordinate role is not significant). LMX is highest when actor and partner both have low anxiety and it is lowest when an actor has high anxiety and the partner does not.

4.3. Moderating effect of emotion regulation We used the same process in our analysis of the interactions of attachment and emotion regulation. Results of the moderation analysis for leader attachment and ER are presented in Table 4. Hypothesis 5, that the use of reappraisal would lessen the negative relationship between attachment anxiety and LMX, was supported for the most part. In assessing the interaction between attachment and anxiety, the actor effect was positive and close to reaching significance (t(284) = 1.91, p = .057), and the partner effect was significant (t(284) = 3.19, p b .01). This suggests that the dyad will achieve a higher quality LMX if leaders and subordinates with attachment anxiety also use reappraisal (see Fig. 3). In contrast, the use of reappraisal did not moderate the association between attachment avoidance and LMX, leaving the sixth hypothesis unsupported. Consistent with Hypothesis 7, the use of suppression significantly moderated the association between attachment anxiety and LMX quality. Both the actor (t(284) = 2.31, p b .05) and partner (t(284) = 2.62, p b .01) effects were significant (see Fig. 4). Those with higher attachment anxiety had higher quality LMX if they or their dyadic partner suppressed negative emotions. Some support for the eighth and final hypothesis was obtained in that a significant partner effect was found for the interaction between suppression and attachment avoidance (t(284) = 3.28, p b .001). LMX is significantly higher when leaders and subordinates with attachment avoidance used suppression to regulate their emotions (Fig. 5). On the other hand, the actor effect was not significant, which is perhaps not surprising, since as a response-focused form of ER, avoidant individuals relying on suppression to regulate their emotion have the felt experience of the negative emotion.

5. Discussion Our study contributes to the attachment and LMX literatures in several ways. First, empirical research concerning individual traits and LMX has entirely neglected the adult attachment literature, despite compelling theoretical grounds to expect that the attachment insecurity of a leader (follower) will negatively influence LMX development (Harms, 2011; Keller, 2003; Keller & Cacioppe, 2001; Martin et al., 2010). We provide empirical support for direct negative actor effects and for a positive leader– follower interaction effect for attachment anxiety. Second, we found support for the moderating effect of ER on the association between attachment style and LMX. This responds to recent calls to examine moderators of trait–LMX relationships (Bauer, Erdogan, Liden, & Wayne, 2006; Sears & Hackett, 2011). Third, Harms (2011) found only three studies that simultaneously considered both leader and follower attachment behaviors/orientations on leadership related outcomes, and none of them examined LMX. We simultaneously examined not only leader and follower attachment, but also leader and follower LMX ratings. Moreover, we used independent dyads as the unit of analysis which, while appropriate, is not typical of the relationship (including leadership) literature (Harms, 2011; Martin et al., 2010).

Table 4 APIM analysis results: Interaction effects attachment and emotion regulation predicting LMX. Actor Coefficient

SE

t

df

Coefficient

SE

t

df

0.04 0.01 0.05 − 0.01

0.02 0.03 0.02 0.04

1.91§ 0.17 2.31⁎

284 284 284 284

0.07 − 0.01 0.06 0.13

0.02 0.03 0.02 0.04

3.19⁎⁎ − 0.29 2.62⁎⁎ 3.28⁎⁎⁎

284 284 284 284

Anxiety × reappraisal Avoidance × reappraisal Anxiety × suppression Avoidance × suppression §

Partner

p = .06. ⁎p b .05. ⁎⁎p b .01. ⁎⁎⁎p b .001.

− 0.25

696

D.A. Richards, R.D. Hackett / The Leadership Quarterly 23 (2012) 686–701

Fig. 3. The interaction between attachment anxiety and reappraisal. LMX is lower when leaders and subordinates with attachment anxiety use less reappraisal. This effect is stronger for partners (p b .01) than actors (p b .057).

