REGULATORY FOCUS

WHEN LEADER-MEMBER EXCHANGE (LMX) MATTERS TO SUBORDINATE CREATIVITY: THE ROLE OF LEADERSHIP

REGULATORY FOCUS

SOM Research and Graduate School, University of Groningen Faculty of Economics and Business

MSc (Res) HRM&OB

Master Thesis

May 2011

Ramzi Said Oosterstraat 86 9711 NZ Groningen

+31 (0)646264763 r.said@student.rug.nl

S1612735

Supervisor:

Prof. Dr. B. A. (Bernard) Nijstad

ABSTRACT

In the present paper we challenge the conventional view in leader-member exchange

(LMX) literature, by proposing that LMX is not necessarily positively related to

subordinate creativity. Rather, we argue, that the nature of this relationship depends

on the specific regulatory focus (i.e. promotion or prevention) adopted by the

supervisor in directing subordinates. More specifically, we reason that a perceived

leadership promotion focus in a high LMX relationship results in more originality and

supervisory rated creativity, whereas a perceived leadership prevention focus in a

high LMX relationship elicits higher subordinate conformity, thereby hampering

subordinates’ creative behaviour. Results of an experimental (study 1) and multi-

source field study (study 2), however, could not support this notion and showed that

LMX was not related to subordinate originality, fluency or supervisory rated

creativity, but rather to conformity (study 1). Perceived regulatory focus did show

some main effect results on (facets of) creativity, but the predicted interactions

between LMX and perceived leadership regulatory focus were not found. Implications

of these results are discussed.

INTRODUCTION

What would be better for creative behaviour of subordinates, a relationship between subordinates and their supervisor based on mutual trust and respect, or one based on strict formal work contracts? As the question goes through your mind, your first reaction probably is: ‘a relationship based on trust and respect, of course!’. And indeed, previous research shows that such a relationship between a supervisor and subordinate, also known as high leader-member exchange (LMX), is best for subordinate creativity (e.g. Tierney, Farmer & Graen, 1999). LMX refers to the reciprocal exchanges between a subordinate and his or her direct supervisor based on trust, respect and obligations (Liao, Liu & Loi, 2010), whereas creativity is defined as

“the production of novel and useful ideas in any domain” (Amabile et al., 1996, p.

1155). Research shows that the relation between LMX and subordinate creativity tends to be positive because, for example, high LMX subordinates engage in more job-related risk taking and receive more task-related resources conducive to their creativity than their low LMX peers (Amabile, 1988; Graen & Cashman, 1975; Graen

& Scandura, 1987). In short, as argued by Tierney et al. (1999), the nature of LMX appears to be compatible with subordinate creativity.

However, there are reasons to believe this is not the whole story. We challenge the conventional view in the current research by proposing that LMX is not by definition positively related to subordinate creativity. On the contrary, we argue that in a high LMX relationship, subordinates are more susceptible to social pressure stemming from the supervisor (Dienesch & Liden, 1986; Graen & Uhl-Bien, 1995) and that this pressure can result in subordinates conforming to their supervisor thereby hampering their creativity (Chirumbolo et al., 2004).

While increased social pressure in a high LMX relationship is not in itself beneficial or detrimental for subordinate creativity, we reason that this depends on the specific behaviour enacted by supervisors in directing subordinates, more specifically their regulatory focus. If this supervisory behaviour is promotion focused (i.e. aimed at obtaining gains or successes), subordinates’ eagerness is increased (Crowe &

Higgins, 1997), making them diverge from their supervisor in their search for success,

which positively influences creativity (Chirumbolo et al., 2004; Friedman & Förster,

2001). If supervisors’ behaviour is prevention focused (i.e. directed at avoiding losses

or failures), this will increase subordinates’ vigilance (Crowe & Higgins, 1997),

which makes subordinates conform more to their supervisor because this conformity will minimize the chance on failures in the eyes of the subordinates. This conformity will then result in less subordinate creativity (Chirumbolo et al., 2004). Therefore, we argue that high LMX with promotion focused leadership will result in subordinates being most creative in their pursuit of successes, whereas high LMX with prevention focused leadership results in less subordinate creativity because subordinates conform more to their supervisor in their quest for avoiding failures.

The contribution of this paper is twofold. First, it adds a new perspective to the current leadership literature, one in which high LMX is not necessarily related to high subordinate creativity. Second, it examines how the regulatory focus the supervisor adopts to direct subordinates interacts with LMX and can be an activator or deactivator of subordinate creativity. Therefore, by combining two different lines of leadership research – LMX and leadership regulatory focus – and examining their individual and joint impact on subordinate creativity and conformity, a new avenue of research is brought into the leadership and creativity literature, arguing that only under specific conditions LMX is positively related to subordinate creativity.

The present study starts with an overview of past research on LMX and leadership regulatory focus, and explains how and why both are important for and related to subordinate creativity and conformity. Thereafter, we combine both streams of leadership research in an experiment (study 1) and a multi-source field study (study 2). Per study, separate hypotheses are formulated.

THEORY

Leader-member exchange (LMX) theory posits that supervisors develop

different exchange relationships with their subordinates through a role-making

process (Dienesch & Liden, 1986). Supervisors establish high quality exchange

relationships with some of their subordinates (called in-group members), whereas

with others the relationship is characterized by low quality exchanges (called out-

group members). With its focus on supervisor-subordinate dyadic interaction, the

LMX model positions leadership as a relational phenomenon serving as a conduit for

several relevant outcomes (Graen & Uhl-Bien, 1995), among which subordinate

creativity (Graen & Scandura, 1987; Tierney et al., 1999).

Creativity is concerned with the production of novel and useful ideas (Amabile et al., 1996). While the construct of creativity is composed of different elements (e.g. originality and fluency, for overview see Rietzchel, De Dreu & Nijstad, 2007), we use the generic term creativity to denote the production of an idea that is novel and useful. Only where important, we will differentiate between the facets of creativity.

