Abstract
Emotion researchers have found that negative mood may either enhance or inhibit employee creativity. Little is known about this conundrum, however, and in particular when and why each effect occurs. To address this concern, we formulate and test hypotheses about likely moderators of the relationship between negative mood and creative process engagement. Results from an experience sampling study with 556 real-time reports from 68 employees support our hypothesis that negative mood is most strongly and positively related to concurrent creative process engagement among employees who (a) have high trait learning goal orientation and (b) perceive that they are empowered. Our hypotheses and findings help to resolve the ongoing controversy surrounding the nature of the negative mood–creativity nexus.
Imagine you are upset by heavy traffic on the way to work, anxious about an upcoming deadline, frustrated by lack of progress on a current task, and angry to find that the photocopier is broken – again! Are you likely to have a creative day? Conventional wisdom might say no, but given the right combination of personal traits and workplace perceptions, a bit of angst might help to kick-start the creative process. In this research, we seek to find out when this might be the case.
Elfenbein (2008: 325) pointed out that ‘[t]here is a temptation in the management literature to argue for . . . the inherent badness of negative emotion.’ Indeed, most management research has assumed that negative feelings produce negative consequences in organizations. Casting doubt on this assumption, however, a special issue of Human Relations (Lindebaum and Jordan, 2014) is devoted to ‘when it can be good to feel bad,’ or the adaptive functions of negative affect at work. More specifically, De Dreu et al. (2008), George and Zhou (2002, 2007) and To et al. (2012), all provide evidence that negative mood can sometimes facilitate creative behaviour.
Although we know that individuals do not always perform at their creative best or choose to engage in behaviours likely to lead to creativity, our understanding of variability in individual creativity is nonetheless limited (Amabile et al., 2005; George, 2008). State mood has long been regarded as a likely predictor of concurrent creativity (for reviews, see Baas et al., 2008; Davis, 2009). While evidence from meta-analyses supports a consistent role of state positive mood as a facilitator of concurrent creativity, negative mood effects are less uniform (Baas et al., 2008; Davis, 2009). In some cases, negative mood seems to inhibit creative behaviour, while in others it facilitates creative behaviour. In the study we report in this article, we answer calls to investigate boundary conditions as to when and why negative mood is associated with creative behaviour by employees (Baas et al., 2008; Davis, 2009; George and Zhou, 2007).
Why might negative mood enhance creative behaviour? Negative feelings can serve as important feedback indicating insufficiency in one’s goal progress, triggering regulatory effort to reduce the discrepancy between the actual state and the desired end state (Carver and Scheier, 1998; Martin and Stoner, 1996). For some individuals, negative mood may act on the one hand as a problem signal that prompts prolonged task-focused effort toward generating new and potentially useful solutions to problems (De Dreu et al., 2008; George and Zhou, 2002, 2007). On the other hand, negative moods can also be seen to be discouraging, intrusive and mentally taxing, thereby interfering with creativity and task performance (Beal et al., 2005). Which of these potential contradictory effects occurs may depend on how the message conveyed by negative mood is interpreted relative to an individual’s goal and his or her perceptions of the extent to which the situation facilitates goal attainment (Martin and Stoner, 1996; Martin et al., 1993; To et al., 2010, 2012). Past mood–creativity research has tended only to focus on the moderating effect of single motivational factors in isolation, however. In this study, we consider how dispositional goal orientation and perceived facilitating situation in the form of psychological empowerment (as between-persons variables) may jointly condition the relationship between negative mood and creative behaviour at the within-person level.
The need for creative effort at work often arises in the course of goal-directed behaviour and, as such, trait goal orientation may prove useful in understanding how these individual self-regulatory propensities may modify responses to negative moods in the form of creative behaviour (Hirst et al., 2011; Janssen and Van Yperen, 2004; To et al., 2012). In this research, we focus specifically on trait learning goal orientation (LGO), which represents a self-regulatory propensity to pursue task mastery and the development of competence by overcoming challenges (DeShon and Gillespie, 2005; Dweck and Leggett, 1988). In the achievement goal literature, LGO is recognized as a distinct form of approach orientation that tends to promote creativity (Gong et al., 2009; Hirst et al. 2011; Janssen and Van Yperen, 2004). In particular, individuals with a strong LGO are more willing to undertake challenging tasks and are more resilient in the face of setbacks and challenges. This may make them more effective in translating the problem signal of negative mood into sustained creative striving (To et al., 2010, 2012).
Reviews of the goal orientation literature by DeShon and Gillespie (2005) and Payne et al. (2007) have pointed to a need to examine how goal orientation effects on behaviour vary as a function of situation. Trait activation theory suggests that latent traits such as dispositional goal orientation are activated when trait-relevant cues are present (Hirst et al., 2011; Tett and Burnett, 2003). In line with this view, a body of laboratory research (see Higgins, 2005 for a review) suggests that task engagement is enhanced when people perceive that the situation permits them to pursue their goals in a manner compatible with their regulatory orientation. Based on this perspective, Seibert et al. (2011: 998) call for ‘greater integration of psychological empowerment theory and theories of motivation based on a self-regulation framework’ to better explain employee work engagement and motivation.
