How to Organize Creative and Innovative Teams: Creative Self-Efficacy and Innovative Team Performance

ABSTRACT

What makes teams creative? We investigated how the diversity and agreeableness of team members moderate the relationship between creative self-efficacy (CSE) and innovative performance at the team level. We found that the educational background diversity is a critical factor to affect the knowledge, perspective, and problem-solving skills and, in turn, has a positive effect on innovative performance. We also showed that the relationship between aggregated CSE and innovative performance was not linear but inverted U-shaped at the team level. We expect our study to provide meaningful managerial implications about team composition and team innovative performance.

Introduction

What makes teams creative? This fundamental question in both academic and organizational contexts has found one of the answers: creative self-efficacy. Creative self-efficacy (CSE) is the belief that one has the necessary knowledge and skills to produce valuable creative outcomes (Tierney & Farmer, 2002), and has been investigated as a key driver of individual creativity (e.g., Karwowski, 2016). In order for one to be creative, one must first believe in their own ability to be creative. While there are a few exceptions, most previous research on CSE have largely been conducted at the individual level (e.g., Tierney & Farmer, 2002 &, 2011). Lacking, however, is research on CSE at the team or organizational levels, or the cross-level interactions that individual CSE has with team creative outcomes. As teams are ubiquitous as the fundamental unit of creative output in most modern-day organizations and firms, it is imperative for managers and researchers alike to understand the influence individual team member CSE has on the creative performance of teams.

The relationship between individual CSE and team creative output is not simple, however. While some studies purport the positive effects of CSE, other studies have demonstrated the negative effects of CSE. These negative effects such as escalation of commitment (Staw, 1997), self-protective attribution patterns (Silver, Mitchell, & Gist, 1995), and seeking less information (Audia, Locke, & Smith, 2000). These traits may decrease team innovative performance. It may be that individual CSE improves team innovative output to a certain point, after which it becomes detrimental. Thus, additional research needs to investigate whether the relationship between individual CSE and team performance may take the shape of an inverted-U.

Furthermore, there may be other factors which influence the relationship between individual CSE and team innovative performance. The personality of team members, and specifically their agreeableness, is expected to moderate this relationship such that higher individual agreeableness will lead to improved creative output at the teal level. This might seem counterintuitive as high agreeableness is associated with low creativity on the individual level (King, Walker, & Broyles, 1996). However, at the team level, we argue that high agreeableness is predicted to improve team cooperation which will in turn increase creative output. Another expected moderating factor is diversity. Team members with diverse educational backgrounds, skills, and knowledge are predicted to complement each other in ways that will improve overall creative output (i.e., Woodman, Sawyer, & Griffin, 1993). We argue that personality and educational background diversity may moderate the relationship between individual CSE and group creativity. Although these factors are not necessary for creative performance, their influences in the team context may improve the creative output of teams. Thus, this study investigates the relationship between CSE of individuals and innovative performance at the team level with the focus on team members’ personality characteristics and team educational background diversity.

Literature review

Creativity has long been a research topic in the psychology field. These days, interdisciplinary interest in creativity has grown and it is now a major research agenda in a wide range of disciplines including education, sociology, engineering, and management. This interdisciplinary growth in creativity research is largely due to the significant influence of creativity in the contexts of various organizations’ In line with previous studies (e.g., Zhou & Shalley, 2003), we define creativity as the production of novel and useful ideas concerning products, services, processes, and procedures by an employee. Creativity in organizations in particular focuses on variables affecting teams, workplaces, and organizations at the macro-level. (Shalley & Zhou, 2008). Woodman et al. (1993) argued that the creativity in the context of organizations can be a source of valuable, useful new products, services, ideas, procedures, or processes since various individuals work together in a complex social system. For example, previous studies highlighted that creativity is most vital during the initial stages of the creative process for the ideation and/or creation of new concepts, and alternative ideas.