Leader-Member Exchange (LMX), Creativity and Conformity

As outlined in the introduction, previous research (e.g. Tierney et al., 1999) found a positive linear relationship between LMX and subordinate creativity (a noteworthy exception is Liao et al., 2010). There are several reasons why such a relation may occur. First, in high quality exchange relationships subordinates are given greater autonomy to carry out their tasks (Graen & Scandura, 1987) and receive more supervisory support, loyalty and trust than their low LMX counterparts (Basu &

Green, 1997; Dienesch & Liden, 1986). By providing autonomy, supervisors create an atmosphere of free thinking in which subordinates can engage in unconventional thoughts and behaviours beneficial to their creativity (Basu & Green, 1997; Graen &

Scandura, 1987; Scott & Bruce, 1994, 1998). Support, loyalty and trust can result in higher subordinate creativity by facilitating an encouraging work environment suitable for creative output (Dunegan, Tierney & Duchon, 1992; Liden, Sparrowe &

Wayne, 1997; Mumford & Gustafson, 1988; Oldham & Cummings, 1996). Second, high LMX relationships are characterized by positive interactions between the subordinate and supervisor. These interactions heighten subordinates’ feelings of energy (Atwater & Carmeli, 2009) and to be creative, individuals need a considerable amount of energy (Amabile, 1983). Finally, Tierney et al. (1999) note that high LMX subordinates are more engaged in job-related risk taking (Graen & Cashman, 1975) and challenging tasks (Liden & Graen, 1980), receive more task-related resources (Graen & Scandura, 1987) and experience a strong sense of advocacy and liking from their supervisors (Duchon, Green & Taber, 1986), which are all positively related to creativity (e.g. see Amabile, 1988; Amabile & Gryskiewiez, 1987; Mumford &

Gustafson, 1988).

Therefore, there are various reasons why LMX should show a positive

relationship with subordinate creativity. However, we argue that there are some

processes inherent to a high LMX relationship that can potentially, under specific

conditions, be detrimental for subordinate creativity. As mentioned, subordinates are more susceptible to social pressure stemming from the supervisor in a high LMX relationship (Dienesch & Liden, 1986; Graen & Uhl-Bien, 1995). We argue that this social pressure can result in subordinates conforming to their supervisor for two reasons. First, the supervisor values and wants desired relational outcomes from the subordinates in a high LMX relationship, such as approval, loyalty, respect, status and esteem (Graen & Uhl-Bien, 1995; Liden & Maslyn, 1998). To show their loyalty, subordinates commit themselves to the goals and visions of their supervisor, thereby conforming themselves to the supervisor (Lee, 2008; Liden & Maslyn, 1998). Second, since subordinates will only receive favourable outcomes from the supervisor when their relationship is at the higher ends of LMX, we argue that in their search to keep a high LMX relationship and receive favourable outcomes, subordinates are pressured to give in to the supervisor (Dienesch & Liden, 1986; Graen & Uhl-Bien 1995; Liden

& Maslyn, 1998). For these two reasons, subordinates in a high LMX relationship are more likely to give in to the supervisor, compared to subordinates working in a low LMX relationship.

Taken together, high LMX has many positive outcomes related to creativity, but can also create an atmosphere of like-mindedness or tranquillity, in which the subordinate conforms to the supervisor to protect the valuable relationship they currently have and to make sure that the supervisor receives the expected relational outcomes from the subordinate. In such a situation, subordinates are less likely to bring forward dissenting opinions and are more likely to seek consensus and uniformity. Conformity, whether seen as lack of dissent from the subordinate or as a search of the subordinate for consensus and uniformity, is negatively associated with creativity (Chirumbolo et al., 2004; Nemeth & Ormiston, 2007).

So when will subordinates take advantage of the possibilities of high LMX to obtain creative output, and when not? To determine when – under which specific conditions – high LMX results in more subordinate creativity or conformity, we take into account the specific regulatory focus adopted by supervisors in directing their subordinates.

Leadership Regulatory Focus, Creativity and Conformity

In general, the leadership literature has paid limited attention to the underlying

processes by which leaders motivate followers, while leaders’ behaviours are an

important component in shaping subordinates’ motivations and behaviours (Kark &

Van Dijk, 2007). According to Brockner and Higgins (2001), subordinates may perceive their leader’s behaviour as being directed at promotion focused or prevention focused concerns. This perception of leaders’ regulatory focus will influence subordinate behaviour because it will elicit a congruent state of regulatory focus in the subordinate, leading subordinates to work according to this perceived leadership regulatory focus. Regardless of subordinates’ chronic regulatory focus, a leader with whom the subordinate interacts heavily, such as a direct supervisor, can make the subordinate promotion or prevention focused by adopting one of these foci in directing subordinates (Wu et al., 2008). Therefore, in the present paper, leadership regulatory focus is conceptualized as specific behaviours of the supervisor in directing subordinates, which are used by the supervisor to influence subordinates to pursue what the supervisor values and wants (i.e. promotion or prevention goals).

A promotion focus allocates attention primarily to the benefits of success and the costs of failing to succeed, by focussing on the achievement of gains or the attainment of ideals (wishes, hopes and aspirations; Crowe & Higgins, 1997; Förster, Higgins & Idson, 1998). A leadership promotion focus therefore encourages subordinate eagerness, which alters their preferences such that subordinates engage more in creativity (Förster, Friedman & Liberman, 2004). When subordinates experience a leadership promotion focus, their attention is allocated to maximizing goals and they are thereby drawn towards the pursuit of approach-oriented ends, such as gains or successes. In this pursuit, the subordinates are more risk-taking, are more likely to violate agreed-upon ways of doing things and are less likely to conform to their supervisor, compared to subordinates experiencing a leadership prevention focus (Wu et al., 2008). Therefore, subordinates perceiving the behaviour of their supervisor highly promotion focused pursue development and change, and explore the benefits of creative behaviours (Kark & Van Dijk, 2007).

A prevention focus, on the other hand, allocates attention primarily to the benefits of avoiding failure and the costs of making mistakes, by emphasizing security needs, avoidance of losses or the fulfilment of obligations (duties and responsibilities, i.e. oughts; Crowe & Higgins, 1997; Förster et al., 1998). A leadership prevention focus therefore encourages subordinate vigilance, which alters subordinates’

preferences to become more conservative and less likely to engage in creative

behaviour (Förster et al., 2004). Thus, a leadership prevention focus results in

subordinates putting effort in minimizing risks and therefore search for avoidance- oriented ends, such as reducing losses and failures. In this search, subordinates will conform more to their supervisor, since this will reduce the chance of making mistakes and minimizes risks. Thus, the purpose of leadership prevention focus is to assure security, to maintain routines and to preserve the status quo, which is likely to induce subordinate preservation, repetitive and conforming behaviour, and less creativity (Kark & Van Dijk, 2007).