Combining these ideas, we propose that high LGO employees may be more likely to express their trait mastery inclination through creative efforts when they feel empowered to do so by a facilitating work context. We adopt Spreitzer’s (1995) conceptualization of psychological empowerment at work, which includes self-referent beliefs about meaning, competence, self-determination and impact in the workplace. Belief in empowerment and high LGO enable a proactive and creative response to transient negative moods that signal something is wrong and that effort and persistence are required to solve the problem. In short, we focus on understanding how person-level LGO interacts with perceptions of psychological empowerment at work to condition the strength of the within-person relationship between negative mood and creative behaviour.
Our work is intended to advance understanding of the relationship between negative mood and creative behaviour in two ways. First, prior research on the relationship between affect and affect-driven behaviour has tended to focus on the moderating role of single traits or situations in isolation, whereas we hypothesize and examine the combined effect of trait LGO and psychological empowerment as a perceived trait-activating condition. As such, our work has the potential to contribute to the emerging body of research on the impact of combinations of motivation-related variables on creative behaviour (George and Zhou, 2001; Hirst et al., 2011; Raja and Johns, 2010). Second, we provide further evidence about the boundary conditions under which negative mood may boost concurrent creative behaviour (Binnewies and Wörnlein, 2011; To et al., 2012). Our hypotheses and supportive findings should serve to clarify the inconsistent negative mood–creativity relationships documented in the literature and produce suggestions about how and by whom negative mood may be harnessed to enhance creative activity.
Creative process engagement
Creativity scholars such as Amabile (1998), Mumford (2000) and Shalley et al. (2004) have emphasized the value of understanding the creative process as well as acts that contribute to the eventual development of novel and useful ideas. Individuals may not be able to ‘be creative’ on demand volitionally, but they can choose to engage in more or less of the behaviours known to lead to eventual creative outcomes. Consistent with this view, the within-person dependent variable we adopted in this research is creative process engagement (CPE). Based on theoretical work by Amabile (1983) and Reiter-Palmon and Illies (2004), Zhang and Bartol (2010a: 108) conceptualized CPE as ‘the process by which creativity occurs.’ They assessed it in terms of three kinds of activities undertaken during the creative idea production phase of creative problem solving: (1) problem identification, (2) information search and encoding and (3) alternative/idea generation. These behaviours supply the information, ideas and perspectives that eventually yield creative outcomes (Nijstad et al., 2010). The utility of conceptualizing CPE as an antecedent of creative outcomes is further supported by Zhang and Bartol (2010a, 2010b), who found that self-rated CPE is positively related to supervisor ratings of creative performance. Past research has also shown that CPE varies meaningfully within-person over short periods of time, making it a suitable dependent variable for research on mood (which also fluctuates within-person throughout the workday, see To et al., 2012).
Our research builds upon recent evidence reported by Binnewies and Wörnlein (2011) and Ohly and Fritz (2010) showing that variance in creativity is attributable to sources at two levels of analysis: (1) within-person (temporal) and (2) between-person. Consistent with this, we adopt a multilevel approach utilizing experience sampling methodology (ESM; Mehl and Conner, 2011) to investigate how between-persons variables (trait LGO and psychological empowerment at work at Level 2) moderate within-person relationships connecting negative mood and concurrent CPE over time (Level 1).
Controversy concerning the negative mood–creativity relationship
A key motivation for our study is to contribute to resolving an on-going controversy in mood–creativity research. Scholars have espoused differing views regarding the role of negative mood in creativity. On the one hand, negative mood may be seen to hinder creativity. In this regard, Isen (1999; Isen et al., 1987) argues that negative moods reduce cognitive flexibility. This is because, unlike positive moods, negative moods do not prime access to extensive and diverse stored memory resources. Baas et al. (2008) found further that activating negative moods such as anxiety tend to reduce cognitive flexibility, but not the other creative components of originality and fluency (number of unique, non-redundant ideas generated). Beal et al. (2005) argue, moreover, that negative feelings such as frustration and anger produce cognitive demands that consume attentional resources and thus interfere with ongoing performance. In addition, there is evidence that people in negative moods recall more negative mood-congruent information and make less optimistic judgements. This creates a defensive rather than a creative frame of mind (Seo et al., 2004). Similarly, threat rigidity theory (Staw et al., 1981) suggests that individuals experiencing aversive states are likely to enact well-learned or habitual routines rather than to generate new approaches to problems.
On the other hand, some researchers suggest that activating negative mood may facilitate behaviour that leads to creative outcomes. A theoretical anchor for this view can be found in the dual pathway model proposed by De Dreu et al. (2008). According to these authors, creativity may be achieved via two routes: (1) the enhanced cognitive flexibility engendered by positive moods, or (2) the increased persistence triggered by negative moods. The persistence pathway leads to creativity through hard work, longer time on task, systematic and effortful exploration of possibilities, and in-depth and focused investigation of perspectives and potential solutions (Nijstad et al., 2010).