Most previous studies on creativity investigated various issues at the individual level, such as knowledge map (e.g., Zhang, Zhang, Yu, & Zhao, 2015), components of creativity (e.g., Costa & McCrae, 1992), personality (e.g., Barrick & Mount, 1991), CSE (Tierney & Farmer, 2002), social context (Amabile, 1983), etc. However, to investigate the creativity in organizations, it is necessary to examine creativity at the team level. Since teams are the fundamental creative unit within organizations, the individual creative contribution to the team is more significant than the contribution of the individual by themselves in organizational contexts. The creative output of organizations is generally produced by teams, rather than directly by individuals (i.e., Damperat, et al. 2016). Creative ideas are first developed by individuals. However, it is then dependent on team processes and composition whether these ideas will be developed into creative output or not (i.e., Zhang, et al. 2019). Teams may have positive effects on both individual and group creativity. In terms of individual creativity, teams can provide information and resources to individuals which benefit individual creativity, especially for those individuals with high CSE. Teams which are empowered may also increase the motivation and the creative performance of their members. (i.e., Liu, Wang & Yao, 2017). In terms of group creativity, CSE of individual members is a determinant of the collective CSE of the group, which in turn positively influences team creative output (Damperat, et al. 2016). With teams playing such an important role in the creative process, it is important to understand the characteristics of teams that may influence efficient creative output. In addition, Ford (1996) highlighted the role of creativity that often plays across multi-levels in organizations of analysis throughout different phases of innovation. While only a few studies investigated the creativity issues at the multi-levels or cross-levels in organizations (i.e., Anderson, Potočnik, & Zhou, 2014), we need more research with an interactionist perspective on creativity.

Creative self-efficacy at the team level

CSE, as an important predictor of individual creativity (e.g., Gong, Huang, & Farh, 2009) may have meaningful influence on team dynamics. Tierney and Farmer (2002) first proposed the construct of CSE and verified the validity of the distinction between CSE and general self-efficacy with an empirical test. Bandura (1997) further suggested different efficacy types to better understand outcomes of various tasks. Some of those specific self-efficacies included computer self-efficacy (Gist, Schwoerer, & Rosen, 1989), accountancy self-efficacy (Saks, 1995), and job-seeking self-efficacy (Caplan, Vinokur, Price, & Van Ryn, 1989). These studies argued that these specific measures of self-efficacy may produce a reliable and valid predicting power about expected behavior.

The relationship between CSE and outputs at the team level may be much more complicated than that at the individual level. Creative outcomes in organizations depend on the CSE of individuals and their creative talents in the context of other people. As Amabile (1988) emphasized the difference between completing a given task effectively and executing it creatively, developing creative outputs in the context of teams may pose real challenges. Previous research found that the effect of CSE depends both personal and environmental factors. They suggested that CSE may be linked to individual characteristics, such as job tenure, education level, learning orientation, and role identity (e.g., Gong et al., 2009; Tierney & Farmer, 2002; Wang, Tsai, & Tsai, 2014). Furthermore, environmental factors including supervisor support, job complexity, transformational leadership, job creativity requirement, and supervisor creativity requirement may be related to CSE (e.g., Gong et al., 2009; Tierney & Farmer, 2011; Wang et al., 2014).

Previous studies have reported somewhat mixed findings about the relationship between CSE and outputs. For example, Bandura (1997) highlighted that strong CSE is necessary for creative productivity and innovation. Ford (1996) suggested CSE as a key motivator to explain individual creative behaviors. These studies show that CSE may be a meaningful antecedent that positively influences creative and innovative performance and innovation in organizations. However, other studies have found that excessive CSE may be detrimental to creative performance, and that this negative influence may also be exacerbated within group environments (Gist, 1987; Janis, 1972; Whyte, 1993). If CSE is too high, it may lead to overconfidence, group think, and escalating commitment in the context of the team. Whilst teams are commonplace with most modern business organizations, these mixed findings about the relationship between CSE and performance at the team level still require further studies to how CSE affects team performance in various contexts.

Personality

Personality is understood as having a significant effect on both individual and team-level creative output. Extensive research on how individuals’ personalities influence their behavior, thoughts, and feelings has been performed. (e.g., Costa & McCrae, 1992). These studies have mainly concerned individuals, however there is a growing interest in how each individual’s personal knowledge, skills, and experiences, or their personality traits, may interact within teams and affect the various performance of the group or entire organizations (e.g., Oh, Kim, & Van Iddekinge, 2015; Ployhart, Van Iddekinge, & MacKenzie, 2011). Recent creativity research pays more attention to understanding how multiple characteristics of individuals, including personality, interact with one another within a complex social system to impact creative performance. House (1988) argued that personality traits can have significant influence on teamwork performance. Previous studies have paid a great deal of attention to team performance because creative organizational tasks are most likely to require capabilities exceeding the talents of single individuals (i.e., Guzzo & Salas, 1995; Kickul & Neuman, 2000).

