Motivation of individuals towards a team goal from the
perspective of goal-framing theory
Master Thesis
Isabella Sigg (11948728)
MSc. in Business Administration – Strategy Track
Faculty of Economics and Business
University of Amsterdam
Supervisor: Siri Boe-Lillegraven
Statement of originality
This document is written by Isabella Sigg who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
II
TABLE OF CONTENT
LIST OF TABLES AND FIGURES ... III ACKNOWLEDGEMENTS ... IV ABSTRACT ... V INTRODUCTION ... 1 LITERATURE REVIEW ... 4 Teamwork literature ... 4 Goal-framing theory ... 8
Conjunction of the two streams of literature and research question ... 10
HYPOTHESIS DEVELOPMENT ... 12
EXPERIMENT DESIGN ... 17
Operationalization ... 17
Procedure treatment group ... 21
Procedure control group ... 22
Pilot work ... 23
Introduction of the observational variable ... 24
Measurement ... 25
DATA ANALYSIS ... 29
Preliminary analysis ... 29
Regression analysis with PROCESS ... 32
RESULTS ... 34
DISCUSSION ... 39
Major findings ... 39
Contributions of this study ... 42
Limitations and future research suggestions ... 43
CONCLUSION ... 47
REFERENCES ... 48
LIST OF TABLES AND FIGURES
Table 1: Means, standard deviations, correlations ... 29
Table 2: Rotated factor loadings components total motivation ... 32
Table 3: Total, direct, and indirect effects for the model presented in Figure 3 ... 34
Table 2: Regression coefficients, standard errors, and model summary information for the model presented in Figure 3 ... 37
Figure 1: Theoretical model ... 13
Figure 2: Example of the task used in the experiment ... 20
IV
ACKNOWLEDGEMENTS
I would like to express special thanks to Lina Bernal Fuentes, who conducted the experiment together with me. Due to our great cooperation and exchange of ideas about the experiment design, I was ultimately able to collect the data for this thesis.
Furthermore, I would like to thank my supervisor, Siri Boe-Lillegraven, who guided me during the six months I spent writing this thesis and always provided valuable feedback. Her constant support during this time made the process of writing this thesis a truly great experience.
ABSTRACT
This study builds on goal-framing theory and applies its insights to the teamwork literature. The crucial concept is that of joint-production motivation, which can be seen as a new motivational explanation in the teamwork literature. The theoretical considerations that combine teamwork literature and goal-framing theory are tested in an experiment designed to find answers to the following research question: Under which circumstances are individuals who are part of a team disposed to contribute to a common goal rather than being motivated to pursue only their individual goal?
The experiment was compiled as a between-subject design and was conducted with 84 students of the University of Amsterdam. I predicted that proper situational cues lead to a normative goal frame. This in consequence should increase the joint-production motivation of the individual which should lead to a higher contribution to the team goal. The data does not empirically support the entire serial-mediation model. However, the data allows us to assume that the normative goal frame is related to higher joint-production motivation and to the higher contribution of individuals to the team goal. In addition, new insights are generated concerning the situational cues that lead to a normative goal frame within a team.
Together, this study paves a new avenue of research within the teamwork literature so as to extend our understanding of how individuals are motivated in teams towards common goals.
Key words: Goal-framing theory, teamwork, normative goal frame, joint-production motivation
1
INTRODUCTION
Several authors have brought out the increased relevance of knowledge and intellectual property for value creation within organizations (Ellemers, De Gilder, & Haslam 2004; Powell, Snellman 2004). The so-called knowledge economy therefore relies more intensively on individuals who make adaptive, intelligent, and interdependent work efforts in exchange with other individuals (Lindenberg & Foss, 2011) than on production of goods. Consequently, teamwork is also growing in popularity in organizations (e.g., Chen & Gogus, 2008; DeShon, Kozlowski, Schmidt, Milner, & Wiechmann, 2004; Van Knippenberg, 2000).
In academia, this development is apparent in the increasing number of papers which are published about teamwork (Mathieu, Hollenbeck, van Knippenberg & Ilgen, 2017; Park, Spitzmüller & DeShon, 2013). In addition, this calls for including insights from psychology and sociology to explain individuals’ behavior (e.g. Bridoux & Stoelhorst, 2016; Bryan & Vinchur, 2012; Lindenberg & Steg, 2014). In particular, enhancement in research is needed with regard to how the behaviors of individuals aggregate to a unity and how these ultimately contribute to the organizational goals (Barney & Felin, 2013).
Along with this necessity, this study investigates the motivation of individuals towards a team goal from the perspective of goal-framing theory. More precisely, the circumstances are examined under which individuals contribute to the team goal.
The first stream of literature considered in this study is the teamwork literature. It provides answers regarding the motivational process of individuals in teams, but the understanding of this process is not yet fully developed. That is because the team context—i.e., how team-level factors affect the motivation of individuals—is open to further investigations (Chen & Gogus, 2008; Chen, Sharma, Edinger, Shapiro, & Farh, 2011). In addition, within the teamwork literature, there is an implicit assumption that there is no conflict between an individual and a
team goal. However, this might not hold true for several reasons. Consequently, to understand the motivation of individuals in teams, a situation with conflicting goals needs to be considered. Another limitation of the teamwork literature is a lack of integration of the latest insights from psychology which indicate that motivational and cognitive processes are intertwined.