Importantly, De Dreu et al. (2008) and Nijstad et al. (2010) also argue that moods must be high arousal or activating if they are to provide the cognitive energy beneficial for creativity by either the flexibility or the persistence route. This is because the arousal accompanying activating moods enhances working memory capacity and therefore an individual’s ability to hold information transiently in mind in the service of comprehension, thinking and planning (Baas et al., 2008). Dual pathway predictions that both positive and negative activating moods foster creativity have been supported in laboratory experiments by De Dreu et al. (2008) and Nijstad et al. (2010).
Further support for the argument that negative moods can enhance creativity comes from cognitive tuning theory (Schwarz and Clore, 2003). In this regard, George and Zhou (2007) make the case that negative moods do more than merely inform people that their environment is problematic. Negative moods also serve to ‘tune’ mental processes to a detail-oriented, analytic approach to understanding the nature of the problem. This, in turn, reduces reliance on pre-existing assumptions and motivates greater efforts toward improvement (Schwarz and Clore, 2003). Using a between-persons design, George and Zhou (2002, 2007) found that activating negative mood was positively related to creativity when facilitating environmental conditions such as supervisory support and reward for creativity were present. More recently, and again consistent with cognitive tuning ideas, To et al. (2012) found that momentary high arousal negative mood was positively related to concurrent CPE within-person among individuals working on a demanding long-term project. This finding is consistent with laboratory evidence that negative moods are beneficial for creativity on tasks that are perceived as important and performance-oriented rather than those framed as being undertaken for fun and enjoyment (Baas et al., 2008; Friedman et al., 2007).
In sum, the evidence indicates that activating negative mood may either facilitate or inhibit creativity. Facilitation is likely to occur when the cognitive and motivational resources provided by an activating negative mood are directed toward increased persistence and prolonged, effortful and analytical probing of alternatives (Baas et al., 2008; De Dreu et al., 2008). We next discuss two contingent factors that might enable the resources derived from negative moods to be directed toward creative endeavours.
Learning goal orientation and negative mood
Prior mood–creativity research from a mood-as-input perspective (Martin and Stoner, 1996; Martin et al., 1993; To et al., 2012) has shown that mood effects on creativity are not universal but are contingent upon the goal of the individuals experiencing the mood states. For example, laboratory evidence reported in Martin and Stoner (1996) and Martin et al. (1993) indicates that the effects of a negative mood can be positive or negative, subject to the goal or objective sought by the individual. This view is in line with self-regulation research by Carver and Scheier (1998), who propose that negative feelings serve as feedback about discrepancies between current and desired states. It seems that negative mood reflects a perceived insufficiency of goal-related progress, triggering a behavioural response that may involve approach and engagement to reduce the discrepancy, or alternatively withdrawal or goal abandonment. Individuals with different goal orientations are thus likely to make different choices in response to the signal provided by a negative mood.
We focus on one approach-goal orientation, trait LGO, 1 which represents the self-regulatory propensity to pursue task mastery and experience intrinsic motivation (DeShon and Gillespie, 2005; Dweck and Leggett, 1988). LGO has also been found to be associated with the use of deep processing strategies, leading to effective performance on complex and unfamiliar tasks (Janssen and Van Yperen, 2004; Steele-Johnson et al., 2000). Thus, high LGO may be beneficial in coping with the ambiguity, challenges and obstacles inherent in complex creative endeavours (Gong et al., 2009; Janssen and Van Yperen, 2004).
Dweck and Leggett (1988) point out further that high LGO individuals believe both that competence is malleable and that effort determines success. For these individuals, negative mood serves as a signal of insufficient goal progress, which in turn stimulates greater effort, but only so long as goal achievement is still seen as possible (Carver and Scheier, 1998; Forgas, 1995). It is likely that the problem signal and arousal of an activating negative mood is compatible with the regulatory motivation of high LGO individuals, who prefer to work on challenging tasks, believe in their ability to master unfamiliar situations and tend to persist in the face of setbacks (Elliot and McGregor, 2001). Thus, and as To et al. (2012) posited, activating negative mood among high LGO individuals is likely to elicit increased CPE. For low LGO individuals, on the other hand, a negative mood state indicating insufficient goal progress is likely to be interpreted as a threat rather than a challenge. For these individuals, engaging in potentially risky creative activities in the pursuit of mastery is not a natural response to negative feedback. Thus:
Hypothesis 1: Trait LGO moderates the relationship between activating negative mood and CPE such that activating negative mood is more strongly positively related to CPE when LGO is high than when LGO is low.