In sum, because creativity requires a cognitive style that is exploratory and flexible (Amabile, 1996), subordinate creative behaviour is more likely under a leadership promotion focus that induces subordinate eagerness, than under a leadership prevention focus that results in subordinate vigilance. Indeed, empirical evidence indicates this is the case. For example, Liberman et al. (1999) found that participants who were promotion-primed showed an exploratory tendency beneficial to their creativity, whereas prevention-primed participants had a more conservative tendency harmful for their creativity. Crowe and Higgins (1997) showed that promotion-focused participants were more flexible, whereas their prevention-focused counterparts were more conservative and less flexible, thereby resulting in less creativity. Furthermore, Friedman and Förster (2001) found that creative insight and creative idea generation were higher under promotion focus, compared to prevention focus. Finally, the findings of Wu et al. (2008) suggested that a leadership promotion focus results in an increase of employees’ creative behaviours.

STUDY 1

Our purpose for Study 1 was to manipulate both LMX (low vs. high) and

leadership regulatory focus (prevention vs. promotion) in an experiment, and to

measure subordinate creativity in an objective manner. This allowed us to draw

conclusions concerning causality, while maintaining a relatively high degree of

mundane realism. We differentiate between the facets of creativity (i.e. originality,

conformity and fluency; see Dijksterhuis & Meurs, 2006; Rietzchel et al., 2007), since

we expect different effects of LMX and leadership regulatory focus on the facets, as

will be argued below. Originality is the number of novel ideas (Reitzchel et al., 2007),

conformity is the number of converging ideas (i.e. ideas similar to that of the

supervisor, thereby signalling noncreativity; Dijksterhuis & Meurs, 2006), and

fluency is defined as “generating as many ideas, options or solutions as possible”

(Rietzchel et al., 2007, p. 856). Therefore, subordinates’ originality (number of novel ideas) and conformity (number of converging ideas) combined form subordinates’

fluency.

Hypotheses

In the present paper, we argue that high LMX can be beneficial for creativity, but that this relationship depends on the regulatory focus adopted by the supervisor in directing subordinates. Given the perceived leadership regulatory focus, subordinates will use the relational and work outcomes of high LMX, such as contribution, loyalty, trust, respect and physical resources (i.e. currencies of exchange; Dienesch & Liden, 1986; Liden & Maslyn, 1998), to pursue different ends.

When subordinates perceive the behaviour of the supervisor as highly promotion focused, their eagerness is increased which will make them take full use of the exchange currencies in a high LMX situation to come up with novel and useful ideas in their pursuit of approach-oriented ends, such as successes. In this pursuit, subordinates do not restrict themselves to the views of their supervisor and explore diverging ways of obtaining desired ends. Since the ability of individuals to be original depends on the ability to approach problems or situations in a new manner without relying on habitual thoughts (Rietzchel et al., 2007), the first hypothesis of this study is:

Hypothesis 1: LMX will be positively related to subordinate originality, and this effect will be more positive when subordinates perceive their supervisor’

behaviour promotion focused compared to prevention focused.

In contrast, subordinates experiencing a leadership prevention focus have their

vigilance increased, making them use the exchange currencies within a high LMX

relationship to avoid failures and to give the supervisor expected relational outcomes,

and not to explore new or divergent ways of achieving desired results. In their quest

for avoiding failures, subordinates conform to their supervisor, thereby stifling

independent thinking and increasing convergent ways of thinking and acting. Since

subordinate conformity depends on the extent to which subordinates have ideas

similar to those of their supervisor (Dijksterhuis & Meurs, 2006), the second hypothesis of this study is:

Hypothesis 2: LMX will be positively related to subordinate conformity, and this effect will be more positive when subordinates perceive their supervisor’

behaviour prevention focused compared to promotion focused.

Finally, since subordinate fluency is composed of the number of original and conforming ideas, and because the effect of LMX on subordinate originality is stronger under a leadership promotion focus and the effect of LMX on subordinate conformity is stronger under a leadership prevention focus, we argue that there will be no differences of the effect of LMX on subordinates’ fluency under the different leadership regulatory foci. Given that high LMX subordinates receive more physical and psychological exchange currencies than their low LMX peers, this enables them to work harder, to be more willing to work beyond what is formally required from them and are thereby more likely to generate many ideas, options or solutions (Amabile, 1988; Basu & Green, 1997; Graen & Scandura, 1987; Graen & Uhl-Bien, 1995; Liden & Maslyn, 1998; Mumford & Gustafson, 1988; Tierney et al., 1999).

Therefore, the third hypothesis of this study is:

Hypothesis 3: LMX will be positively related to subordinate fluency.

Method

Participants and design. One hundred and four undergraduate business students (66 men and 38 women; average age = 21.86 years, SD = 1.90) at the University of Groningen participated in this study for partial course credits or for money (€4). Participants were randomly assigned to a 2 (LMX: low vs. high) x 2 (leadership regulatory focus: preventions vs. promotion) between-subjects design.

Experimental procedure. After reading and signing a consent form, participants were placed in separate cubicles with a computer. The experiment was fully computer-based and all information was provided to participants via the computer.

To manipulate LMX – which is relationship based – participants were led to

believe they were paired with another participant who was also currently in one of the

cubicles, and that one of both individuals would become the supervisor of the other.

In reality however, all participants were assigned the role of subordinate, and all information given to the participants by their supervisor was in fact pre-programmed.

Before participants were given information about/from their supervisor, they answered several questionnaires, measuring their gender, age and personality (Big 5).

While participants filled out the personality questionnaires, the supervisors allegedly filled out a number of questionnaires regarding their management style and philosophy. Participants were given this information in the form of the following scenarios in which LMX and leadership regulatory focus were manipulated. The LMX scenarios were based on previous research of Dienesch and Liden (1986), Liden and Graen (1980) and Liden and Maslyn (1998), whereas the leadership regulatory focus scenarios were based on research of Higgens et al. (2001), Kark and Van Dijk (2007) and Wu et al. (2008).

In the low LMX condition, the scenario (i.e. management philosophy) said:

High performance is important to me and I want us, together, to perform well.

Therefore, I see our relationship as one based on a formal contract. I will make sure you perform your job properly, and I will only give you the resources and work-related help I think you need to do your job. Your work tasks are exactly defined, therefore we do not need to discuss your work or performance on a regular basis. To me it would be best if we just work conform our job descriptions, nothing more, nothing less.

Participants were also shown a graph with the score (1-10) of their supervisor on management style and the average score of all supervisors. The higher the score, the more the supervisor valued a good relationship with the subordinate. In the low LMX condition, this score was set at 3.3 for the supervisor and at 6.4 as average score of all supervisors, which gave participants the impression that their supervisor did not value a good relationship with them to great extent.

In the high LMX condition, the scenario (i.e. management philosophy) said:

High performance is important to me and I want us, together, to perform well.

Therefore, a good relationship and cooperation with you are important to me.