The second stream of literature considered in this study is that of goal-framing theory (Lindenberg & Foss, 2011; Lindenberg & Steg, 2014). Goal-framing theory is applied to teamwork and can offset the above-mentioned limitations of the teamwork literature with regard to the motivation of individuals in teams. First of all, goal-framing theory breaks with the assumption that individuals in teams always have a common goal (Foss & Lindenberg, 2012) and states instead that different goals are present at any time but that situational cues can push individual goals to the background and can make a common goal become the overarching one (Lindenberg & Foss, 2011). This theory also suggests that a normative goal-frame is especially capable of pushing an individual goal into the background and making the common goal focal. Second, goal-framing theory considers the team level for the explanation of individuals’ motivation, as the framing ultimately comprises the team environment. Third, goal-framing theory builds on the notion that motivational and cognitive processes are intertwined such that, by applying goal-framing to the teamwork literature, the latest insights from psychology are acknowledged and translated into the teamwork literature.
Having combined the two streams of literature, and having derived the observation that a normative goal frame is important to the decision of individuals who would contribute to the team goal from goal-framing theory, we must now ask how a normative goal frame can be established in a team. Situational cues that foster the normative goal frame in a team need to be identified and empirically tested.
3
and treatment groups in which the manipulation of situational cues was tested. The data collected during the experiment was analyzed in SPSS. The macro PROCESS by Hayes (2012) was used to conduct an ordinary least-squares regression analysis. The superordinate hypothesis—which proposed a complete serial mediation between the four variables of the theoretical model—could not be supported. In particular, the first sub-hypothesis which suggested and integrated tasks and team design in combination with signals of other team members towards a normative goal frame as the situational cues that positively influence the creation of a normative goal frame could not be statistically approved. However, the second and third sub-hypotheses that describe the second part of the serial mediation are supported. Therefore, we may infer that the normative goal frame is related to higher joint-production motivation and to a higher contribution of individuals towards a team goal. In other words, the mediation effect of the second part of the model holds regardless of the difference between the control and the treatment group.
This study contributes both the team work literature and the goal-framing theory. By applying the ideas of goal-framing theory to the teamwork literature, an alternate explanation of how people can be motivated to adopt team goals is presented. Though the results of the experiment do not support all of the hypotheses of this study, new insights with respect to the situational cues in teams that lead to a normative goal frame are generated, and the ideas pave a new avenue of research within the teamwork literature.
The main contribution made here to goal-framing theory is that the main constructs of both the normative goal-frame and joint-production motivation were tested empirically for the first time, which plays an essential role in academia.
LITERATURE REVIEW
Going back to the classic economic perspective, the basic assumption about individuals is that they possess full rationality, self-interest, and self-control (e.g. Eisenhardt, 1989; Osterloh & Frey, 2013). Therefore, individuals are assumed not to care about the collective but only about themselves. If individuals contribute to the collective goal, they do so only because the individual goal is identical to the collective goal (Alchian & Demsetz, 1972; Foss & Lindenberg, 2012). This involves the perception that the individual and the collective goal need to be aligned and optimally congruent. Hence, incentives are used to direct the behavior of the individual to the best outcome for the collective.
However, this is a narrow perspective of human behavior. It oversimplifies our perception of humans and excludes potentially new insights which might lead to more realistic explanation of human behavior in organizations and teams.
To take a step forward to a more realistic view of individuals, many authors in the field of strategy and management have started to include insights from psychology and sociology to explain individuals’ behavior (e.g., Bridoux & Stoelhorst, 2016; Bryan & Vinchur, 2012; Lindenberg & Steg, 2014). The one axiom of their theories is that human behavior is an outcome of motivation and cognition (Locke, 2000). The following literature review shows that the concepts of motivation and cognition are delineated in both the teamwork literature and goal-framing theory. The limitations in the team work literature with regard to motivational and cognitive processes are presented in what follows. Goal-framing theory is described as one way to enhance our understanding of those limitations.
Teamwork literature
The one framework that guides the literature on team effectiveness is the framework of input, process, and outcome known as the IPO framework. It was first described by McGrath (1964)
5
and Hackman and Morris (1975). More recent papers by Chen and Gogus (2008), Kozlowski (2017), and Mathieu, Maynard, Rapp, and Gilson (2008) summarize several aspects of the model. Inputs are seen as individual-, team-, and organizational-level resources that can be variously combined. Research includes, for example, members’ characteristics, team design, and leadership characteristics. Processes are all kinds of interactions that occur in a team. Communication and coordination between members are parts of the processes and the strategy formulation. In addition to these dynamic processes, emergent states are highlighted by the the literature related to the middle part of the IPO framework. Outputs in this framework are considered as performance or attitudes and are studied on individual, team, and organizational levels.
Within the teamwork literature, cognition and motivation—identified as the core team processes by Kozlowski (2017)—are aspects of the middle part in the IPO model, the processes. Hence, this study concentrates on this part of the teamwork literature.
Motivation is commonly used as an umbrella term to explain observable changes in peoples’ actions (Gollwitzer & Oettingen, 2012; Kanfer, Frese, & Johnson, 2017). The factors that energize, direct and make people persistent in putting effort into one direction over time have been identified as the most relevant to those changes (Chen & Kanfer, 2006; Ellemers et al., 2004). These factors have also been experimentally researched (Mathieu et al., 2008), and examples of extensive research can be found within the two streams that have been established to explain individual motivational processes. This is, on the one hand, goal-setting theory, which combines the factors of direction and persistence (Ellemers et al., 2004) by focusing on the evaluation and the selection of different goals (Locke & Latham, 1990). Goal setting implies that specific and high goals lead to better task performance than easy and non-specific goals do (Locke & Latham, 2006).
implemented effectively (Gollwitzer, 1999). The ultimate goal attainment depends, besides the goal setting of the first step, on self-regulation (Gollwitzer & Oettingen, 2012).