5.1. Attachment orientation and LMX: Direct effects While leader and follower attachment insecurity both were negatively associated with self evaluation of LMX, insecurity was not directly related to their dyad partner's perceptions of relationship quality. This suggests that there may be some “compensating” or “neutralizing” of one's attachment insecurity when interacting with the dyadic partner (e.g., possibly through ER). However, that attachment insecurity was negatively related to one's own appraisal of LMX is consistent with the influence that attachment orientation is believed to have on one's relationship related expectations. As explained earlier with respect to SVT, attachment insecure individuals, for example, will seek to affirm a self-concept of being incapable of forming mutually enriching and trusting relationships (Swann et al., 2003). This self-defeating perspective may negatively impact their perceptions of their current relationships, even if it does not align with reality. Indeed, Harms (2011) argued that such differences in expectation-based perceptions between leaders and subordinates might help account for their typically poor agreement on LMX quality (corrected for unreliability, r = .37; Gerstner & Day, 1997). 5.2. Attachment orientation: Leader and subordinate interaction effects The interaction of leader and subordinate attachment anxiety is important in that the direct negative actor effects can be vitiated when leaders and subordinates have similar levels of attachment anxiety. Thus LMX may be enhanced through mutual levels of proximity seeking. The fact that leader and subordinate attachment avoidance did not interact is noteworthy because it suggests that the negative effect of one dyad partner deactivating proximity seeking is sufficient to inhibit LMX development. 5.3. Attachment and emotion regulation: Moderated effects Since ER moderates associations between attachment orientation and LMX, these effects lend themselves to more substantive interpretation (Aiken & West, 1991). Our findings suggest that individuals with an attachment anxiety orientation will form higher quality LMX relationships when they use ER strategies. In general, those who did not use either suppression or reappraisal had a dyad partner who reported lower LMX.

Fig. 4. The interaction between attachment anxiety and suppression. LMX is lower when leaders and subordinates with attachment anxiety use less suppression. The use of suppression by a partner has a stronger effect than when it is used by an actor.

D.A. Richards, R.D. Hackett / The Leadership Quarterly 23 (2012) 686–701

697

Fig. 5. The interaction between attachment anxiety and suppression. LMX is lower when leaders and subordinates with attachment avoidance use less suppression. The use of suppression by an actor is not significant.

5.3.1. Attachment anxiety and reappraisal As explained earlier, for individuals with higher attachment anxiety, persistent proximity seeking and overdependence likely play havoc with LMX building in the absence of ER. The dependency that accompanies proximity seeking likely “gets in the way” of more rationally directed (versus emotionally motivated) social exchanges that underlie high LMX. In the workplace, difficult job demands, negative treatment by a coworker or client, or other work-related stressors may result in the need to either reframe the threats that “trigger” the attachment system or to reinterpret the experiences associated with unmet attachment needs. Through reappraisal, anxiously attached individuals, appear to be better able to reframe the cognitions surrounding an event (Gross, 1998b; Gross & John, 2003) and, reduce negative emotions that would otherwise trigger hyperactivation of the attachment system. In any case, the underlying processes linking attachment style, ER and LMX quality need to be investigated. For example, “experience sampling” through repeated brief reporting could help capture “in the moment” accounts (Beal & Weiss, 2003; Csikszentmihalyi & Larson, 1987). 5.3.2. Attachment avoidance and reappraisal The interaction between attachment avoidance and reappraisal was not significant for either actors or partners. Although counter to our hypothesis, it is not entirely surprising in that high attachment avoidant individuals are more likely to divert their attention from — or to deny the emotional aspects of — difficult experiences (Mikulincer et al., 2003; Shaver & Mikulincer, 2002) which may circumvent the opportunity to use reappraisal. Indeed, avoidance is negatively related to reappraisal (see Table 1) for both leaders and subordinates. This suggests that avoidant individuals use reappraisal less often than nonavoiders when facing threat-related cues at work, such as emotionally-charged interactions with coworkers where for example, their competence is challenged. 5.3.3. Attachment insecurity and suppression Use of suppression by individuals high in attachment insecurity (as reflected by either anxiety or avoidance) was positively associated with their relationship partner's rating of LMX quality. This suggests that suppression helps keep one's negative emotions “in check” so as to minimize any adverse effects on one's exchange partner. While suppression can also constrain relationship-building expressions of positive emotions (Butler et al., 2003; Gross, 2002), attachment insecure individuals are inherently less likely to experience them in the first place (Mikulincer & Sheffi, 2000). Thus, the net effect of suppression on LMX quality for attachment insecure individuals is likely to be positive. Importantly, there was no actor effect for attachment avoidance, suggesting that suppression when used by avoidant individuals did not elevate their own evaluation of LMX quality. The suppression of experienced emotions may provide relief for anxiously attached individuals who are hyper-vigilant to perceived threats, yet may not be of help to attachment avoidant persons who tend to disavow negative emotions. These findings should be considered in light of the ER research showing that use of suppression comes at a personal cost with respect to one's own emotional and psychological well-being. That is, the negative relationship-focused emotions underlying attachment insecurity, though suppressed from public display, are likely to continue to negatively impact one's own expectations, perceptions and evaluations of LMX, which may account for the lack of an actor effect in relation to suppression and attachment insecurity. 5.4. Contributions to scholarship Our findings contribute to the growing intersecting literatures concerning attachment and leadership. Martin et al. (2010, p. 59) noted that “there is potential for the LMX literature to be informed by (and to inform) the interpersonal relationships literature”. Our findings underscore this point, and are generally consistent with SVT (Swann, 2012), which suggests that attachment style may affect one's propensity to form strong LMX relationships (Keller, 2003; Keller & Cacioppe, 2001; Martin et al., 2010). Apart from leader–follower similarity (actual and perceived), few relationship-focused individual difference variables have been examined as antecedents to LMX (Martin et al., 2010). We help fill this void. Further, the moderating effects of ER uncovered through analyzing the dyads may help account for the inconsistent individual-level findings involving leader and/or follower personality traits and LMX quality (Bernerth, Armenakis, Feild, Giles, & Walker, 2008; Harris, Harris, & Eplion, 2007; Martin, Thomas, Charles, Epitropaki, & McNamara, 2005; Phillips & Bedeian, 1994).