I will support and help you as much as I can with work-related and personal problems, so you can perform properly. However, you are also responsible for your own work and you should have a drive to achieve. I will discuss your work and our cooperation on a regular basis with you, to improve them. In sum, I will go beyond my formal job description to work together and to achieve, and I hope you will do the same.

The graph showed a score of 8.1 for their supervisor on valuing relationship quality, whereas the average of all supervisors was set at 6.4, thereby giving participants the impression that their supervisor valued a good relationship with them to greater extent than other supervisors do.

In the leadership prevention focus condition, the scenario said:

To me it is important that people who work for me feel responsible and do their job the right way. That is why I want you to focus on completing your work tasks correctly, and to do your duty and fulfil your obligations at work.

When you do make mistakes, this will have consequences for you.

In the leadership promotion focus condition, the scenario said:

To me it is important that people who work for me achieve their goals and have the drive to do their work even better. This means that you accomplish a lot of work, to make sure you always get ahead at work. This usually results in success and you will be rewarded when that success is achieved.

After these manipulations, participants answered questions concerning their

perceived LMX and leadership regulatory focus (see measures below), to examine

whether the manipulations were successful. Before participants filled out these

questionnaires, they were given information that their supervisor was asked to give

them instructions for the upcoming task.

Next, participants had to do the task, meant to measure creativity: generate new names for pasta (Dijksterhuis & Meurs, 2006; see measures below). In all conditions, participants were informed that they had to complete this task on behalf of their supervisor, that their supervisor would evaluate them and that they had to think about what their supervisor values within their relationship. To refresh their memory, participants again received information, allegedly from their supervisor, about management style and philosophy.

In the low LMX condition, the scenario said:

To make sure that you will do well at this task, I will watch you closely and give support according to your contract, nothing more, nothing less.

In the high LMX condition, the scenario said:

I know you will do fine on this task, I trust, support and help you, and together we will complete this task successfully.

In the leadership prevention focus condition, the scenario said:

To me it is important that people who work for me do so correctly and fulfil their obligations. You are responsible for your task: generating pasta names.

Too few or bad pasta names will thus lead to a negative evaluation.

In the leadership promotion focus condition, the scenario said:

Success is very important to me. That is why I want you to think of as much pasta names as possible, to get ahead at work and at others. Generating as many good pasta names as possible will thus lead to a positive evaluation.

In all conditions, the supervisor listed five examples of non-existing names in

the instructions, all ending with the letter ‘i’. These names were supposedly given by

the supervisor to help participants come up with new pasta names. Participants had

three minutes for this task.

After participants completed the task, they answered questions concerning their mood and task engagement. Finally, participants were debriefed, thanked and received their partial course credits or money.

Measures. The leader-member exchange manipulation check was adapted from Liden and Maslyn (1998). This measure consisted of 4 items, based on the highest factor loadings per dimension (i.e. affect, loyalty and contribution, see Liden

& Maslyn, 1998) and one overall item. No item was included for the factor professional respect. The items were answered on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree) and an example item is: ‘I like my supervisor very much as a person’. Chronbach’s alpha was .90.

The leadership regulatory focus manipulation check was adapted from Wallace, Johnson and Frazier (2009). Two prevention and two promotion items were included in this measure, again based on highest factor loadings per dimension (see Wallace et al. 2009). To test the dimensionality of this scale, exploratory factor analysis with varimax rotation was performed on these 4 items, which yielded support for a two-factor structure with prevention items loading on one factor and promotion items loading on the other (all loadings were above .85). Reliability analyses showed that the 2-item promotion scale had a Chronbach’s alpha of .68, whereas the 2-item prevention scale had an alpha of .70. These scales were both sufficient, given the fact that they were composed of only 2 items each. The items were answered on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree) and example items are: ‘My supervisor wants me to focus on accomplishing a lot of work’ (promotion) and ‘My supervisor wants me to focus on doing my duty and fulfilling my obligations at work’

(prevention).

Creativity was assessed using the pasta names generated by the participants

(see Dijksterhuis & Meurs, 2006). Since creativity is composed of different facets, we

assessed participants’ fluency (total number of pasta names), originality (total and

relative number of pasta names not ending on the letter ‘i’, diverging items) and

conformity (total and relative number of pasta names ending on the letter ‘i’,

converging items). The relative number of original and conforming pasta names is the

total number of original and conforming pasta names divided by the total number of

pasta names (i.e. fluency). Since the supervisor already listed 5 pasta names ending on

the letter ‘i’, pasta names generated by participants ending with the letter ‘i’ signal

conformity, whereas names not ending with the letter ‘i’ signal originality (see also

Dijksterhuis & Meurs, 2006). Higher fluency and originality signal creativity, whereas higher conformity signals noncreativity.

To explore possible mediation effects, we also assessed participants’ mood and task engagement after the manipulations and creativity task. The Positive Affect Negative Affect Schedule (PANAS; Watson, Clark & Tellegen, 1988) was used as process measure to assess participants’ mood after the manipulations and creativity task. In line with Watson et al. (1988), exploratory factor analysis with varimax rotation indicated that positive affect and negative affect were two different factors.

The positive affect factor (10 items) had a reliability coefficient of .88, whereas this coefficient was .90 for the negative affect factor (10 items). All items started with ‘At this moment I feel…’ and were followed by, for example, ‘…enthusiastic’ (positive affect) or ‘…scared’ (negative affect). Answering was on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree).

The second process measure, task engagement, was adapted from Schaufeli, Bakker and Salanova (2006). This scale consisted of 9 items, answered on a 7-point Likert scale (1 = strongly disagree, 7 = strongly agree), and showed good internal consistency (α = .91). An example item is: ‘The task inspired me’.

Results

Manipulation checks. Three 2 (LMX: low vs. high) x 2 (leadership regulatory focus: prevention vs. promotion) analyses of variance (ANOVAs) were performed on the LMX, prevention and promotion measures (for overview results see table 1).

Based on the factor analysis, prevention and promotion focus were treated as two different variables.

--- Insert table 1 about here ---

A two-way ANOVA on the 4-item LMX measure yielded a significant main

effect for LMX, F(1, 100) = 375.14, p < .001, η

= .79, indicating that participants in

the high LMX condition indeed gave higher LMX-ratings than participants in the low

LMX condition (M = 4.13 vs. 2.06, SD = .66 and .45, respectively). While less strong,

there was, however, also a significant main effect for leadership regulatory focus, F(1,

100) = 6.36, p < .05, η

= .06, and a significant interaction, F(1, 100) = 4.68, p < .05,

η

= .05. The main effect of leadership regulatory focus indicated that the participants

in the promotion condition gave higher LMX-ratings than participants in the prevention condition (M = 3.23 vs. 2.96, SD = 1.27 and 1.08, respectively). The interaction effect was caused by a stronger effect of LMX in the prevention than in the promotion focus condition. As simple effect tests showed, the effect of LMX on the 4-item LMX measure was stronger in the leadership prevention, F(1, 50) = 248.57, p < .001, η

= .83, compared to the leadership promotion condition, F(1, 50) = 138.70, p < .001, η

= .74.