Besides these general models of motivation, self-determination theory counts as one of the most frequently used models of motivation with respect to work organizations. It has evolved from research on intrinsic and extrinsic motivation (Deci, Olafsen, & Ryan, 2017). Intrinsic motivation refers to the motivation derived from interest and enjoyment in the activity itself. Intrinsic motivation is often seen as more valuable because performance is expected to be higher given intrinsic motivation that it is expected to be given extrinsic motivation. Extrinsic motivation is motivation based on separable tangible or intangible consequences.
The idea to translate the above-explained concepts regarding individual motivation to the team-motivation process is rather new. One of the first approaches to explaining individual motivation within teams is the social-identity approach (Ellemers et al., 2004). The main rationale behind this approach is that people who feel oneness with a team also experience the group goals as their own (Van Knippenberg, 2000). The first explicitly named integration of the concepts of motivation and teamwork was proposed by Chen and Kanfer (2006), who introduce the multilevel model of motivation. This model is based on the ideas of DeShon et al. (2004) concerning the mutual effects of the individual and the team level. Chen and Kanfer (2006) state that motivational processes are functionally similar on both levels and, therefore, that goal-setting and goal-striving theories can be applied on a team level. The multilevel model has been empirically tested by Chen, Kanfer, DeShon, Mathieu, and Kozlowski (2009), which advanced our overall understanding of the similarities between motivational processes at different levels (Park et al., 2013).
Despite these recent enhancements to our understanding of motivational processes at different levels, the literature on teamwork still lacks an extensive explanation of motivation of individuals in teams that considers the team context. As stated by Chen and Gogus (2008), the
7
main focus is to support the generalizability of the motivational processes of individuals to the team level. However, the team context, that is, how team-level factors affect the motivation of individuals, and the question whether the generalizability of motivation is more legitimate in certain situations is still open (Chen & Gogus, 2008; Chen et al., 2011). Hence, the team context is not adequately considered as a factor for individual motivation for a team goal.
In addition to this first limitation, the teamwork literature does not explicitly distinguish between individual and collective goals. Though some authors in the field of teamwork (e.g., Bornstein, 2003; DeShon et al., 2004) recognize that individual and collective goals might contradict each other, the general assumption is one common goal within teams. This is exemplified by the definition of teams. It indicates that socially interacting individuals share one or more common goals and interact to achieve these common and valued goals (Kozlowski & Ilgen, 2006; Mathieu et al., 2017; Salas, Dickinson, Converse & Tannenbaum, 1992).
Besides these motivational aspects of human behavior in teams, the complexity of human behavior has been addressed within this stream of literature since the beginning of the twenty-first century by including consideration of cognitive aspects of teamwork (Mathieu et al., 2017).
The two cognitive constructs that are generally explored in a team are shared mental models and transactive memory systems (DeChurch & Mesmer-Magnus, 2010; Grand, Braun, Kuljanin, Kozlowski, & Chao, 2016). The first implies that knowledge is commonly held in teams whereas the latter claims that knowledge is distributed among team members (DeChurch & Mesmer-Magnus, 2010; Fernandez, Shah, Rosenman, Kozlowski, Parker, & Grand, 2017; Kozlowski & Ilgen, 2006). DeChurch and Memsmer-Magnus (2010) provide meta-analytical evidence for these cognitive constructs. But what is not provided in the
literature is a combination of the motivational and the cognitive constructs. In the team context, motivation and cognition are at large seen as different processes leading to human actions. For example, by developing the social-identity approach that causes individuals to take the group perspective, Ellemers et al. (2004) focus only on the motivational aspect of teamwork. As Lindenberg and Foss (2011) state, the individuals are interchangeable in this setting, and the collaborative activities are not based on a cognitive process.
Consequently, these theories cover the aspects of motivation and cognition but so far do not fully include the latest insights from cognitive science, behavioral economics, and social psychology (Foss & Lindenberg, 2013): i.e., that cognition and motivation are intertwined such that cognitive processes influence motivation and vice versa (Kruglanski & Köpetz, 2009).
A different theory that does include both cognitive and motivational aspects and hence reacts to the need to include sufficient insights from cognitive psychology into organizational theory (Powell, Lovallo, & Fox, 2011) is the goal-framing theory.
Goal-framing theory
Goal-framing theory builds on insights from Bacharach (2006), who was one great thinker in decision and game theories. His insights include the finding that members of a team can be motivated to achieve a common goal when they perceive a so-called We frame. But, according to Bacharach (2006) this motivation does not exist in I frames. The logic is justified by evolutionary anthropology. Therein, human beings are seen as capable of doing group work because they are equipped with cognitive and motivational faculties for this purpose (Tomasello, Carpenter, Call, Behne, & Moll, 2005). For example, people are able to cooperate within groups and also between groups of people (Cordes, Richerson, McElreath, & Strimling, 2008).
9
A proof of the phenomenon of framing can also be found in research conducted by Liberman, Samuels and Ross (2004). They find, in two large experiments, that cues in the environment in the form of situation labeling define whether an individual cooperates or defects. A situation can either be labeled as a Wall Street game or as a community game which impacted the participants’ decision to cooperate or defect. A similar study was conducted by Pillutla and Chen (1999). They found that people behave more competitively in situations that include an economic decision compared to situations that include a noneconomic decision.