698

D.A. Richards, R.D. Hackett / The Leadership Quarterly 23 (2012) 686–701

As noted earlier, the attachment secure orientation associates positively with leadership emergence (Popper et al., 2004), the use a constructive leadership style, including, transformational leadership (Popper et al., 2000), socialized charismatic leadership (Popper, 2002), and relationship (versus task) oriented leadership (Davidovitz et al., 2007). Attachment security is also positively associated with leadership effectiveness (Davidovitz et al., 2007; Popper, 2002; Popper et al., 2000). Importantly, the significant interactions among attachment and ER appear to be quite robust given the difficulty of detecting interaction effects in field studies. The likelihood of detecting moderating effects is lessened for example, by restriction of range and the clustering of responses in the center as opposed to the extremes of the distribution such that the residual variance of the interaction variable is reduced (McClelland & Judd, 1993). Finally, our findings have important implications for Keller (2003), who proposed that leader and follower will experience consistently positive relationships only where their attachment styles match. Keller (2003) reasoned that congruence in attachment styles would give rise to shared implicit leadership theories (Engle & Lord, 1997), which enhance perceived similarity and mutual identification. In turn, this is thought to “promote common expectations, provide a basis for common understanding, and permit more automatic, intuitive social interactions” (Keller, 2003, p. 53). In the end, the behavior of each party conforms to mutual expectations, producing greater liking and high quality LMX. Ultimately however, this reasoning belies the very essence of a high quality LMX relationship; one marked by strong social bonding expressed through loyalty, trust, and respect, in addition to affect (Graen & Uhl-Bien, 1995). Further, the inability to trust one's relationship partner is a defining characteristic of insecure attachment style; yet trust is fundamental to high quality LMX (Deluga, 1994, 1995; Dirks & Ferrin, 2002; Sue-Chan et al., 2012). Although we found a congruency effect for attachment anxiety, the lack of significant findings involving attachment avoidance, together with the significant interactions with emotion regulation, are inconsistent with Keller's congruency model. The moderating effects of ER, including their varying impact on actor and partner, suggest that more complex interpersonal dynamics underlie the relationship between attachment style and the experience of LMX than is captured by Keller's model.

5.5. Practical Implications As LMX positively predicts a variety of positive individual and organizational work outcomes (Gerstner & Day, 1997), efforts to promote the development of strong LMX relationships have practical value. Unfortunately, since attachment dispositions are relatively stable, it is likely unrealistic for businesses to attempt to modify them. On the other hand, since our findings suggest that the ability to regulate emotion is beneficial in fostering strong leader–subordinate relationships, it may be advantageous for organizations to consider ways to develop ER capabilities. Since skills in recognizing and regulating one's own emotions can be developed through experience and learning (Côté & Miners, 2006), they could be incorporated into leadership coaching, mentoring, and development programs. Such interventions could also help rank and file employees improve their relationships with coworkers (e.g. important for team-building) and clients alike.