Personality traits that improve dynamics of teamwork may therefore increase the team creative output. The five-factor model (FFM, Goldberg, 1993) describes the personality using five factors: extraversion, agreeableness, conscientiousness, emotional stability and openness to experience, which are assumed to be relatively stable over one’s lifetime, and to cause individuals to behave in particular ways (McCrae & Costa, 1997). Kickul and Newman (2000) posited that certain personality traits, such as extroversion and openness to experience, may be more effective in increasing team-based knowledge, skills, and abilities, consequentially improving creative capabilities of teams. McCrae & Costa, (1989) understood that extroversion and openness may be effective at encouraging peers to work together to accomplish challenging goals and increase the quality of social interaction. Others (Kickul & Newman, 2000; McCrae & Costa, 1987) further argued that open individuals prefer variety, are intellectually curious, and demonstrate independence of judgment. In addition, openness is related to divergent and novel thinking, creativity, and curiosity that may provide unique insight in creative problem-solving situations (Stevens & Campion, 1994).

Despite previous research on personality, little research has examined the effect of agreeableness on CSE and team creative outputs. Agreeableness is the tendency for a person to be cooperative, helpful, or compassionate toward others (Barrick & Mount, 1991). Previous studies showed that, at the individual level, agreeableness had a negative impact on creative or innovative performance (King et al., 1996). Karwowski and Lebuda (2016) argued that agreeableness is the weakest of the five FFM traits at predicting creative self-beliefs. While agreeableness had a negative effect on creativity at the individual level, previous studies did not answer whether it may have a different effect on creative output at the team level.

Diversity

Team diversity refers to the heterogeneity of demographic characteristics of members who interact with other members in a team (e.g., Jackson, 1992). These characteristics can range from inherent, low job-relevancy traits such as gender, age, and ethnicity, to fluid, high job-relevancy traits such as educational background or career experience (e.g., Jackson, 1992). In addition, non-tangible characteristics including personality and beliefs may contribute to the diversity of groups (Bowers, Pharmer, & Salas, 2000; Harrison, Price, & Bell, 1998).

Previous studies have investigated the effects of intra-group diversity on work processes and performance through two lenses: the social categorization perspective and the information/decision-making perspective (Van Knippenberg & Schippers, 2007; Williams & O’Reilly, 1998). Studies following the social categorization perspective argued that both similarities and differences among group members may affect the formation of internal subgroups, as people may tend to seek out and cooperate with members sharing similar characteristics (Brewer, 1979). They further suggested that teams with less intra-group diversity may perform more efficiently than teams with higher diversity. On the other hand, the information/decision-making perspective suggested that high intra-group diversity may have positive effects on team performance. A team comprised of diverse individuals will likely have a broader knowledge base, more diverse skill sets, and different experiences to draw from which can be beneficial to performing creative tasks (e.g., Jackson, 1992). Previous research on the link between diversity and creativity assumed that creative tasks often require new ideas, solutions, and the synthesis of prior experiences. They showed that teams organized with diversity in mind may fulfill these requirements more effectively than less diverse teams (e.g., Mumford & Gustafson, 1988). Consequently, organizations in many fields have begun to organize their teams to ensure higher member diversity (e.g., Williams & O’Reilly, 1998). This existing literature reinforces the need to further investigate the effect of member diversity at the team-level creativity.

Hypotheses

Individuals with high CSE can better deal with possible barriers of creative performance in the context of team dynamics. One such barrier is production blocking, which may occur when individual(s) are unable to contribute to a discussion, either because of insufficient time, lack of confidence, or domineering coworkers (Diehl & Stroebe, 1987; Nijstad, Stroebe, & Lodewijkx, 2002). A person with a high CSE may have greater confidence in their abilities and knowledge, and thus reduce their concerns of negative evaluations. Diehl and Stroebe (1987) argued that evaluation apprehension may cause a loss of creative productivity within teams. Evaluation apprehension refers to the fact that individuals may not share their ideas due to worries of negative feedback from their peers. In addition, free riding or social loafing may occur due to the evaluation apprehension and the difficulties in identifying individual contributions (Diehl & Stroebe, 1987). Previous studies suggested that employees with high CSE may mitigate these negative effects since they are more likely to prove their excellence rather than take a free ride (Bandura, 1977).