Goal-framing theory adopts this view by saying that an overarching goal frames the situation such that different cognitive and motivational processes are stimulated; hence, those goals influence human behavior. The stimulation includes the fact that people differ in what they attend to, which alternatives they consider, and what knowledge they draw on (Foss & Lindenberg, 2013). In other words, people are selecting different inputs for the cognitive processes (Lindenberg, 2013). The overarching goal needs to be activated by situational cues and is focal, if this activation has happened. During these selective cognitive processes, individuals may or may not adopt team goals. The one important assumption is that overarching goals always exist but can be pushed into the background by the situational cues that make other goals focal. (Foss & Lindenberg, 2013).
There are three overarching goals according to Lindenberg and Foss (2011). A hedonic goal responds to one’s personal need and is related to how one feels right now. It is considered to be the strongest of all three goals because of the direct connection to one’s need satisfaction. The gain goal is related to one’s desire to improve. Especially with respect to resources and is therefore often a monetary entity. The gain goal is also a strong goal, but it can be pushed into the background by a hedonic goal. The third goal and the weakest goal is the normative. It expresses the desire to act appropriately in a group and is oriented to what is important for achieving collective goals. Though it is the weakest, the normative goal is the most important
goal with regard to teamwork.
Lindenberg and Foss (2011) suggest several antecedents of a normative goal frame in their theoretical overview of the relationship between normative goal framing and joint-production motivation. They focus especially on structural and governance features that interplay with the normative goal frames of individuals, and they argue that these improve joint-production motivation. They emphasize that members need to feel the interdependence in terms of joint-production motivation. The second and direct support for normative goal frames stems, according to Lindenberg and Foss (2011), from cognitive and symbolic management. The third antecedent for a normative goal frame is the indirect support it receives from individual hedonic and gain frames regarding the reward structure. And the last antecedent in their theoretical approach is an authority structure which is supportive of normative goal frames. However, these suggested relations have not yet been empirically tested yet (Lindenberg & Foss, 2011). To enhance future knowledge in joint-production motivation, an important next step, however, is to test these relations.
Conjunction of the two streams of literature and research question
Having explained both streams of literature, three conjunctions can be derived. First of all, goal-framing theory breaks with the assumption that individuals in teams always have a common goal (Foss & Lindenberg, 2012). This brings new possibilities regarding new ways of reasoning within the teamwork literature, which focuses on teams in which socially interacting individuals share one or more common goals and interact toward this common and valued goal (Kozlowski & Ilgen, 2006; Mathieu et al., 2017; Salas et al., 1992). Goal-framing theory, in contrast, states that different goals are present at any time and that situational cues can push individual goals into the background and can make common goals into one overarching one (Lindenberg & Foss, 2011). This theory also suggests that a normative goal-frame is especially able to push an individual goal into the background and make the common
11
goal focal. This leads to the second conjunction, that the goal-framing theory considers the team level for its explanation of individual motivation. The normative goal-frame reflects the desire to act appropriately in a group and is oriented to what is important for achieving collective goals (Lindenberg & Foss, 2011) and therefore comprises the team environment. Third, goal-framing theory builds on the notion that motivational and cognitive processes are intertwined; hence, by applying goal-framing to the teamwork literature, these latest insights from psychology are acknowledged and translated into the teamwork literature.
Combining those two streams of literature, and deriving from goal-framing theory the conclusion that the normative goal frame is important to the decision of individuals to contribute to a team goal, we must now ask how a normative goal frame can be established in a team. In other words, this is expressed by the research questions. Under which circumstances are individuals who are parts of a team disposed to contribute to a common goal rather than being motivated to pursue only their individual goals? In the following part of this study, the theoretical considerations of goal-framing theory are applied to identify the situational cues that foster the normative goal-frame in teams. Furthermore, these predictions are empirically tested, which has not been done so far (Lindenberg & Foss, 2011).
HYPOTHESIS DEVELOPMENT
According to Foss and Lindenberg (2013), the best condition for an organization occurs when individuals see themselves in a normative goal frame. This constitutes a situation in which the gain and hedonic frames that mostly incentivize individual behavior are less relevant such that collective goals are in the foreground. This leads to higher value creation both because individual incentives are less relevant and are not needed in the form of monetary rewards and because the performance of individuals is higher as they engage in adaptive efforts and exert intelligent effort (Foss & Lindenberg, 2013).
This essential idea of goal-framing theory is applied to the team level in the following chapter. Furthermore, to empirically test the predictions of goal-framing theory in the team context and find answers to our question—i.e., under which circumstances individuals are disposed to contribute to a team goal—a theoretical framework and testable hypotheses are developed in the following part.
The proposed association between the variables is shown in Figure 1. Besides the question of whether a causal relation between the independent and dependent variable exists, the question of how the underlying process is also of high interest. The underlying mechanism is presented by two variables: namely, normative goal frame and joint-production motivation (Lindenberg & Foss, 2011). Because these variables relate to the underlying psychological process that causally links the situational cues and the contribution to the team goal, they are treated as the mediating variables in the model (Hayes, 2013).
The expected connection between the variables is that the proper situational cues lead to a normative goal frame. This in consequence increases the joint-production motivation of the individual and, lastly, joint-production motivation leads to higher performance. This depiction also emphasizes the connection of the cognitive and motivational parts of psychological
13
processes. It shows that these processes are intertwined because both are ultimately needed to encourage individuals to contribute to the team goal. This multi-staged, underling psychological process is formulated as the superordinate hypothesis.
Hypothesis 1. Individuals in a team are disposed to contribute to the team goal if particular situational cues lead to a normative goal frame and this frame in turn leads to higher joint-production motivation.