5.6. Limitations and future research Our sample drew from a variety of work settings and locations, enhancing the generalizability of our findings. However, our design precluded an examination of contextual factors that may influence specific work sectors or industries. Additionally, since the subordinates invited their supervisors to participate, there could be restriction of range in the quality of relationships within our dyads. On the other hand, as noted above, there was enough variation to allow for the detection of moderators. Similarly, our sampling method may have resulted in relatively large correlations between leaders and subordinates on the measure of the attachment dimensions (r = .61 for anxiety and .40 for avoidance). It is not clear if this result was due to a general attraction between similar individuals or a sampling issue. Research on attachment in other contexts has found that there is an attraction to secure individuals regardless of attachment disposition; however attachment insecure individuals are more likely to be attracted to similar others in romantic relationships (Klohnen & Luo, 2003) and other social relationships (Brumbaugh & Fraley, 2006). Since there are virtually no data on attachment similarity in leader–subordinate dyads, further research is warranted to determine whether there is an attraction effect. Single source bias is also a potential concern for the actor effects we observed. APIM does not address the issue of common method variance (Conway & Lance, 2010; Podsakoff, MacKenzie, Lee, & Podsakoff, 2003); however, the main focus of the study was on the cross-sourced data. Since we asked participants to indicate how they typically regulate their emotions our study design did not benefit from proximal assessments of emotions. Further, while the process model of ER presents reappraisal and suppression as distinct constructs, in everyday life they are less likely to occur in a compartmentalized way (Gross, 2010). ER can happen both automatically and intentionally (Gyurak, Gross, & Etkin, 2011), which was not reflected in our study design. As explained earlier, important strengths of our study, include ratings of attachment style and LMX from both leaders and their followers, dyads that were completely independent from one another, and the dyadic level analysis. In general terms, future research should look to (a) replicate our findings on a broader array of populations; (b) employ longitudinal and experience sampling designs that proximately capture the dynamics underlying LMX development, attachment style and ER; and (c) determine optimal strategies for regulating emotions among leaders and their followers.