A two-way ANOVA on the 2-item leadership prevention measure yielded a significant main effect for leadership regulatory focus, F(1, 100) = 9.09, p < .005, η

= .08, indicating that participants in the leadership prevention condition experienced this prevention focus more than participants in the leadership promotion condition (M

= 4.52 vs. 4.17, SD = .56 and .65, respectively). There was, however, also a significant main effect for LMX, F(1, 100) = 9.09, p < .005, η

= .08. Participants in the low LMX condition gave significantly higher ratings on leadership prevention than participants in the high LMX condition (M = 4.52 vs. M = 4.17, SD = .50 and .69, respectively). The interaction was not significant, F(1, 100) = .11, p = .74, η

= .001. Simple effect tests showed that the effect of leadership regulatory focus on the 2-item leadership prevention measure was significant in the low LMX, F(1, 50) = 5.24, p < .05, η

= .10, and the high LMX condition, F(1, 50) = 4.27, p < .05, η

= .08.

A two-way ANOVA on the 2-item leadership promotion measure revealed the expected significant main effect for leadership regulatory focus, F(1, 100) = 43.36, p

< .001, η

= .30, indicating that participants in the leadership promotion condition thought there supervisor was more promotion focused, compared to what participants in the prevention condition thought about their supervisor (M = 4.31 vs. 3.33, SD = .77 and .79, respectively). Also here, there were a significant main effect for LMX, F(1, 100) = 6.70, p < .05, η

= .06, and a non-significant interaction, F(1, 100) = .00, p = 1.00, η

= .00. The main effect of LMX indicated that in the low LMX condition, participants gave significantly higher promotion focus ratings, compared to participants in the high LMX condition (M = 4.01 vs. 3.63, SD = .87 and .93, respectively). Simple effect tests showed that the effect of leadership regulatory focus on the 2-item leadership promotion measure was significant in the low LMX, F(1, 50)

= 23.38, p < .001, η

= .32, and the high LMX condition, F(1, 50) = 19.89, p < .001,

η

= .29.

In sum, manipulation checks showed that the manipulations were successful, but that LMX and leadership regulatory focus could not be manipulated independently from each other. Simple effect tests showed that the relevant effects were present within both conditions of the other variable.

Creativity. Two-way ANOVAs were performed with as dependent variables the total and relative number of pasta names not ending on the letter ‘i’ (originality), total and relative number of pasta names ending on the letter ‘i’ (conformity), and total number of pasta names (fluency). Results (see table 2) revealed there were no significant interactions for any of the dependent variables, but there were some significant main effects.

--- Insert table 2 about here ---

Concerning originality in absolute terms, results only showed a main effect of leadership regulatory focus. Leadership promotion focus resulted in participants being more original than leadership prevention focus, F(1, 100) = 5.54, p < .05, η

= .05.

Participants in the leadership promotion condition generated more pasta names not ending on the letter ‘i’ (M = 6.65, SD = 6.44), compared to participants in the leadership prevention condition (M = 3.92, SD = 5.26).

For conformity in absolute terms, results only indicated the expected main effect of LMX. For the total number of pasta names ending on the letter ‘i’, there was a marginally significant main effect of LMX, F(1, 100) = 2.83, p < .10, η

= .03. As predicted, high LMX resulted in higher conformity (M = 9.35, SD = 6.32) than low LMX (M = 7.54, SD = 4.38). Thus, participants experiencing a high LMX- relationship are more likely to conform to their supervisor and thereby show noncreativity, than participants who experience a low LMX-relationship with their supervisor.

Regarding conformity in relative terms there was a significant main effect of

leadership regulatory focus, F(1, 100) = 5.29, p < .05, η

= .05. As expected, a

leadership prevention focus resulted in more conformity (M = .70, SD = .27)

compared to a leadership promotion focus (M

= .58, SD

= .26). Since originality in

relative terms is the opposite of conformity in relative terms, there was also a

significant main effect of leadership regulatory focus on originality, in which

participants in the promotion condition were more original than participants in the

prevention condition. Taken together, above results concerning subordinates’

originality and conformity, in absolute and relative terms, showed that hypothesis 1 and 2 were not supported.

Finally, for fluency, results revealed there was no significant main effect of LMX. While results showed that in the high LMX condition, participants generated more new names for pasta (M = 14.23, SD = 6.77) than participants in the low LMX condition (M = 13.23, SD = 8.43), these numbers were not significantly different from each other. Thus, high LMX does not result in higher subordinate fluency than low LMX, and therefore hypothesis 3 was rejected.

Process variables. To test whether the manipulations induced different moods or task engagement, the PANAS and task engagement measure were included as dependent variables in two-way ANOVAs. Results (see table 2) indicated two marginally significant interactions and a significant main effect.

For participants’ mood after the manipulations and creativity task, results revealed a significant main effect of LMX on positive affect (F(1, 100) = 5.60, p <

.05, η

= .05) and a marginally significant interaction effect for positive affect (F(1, 100) = 3.06, p < .10, η

= .03), whereas for negative affect no significant effects were found. For the main effect of LMX, results showed that the high LMX condition resulted in participants experiencing higher positive affect (M = 3.56, SD = .64), compared to the low LMX condition (M = 3.27, SD = .61). Concerning the marginally significant interaction, results indicated that participants in the high LMX-promotion condition experienced higher positive affect (M = 3.68, SD = .68) than participants in the low LMX-promotion condition (M = 3.18, SD = .64). The high LMX-prevention condition (M = 3.44, SD = .59) and low LMX-prevention condition (M = 3.37, SD = .58) were not significantly different from each other or from the other conditions.

For participants’ task engagement, a marginally significant interaction was found, F(1, 100) = 3.74, p < .10, η

= .05. Results showed that participants in the high LMX-promotion condition were more task engaged (M = 4.70, SD = 1.14) than the other three conditions: high LMX-prevention: M = 4.01, SD = 1.07; low LMX- promotion: M = 3.97, SD = .97; low LMX-prevention: M = 4.03, SD = .82. These three conditions did not significantly differ from each other on task engagement.