To capture in greater detail how this complete process works, three essential questions need to be answered. The first concerns how specific situational cues can create a normative goal frame. The second concerns how a normative goal frame leads to higher joint-production motivation. The third is this: How does joint-production motivation lead to the decision to contribute to the team goal? A theoretical explanation of these questions is given in the following paragraphs. The three processes also reflect the three sub-hypotheses.
The first part—how the normative goal frame is established and supported—was described by Lindenberg and Foss (2011) on an organizational level by means of structural and governance features. To translate these ideas to a team level, the antecedents of a normative goal frame described above were modified to operate as attributes of the independent variables that set
the normative goal frame. To feel the interdependence of the individual’s own task and the team goal, there must be an integrated task and team design (Foss & Lindenberg, 2011). This means that the individuals need to understand what other members are doing and how the individual contributions converge in the end to the outcome and accordingly to the team goal. This perceived interdependence is actually very similar at both the organization and the team level. On the team level, however, interdependence is easier to generate, as the size of a team is smaller than that of most organizations.
In addition to this precondition of a normative goal frame, there must be clear signals from other team members that they are behaving in a normative goal frame (Steg, Lindenberg & Keizer, 2015). This support of the normative goal frame is described by Lindenberg and Foss (2011) on the organizational level as the support of the normative goal frame via visions or missions. These are spread throughout the organization at the management level.
Lindenberg and Foss (2011) also include the two other antecedents mentioned above: the indirect support of an aligned reward system and the authority structure that both contribute to a normative goal frame at the organizational level. But, these two factors are meant to maintain the normative goal frame and are not relevant to its initiation. Furthermore, the authority structure is not especially relevant to the teamwork context because the authority structure is usually not fully integrated in one team. Monetary reward could be applied to some teams; however, in this thesis, it is not considered as an active antecedent to simplify the model. For these reasons, the two factors are not included in the framework and the experiment.
In addition, an integral part of goal-framing theory, and the first part of the theoretical model is that the hedonic, the gain, and the normative goals all exist at the same time. However, the situational cues have to push the normative goal frame into the foreground. This integral part is directly linked to another important fact: that goal-framing is really situational. This means
15
that no change of preferences is taking place but that everything depends on the situational cues in the moment of action (Lindenberg, 2006). Consequently, if individuals do not perceive a normative frame, they do not act upon it but act within the hedonic or gain frame. Situational cues are especially needed, because the normative goal frame is the weakest according to goal-framing theory, and it does not prevail without sufficient initiation and support. This leads to the following hypothesis.
Hypothesis 1a. The situational cues that positively influence the creation of a normative goal frame are an integrated tasks and team design in combination with signals from other team members with respect to a normative goal frame.
The second part of the theoretical framework concerns the relation between a normative goal frame and joint-production motivation. This is the integral part in which the cognitive and the motivational processes interact. Individuals need to perceive the normative frame and activate specific knowledge or be reminded of specific norms. By perceiving this more actively, one can become more motivated to contribute to the team goal.
Hypothesis 1b. The perception of a normative goal frame is positively linked to the higher joint-production motivation of the individuals.
The third part in the model explains how joint-production motivation ultimately leads to a higher contribution to the team goal, which is directly linked to higher performance. The important underlying construct in this relation is joint production, which can be any activity that involves heterogeneous but complementary resources. Furthermore, joint production involves a high degree of task and outcome interdependence (Lindenberg & Foss, 2011). Motivation is a term that expresses the purpose and reasons for an underlying decision process and goal selection (Kanfer & Chen, 2016). In combination, the joint-production motivation is an expression for the human capacity to engage actively in collaborative activities (Lindenberg & Foss, 2011).
Lindenberg and Foss (2011) explain the higher performance on an organizational level by citing a few factors. For example, they state that joint-production motivation influences the decision of which task people are willing to do and how much effort they put into it. In addition, the joint-production motivation influences how people coordinate their actions and how much they engage in prosocial behavior such as knowledge sharing.
Alongside the factors that increase productivity, motivation and coordination costs decrease as joint-production motivation suspends opportunism and thereby lowers the cost of control. In addition, if people achieve joint goals, the need for planning and formalization is lower. (Lindenberg & Foss, 2011)
On a team level, all those factors also hold, and the overall joint-production motivation implies a willingness to put more effort into the team task and give up individual goals.
Hypothesis 1c. Individuals with higher joint-production motivation contribute to the common goal to a higher degree and disregard their individual goals.
17
EXPERIMENT DESIGN
Operationalization
To test the hypotheses stated in the theoretical part above, an experiment was conducted, as this is an excellent way to find answers to questions of causality (Bono & McNamara, 2011). The experiment has a between-subject design. A between-subject design offers the advantage that there is no carryover or learning effect compared to a within-subject design. Participants are assigned to only one condition in the experiment and are therefore not affected by anything from another condition of the experiment (Bausell, 2015). A disadvantage of between-subject design is in the differences between the participants caused by the natural variation of individuals. However, this is assumed to be low when the number of people is large enough and the participants are assigned randomly to groups. Nonetheless, control variables were included to control for potential differences in the participants’ interest, skills, and abilities. The following chapter describes how the the theoretical model described above was operationalized and how the experiment was conducted.
Sample
All participants (N=84) were students of the University of Amsterdam who were enrolled in the Master’s of Business Administration Program. The advantage of having a student sample is that it is possible to conduct an experiment with a large number of people, which is needed for any experiment examining behavior in groups. In addition, students enrolled in business-administration courses comprise a representative sample of people who later on work in organizations.