Psychological empowerment and negative mood
Individuals may not always believe that they have the control and capacity to respond effectively when faced with the problem signal of negative mood. We propose that psychological empowerment can facilitate creative task persistence resulting from an activating negative mood. Empowerment concerns an employee’s power to take initiative and effective action (for reviews, see Maynard et al., 2012; Seibert et al., 2011). Earlier researchers regarded empowerment as a set of structural practices designed to share power and authority throughout the organization (e.g. Bennis and Nanus, 1985; Kanter, 1983). Conger and Kanungo (1988) and Thomas and Velthouse (1990) subsequently introduced a psychological perspective on empowerment, pointing out the importance of employees’ perceptions or beliefs about their degree of empowerment. According to this view, ‘psychological empowerment is not an organizational intervention or a dispositional trait but rather a cognitive state achieved when individuals perceive that they are empowered’ in a particular job setting (Maynard et al., 2012: 1235).
Following on from this theorizing, Spreitzer (1995) developed a multidimensional construct and measure, suggesting that psychological empowerment is comprised of four specific cognitions related to the self in the work role: meaning, competence, self-determination and impact. Meaning refers to the fit between work goals and personal beliefs or values (Spreitzer, 1995). Competence refers to the match between ability/skills and task demands and subsequent self-efficacy (Cable and DeRue, 2002; Spreitzer, 1995). Self-determination is an individual’s sense of having autonomy and independence in doing tasks; and impact refers to perceptions that important outcomes are contingent upon the individual’s own behaviour or effort (Spreitzer, 1995). Psychological empowerment also sets the stage for intrinsic motivation (Seibert et al., 2011). In a recent meta-analysis, Seibert et al. (2011) confirmed that this four-component conceptualization of overall psychological empowerment is related to favourable work attitudes and outcomes, including notably to innovation.
A positive relationship between empowerment and CPE has also been reported by Zhang and Bartol (2010a) at the between-persons level. They reason that this is because empowerment enables employees to ‘focus on an idea or a problem longer and more persistently’ (2010a: 112). Empowerment as a sense of control and efficacy may thus enable employees to take greater advantage of their activating negative mood through persistence (Binnewies and Wörnlein, 2011). In addition to this evidence, Baas et al. (2008) found meta-analytic evidence that negative mood fosters creativity when the task is perceived as important, which is consistent with the tendency of individuals with high psychological empowerment to perceive their work as meaningful and important (Spreitzer, 1995). Overall, it seems that a negative mood state, which functions to draw attention to a problematic situation, may engender more creative striving among those who generally experience high rather than low empowerment at work. Thus:
Hypothesis 2: Psychological empowerment moderates the relationship between activating negative mood and CPE such that activating negative mood is more strongly positively related to CPE when empowerment is high than when empowerment is low.
Combining learning goal orientation, empowerment and negative mood
Our final hypothesis specifies the role of LGO and psychological empowerment in jointly conditioning negative mood–CPE relationships. In this regard, trait activation theory (Tett and Burnett, 2003) suggests that traits such as LGO are latent potentials residing in the person, and are expressed in behaviour when trait-relevant cues are present. According to this theory, perceptions of the environment can lead individuals to behave in trait-consistent ways (see also Kamdar and Van Dyne, 2007; Shoda et al., 1994). Indeed, a large body of laboratory research based on regulatory fit theory (see Higgins, 2005 for a review) supports the idea that favourable effects occur when people believe that they can pursue goals in a manner compatible with their regulatory orientation.
Based on the forgoing, we propose that psychological empowerment will allow high LGO individuals to exercise their penchant for learning, experimentation and persistence when faced with challenges and setbacks, resulting in a stronger negative mood–CPE relationship. Empowerment involves a sense of impact and autonomy, which is congruent with the regulatory preferences of high LGO individuals to pursue competency development and task mastery through increased effort (Dweck and Leggett, 1988). Further, the felt ownership, meaningfulness and intrinsic motivation associated with empowerment fit well with the mastery inclination of high LGO individuals to engage in task activities for their own sake (Spreitzer, 1995). The competency component of empowerment provides the self-efficacy to believe that success is possible through sustained effort. Thus, fuelled by an activating negative mood, high LGO individuals may most effectively stay focused, persist and try out new alternatives if they also perceive high psychological empowerment. In the case of those low on LGO, there is little mastery or creative propensity available to be activated by high empowerment. In this situation, a negative mood may represent a threat, likely failure and a stimulus to avoid risk.
Negative mood may also be less likely to prompt persistent effort toward creativity when LGO is high but empowerment is low. In this regard, De Dreu et al. (2011) found in a laboratory study that approach-oriented people’s creativity is inhibited when they experience poor fit between their natural inclinations and situational affordances. Low psychological empowerment implies limited independence, low personal control and weak ownership (Brockner et al., 2004; Ohly and Fritz, 2010), which may be a poor fit for individuals high on trait LGO. Such individuals’ mastery inclination may be inhibited (Hirst et al., 2011), and so they may be less effective in translating negative mood into CPE. In sum, we hypothesize a three-way interaction such that activating negative mood is most beneficial to CPE when both LGO and psychological empowerment are high. As such, the relative absence of either LGO or psychological empowerment may be sufficient to disrupt the persistence pathway to creativity in the face of a negative mood. Thus:
Hypothesis 3: There is a three-way interaction between activating negative mood, LGO and psychological empowerment in predicting CPE, such that activating negative mood has the strongest positive relationship with CPE when both LGO and empowerment are high.