However, the relationship between CSE and creative performance at the team level may not be a simple linear association. Performing tasks within a team inevitably leads to coordination costs, decreasing productivity (Hackman & Morris, 1975). Furthermore, excessive CSE may exacerbate these coordination costs, since it may hinder group dynamics and the ability to make a final proposal (Diehl & Stroebe, 1987; Jehn & Mannix, 2001). To develop creative team performances, most team members may need to evaluate their strengths and weaknesses as objectively as possible. Individuals with remarkably high CSE may further experience an escalation of commitment, which is a pattern of behavior in which participants continue to invest more efforts into their current task rather than altering or changing their behavior in the presence of negative consequences of their previous choices (Straw, 1981). Previous studies argued that this behavior is more prevalent within individuals and groups who perceive themselves to have exceedingly high self-efficacy (Whyte, Saks, & Hook, 1997). Previous studies on attribution theory have suggested that individuals with extremely high CSE are more likely to deal with escalating commitment by avoiding self-serving attributions for unsuccessful performances (Silver et al., 1995). These self-serving attributions may blame external factors for their failure, such as environmental changes or bad luck. Audia et al. (2000) argued that individuals with remarkably high CSE and previous successes may also lead to overconfidence or arrogance, causing teams and individuals to seek out less external information. If teams frequently blame external factors, this may eventually decrease the probability to produce creative outputs. We posit that CSE will have a positive effect on team creative performance up to a certain point, after which CSE will begin to have a negative influence. Given the above trade-offs between CSE and creative performance, we predict as follows:

Hypothesis 1. There will be an inverted U-shaped relationship between CSE and innovative performance at the team level.

Role of educational background diversity at the team level

While CSE may significantly affect team performance, the objective skills and capabilities of teams may further improve the innovative outputs at the team level. While these skills are not necessary for innovative outcomes to occur, they may positively moderate the relationship between CSE and innovative performance if they are present. The strength and capabilities of teams required for creative tasks can be the result of multiple individuals combining their knowledge, skills, and abilities to generate and synthesize new ideas, processes, and solutions (Mumford & Gustafson, 1988). Individual knowledge, skills, and abilities are significantly influenced by educational background, specifically types, levels, and contents of education while Individuals’ cognitive style, personality, and values are also reflected in their selection of educational major (i.e., Bantel, 1993; Holland, 1973). For college students with no professional experience, their educational background may provide some indication of certain skill bases (i.e., Thatcher, 1999) and further affect the likelihood to collaborate other team members. Ojha (2005)suggested that differences in educational background among team members affected the likelihood of participation in team collaborations. Harrison et al. (1998) suggested that the quality and process of team dynamics is also strongly affected by the team members’ personal characteristics and background.

This paper specifically predicts educational diversity to have a moderating effect because while educational diversity may improve the creative outcomes of a team, it is not a necessary condition for creative outcomes to occur. A team with a high diversity of these characteristics has a larger pool of resources to draw from to foster creativity (Bantel & Jackson, 1989). Some studies have also related diversity to greater cognitive processing and information processing capabilities (Watson, Kumar, & Michaelsen, 1993). Among others, Williams and O’Reilly (1998) identified educational diversity as one of the most pertinent factors which affects creativity. Knowledge, skills, and different perspectives are acquired throughout an individual’s lifetime, predominantly via education and work experiences (Tierney & Farmer, 2002). Bell, Villado, Lukasik, Belau, and Briggs (2011) demonstrated that functional and educational background diversity has a positive relationship with team performance, creativity, and innovation. They further showed that the relationship was stronger for open-ended, non-routine tasks such as product development or design, which require creative solutions and/or problem-solving skills. Educational background diversity also assists in evaluating ideas because teams with diverse knowledge may be more capable of appropriately assessing their feasibility (Larey & Paulus, 1999). In addition, a group with a greater educational background diversity will likely have more external contacts across various industries, such as former colleagues or classmates, which can be beneficial toward projects requiring high creativity (i.e., Carpenter, 2002). These positive impacts of educational diversity may significantly moderate the relationship between CSE and innovative performance at the team level. Thus, we predict as follows:

H2. The educational background diversity will positively moderate the relationship between CSE and innovative performance at the team level.