However, there are also some disadvantages that should be mentioned. For example, the teams of students are not professional work teams in a real-life situation in an organization—a
fact that decreases the external validity of the experiment. In addition, the random assignment is not completely fulfilled, as the participants were chosen by their status as students.
Setting
The experiment was conducted within two weeks with different lecture and tutorial groups at the University of Amsterdam. The main reason for this was to have a group of people in one room at the same time so that we would not be dependent on people signing up for the experiment and coming to a lab at a specific time. To minimize possible differences between the different groups tested, the rooms for the experiment were chosen very carefully in terms of size and seating. For example, we considered the possibility that the teams would sit closely together and would not sit only in one row.
Teams
For each tutorial group, it was random which experimental condition was tested. Overall, two experimental conditions were used to run the experiment. In both groups, the participants were randomly assigned to teams of five people.
The size of the team has been examined previously by many scholars. Several studies indicate that team size influences team effectiveness. For example, in large teams, coordination and motivation problems occur, and conflicts within the groups may arise (Hackman, 2002; Levine & Moreland, 1998; Staats, Milkman, & Fox, 2012). However, Hoegl (2005) points out that optimal team size ultimately depends on the task the team is performing. Speaking of the experiment, the task actually did not require participants to collaborate or coordinate because they were not performing one task together and were not asked to make decisions together. They were only working towards one common goal, and the question was whether they individually decided to contribute to the team goal or not. Consequently, the team size was not the most relevant, but the focus was on creating a team context. In the experiment,
19
therefore, teams of five people were created. In less than 20% of the groups, there were also four or six people in one team.
Task
The task was chosen so as to meet different requirements for this experiment. First, the task must be time consuming so that performing it is in the end associated with more time contributed to the team goal. This goes together with the second characteristic: the task has to be mentally challenging and must require the participants to forth some effort. Third, the task must not be too fun. This eliminates the possibility that people are only putting more effort into the team goal because they like it. Fourth, the task is designed so that there is no learning effect from solving more calculations. Hence, no long instructions were needed in the beginning that can be seen as an upfront investment. The last criterion to be met is that the participants do not require previous knowledge to perform the task.
The calculation of sums meets all these criteria. In the experiment, the task was presented to the students as shown in Figure 1. Numbers from 1 to 99 were given in two matrices, and the sum of the two matrices had to be calculated by the participants. A further advantage of this task is that participants are unlikely to successfully guess the right answer. In addition, time could be saved at the beginning of the experiment, as no long instructions were needed to explaining the task. Nevertheless, there are some disadvantages to this task. First of all, individuals have different calculation skills. This implies that some people need more time to build the sums and others less. As ability is often connected to the degree someone likes the task, one question in the questionnaire was included to control for this difference in the data analysis (i.e., “I liked doing the calculations”).
Figure 2: Example of the task used in the experiment
Goals
In the experiment, the participants had two predetermined goals that were conflicting. Hence, they had to decide between the individual and the team goal. The individual goal for each participant was to complete one matrix. The team goal for the five people was to collectively complete 15 matrices. Divided by the five participants, this means that, by completing three matrices per person, the team goal could be reached. If a participant decided to contribute to the team goal by completing three matrices, they could not reach their individual goal.
The fact that the two goals conflicted—that only the team goal and not automatically the individual goal could be reached by completing three matrices—was explicitly mentioned several times during the instruction. In addition, the slide that was presented during the experiment clearly stated the difference.
To control for this problem in the data, one question was included that asked about the instruction and whether the difference between the individual and the collective goal was clear (i.e., “The instructions were clear to me and I was aware of the difference between the individual goal and the team goal”).
21 General remarks
The instructions for both groups also covered additional aspects that were included based on the questions the participants of the pilot studies asked. Specifically, no competition was raised in any of the experimental groups, as we did not offer any prize for the best or fastest group. Also, no time limit was set so as to avoid the possibility that participants would not contribute to the team goal because of time restrictions.
A €30 Amazon voucher was raffled between all the participants of the experiment and the pilot studies. To take part, the participants could add their e-mail addresses on the consent sheet.
Procedure treatment group
The experiment started with basic instructions that were given to all participants collectively. After this, the participants were allowed to ask questions to ensure that the procedure of the experiment was clear to everyone. After the instructions, all participants signed a consent sheet.
The students had to solve the task described above on their own. Each student received a sheet of paper with three printed matrices and, as explained above, they were clearly told that there is a conflict between the two goals and that they can either contribute to the individual goal by completing one matrix or contribute to the team goal by completing three matrices.
The participants were not limited in time for the completion of the task. The intention was not to put time pressure on the participants that could influence their decision to complete one or three matrices.
As the manipulation of the independent variable is essential in an experiment, the two situational cues were manipulated in the treatment group.
The first situational cue, the integrated task and team design, was manipulated in the treatment group by sharing the outcome within the team. The calculations solved by each member were collected within the team at the end of the experiment. Already during instruction, they were told that they must make one pile of the final results. In the control group, the participants put their solved calculations into an envelope and did not share the number of solved matrices. Consequently, in the treatment group, each member could track who in the team finished how many matrices. This refers to the perceived interdependence described by Lindenberg and Foss (2011). People should now feel more interdependent due to the increased transparency of the accomplishment of each team member.
The second situational cue, peer signaling, was manipulated in the treatment group by allowing participants to talk to each other throughout the experiment. Talking allows participants to share information and to signal their motivations to other team members. The signals work like a contagion with respect to the other members of the team. They are reminded of the social norm to act appropriately (Lindenberg & Foss, 2011).
Consequently, in the treatment group, participants should overall perceive the normative goal as overarching, which favors appropriate action in this group and hence makes them contribute to the team goal.