Unfortunately, the process variables (i.e. mood and task engagement) could

not explain the relationship between LMX, leadership regulatory focus and the facets

of creativity, because of weak correlations between these process variables and the creativity measures (i.e. correlations ranging between -.08 and .18, NS).

Discussion

Taken together, the experimental study could not support our hypotheses.

First, results regarding subordinates’ originality (hypothesis 1) showed that participants were more original when they perceived their leader’ behaviour to be promotion focused, compared to prevention focused. This result is in line with previous literature and empirical evidence (e.g. Crowe & Higgins, 1997; Friedman &

Förster, 2001; Liberman et al., 1999; Wu et al., 2008). However, LMX was not positively related to subordinate originality and also the anticipated interaction was not found. It seemed that, contrary to existing research (e.g. Tierney et al., 1999), a high LMX relationship could not facilitate a comforting work environment in which free thinking is stimulated and in which subordinates can come up with novel and unconventional ideas and solutions to problems.

Second, our results for hypothesis 2 showed that for conformity in absolute terms there was a significant main effect of LMX, indicating there was higher conformity when participants experienced a high LMX relationship. This result was expected, given that in a high LMX relationship, subordinates are more likely to give in to the supervisor to protect the valuable relationship and to give the supervisor expected relational outcomes, thereby receiving favourable outcomes in return. In relative terms there was a significant main effect of leadership regulatory focus, indicating there was higher conformity when subordinates experience a prevention focus. Indeed, previous research showed that subordinates’ prevention focus results in subordinates being more conservative and vigilant (Crowe & Higgins, 1997; Förster et al., 2004; Liberman et al., 1999) and more likely to conform in their quest for avoiding failures (Wu et al., 2008). Nevertheless, also for subordinate conformity, the expected interaction between LMX and leadership regulatory focus was not found.

Finally, concerning hypothesis 3, our results showed that LMX was not positively related to subordinate fluency, although the means were in the expected direction. This was a somewhat surprising result, given that subordinates in a high LMX relationship work in an environment in which they get the exchange currencies necessary to work hard and to come up with many new ideas (Basu & Green, 1997;

Graen & Scandura, 1987; Liden & Maslyn, 1998; Tierney et al., 1999).

STUDY 2

Study 2 is a field study including leader-member dyads in a variety of organizations and industries in the Netherlands. Since the experiment (study 1) did not yield the predicted interactions using manipulations of LMX and leadership regulatory focus, we took a different approach for this field study. Instead of examining leadership promotion focus versus leadership prevention focus as in the experiment, we looked at promotion (low, high) and prevention (low, high) focus separately. Furthermore, we did not differentiate between the facets of creativity and we used expert ratings of overall subordinate creativity provided by subordinates’

direct supervisors (e.g. see Van Dyne, Jehn & Cummings, 2002)

Hypotheses

When subordinates experience a leadership promotion focus to a high extent (as opposed to low extent) in a high LMX relationship, their eagerness is increased more, they are more willing to take risks and to do things differently than their supervisor and they will thereby use the exchange currencies to greater extent to obtain creative output (Crowe & Higgins, 1997; Dienesch & Liden, 1986; Förster et al., 1998, 2004; Tierney et al., 1999; Wu et al., 2008). Therefore, the fourth hypothesis of this study is:

Hypothesis 4: LMX will be positively related to subordinate creativity when subordinates experience a leadership promotion focus, such that this effect will be stronger when subordinates perceive their supervisor’ behaviour highly promotion focused compared to low promotion focused.

Given that experimental results showed that high LMX and a perceived

leadership prevention focus can result in subordinate conformity (thereby signalling

noncreativity), we secondly argue that when subordinates experience a prevention

focus to high extent (as opposed to low extent) in a high LMX relationship, their

vigilance is increased more, they are less likely to adopt a diverging view than their

supervisor and they use the exchange currencies to avoid failures and losses, thereby

restricting themselves in their ability to be creative (Chirumbolo et al., 2004;

Dienesch & Liden, 1986; Crowe & Higgins, 1997; Förster et al., 1998, 2004; Nemeth

& Ormiston, 2007). Therefore, the fifth and final hypothesis of this study is:

Hypothesis 5: LMX will be negatively related to subordinate creativity when subordinates experience a leadership prevention focus, such that this effect will be stronger when subordinates perceive their supervisor’ behaviour highly prevention focused compared to low prevention focused.

Method

Participants. For this study, 10 companies from different industries (e.g.

service, banking, production, construction and education) were approached. Per company, the highest senior manager was contacted who directed several subordinates. In every company, the senior manager was the direct supervisor of the subordinates. For distributing the paper-and-pencil questionnaires, we made appointments with the senior managers in which we explained the purpose of the current study and in which we handed out the supervisor questionnaire (in which the supervisor had to rate their subordinates on creativity) and the questionnaires for their subordinates. The senior managers distributed the questionnaires among their subordinates, and the subordinates returned the completed questionnaires in a sealed envelop back to their manager. We collected all questionnaires from the managers. In total, we received 61 completed questionnaires (out of 73; 83.6% response rate) from the subordinates, and all 10 contacted senior managers returned the questionnaire.

Thus, the present sample consists of 61 leader-member dyads, and per senior manager there were on average six subordinates. Of the senior managers, 90% were men, the average age was 48.30 years (SD = 11.34) and most managers (7) held a University degree. Average total tenure was 26.40 years (SD = 9.94), whereas tenure in the current organization was 13.20 years (SD = 10.78). Of the subordinates, 60% were men, the average age was 38.78 (SD = 12.98) and most employees (30) held a college degree. Average total tenure was 17.53 (SD = 11.61), whereas current organizational tenure was 8.77 (SD = 8.94).

Measures. All answers were provided on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree), except where otherwise noted.

We assessed perceived leader-member exchange using the 11-item scale of

Liden and Maslyn (1998). Following Liden and Maslyn’s (1998) study in which they

provide support for a higher-order factor, we used a composite of all items to measure subordinates’ perceived LMX. Example items are: ‘My supervisor is the kind of person one would like to have as a friend’ and ‘I respect my supervisor’s knowledge of and competence on the job’. Chronbach’s alpha was .83.

For measuring perceived leadership regulatory focus, an adapted version of the 12-item scale of Wallace et al. (2009) was used. This scale consists of 6 promotion and 6 prevention items. Exploratory factor analysis with varimax rotation showed, as in the experimental study, that a two-factor structure was optimal.