D.A. Richards, R.D. Hackett / The Leadership Quarterly 23 (2012) 686–701

699

References Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Thousand Oaks, CA: Sage Publications, Inc.. Ainsworth, M. D., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of attachment: A psychological study of the strange situation. Hillsdale, NJ: Erlbaum. Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. Barsade, S. G. (2002). The ripple effect: Emotional contagion and its influence on group behavior. Administrative Science Quarterly, 47, 644–675. Bauer, T. N., Erdogan, B., Liden, R. C., & Wayne, S. J. (2006). A longitudinal study of the moderating role of extraversion: Leader–member exchange, performance, and turnover during new executive development. Journal of Applied Psychology, 91, 298–310. Beal, D. J., & Weiss, H. M. (2003). Methods of ecological momentary assessments in organizational research. Organizational Research Methods, 6, 440–464. Bernerth, J. B., Armenakis, A. A., Feild, H. S., Giles, W. F., & Walker, J. H. (2008). The influence of personality differences between subordinates and supervisors perceptions of LMX: An empirical investigation. Group and Organization Management, 33, 216–240. Berson, Y., Dan, O., & Yammarino, F. (2006). Attachment style and individual differences in leadership perceptions and emergence. Journal of Social Psychology, 146, 165–182. Blau, P. M. (1964). Exchange and power in social life. New York: John Wiley & Sons, Inc. Bowlby, J. (1973). Attachment and loss: Vol. 2. Separation: Anxiety and anger. New York: Basic Books. Bowlby, J. (1980). Attachment and loss: Vol. 3. Sadness and depression. New York: Basic Books. Bowlby, J. (1982). Attachment and loss: Vol. 1. Attachment (2nd ed.). Basic Books: New York. Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Self-report measurement of adult attachment: An integrative overview. In J. A. Simpson, & W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 46–76). New York: Guilford Press. Brower, H. H., Schoorman, F. E., & Tan, H. H. (2000). A model of relational leadership: The integration of trust and leader–member exchange. The Leadership Quarterly, 11, 227–250. Brumbaugh, C. C., & Fraley, R. C. (2006). Transference and attachment: How do attachment patterns get carried forward from one relationship to the next? Personality and Social Psychology Bulletin, 32, 552–560. Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models: Applications and data analysis methods. Newbury Park, CA: Sage. Butler, E. A., Egloff, B., Wilhelm, F. H., Smith, N. C., Erickson, E. A., & Gross, J. J. (2003). The social consequences of expressive suppression. Emotion, 3, 48–67. Campbell, L. J., & Kashy, D. A. (2002). Estimating actor, partner, and interaction effects for dyadic data using PROC MIXED and HLM5: A brief guided tour. Personal Relationships, 9, 327–342. Cassidy, J., & Kobak, R. R. (1988). Avoidance and its relationship with other defensive processes. In J. Belsky, & T. Nezworski (Eds.), Clinical implications of attachment (pp. 300–323). Hillsdale, NJ: Erlbaum. Conway, J. M., & Lance, C. E. (2010). What reviewers should expect from authors regarding common method bias in organizational research. Journal of Business and Psychology, 25, 325–334. Cook, W. L., & Kenny, D. A. (2005). The actor-partner interdependence model: A model of bidirectional effects in developmental studies. International Journal of Behavioral Development, 29, 101–109. Cooper, M. L., Shaver, P., & Collins, N. (1998). Attachment styles, emotion regulation, and adjustment in adolescence. Journal of Personality and Social Psychology, 74, 1380–1397. Côté, S., & Miners, C. T. H. (2006). Emotional intelligence, cognitive intelligence, and job performance. Administrative Science Quarterly, 51, 1–28. Cropanzano, R., & Mitchell, M. S. (2005). Social exchange theory: An interdisciplinary review. Journal of Management, 31, 874–900. Csikszentmihalyi, M., & Larson, R. (1987). Validity and reliability of the experience-sampling method. The Journal of Nervous and Mental Disease, 175, 526–536. Dasborough, M. T., & Ashkanasy, N. M. (2002). Emotion and attribution of intentionality in leader–member relationships. The Leadership Quarterly, 13, 615–634. Davidovitz, R., Mikulincer, M., Shaver, P. R., Izsak, R., & Popper, M. (2007). Leaders