Role of agreeableness at the team level

While agreeableness has been correlated with low creativity at the individual level, we posit that high agreeableness in group members may increase the creative performance of a team. This is because the agreeableness of team members may improve team dynamics and the ability of team members to collaborate, thus improving creative performance. Studies have shown that agreeableness most strongly influence an individual’s interactions with others (e.g., Barry & Stewart, 1997). Barrick, Stewart, Neubert, and Mount (1998) argued that a high degree of agreeableness positively impacts the efficacy of a team, and that secondary characteristics such as helpfulness, trustworthiness, friendliness, tolerance, and/or non-competitiveness have a favorable effect on team processes too (e.g., Costa & McCrae, 1992).

The creative performance of a team is significantly influenced by the free and open exchange of ideas and constructive communications and interactions between individuals (Goncalo, Flynn, & Kim, 2010; Nijstad et al., 2002). Jackson, Stone, and Alvarez (1992) suggested that people who have high level of agreeableness can effectively mediate between competing ideas. These characteristics may reduce the negative effects of a high CSE mentioned above and improve overall team performance. Team members with high agreeableness may promote collaborations within their team and increase creative performance. As such, the agreeableness of individual team members may improve team dynamics, in turn improving the creative output of the team. Thus, we predict at follows:

H3. Agreeableness will positively moderate the relationship between CSE and innovative performance at the team level.

Method

We utilized a randomized field study method, common in educational research, in order to test our hypotheses. 200 undergraduate students and 223 MBA students enrolled at a College of Business Administration in Korea participated in this study. The length of the study was 15 weeks for undergraduate courses and 8 weeks for MBA courses, corresponding with the length of the respective semesters. 72.3% of the participants were female and the average age of all participants was 31.39 years. Seven to ten project teams were formed per course in each semester from 2015 to 2017 and a total of 101 teams participated in this study. The goal of the group project was to develop a new business idea that can be commercially successful (feasibility) and creative (originality) in a relevant market. Student teams had to evaluate multiple ideas, discern the most feasible and appropriate options, and reach a consensus on developing their final creative ideas. By identifying problems or weaknesses in existing products and services, each team had to generate innovative ideas and concepts. Participants were randomly assigned to each team. The students had meaningful motivation to actively participate in their group projects since 30% of each participant’s overall grade was allocated to their team’s output. Every team was requested to pursue the same task: to generate creative business ideas throughout the semester. To maintain comparable team collaboration dynamics, each team was requested to submit very detailed progress reports at least four times per semester and to have at least two team meetings per each report. At the end of each semester each team presented their final ideas to their entire class.

Previous laboratory studies often evaluated creative team output based on their performance in single-part tasks. Single-part tasks have often required team members to “ideate names or uses or consequences of a thing or ideate ways to achieve a goal” (Brophy, 1998). These teams were often short-lived and not necessarily motivated to put forth their best efforts. This study has several advantages over previous laboratory studies. In this study, teams completed a multi-part and open-ended assignment over a relatively long period of time. While all teams had a pre-determined deadline for their outputs, participants were given flexibility in terms of time management and resource allocation within their teams over the course of the study.

We evaluated innovative team performance by assessing the feasibility and originality of the final group project output. Amabile (1983) argued that output-based operationalization may be more appropriate than process-based operationalization for measuring creativity, since it is relatively more demanding to quantify the qualitative and/or subjective nature of creative processes. Following previous studies (i.e., Amabile, Conti, Coon, Lazenby, & Herron, 1996), the final project outputs were assessed by a professor and two teaching assistants at the end of each semester. Final projects were evaluated on a 10-point scale with the minimum of 4-points with given descriptors for all numbers: 9–10 very creative and excellent; 7–8 creative and good; 5–6 not creative and average; 4–5 needs improvement. These assessments were also reviewed by two other professors at other business schools and two external experts in the venture capital industry in Korea. While some studies argued that high interrater reliability alone is not a sufficient condition to ensure assessment validity (Kaufman & Baer, 2012), we checked interrater reliability by computing rwg for professors, teaching assistants, and external experts. The value of rwg ranged from 0.76 to 0.81.

In this study teams were the basic unit of analysis. All hypotheses were presented at the team level, and the analyzed variables were re-calculated at the team level by converting variables measured at the individual level. Since all participants were randomly organized and no one was selected as a leader, we assumed that all participants have equal power. Thus, to aggregate individual-level measures to team-level measures, we used a team profile model that applied equal weight to each team member (e.g., Rousseau, 1985).