Procedure control group
The second group is the control group. It is a no-treatment control group because, in this group, no frame was intentionally established. A control group was chosen, as this resembles the organization in the best way. Aware, that people might choose any frame, this is also the case in organizations when people are told to perform a task in a team.
The setting was the exactly the same in both the control and treatment groups. Also, the procedure was the same. However, as mentioned above, in the control group, participants
23
were not allowed to talk. In addition, they were not obliged to share the number of matrices solved with the team because they had envelopes in which the sheet of paper was put. Already during the instructions, they were told to do this, and it was emphasized that they must not agree on anything or share the number of matrices each team member wants to solve. After completing the calculations, the participants in the control group filled in the same questionnaire as the participants in the treatment group.
Pilot work
The final experiment design described above was the result of an iterative process. Previously to the data collection, two pilot studies were conducted to test the design of the experiment. Those pilot studies provided important insights and led to several modifications of the initial experiment design. The process and the modifications are shortly described in the following paragraph.
The first pilot study was conducted with four students and was a test of the control group. Participants were allowed to talk in the control group and the initial concerns were affirmed that participants create the normative goal frame on their own by signaling their motivation and reminding each other to the appropriateness of contributing to the team goal. This led to the change of the manipulation of the peer signaling as described above. The initial idea was to allow talking in both groups and manipulate peer signaling by bringing additional briefed participants to the experiment. Those would have been told to clearly signal their joint-production motivation to contribute to the team goal. However, due to the experience gathered in the first pilot study, this idea was discarded. Consequently, this change also made the design more feasible, because bringing people who signal motivation would have limited the scope for data collection. Both for the treatment and for control groups, the recruiting of additional participants would have been necessary to control for the effects of bringing people
to the experiment. A measurement for peer signaling was then included to compensate for the active peer signaling of additional people.
The second pilot study was conducted with five participants. Again, the control group was tested so as to be able to test the adjustments after the first pilot study. Afterwards, some additional details were changed in the design. For example, the way of surrendering the task was changed. In the first stage, separate question cards were handed out to the students. However, this made it very obvious in the control group how many matrices a participant had actually completed. After the second pilot study, this was changed, and the matrices in the experiment were then handed out on one sheet of paper. Consequently, the participants did not see whether a team member was calculating the first matrix or a second or third. Regardless, in the final experiment, individuals did not have a private work space.
Introduction of the observational variable
During the experiment, the observation was made that participants started to agree on not contributing to the team goal. This was contrary to the pilot study, which clearly indicated that talking induces participants to agree on the team goal. Because of the initial weakness— i.e., that it is not possible during the experiment to actively manage peer signaling—one question was included that could be used to measure the signaling. However, the negative peer signaling was not foreseen, and the question for the measurement was not appropriate in this case. Consequently, an additional variable was also included that is based on the observations made in the treatment group. In the control group, this observation was not included as talking was not allowed; therefore, neither positive nor negative peer signaling could take place. Because the overall group size was quite small in each group for data collection, it was possible to observe each team and see what the participants communicated to each other. This additional observational variable is used for the post hoc analysis.
25 Measurement
The measurement procedure was exactly the same in both the control and treatment groups. Upon completing the calculations, all participants filled in a questionnaire that was comprised of questions regarding the scale with which to measure the normative goal frame and joint-production motivation. Further questions were included for the control variables. In addition, the questionnaire comprised a second part that was added for another thesis by Lina Bernal Fuentes, for which data collection was undertaken together with this thesis. However, the influence of this measurement on the one used for this thesis can be assumed to be low, as the questions were asked afterwards. The measures of all variables are explained in more detail in the following paragraph. Furthermore, the descriptive statistics are included to report on the reliability of the measurement scales.
Independent variable: Situational cues
The independent variable was manipulated in the treatment group; therefore, no measurement scale is needed. The dichotomous independent variable reflects the two experimental conditions.
Mediator 1: Normative goal frame
The scale used to measure the normative goal-frame (M = 2.38, SD = 0.92, α = 0.68) was newly developed for the experiment. The main reason for this is that no scale of the normative goal-frame can be found in the literature. To come up with the items for the scale, a theory-driven approach (as described by Kline 2005) was applied. According to this approach, the construct needs to be determined first. For the construct underlying the normative goal-frame, goal-framing theory was used. The second step includes coming up with the items that assess each aspect of the construct. In this step, five items were created. All of the items of the scale are exclusively declarative statements, as advised in the literature (e.g., Kline, 2005). The first
two items ask for the perceived task interdependence (“I felt that my effort was needed to achieve the team goal” and “What I was doing was visible to the team”). Two other items ask for the team signaling (“Someone in the team showed special interest in completing the team task” and “Overall I perceived that the other team members were motivated to achieve the team goal”). One further question about the overall assessment of the situation rounds up the scale (“I felt that it is appropriate to contribute to the team goal”).
The items were assessed in the questionnaire by the participants on a scale from 1 (strongly disagree) to 5 (strongly agree).
Though it is more common and more often advised to use a rational model that combines theoretical and empirical work so as define items for a measurement scale (e.g., Kline, 2005), a theory-driven approach was used, mainly because the advantage of this approach is that no time-intensive testing of the items is needed. With regard to the scope of the project, this preponderated the disadvantages of the theory-driven approach that the reliability and validity of the scale is not empirically confirmed before the final data collection.