Method
Participants and procedure
Using ESM (Fisher and To, 2012; Mehl and Conner, 2011), we collected real-time data from a sample of 68 employees from 12 R&D and marketing organizations in Australia. Participants held jobs in research (42%), administration (30.4%) and other areas such as graphic design and software development (27.6%). Their mean age was 36.6 (SD = 10.7) and mean organizational tenure was 4.5 years (SD = 5.1). Thirty-four percent of participants were male, 43 percent held a bachelor’s degree and 44 percent held a masters or higher degree.
We collected our data in two stages. In Stage 1, participants completed a one-time online survey assessing the between-persons (Level 2) variables. In Stage 2, they responded to an online questionnaire twice daily (morning and afternoon) for one week. These surveys assessed within-person (Level 1) mood and CPE. Participation in the study was voluntary, and participants received a gift voucher (AU$40) as an incentive. Five hundred and fifty-six reports were received. The mean number of responses per participant (out of a possible 10) was 8.26 (SD = 1.26).
Measures
Creative process engagement
CPE entails mental effort and activities that are not normally visible to external observers throughout the working day. In view of this, it was necessary for us to measure CPE using real-time self-reports. In order to capture within-person variation in CPE, we collected these reports twice per day using an adapted version of Zhang and Bartol’s (2010a) 11-item measure of CPE. Their CPE scale contains items related to the three areas of problem identification, information seeking and idea/alternative generation. Zhang and Bartol (2010a) found that the three subscales comprised a second order CPE factor and that a scale comprised of all the items had high reliability. They used a single summary CPE score in subsequent research (Zhang and Bartol, 2010b), and we follow this precedent. In view of its repeated administration and the fact that respondents were reporting on very recent behaviour, we followed typical ESM practice (Fisher and To, 2012) by shortening the scale to two items per sub-factor. To et al. (2012) found that this six-item CPE measure correlated .95 with the full 11-item version.
Participants rated each of the six CPE items on a five-point scale ranging from 1 (not at all) to 5 (very much) twice per day. The items were phrased with half-day time instructions (e.g. ‘This morning/ afternoon . . . I spent considerable time trying to understand the nature of a matter’ (problem identification), ‘I searched for information from multiple sources, e.g. personal memories, others’ experience, documentation, Internet, etc.’ (information-seeking), and ‘I tried to devise potential solutions that move away from established ways’ (idea generation). Alpha, computed at each signal and then averaged over all measurements periods, was .93, confirming the appropriateness of using a summary scale to represent CPE.
To validate our use of self-report CPE, we also asked participants’ peers or supervisors to rate their recent creativity using 10 items from Zhou and George’s (2001) creativity scale. The correlation between weekly average self-rated CPE and observer-rated creative performance was .43 (p < .05, n = 32 pairs). 2 Given that these results are based on different measures, different time frames and different sources, this correlation provides both evidence of the predictive utility of CPE and of the appropriateness of the self-rating methodology. These results are consistent with Zhang and Bartol (2010a, 2010b), who also report that self-reported CPE is significantly correlated with supervisor-rated creative performance.
Activating negative mood
Similar to De Dreu et al. (2008) and To et al. (2012), we used five mood adjectives (upset, anxious, ashamed, angry, frustrated; average alpha = .78) to measure activating negative mood on a five-point scale, ranging from 1 (not at all/very slightly) to 5 (very much). Mood items were phrased with half-day (morning or afternoon) time instructions.
Trait learning goal orientation
We measured LGO once in the Stage 1 survey using four items from VandeWalle’s (1997) work domain goal orientation instrument. A sample item is: ‘I am willing to select a challenging task that I can learn a lot from.’ Ratings were made on a five-point Likert scale, and alpha reliability was .74. 3
Psychological empowerment
We used Spreitzer’s (1995) 12-item scale to measure psychological empowerment (alpha = .91) once in the Stage 1 survey. Sample items for each dimension are: ‘I can decide on my own how to go about doing my work (self-determination),’ ‘I have a great deal of control over what happens in my work (impact),’ ‘The work I do is very important to me (meaning),’ and ‘I am confident about my ability to do my job (competence).’
Controls
Consistent with past mood–creativity research (e.g. Binnewies and Wörnlein, 2011; George and Zhou, 2007; To et al., 2012), we measured positive mood as a control variable. The five adjectives measuring activating positive mood at each ESM report were excited, enthusiastic, elated, inspired and proud (average alpha = .88). Finally, as Judge and Ilies (2004) recommend, a time index variable was included in the analyses to control for possible time trends in CPE. This variable was coded for day and time of each measurement.
Results
Descriptive statistics
Table 1 shows descriptive statistics and intercorrelations. The mean of negative mood was 1.52 (SD = .66) on a five-point scale. Low means on transient negative affect are common in experience sampling studies, with most people only occasionally feeling negative feelings. Negative feelings may be infrequent enough that they gain attention and signal the need for corrective action and thus play a role in facilitating creative efforts toward problem solving.