as attachment figures: Leaders' attachment orientations predict leadershiprelated mental representations and followers' performance and mental health. Journal of Personality and Social Psychology, 93, 632–650. Day, D. V., & Crain, E. C. (1992). The role of affect and ability in initial exchange quality perceptions. Group & Organization Management, 17, 380–397. Deluga, R. J. (1994). Supervisor trust building, leader–member exchange and organizational citizenship behaviour. Journal of Occupational and Organizational Psychology, 67, 315–326. Deluga, R. J. (1995). The relation between trust in the supervisor and subordinate organizational citizenship behavior. Military Psychology, 7, 1–16. Dienesch, R. M., & Liden, R. C. (1986). Leader–member exchange model of leadership: A critique and further development. Academy of Management Review, 11, 618–634. Dirks, K. T., & Ferrin, D. L. (2002). Trust in leadership: Meta-analytic findings and implications for research and practice. Journal of Applied Psychology, 87, 611–628. Doverspike, D., Hollis, L., Justice, A., & Polomsky, M. (1997). Correlations between leadership styles as measured by the Least Preferred Co-Worker scale and adults' attachment styles. Psychological Reports, 81, 1148–1150. Engle, E. M., & Lord, R. G. (1997). Implicit theories, self-schemas, and leader–member exchange. Academy of Management Journal, 40, 988–1010. Erdogan, B., Kraimer, M. L., & Liden, R. C. (2004). Work value congruence and intrinsic career success: The compensatory roles of leader–member exchange and perceived organizational support. Personnel Psychology, 57, 305–332. Ferris, G. R., Liden, R. C., Munyon, T. P., Summers, J. K., Basik, K. J., & Buckley, M. R. (2009). Relationships at work: Toward a multidimensional conceptualization of dyadic work relationships. Journal of Management, 35, 1379–1403. Fisk, G. M., & Friesen, J. P. (2012). Perceptions of leader emotion regulation and LMX as predictors of followers' job satisfaction and organizational citizenship behaviors. The Leadership Quarterly, 23, 1–12. Fraley, R. C., Niedenthal, P. M., Marks, M., Brumbaugh, C., & Vicary, A. (2006). Adult attachment and the perception of emotional expressions: Probing the hyperactivating strategies underlying anxious attachment. Journal of Personality, 74, 1163–1190. George, C., & Solomon, J. (1999). Attachment and caregiving: The caregiver behavioral system. In J. Cassidy, & P. R. Shaver (Eds.), Handbook of Attachment. New York: Guilford Press. Gerstner, C. R., & Day, D. V. (1997). Meta-analytic review of leader–member exchange theory: Correlates and construct issues. Journal of Applied Psychology, 82, 827–844. Glasø, L., & Einarsen, S. (2008). Emotion regulation in leader–follower relationships. European Journal of Work and Organizational Psychology, 17, 482–500. Goldin, P. R., McRae, K., Ramel, W., & Gross, J. J. (2008). The neural bases of emotion regulation: Reappraisal and suppression of negative emotion. Biological Psychiatry, 63, 577–586. Graen, G. B., & Scandura, T. A. (1987). Toward a psychology of dyadic organizing. In L. L. Cummings, & B. M. Staw (Eds.), Research in organizational behavior, Vol. 9. (pp. 175–208)Greenwich CT: JAI Press. Graen, G. B., & Uhl-Bien, M. (1995). Relationship-based approach to Leadership: Development of leader–member exchange (LMX) theory of leadership over 25 years: Applying a multi-level, multi-domain perspective. The Leadership Quarterly, 6, 219–247. Grandey, A. A. (2003). When “the show must go on”: Surface and deep acting as determinants of emotional exhaustion and peer-rated service delivery. Academy of Management Journal, 46, 86–96. Gross, J. J. (1998). The emerging field of emotion regulation: An integrative review. Review of General Psychology, 2, 271–299. Gross, J. J. (1998). Antecedent- and response-focused emotion regulation: Divergent consequences for experience, expression, and physiology. Journal of Personality and Social Psychology, 74, 226–237. Gross, J. J. (2002). Emotion regulation: Affective, cognitive, and social consequences. Psychophysiology, 39, 281–291. Gross, J. J. (2010). Emotion regulation. In M. Lewis, J. M. Haviland-Jones, & L. Feldman Barrett (Eds.), Handbook of emotions (pp. 497–512). (3rd ed.). New York, NY: Guilford Press.