Perceived leadership regulatory focus was assessed by statements beginning with ‘My supervisor wants me to focus on…’. Example items are ‘My supervisor wants me to focus on my work responsibilities’ (prevention) and ‘My supervisor wants me to focus on my work accomplishments’ (promotion). Reliability analysis showed that the 6-item perceived leadership promotion focus scale had a Chronbach’s alpha of .75, whereas this was .80 for the perceived leadership prevention focus scale.

Given the absence of archival data in the companies, creativity was measured using supervisor ratings, as has been done in previous research (e.g. George & Zhou, 2001; Oldham & Cummings, 1996; Tierney et al., 1999; Tierney & Farmer, 2002;

Van Dyne et al., 2002). This rating consisted of 4 items, adapted from Ettlie and O’Keefe (1982; see also Tierney et al., 1999). Example items were ‘This employee shows originality in his/her work’ and ‘This employee tries out new ideas and approaches to problems’. Chronbach’s alpha was .91.

Because in a high LMX relationship, leadership ratings of employee creativity may be positively biased, we also measured supervisors’ affect towards the subordinates to be able to control for this bias (see also Landy & Farr, 1980). We used an adapted version of the 3-item affect scale of Liden and Maslyn (1998) for this purpose. Chronbach’s alpha was .82 and an example item is ‘This employee is a lot of fun to work with’.

Also for the field study, we assessed two process variables (i.e. conformity and task engagement), to examine possible mediation effects. Subordinates’

conformity to their supervisor was assessed using a modified version of the 11-item

scale of Mehrabian (2005), measured on a 9-point scale (-4 = very strong

disagreement, +4 = very strong agreement). A sample item is: ‘I often rely on, and act

upon, the advice of my supervisor’. Total conformity scores were calculated by

subtracting the sum of negatively worded items from the sum of positively worded

items (α = .72). Higher scores indicated a greater tendency to conform. Previous work (e.g. Mehrabian & Stefl, 1995) has shown that this conformity scale has good test-retest reliability and construct validity.

Given the marginally significant interaction effect found in the experimental study for task engagement, we also included this process measure in our field study.

This measure was again taken from the study of Schaufeli et al. (2006) and consisted of 9 items. The scale showed good internal consistency (α = .89) and a sample item is: ‘I am immersed in my work’.

Since we expect subordinates who are more interdependent in general to conform more to their supervisor, we controlled for subordinates’ self-construal in(ter)dependency by using the 11-item interdependent self construal scale of Cross, Bacon and Morris (2000). Respondents scoring high on this scale are more interdependent. An example item is: ‘My close relationships are an important reflection of who I am’. Internal consistency was sufficient (α = .75).

Analytic Strategy

Given the multi-level nature of the data, we applied multi-level analyses with the software MLwiN 2.22 to test our hypotheses (Rasbash et al., 2010). We first ran null models with no predictors and creativity, task engagement or conformity as dependent variable. Results showed an intraclass correlation (ICC) of .03 for creativity, an ICC of .02 for conformity and an ICC of .02 for task engagement, indicating little variance at the organization-level. This justifies the use of ordinary least squares (OLS) regression as the appropriate analytic technique.

Results

Means, standard deviations, intercorrelations and reliabilities for the study

variables are displayed in table 3. To test our hypotheses, we conducted a hierarchical

regression analysis in which creativity was predicted by the main effect terms (model

2), the interaction terms (model 3) and the process variables (i.e. task engagement and

conformity; model 4), while controlling for gender, self-construal and affect (model

1). We also ran models with task engagement and conformity as dependent variables,

to check for possible mediation effects. Following Aiken and West (1991), we

centered perceived leadership promotion focus, perceived leadership prevention focus and LMX, and we based the interaction terms on these centered scores.

--- Insert table 3 about here ---

Table 4 shows the regression results. Results concerning subordinates’

creativity showed that LMX was not significantly related to creativity (β = .04, NS), nor were perceived leadership promotion (β = -.06, NS) or prevention focus (β = -.19, NS). Furthermore, the hypothesized interactions between LMX and perceived leadership promotion focus (β = -.01, NS) and between LMX and perceived leadership prevention focus (β = -.03, NS) were not found (see table 4, dependent variable creativity, model 3). Also the process variables, task engagement (β = .17) and conformity (β = -.25), were not significantly related to subordinates’ creativity (see table 4, dependent variable creativity, model 4). There was, however, a strong positive relationship between supervisors’ affect with their subordinates and the creativity scores the supervisors gave to their subordinates (β = .42, p < .01), but including affect as control variable did not change the main or interaction effect results on subordinates’ creativity.

--- Insert table 4 about here ---

For task engagement as dependent variable, results showed that between LMX and task engagement there was a highly significant positive relationship (β = .59, p <

.01), which indicated that when subordinates experienced a high quality relationship with their supervisor they were more task engaged. Perceived leadership promotion (β

= .12, NS) and prevention focus (β = -.00, NS), on the other hand, were not

significantly related to subordinates’ task engagement. Also here, no interaction

effects were found: neither promotion focus could moderate the relationship between

LMX and task engagement (β = -.13, NS), nor prevention focus (β = .20, NS; see

table 4, dependent variable task engagement, model 3). In model 4, gender had a

marginally significant effect on task engagement (β = .23, p < .10), indicating that

women were more task engaged than men.

Finally, results concerning conformity as dependent variable showed that LMX (β = .05, NS) and perceived leadership promotion focus (β = -.16) were not significantly related to subordinates’ conformity to their supervisor, but perceived leadership prevention focus was (β = .38, p < .01; see table 4, dependent variable conformity, model 3). Thus, the more subordinates experienced their supervisor’

behaviour to be prevention focused, the more subordinates conformed to their supervisor. Again, no significant interaction effects were found between LMX and perceived leadership promotion focus (β = .13, NS) and between LMX and perceived leadership prevention focus (β = -.22, NS; see table 4, dependent variable conformity, model 3). In the final model, both gender (β = .28, p < .05) and self-construal (β = .33, p < .05) were significantly related to subordinates’ conformity. Hence, female subordinates and subordinates whose personality was characterized by an interdependent self-construal were more likely to conform to their supervisor.

The process variables (i.e. task engagement and conformity), however, could not mediate the relationship between LMX, leadership promotion and prevention focus and creativity, because LMX, leadership promotion and prevention focus and the process variables were not significantly related to subordinates’ creativity.

Discussion

Altogether, our hypotheses could not be supported. For both hypothesized interactions, neither the main effects nor the interactions terms were significant.

Results showed that a perceived leadership promotion focus in a high LMX relationship did not lead subordinates to pursue creative output, whereas subordinates perceiving their leader’ behaviour to be prevention focused did not show less creativity.