Descriptive statistics, correlations and reliabilities.
SD = standard deviation.
The correlations below the diagonal represent between-persons associations (for variables 1 through 3, we computed the between-persons correlations using individuals’ aggregated scores over time; n = 68). The correlations above the diagonal represent within-person associations and were estimated from HLM models with single standardized Level 1 variables and no Level 2 variables. Reliabilities are reported on the diagonal in parentheses; for the Level 1 variables, reliability values were averaged over measurements.
p < .05, **p < .01, ***p < .001 (two-tailed tests).
Before testing our hypotheses, we estimated the within- and between-person variance components from the null hierarchical linear model for all Level 1 variables. The percentage of total variance that was within-person was 53 percent for CPE, 66 percent for activating negative mood and 46 percent for activating positive mood.
Hypothesis tests
In accord with hierarchical linear modelling (HLM; Bryk and Raudenbush, 1992) principles, we regressed individuals’ within-person slopes for predicting CPE from activating negative mood (Level 1: slopes-as-outcomes) on the between-persons variables (Level 2: trait LGO, psychological empowerment and their cross-products). Level 2 variables were centred on the grand mean and Level 1 variables were centred relative to each respondent’s average score to eliminate between-individual variance (Bryk and Raudenbush, 1992). Finally, because time-series data violate the assumption of residual independence at Level 1 (Hofmann et al., 2000), we followed standard ESM practices (Ilies and Judge, 2002; Judge and Ilies, 2004) and controlled for lagged CPE (Time t–1) in all analyses of our hypothesis testing.
Table 2 shows the succession of multilevel models required to test our hypotheses, with Model 3 being most relevant to the hypothesized cross-level moderation effects (Hofmann et al., 2000; Snijders and Bosker, 2012). The random coefficient model (Model 1) involves only Level 1 variables. The Level-1 variance accounted for was 19 percent. Interestingly, and as one would expect from the literature, the control variable positive mood is positively related to CPE at Level 1, while negative mood is not related to concurrent CPE. 4 The latter finding is expected, because the negative mood–CPE relationship should depend on the level of the hypothesized moderators. Level 2 variables were added next in Model 2, the intercept as outcome model. The Level-2 variance accounted for was 22 percent. There were no Level 2 main effect hypotheses but, as we expected, LGO and empowerment each predicted mean CPE. Further, the Level 2 LGO by empowerment interaction also significantly predicted average CPE, such that LGO was positively associated with average CPE only when psychological empowerment was high at + 1 SD (coefficient = .73; p < .001).
Hierarchical linear modelling estimates for the interaction effects on creative process engagement.
Values are unstandardized coefficients, with standard errors in parentheses.
The models were based on 556 real-time reports from 68 employees.
p < .05; **p < .01; ***p < .001(two-tailed tests).
The slope as outcome model (Model 3) tests the three hypothesized cross-level interactions. As shown in Model 3, the two-way interaction of LGO and negative mood in explaining CPE (Hypothesis 1) was not significant (coefficient = .01; NS), while the interaction of psychological empowerment and negative mood in explaining CPE (Hypothesis 2) was significant (coefficient = .35; SE =.17; p <.05). Simple slope testing showed that activating negative mood was not significantly related to CPE when empowerment was low at −1 SD, but was positively related to CPE when psychological empowerment was high at +1 SD (coefficient = .39; SE = .13; t = 3.07; p < .01). Finally, the three-way interaction of LGO, psychological empowerment and activating negative mood in explaining CPE was significant (coefficient = .35; SE = .15; p < .05), offering support for Hypothesis 3. The variance accounted for by the three-way interaction was 4 percent. The significant higher order interaction renders the interpretation of the lower order interactions incomplete, so they will not be discussed further (Cohen et al., 2003). We examined the three-way interaction results, and simple slope tests revealed that, as expected, activating negative mood was positively associated with concurrent CPE when LGO and psychological empowerment were both high at +1 SD (coefficient = .50; SE = .17; t = 3.01; p < .01). Activating negative mood was not significantly related to CPE when LGO was high at + 1SD and psychological empowerment was low at −1 SD (coefficient = −.14; SE = .22; t = −.63; NS). The relationships between activating negative mood and CPE under low LGO and low psychological empowerment (coefficient = .08; SE = .11; t = .78; NS), and low LGO and high psychological empowerment (coefficient = .27; SE = .19; t = 1.43; NS) were also non-significant. Figure 1 displays these slopes. In sum, the data are consistent with Hypothesis 3 – that negative mood is associated with greater CPE but only when individuals are high on LGO and feel empowered to act in the workplace.
Interactive effects of trait learning goal orientation, psychological empowerment and activating negative mood on creative process engagement.