Measures

CSE. To assess participants’ CSE, we used Tierney and Farmer (2002)’s three-item scale. Tierney and Farmer (2002) developed an initial questionnaire based on the self-efficacy and creativity literature (e.g., Amabile, 1988; Bandura, 1997; Woodman et al., 1993). The questionnaire was comprised of generalized questions in accordance with general creative skill suggested by Amabile (1983). Participants were asked to assess their belief in their own ability to perform creative behavior successfully, with responses ranging from 1, “very strongly disagree”; to 7, “very strongly agree.” Sample statements included “I have confidence in my ability to solve problems creatively”, and “I feel that I am good at generating novel ideas.” This scale has been used extensively in previous research and recognized to be linked to various indicators of creative performance of individuals (i.e., Hammond et al., 2011). Cronbach’s alpha was 0.81 and comparable in size to those reported by previous studies (i.e., Ohly, Pluckthun, & Kissel, 2017). The CSE score of each team member was averaged and converted into a team-level variable. Given that CSE may affect the level of participation of team members, it is appropriate to regard team CSE as the sum of individual members’ CSE. Averaging team member characteristics is the most common operationalization method for calculating team-level variables in multi-level research designs (e.g., Taggar, 2002). For the team-level CSE, we tested the within-team agreement for CSE by computing rwg, which yielded the median value of 0.84. We also calculated ICC estimates and their 95% confident intervals using SPSS statistical package. The ICC value was 0.892 and its 95% confidence interval ranged between 0.843 and 0.936.

Educational background Diversity. Blau (1977)’s index of heterogeneity was used to measure the categorical differences in the educational background of team members. Following previous studies (i.e., Harjoto, Laksmana, & Yang, 2019), we divided the educational background diversity of participants into nine categories, such as social sciences, natural sciences, humanities, and law. Participants were asked to identify their undergraduate major in a paper-based questionnaire. Then, the team’s educational background diversity was calculated using Blau (1977)’s index formula, $1-{\sum }_i^{p}2$, where $p_i$ denotes the proportion of a team’s members in the ith category. The maximum and minimum diversity ranged from 0 to 0.889.

Agreeableness The Big Five Inventory (BFI) was used to measure agreeableness, as an alternative to the NEO-PI scales (John, Donahue, & Kentle, 1991). The BFI was developed as an efficient measurement tool where additional differentiation of the big five factors is not required. While BFI has a reduced number of questionnaire items, it is still adequate for measuring psychometric properties (John & Srivastava, 1999). The BFI measures personality traits by using short phrases which include the typical adjectives associated with the five characteristics. Using these short phrases instead of single-word adjectives can improve the consistency of results (Goldberg & Kilkowski, 1985). Participants responded to each questionnaire item on a 5-point Likert scale ranging from 1 “not at all descriptive of me” to 5 “very descriptive of me.” Consistent with previous studies, the mean scores of agreeableness were computed for each team to operationalize team-level variance in personality (e.g., Oh et al., 2015).

Control Variables We controlled for team size, the length of the team project, gender, and team members’ age heterogeneity. Previous studies demonstrated that team size may influence the performance of ill-defined tasks (Hülsheger, Anderson, & Salgado, 2009). Theoretically, a larger team has a higher diversity of perspectives, knowledge, and skills. Past studies also suggested that gender may impact team dynamics, thus affecting creative output (e.g., Pearsall, Ellis, & Evans, 2008). We also controlled for the heterogeneity of age among team members. If gender and age become salient factors, they may lead to emotional conflict or the formation of subgroups (e.g., Pearsall et al., 2008). These may impede team processes and reduce creative output. We also controlled for the length of the team project. As cognitive thinking ability may significantly influence creative performance (e.g., Amabile, 1983), we controlled for the creative thinking capability (CTC) of individuals by measuring their divergent and convergent thinking abilities. Finally, following previous studies (e.g., Barrick et al., 1998; Goncalo et al., 2010), we controlled for CSE covariance amongst team members. Even when teams may have similar aggregate CSE scores, the distribution of individual CSE scores within teams may be significantly different, either highly concentrated or widely distributed.