We did find, however, that LMX was strongly positively related to subordinates’ task engagement, whereas perceived leadership prevention focus and interdependent self-construal were strongly positively related to subordinates’

conformity to their supervisor. However, the process variables (i.e. task engagement

and conformity) could not explain the pathway(s) between LMX and subordinate

creativity. Hence, there remains a research gap in this relationship and the specific

processes by which LMX facilitates or hampers subordinate creativity are still to be

understood.

DISCUSSION AND CONCLUSIONS

The present article adopted a theoretical framework in which the relationship between LMX, leadership regulatory focus and subordinate creativity was examined.

Unlike other authors (e.g. Atwater & Carmeli, 2009; Tierney et al., 1999), we argued that LMX is not necessarily positively related to subordinate creativity, but that this depends on the specific regulatory focus adopted by the supervisor in directing subordinates. Taken together, the two studies described here do not provide strong evidence for this main assumption of our paper. We examined this relationship in one experimental and one field study, in which we respectively manipulated and measured LMX and leadership regulatory focus. Moreover, we measured the different facets of creativity (i.e. originality, conformity and fluency) in the experimental study and we measured overall subordinate creativity by means of supervisory ratings in the field study. In the following paragraphs, we discuss in detail our findings and their implications.

Summary of Results

Starting with the relationship between LMX and (facets of) creativity, results from the experimental study indicated that LMX resulted in more conformity in absolute terms, where conformity was measured as the number of ideas generated by participants that converge to the ideas of the supervisor (i.e. pasta names ending on the letter ‘i’), and thereby indicating noncreativity (see also Dijksterhuis & Meurs, 2006). As reasoned in the present paper, social pressure stemming from the supervisor in a high LMX relationship, whether manifested in supervisors wanting desired relational outcomes from the subordinates or subordinates wanting to keep the relationship of high quality to receive favourable outcomes, can result in subordinates conforming to their supervisor. The experimental results also showed that LMX did not result in higher fluency (i.e. the total number of ideas generated; Dijksterhuis &

Meurs, 2006). This was a surprising result since subordinates in high LMX relationships often report that they receive the physical (e.g. time, space, materials) and psychological (e.g. trust, autonomy) resources necessary to work and perform above and beyond the call of duty (Basu & Green, 1997; Graen & Scandura, 1987;

Liden & Maslyn, 1998; Tierney et al., 1999). The field study results showed that

LMX was not significantly related to overall subordinate creativity. While we did not

explicitly hypothesize this main effect, it was somewhat expected given that previous research (e.g. Tierney et al., 1999) found a clear positive relationship between LMX and subordinate creativity, when subordinates’ creativity was rated by the direct supervisor.

Second, concerning the relationship between leadership regulatory focus and subordinates’ creativity, the results of the experimental study indicated that a perceived leadership promotion focus resulted in higher absolute and relative originality (i.e. the number of novel ideas; Rietzchel et al., 2007), whereas a prevention focus resulted, relatively, in more conformity. These results were expected and in line with previous research (e.g. Förster, Friedman & Liberman, 2004;

Friedman & Förster, 2001; Higgins, 1997; Wu et al., 2008). A promotion focus, a promotion focus cue or a perceived leadership promotion focus may enhance creative thoughts and originality by raising eagerness, whereas a prevention focus, a prevention focus cue or a perceived leadership prevention focus heightens conformity by elevating vigilance and can thereby hamper creative output. The results of the field study showed, however, that neither a perceived leadership promotion focus nor a perceived leadership prevention focus had a significant relationship with subordinate creativity. Hence, the results from the field study could not support the results found in the experimental study.

Third, for the interactions in the experimental and field study, the predicted effects on creativity were not found. While we, and previous theory for that matter (e.g. Atwater & Carmeli, 2009; Brockner & Higgins, 2001; Kark & van Dijk, 2007), argued that leaders’ behaviour is important for shaping subordinates’ motivations and behaviours, no stronger effects of leadership regulatory focus were found in a high LMX relationship, compared to a low LMX relationship. Neither a perceived leadership promotion, nor a perceived leadership prevention focus in a high LMX situation could alter the originality (experiment), the conformity (experiment) or the overall creativity (field study) of subordinates.

Finally, the field study showed two notable main effect results concerning the

process variables (i.e. task engagement and conformity). First, results showed that

LMX had a strong positive effect on task engagement, which indicated that the more

subordinates perceived their relationship with their supervisor to be of high quality,

the more they were engaged in their tasks and work. This effect complemented the

marginally significant interaction effect found in the experimental study, in which

task engagement was the highest under a high LMX-promotion condition. Thus, together, these results suggest that LMX may have an important direct effect on task engagement, alone or in combination with perceived regulatory focus. The second notable process result from the field study is the positive relationship between perceived leadership prevention focus and conformity, which reproduced the results found in the experiment. The more subordinates perceived their supervisor to be prevention focused, the more they conformed to him/her. However, both process variables (i.e. task engagement and conformity) had no significant effect on creativity.

Theoretical Implications

Overall, the results of the present paper suggest, in contrary to previous empirical results (e.g. see Atwater & Carmeli, 2009; Scott & Bruce, 1994; Tierney et al., 1999), that LMX is not necessarily positively related to (facets of) subordinate creativity. While we should be careful in interpreting null-effects, it is striking that neither the lab nor the field study showed that LMX was related to the creative output of subordinates. Not for originality, fluency or supervisory rated creativity, there was a significant effect of LMX. It therefore seems that there is more to the concept of LMX than is currently being argued in research, which can potentially alter the outcomes of LMX for subordinate creativity. It appears that, besides a high quality relationship with a supervisor, more is needed to become creative as subordinate.

While we examined leadership regulatory focus as potential activator or deactivator of subordinate creativity in a high LMX relationship, it did not work as planned. Being more eager by perceiving a leadership promotion focus, or being more vigilant by perceiving a leadership prevention focus did not result subordinates in high LMX relationships to become more or less creative. Thus, while theoretically it seemed feasible to include leadership regulatory focus, empirical results proved differently.

With the present findings, we found some evidence for certain processes by which the leadership variables were related to subordinate creativity, and we thereby add new elements to the leadership and creativity literature. First, we showed that a high LMX relationship can elicit subordinate conformity. Thus, while previous research focused mainly on the creativity of subordinates (e.g. Basu & Green, Scott &

Bruce, 1994; Tierney et al., 1999), subordinates’ conformity to their supervisor might

be an equally important outcome of a high quality exchange relationship. This is a