Post hoc analysis
Following the recommendation of Maynard et al. (2012) to explore potential differential effects of the four empowerment sub-factors, we repeated our analyses to determine whether trait LGO interacted with each of the sub-factors to moderate negative mood–CPE relationships as expected. The three-way interaction terms of activating negative mood, trait LGO and three empowerment sub-factors were significant and of similar magnitude and shape: self-determination (coefficient = .40; SE = .13; p < .01), meaning (coefficient = .51; SE = .18; p < .01) and impact (coefficient = .37; SE = .10; p < .001). The interaction with competence was not significant (coefficient = −.01; NS).
Discussion
Writers in the management literature have traditionally presumed that negative feelings inherently produce negative work outcomes (Elfenbein, 2008; Lindebaum and Jordan, 2014). This is not necessarily so. While negative moods can lead to maladaptive responses, they can also provide motivational and cognitive resources helpful for solving problems. A case in point is the relationship of negative mood and creativity, where research supports a more nuanced explanation (Baas et al., 2008; George and Zhou, 2002; To et al., 2012). Specifically, we hypothesized and found that activating negative mood can have a positive relationship to CPE in particular circumstances. Negative mood provides the problem signal and impetus that calls forth effort to solve a problem. LGO is the trait tendency to respond with engagement when faced with the problem signal, in an attempt to learn and gain mastery. Psychological empowerment is a job specific perception of the work situation as affording meaning, impact, competence and self-determination, which may activate the learning goal tendency by suggesting that mastery is possible through effort. In support of our ideas, our data show that, when trait LGO and perceived psychological empowerment are both high, negative mood is positively related to CPE. These results are consistent with Higgins’ (2005) regulatory fit theory and Tett and Burnett’s (2003) trait activation model suggesting that empowerment fits and permits the trait-derived inclinations of high LGO employees to be expressed in mastery efforts. Under this favourable fit condition, activating negative mood appears to function as a problem signal that stimulates increased effort and persistence toward creative solutions (De Dreu et al., 2008).
Theoretical implications and future research on mood–creativity relationships
Research on real time determinants and correlates of creative behaviour at work is in its infancy and much remains to be done both theoretically and empirically. Our work sheds light on the need to integrate the psychological empowerment and self-regulation literatures to explain the relationship between negative mood and creative behaviour at work. In an extension to earlier ideas (To et al., 2010, 2012; Zhang and Bartol, 2010a), our work demonstrates that it is the motivational alignment of psychological empowerment and trait LGO rather than their independent effects that produces the adaptive effect of activating negative mood on creative behaviour.
We believe that an important outcome of our research lies in our contribution to the debate as to whether or not psychological empowerment is truly motivating in organizations (Argyris, 1998; Forrester, 2000). Seibert et al. (2011) noted that empowerment may have unintended harmful outcomes if it prompts a person to approach a goal in a manner that does not naturally fit the individual’s motivational orientation. On the other hand, goal orientation research (DeShon and Gillespie, 2005) has highlighted the importance of considering the factors that may influence the expression of dispositional goal orientation. Synthesizing these literatures, our findings suggest that psychological empowerment plays an important role in facilitating the behavioural manifestation of trait LGO, seen in this research as the persistence pathway to creativity (De Dreu et al., 2008).
Implicit in our hypotheses, and in line with trait activation theory, is the idea that goal orientation may be switched on or off by cues in the environment (Tett and Burnett, 2003). DeShon and Gillespie (2005: 1115) state that, ‘[t]he activation of goals can be thought of in terms of chronic and transient accessibility. Goals that are frequently stimulated or activated become highly accessible and, therefore, are chronically ready to be activated.’ This implies that state goal orientation may act as a mediator underlying the cross-level relationship between trait goal orientation and perceived empowerment (between-persons level), and momentary mood and creative behaviour (within-person level). Future researchers may wish to test hypotheses about how state goal orientations and shorter-term fluctuations in perceived empowerment affect creative and other performance-related responses to moods. For instance, activating negative mood in an empowering context may signal to a high LGO individual that harder work toward skill acquisition is needed, and therefore may activate a transient LGO that facilitates creative effort. In contrast, the same mood may inform a low LGO individual that the situation is too difficult, resulting in withdrawal behaviour. This may, in turn, inhibit the persistence pathway.
Another interesting research avenue may be to explore the social effects of team member negative mood on team creativity. Van Kleef (2014) points out that feeling states provide the social information necessary to guide interpersonal behaviour at work. While experiencing negative feelings may promote idea generation via persistence among those high in LGO and empowerment, how might expressing these negative feelings affect co-workers who may differ in goal orientation and workplace perceptions? Future researchers may wish to explore the effects of these affective dynamics in teams.
Based on their review of empowerment research over the past two decades, Maynard et al. (2012) recommend exploring possible differential effects of the four empowerment sub-factors. Ergeneli et al. (2007) and Wang and Lee (2009) have raised similar concerns about whether the potency of empowerment is based on the entire construct or a subset of its dimensions. Although some evidence indicates that the sub-factors have different antecedents and predict different outcomes (see Maynard et al., 2012 for a review), a recent meta-analysis by Seibert et al. (2011) demonstrated the predictive utility of a composite construct of psychological empowerment.