Results

Table 1 showed the means and variances of innovative team performance, CSE, educational background diversity, agreeableness, and other control variables. It also presented the correlation coefficients of the independent and dependent variables. We did not find multicollinearity problems because while some of the correlations are statistically significant, their magnitudes were not large. We used the hierarchical regression analysis to test our hypotheses. First, we tested the effects of the control variables (Model 1) and then the main effect variables (Model 2). Since we predicted a curvilinear relationship between CSE and innovative performance at the team-level (H1), a quadratic term by squaring the score of aggregated CSE was inserted in the regression equation (Model 3). Next, we tested the effects of educational background diversity, agreeableness, and their moderating effects (H2, H3) by adding the relevant interaction terms (Model 4 and 5). In order to avoid multi-collinearity problems of interaction variables, we used mean-centering procedures for the relevant variables. We first reported the descriptive statistics in Table 1 and the result of the hierarchical regression analysis in Table 2.

Table 1. Descriptive statistics and correlation matrix

	M	SD	Min.	Max.	1	2	3	4	5	6	7	8	9
1. Performance	7.03	1.11	4	10	-
2. CSE	5.43	0.36	4.48	6.54	0.17	-
3. Major diversity	0.58	0.15	0.00	0.78	0.17	0.08	-
4. Agreeableness	2.44	0.83	0.46	4.22	0.16	0.62*	−0.12	-
5. Age diversity	0.08	0.05	0.00	0.24	−0.08	0.09	0.02	−0.07	-
6. Team size	4.19	0.61	3.00	6.00	−0.15	0.15	0.14	0.15	0.15	-
7. Project period	7.93	3.54	3	12	0.23*	−0.05	0.32*	0.03	−0.21*	−0.03	-
8. Gender diversity	0.27	0.21	0.00	0.50	0.01	0.07	−0.08	0.28*	−0.10	0.05	0.10	-
9. Creative thinking capability	59.31	14.09	29.50	88.38	0.34*	0.06	0.15	0.13	−0.23*	−0.28*	0.53*	0.08	-

Table 2. Results of hierarchical regression predicting team innovative performance

	Model 1	Model 2	Model 3	Model 4	Model 5
CSE – (A)		0.556	0.715*	0.626*	1.072*
		(0.071)	(0.023)	(0.042)	(0.035)
(CSE)^2 – (B)			−1.214*	−2.309***	−0.0328
			(0.038)	(0.001)	(0.970)
Major				−0.128
Diversity-(C)				(0.878)
(A)*(C)				1.349
				(0.625)
(B)*(C)				18.92**
				(0.005)
Agreeableness					0.157
-(D)					(0.375)
(A)*(D)					−0.943
					(0.105)
(B)*(D)					−1.263
					(0.108)
Age	0.261	−0.102	0.780	1.105	1.310
Heterogeneity	(0.915)	(0.967)	(0.748)	(0.637)	(0.593)
Team Size	−0.135	−0.196	−0.304	−0.339	−0.456*
	(0.467)	(0.295)	(0.110)	(0.069)	(0.026)
Project Length	0.0262	0.0336	0.0340	0.0322	0.0373
	(0.470)	(0.351)	(0.336)	(0.370)	(0.288)
Gender	−0.0718	−0.136	−0.263	−0.231	−0.0701
Heterogeneity	(0.891)	(0.793)	(0.613)	(0.646)	(0.898)
CTC	0.0221*	0.0193*	0.0203*	0.0215*	0.0194*
	(0.021)	(0.043)	(0.031)	(0.018)	(0.043)
CSE cov.			2.894	3.465	2.369
			(0.159)	(0.082)	(0.247)
N	101	101	101	101	101
adj. R^2	0.07958	0.10173	0.13411	0.20328	0.15103
F	2.729*	2.887*	2.936**	3.319***	2.617**

p-values in parentheses

* p < 0.05, ** p < 0.01, *** p < 0.001

The results in model 3 showed that the CSE coefficient was positive and statistically significant (β = 0.72, p < .05). The coefficient of the quadratic term of aggregated CSE was negative and significant (β = −1.21, p < .05). These results were largely consistent with our hypotheses. Hypothesis 1 proposed that the relationship between CSE and innovative performance at the team level had an inverted U-shape. Hypothesis 1 was supported. Model 4 showed that the coefficients testing the moderating effects of educational background diversity were statistically significant (β = 18.92, p < .01), supporting our Hypothesis 2. Model 5 reported the moderating effect of agreeableness and their coefficients were not significant. Hypothesis 3 was not supported.