700

D.A. Richards, R.D. Hackett / The Leadership Quarterly 23 (2012) 686–701

Gross, J. J., & John, O. P. (2003). Individual differences in two emotion regulation processes: Implications for affect, relationships, and well-being. Journal of Personality and Social Psychology, 85, 348–362. Gyurak, A., Gross, J. J., & Etkin, A. (2011). Explicit and implicit emotion regulation: A dual-process framework. Cognition & Emotion, 25, 400–412. Hackett, R. D., Farh, J. -L., Song, L. J., & Lapierre, L. M. (2003). LMX and organizational citizenship behavior: Examining the links within and across Western and Chinese samples. In G. Graen (Ed.), Dealing with diversity, LMX leadership: The series, Vol. 1. (pp. 219–264)Greenwich, CT: Information Age Publishing. Hardy, G. E., & Barkham, M. (1994). The relationship between interpersonal attachment styles and work difficulties. Human Relations, 47, 263–281. Hareli, S., & Rafaeli, A. (2008). Emotion cycles: On the social influence of emotion in organizations. Research in Organizational Behavior, 28, 35–59. Harms, P. D. (2011). Adult attachment styles in the workplace. Human Resource Management Review, 21, 285–296. Harris, K. J., Harris, R. B., & Eplion, D. M. (2007). Personality, leader–member exchanges, and work outcomes. Journal of Behavioral and Applied Management, 8, 92–107. Hazan, C., & Shaver, P. (1987). Romantic love conceptualized as an attachment process. Journal of Personality and Social Psychology, 52, 511–524. Hazan, C., & Shaver, P. (1990). Love and work: An attachment–theoretical perspective. Journal of Personality and Social Psychology, 59, 270–280. Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley. Johnson, S. K. (2008). I second that emotion: Effects of emotional contagion and affect at work on leader and follower outcomes. The Leadership Quarterly, 19, 1–19. Kahn, W., & Kram, K. (1994). Authority at work: Internal models and their organizational consequences. Academy of Management Review, 19, 17–50. Kashy, D. A., & Kenny, D. A. (2000). The analysis of data from dyads and groups. In H. T. Reis & C. M. Judd (Eds.), Handbook of research methods in social psychology (pp. 451–477). New York: Cambridge University Press. Keller, T. (2003). Parental images as a guide to leadership sensemaking: An attachment perspective on implicit leadership theories. The Leadership Quarterly, 14, 141–160. Keller, T., & Cacioppe, R. (2001). Leader–follower attachments: Understanding parental images at work. Leadership & Organization Development Journal, 22, 70–75. Kelley, H. H. (1967). Attribution theory in social psychology. In D. Levine (Ed.), Nebraska Symposium on Motivation, Vol. 15. (pp. 192–238)Lincoln, NE: University of Nebraska Press. Kenny, D. A., Kashy, D. A., & Cook, W. L. (2006). Dyadic data analysis. New York, NY: Guilford Press. Klohnen, E. C., & Luo, S. (2003). Interpersonal attraction and personality: What is attractive — self similarity, ideal similarity, complementarity, or attachment security? Journal of Personality and Social Psychology, 85, 709–722. Liden, R. C., & Maslyn, J. M. (1998). Multidimensionality of leader–member exchange: An empirical assessment through scale development. Journal of Management, 24, 43–72. Liden, R. C., Sparrowe, R. T., & Wayne, S. J. (1997). Leader–member exchange theory: The past and potential for the future. In G. Ferris (Ed.), Research in personnel and human resources management, Vol. 16. (pp. 47–119)Greenwich, CT: JAI Press. Martin, R., Epitropaki, O., Thomas, G., & Topaka, A. (2010). A critical review of leader–member relationship (LMX) research: Future prospects and directions. In G. Hodgkinson, & K. Ford (Eds.), International review of industrial and organizational psychology. Chichester, UK: Wiley. Martin, R., Thomas, G., Charles, K., Epitropaki, O., & McNamara, R. (2005). The role of leader–member exchanges in mediating the relationship between locus of control and work reactions. Journal of Occupational and Organizational Psychology, 78, 141–147. McClelland, G. H., & Judd, C. M. (1993). Statistical difficulties of detecting interactions and moderator effects. Psychological Bulletin, 114, 376–390. Merton, R. K. (1968). Social theory and social structure (1968 enlarged edition). New York: The Free Press. Mikulincer, M. (1995). Attachment style and the mental representation of the self. Journal of Personality and Social Psychology, 69, 1203–1215. Mikulincer, M., & Florian, V. (1998). The relationship between adult attachment styles and emotional and cognitive reactions to stressful events. In J. A. Simpson & W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 143–165). New York: The Guilford Press. Mikulincer, M., & Shaver, P. R. (2005). Attachment theory and emotions in close relationships: Exploring the attachment-related dynamics of emotional reactions to relational events. Personal Relationships, 12, 149–168. Mikulincer, M., & Shaver, P. R. (2007). Attachment in adulthood: Structure, dynamics, and change. New York: Guilford Press. Mikulincer, M., Shaver, P. R., & Pereg, D. (2003). Attachment theory and affect regulation: The dynamics, development, and cognitive consequences of attachment related strategies. Motivation and Emotion, 27, 77–102. Mikulincer, M., & Sheffi, E. (2000). Adult attachment style and cognitive reactions to positive affect: A test of mental categorization and creative problem solving. Motivation and Emotion, 24, 149–174. Newcombe, M., & Ashkanasy, N. (2002). The role of affect and affective congruence in perceptions of leaders: An experimental study. The Leadership Quarterly, 13, 601–614. Noftle, E., & Shaver, P. (2006). Attachment dimensions and the big five personality traits: Associations and comparative ability to predict relationship quality. Journal of Research in Personality, 40, 179–208. Pellegrini, E. K., & Scandura, T. A. (2006). Leader–member exchange (LMX), paternalism, and delegation in the Turkish business culture: An empirical investigation. Journal of International Business Studies, 37, 264–279. Phillips, A. S., & Bedeian, A. G. (1994). Leader–follower exchange quality: The role of personal and interpersonal attributes. Academy of Management Journal, 37, 990–1001. Podsakoff, P., MacKenzie, S., Lee, J., & Podsakoff, N. (2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88, 879–903. Popper, M. (2002). Narcissism and attachment patterns of personalized and socialized charismatic leaders. Journal of Social and Personal Relationships, 19, 797–809. Popper, M., & Amit, K. (2009). Attachment and leader's development via experiences. The Leadership Quarterly, 20, 749–763. Popper, M., Amit, K., Gal, R., Mishkal-Sinai, M., & Lisak, A. (2004). The capacity to lead: Major psychological differences between leaders and non-leaders. Military Psychology, 16, 245–263. Popper, M., Mayseless, O., & Castelnovo, O. (2000). Transformational leadership and attachment. The Leadership Quarterly, 11, 267–289. Richards, J. M., & Gross, J. J. (2000). Emotion regulation and memory: The cognitive costs of keeping one's cool. Journal of Personality and Social Psychology, 79, 410–424. Richards, D. A., & Schat, A. C. H. (2011). Attachment at (not to) work: Applying attachment theory to explain individual behavior in organizations. Journal of Applied Psychology, 96, 169–182. Ross, L. R., McKim, M. K., & DiTommaso, E. (2006). How do underlying “Self” and “Other” dimensions define adult attachment styles? Canadian Journal of Behavioural Sciences, 38, 294–310. Sears, G. J., & Hackett, R. D. (2011). The influence of role definition and affect in LMX: A process perspective on the personality–LMX relationship. Journal of Occupational and Organizational Psychology, 84, 544–564. Shaver, P. R., Collins, N., & Clark, C. L. (1996). Attachment styles and internal working models. In G. Fletcher & J. Fitness (Eds.), Knowledge structures in close relationships: A social psychological approach. Hillsdale, N.J: Erlbaum. Shaver, P. R., & Mikulincer, M. (2002). Attachment-related psychodynamics. Attachment & Human Development, 4, 133–161. Sparrowe, R. T., & Liden, R. C. (1997). Process and structure in leader–member exchange. Review, 22, 522–552. Stanton, J. M., & Weiss, E. M. (2002). Online Panels for Social Science Research: An Introduction to The StudyResponse Project. (Tech. Rep. No. 13001). Syracuse University, School of Information Studies (Retrieved from: http://www.studyresponse.net/TechRpt13001.pdf) Sue-Chan, C., Au, A. K., & Hackett, R. D. (2012). Trust as a mediator of the relationship between leader/member behavior and leader–member-exchange quality. Journal of World Business, 47, 459–468. Swann, W. B., Jr. (2012). Self-verification theory. In P. A. M. Van Lange, A. W. Kruglanski, & E. T. Higgins (Eds.), Handbook of theories of social psychology, Vol. 2. (pp. 23–42)London: Sage.

D.A. Richards, R.D. Hackett / The Leadership Quarterly 23 (2012) 686–701

701

Swann, W. B., Jr., Rentfrow, P. J., & Guinn, J. S. (2003). Self verification: The search for coherence. In M. R. Leary & J. P. Tangney (Eds.), Handbook of self and identity. New York: The Guilford Press. Sy, T., Côté, S., & Saavedra, R. (2005). The contagious leader: Impact of the leader's mood on the mood of group members, group affective tone, and group processes. Journal of Applied Psychology, 90, 295–305. Tett, R. P., & Burnett, D. D. (2003). A personality trait-based interactionist model of job performance. Journal of Applied Psychology, 88, 500–517. Wayne, S. J., Shore, L. M., & Linden, R. C. (1997). Perceived organizational support and leader–member exchange: A social exchange perspective. Academy of Management Journal, 40, 82–111. Wong, C. S., & Law, K. S. (2002). The effects of leader and follower emotional intelligence on performance and attitude: An exploratory study. The Leadership Quarterly, 13, 243–274. Yakovleva, M., Reilly, R. R., & Werko, R. (2010). Why do we trust? Moving beyond individual to dyadic perceptions. Journal of Applied Psychology, 95, 79–91.