Offering a more fine-grained view, our post hoc analyses show that three of the four empowerment sub-factors acted similarly in enabling high LGO individuals to capitalize on activating negative mood in the service of CPE. The sub-factor of competence did not have this effect. In retrospect, this finding is consistent with Elliot and Dweck (1988), who found that children high on LGO responded to failure feedback by increasing mastery efforts, regardless of beliefs about their current skill level. In addition, within-person research has revealed a possible negative effect of high self-efficacy (competence beliefs) on performance, contrary to the conventional view of high self-efficacy as motivational (Vancouver and Kendall, 2006). As shown by Schmidt and DeShon (2010), for tasks with high ambiguity, a high efficacy belief may inflate perceptions of performance and thereby hinder effort and subsequent performance. This opens a potentially meaningful avenue for future research to explore whether an empowering belief about existing competence is beneficial or potentially detrimental for motivating exploration and creative efforts over time. It is possible that individual differences (e.g. high versus low LGO) and task types (e.g. creative tasks versus routine tasks) may play a moderating role in this relationship.
Implications for practice
Our findings also have implications for practice. In general, empowering high LGO employees seems wise, since the perception of empowerment may activate their trait tendency to exert effort toward mastering difficult tasks and building skills. For both employees and managers, it is important to note that creative behaviour varies substantially within-person over time. Positive mood is consistently positively related to CPE, but transient negative mood may either facilitate or suppress creative efforts. We recommend that, in order to maximize creative behaviour, supervisory approaches should be tailored to employees’ trait goal orientation and empowerment perceptions. While we did not study or manipulate the source of negative moods in this study, a likely on-the-job source is negative feedback or failure to progress on a task. For individuals high in LGO who feel empowered, immediate attempts to remedy the problem via creative efforts following negative feedback or failures may bear fruit given their tendency to demonstrate resilience and persistence in converting the energy supplied by activating negative moods into creative efforts. For individuals who are not high in LGO or those who do not feel empowered, it is possible that requests to creatively solve a problem would be best postponed for a while until the negative mood has faded or been replaced with a more positive mood.
Limitations
Concerns may be raised about self-report, common method bias, sample size and causality in this research. We address each issue below. As is typical in ESM research, we used real-time self-report measures for all state variables (e.g. Dalal et al., 2009; Ilies and Judge, 2002). It makes sense to use self-reports to study mood–creativity links within-person because there are no external sources who can consistently monitor and accurately report either mood or creative behaviour on a short-term basis (Ng and Feldman, 2012). As we demonstrated, mean CPE scores converged with observer ratings of overall creative performance.
While self-reporting may raise concerns about common method variance, repeated-measure designs involving within-person analyses are less subject to this problem (Foo et al., 2009; Williams and Alliger, 1994). This is because participants report immediate situations rather than relying on error-prone recollections to construct descriptions of typical feelings or behaviour over longer periods of time (Robinson and Clore, 2002). Further, because all variables in the within-person analyses were centred at each person’s mean, the usual problems with self-reported data such as personality confounds and response set tendencies were eliminated (Judge and Ilies, 2004; Scott and Judge, 2006). In addition, Level 1 and Level 2 variables were collected in different questionnaires at different times, thus reducing the potential for common method bias, demand characteristics and hypothesis guessing (Podsakoff et al., 2012). Finally, Lai et al. (2013) recently demonstrated in a Monte Carlo study that, in the absence of true effects, it is extremely unlikely for common method variance to generate significant cross-level interactions in HLM.
By typical between-person design standards, our sample size of 68 would not be considered large. Nonetheless, it easily surpasses Scherbaum and Ferreter’s (2009) Level 2 minimum sample size requirement for multi-level analysis, which is N = 30. A review by Mathieu et al. (2012), including 79 multilevel investigations published in the Journal of Applied Psychology between 2000 and 2010, reported that the median Level 2 sample size was 51. Our analyses were based on a total of 556 Level 1 reports as well as 68 Level 2 reports by 68 participants, suggesting a sufficient sample size at both levels.
Finally, we note that the correlational design used here does not allow definitive conclusions about causal relationships between mood and CPE. Nevertheless, support for a causal impact of mood on creativity can be found in decades of laboratory research in which mood has been manipulated and subsequent creativity measured. Further, a large body of experimental research identifies specific mechanisms by which mood may influence creativity (Baas et al., 2008). Our significant multilevel interactions draw on this body of theory and are consistent with these mechanisms, lending credence to causal interpretations of our data.
Conclusion
In conclusion, we hypothesized and found that combinations of trait LGO and psychological empowerment constitute joint boundary conditions for the negative mood-creativity relationship. Our study contributes to resolving the ambiguity surrounding this relationship. Returning to our opening question: Yes, it is possible that the employee experiencing angst will have a more creative day than usual. It seems that negative mood can be unleashed in creative striving, but only if high LGO and psychological empowerment are aligned to facilitate creative behaviour.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.