Discussion

This study investigated the relationship between CSE and innovative performance at the team level and the moderating effects of educational background diversity and agreeableness. The results demonstrated an inverted U-shaped relationship between CSE and team innovative performance. While most creativity studies have been conducted at the individual level (e.g., Espedido & Searle, 2018; Sternberg & Lubart, 1999) and most innovation research at the organization level (e.g., Boh, Huang, & Wu, 2020), this study endeavored to combine these previous studies at the team level. Due to the challenging research design that needs to integrate multiple individuals into a single team at the cross-level (e.g., Damperat et. al., 2016; Liu et. al., 2017), there have been fewer studies of organizational creativity at the team level. To develop a better understanding of the possible linkage between individual creativity and innovative performance in the context of organizations, this study created an empirical linkage between the creativity and innovation issues at the team level.

Some previous team-level studies have focused on procedural or environmental variables rather than the individual characteristics of team members (e.g., Zhang et al., 2019). However, as the creative process is ultimately driven by individual team members and their ongoing interactions and collaborations, it is important to study the relationship between team member’s individual characteristics and team innovative performance. Previous studies largely espoused the benefits of higher CSE at the individual level (e.g., Karwowski & Lebuda, 2016); however, this study has demonstrated that excessive CSE at the team level may have negative impacts on innovation performance. At an individual level, a high CSE may positively impact performance since it may increase motivation and self-confidence. However, at the team level, an excessively high CSE may lead to escalation of commitment, a lack of critical awareness, and overconfidence, negatively impacting team interactions and creative processes.

This study also examined the moderating effect of educational background diversity and agreeableness on the inverted U-shaped relationship between CSE and innovative performance. Our results demonstrated that the education background diversity of team members is a critical factor. It had a positive impact on innovative performance, indicating that it may increase knowledge, perspective, and problem-solving skills in the context of teams. We originally theorized that in team contexts, agreeableness may improve team dynamics and thus result in increased innovative performance. However, the moderating effect of team members’ agreeableness was not supported. Previous studies suggested that agreeableness may lead to conformity and impede individual creativity and idea generation (e.g., King et al., 1996). Our results suggest, however, that the benefits of agreeableness in group project contexts does not outweigh the negative effect at the individual level. As previous studies suggested that the effect of personality may depend on the difficulty of task and the nature of reward (e.g., Espedido & Searle, 2018; Koswara, Widyanti, & Park, 2020), future studies may still find some effect of agreeableness on team outputs in other team contexts.

This study is not free from potential limitations. First, we measured the CSE of individual team members at the beginning of the semester but did not measure the CSE again thereafter. Participants may have responded differently at the end of the study because the CSE measure may vary over the semester-long project duration. This would significantly impact our findings. Future research should consider the timing and number of CSE measurements especially for studies with a long observation period. Second, we assumed that team members’ characteristics influence the team processes, and in turn, affect overall innovative performance. However, we could not directly test the relationship between team members’ characteristics and team-level process variables such as dealing with conflicts, maintaining ongoing communication, developing cohesiveness, and providing psychological safety. These process-related variables may provide meaningful research opportunities to better understand team dynamics and team-level performance. Third, we relied on the subjective assessment of three independent evaluators to measure the innovative team performance in this study. Thus, the criteria by which the evaluators judged creativity are liable to be inconsistent, limiting us from providing a more descriptive account of the parameters used in our rating scale. Future studies may need to further consider how to triangulate creative and/or innovative performance measures and ensure their consistency (e.g., Park, Chun, & Lee, 2016). Fourth, although our hypotheses were supported, the external validity and real-world applicability may be limited due to the use of the randomized filed study method. For instance, in a real business context, the surface-level diversity of employees may cause the formation of cliques or factions, which can reduce the interactions of employees across different groups and ultimately decrease team-level performance. Fifth, future studies can also investigate whether our results are generalizable across different organizational and/or national cultures. Different cultures may emphasize different values, norms, and behaviors which may affect possible interactions and team dynamics. Our study can provide the following implications for managers and scholars. First, we need to consider the CSE of employees when organizing individuals with various backgrounds in order to create innovative teams in real organizations. In addition to improving the creative thinking capability of employees, inspiring self-confidence in their creative abilities may be an effective way to create innovative team performance. Second, we should clearly understand the negative effects of a high CSE on team interactions. Whilst high CSE certainly has advantages in terms of motivation, it may hamper potential interactions among team members. Lastly, managers should not be over-confident in our findings. They need to understand that our findings may be sensitive to hidden factors not tested in our study.

Disclosure statement

No potential conflict of interest was reported by the authors.