Mediator 2: Joint-production motivation
The measurement of the mediating variable, joint-production motivation (M = 2.79, SD = 1.13, α = 0.92), was based on the multidimensional work-motivation scale developed by Gagné et al. (2014). This scale fits best with the conducted experiment and to measure motivation towards a team goal because it assesses motivation at a general level and is not specific on a task level like the scale by Fernet, Senécal, Guay, Marsh, and Dowson (2008) for example does.
The multidimensional work-motivation scale was widely tested and shows high factorial validity in several languages, but it was specifically designed for work environments. Therefore, some items on the scale were adjusted. For example, the questions were modified
27
to the past tense because the students were asked about their motivation in hindsight. In addition, the questions were applied to the team level and phrased towards the team goal. As part of this, the superordinate question to answer the items was changed from “Why would you put effort in the current job?” to “Why did you put effort into the team goal?”. It is important to mention, that only the question regarding motivation with respect to the team goal was asked independent of whether participants contributed to the team goal or not. Consequently, people who did not contribute to the team goal were expected to fill in the questions with a low score.
Overall, the scale comprises 18 items distributed on six sub-scales that reflect different types of motivation. Questions were asked, for example, about different aspects of the participants’ intrinsic and extrinsic motivations. For the total score of joint-production motivation all measures were used in combination, because theory predicts that all the aspects (of intrinsic and extrinsic motivation) exist at the same time. It is not assumed that one motivation is overruled by another but that all aspects need to be aligned and can support each other (Lindenberg, 2001). Hence, joint-production motivation can be seen as a product of all of the different kinds of motivation that direct to the team goal.
Compared to the original scale of Gagné et al. (2014), one aspect of motivation was left out. This is the aspect regarding monetary incentives, as this did not apply to the experiment. To further make the scale applicable to the experiment, one more aspect that reflects the normative frame was included. The complete scale used in the experiment is presented in the appendix.
Dependent variable: Contribution to the team goal
The sheets of paper with the calculations were collected at the end of the experiment and used to measure the dependent variable. The performance was coded in three different levels; more precisely, it was coded in a scale from one (for one matrix) to three (for three matrices). The
number of matrices reflects whether participants contributed to the team goal or not, as the instructions indicate that one matrix equals the individual goal and that three matrices equal the team goal. In addition, the questionnaire included a question concerning whether participants contributed to the team goal or not (a yes or no question). The mean of the calculations solved is M = 1.8; the standard deviation SD = 0.94.
Control Variables
To allow control for several factors that could possibly influence the effect sizes, additional items were added in the questionnaire. General information concerning age, gender, and study program was requested in the questionnaire. Furthermore, two measures were included so as to control for the possible influences previously mentioned via the task and by the clarity of the instructions. Two declarative statements – “I liked doing the calculations” and “The instructions were clear to me and I was aware of the difference between the individual goal and the team goal” – were presented, and participants indicated their level of agreement from 1 (strongly disagree) to 5 (strongly agree).
29
DATA ANALYSIS
Preliminary analysis
The data collected during the experiment was analyzed in SPSS. 84 people participated in the experiment, of which 45 were assigned to the control group and 39 to the treatment group. Slightly more women (48) participated in the experiment than men (34). The age of the participants ranges from 21 to 31 with a participant’s average age of 24. All participants were master’s students in business administration at the University of Amsterdam, of which 80% also did a bachelor program in business administration and/or economics.
Before testing the stated hypothesis, a detailed screening of the data was conducted. The results of the data screening are presented in the following section.
Frequencies indicated no pattern in the missing data. For 12 items of the whole questionnaire, one person did not fill in a value. For three other items, two people respectively did not fill in a value. Consequently, all missing data points are at random, and those were deleted pairwise for the correlation matrix and listwise for all other analyses.
Analysis normative goal frame measure
The Cronbach’s alpha for the total scale normative goal-frame indicates an acceptable level of overall reliability (α = 0.68). This includes all five items, and the corrected item-total correlations indicate that four of the five items exhibit good correlation with the total score of the scale (above 0.30). The corrected item-total correlation for the item, “What I was doing was visible to the team” is 0.29. However, the difference in the Cronbach’s alpha if the item is deleted is below 0.10 and is therefore not deleted. Of all other items, also none would substantially affect reliability if they were deleted (Δ<.10).
The univariate normality for the distribution of answers to the scale normative goal-frame was tested via skewness and kurtosis. Normality for the variable can be assumed, as zero lies within the boundaries of the 95% confidence interval. The same result was obtained with the histogram of the frequency distribution of the variable.
Analysis joint-production motivation measure
Before calculating the reliability of the scale, the three negatively coded items that represent amotivation were recoded into new variables. For all sub-scales, the Cronbach’s alphas were calculated; and for all of the six sub-scales, the values show excellent reliability. For further analysis, all scores obtained from all sub-scales were combined by calculating the means. As suggested by Gagnè et al. (2014), this is a legitimate option, and it was done to make further analyses easier. In addition, all the sub-scales were combined (as means) because goal-framing theory suggests, that all motivational parts are combined and that they do not need to be balanced.
The Cronbach’s alpha for the total scale indicates an excellence level of overall reliability (α = 0.92). This includes all 18 items, and the corrected item-total correlations indicate that all
31
items exhibit good correlation with the total score of the scale (all above 0.30). In addition, none of the items would substantially affect reliability if they were deleted.
The univariate normality for the distribution of answers to the joint-production motivation scale was tested via skewness and kurtosis. Normality for this variable can be assumed, as zero lies within the boundaries of the 95% confidence interval. The same result was obtained with the histogram of frequency distribution of the variable.
For the six components of the joint-production motivation scale a principal-axis factorial analysis (PAF) was conducted. The Kaiser-Meyer-Olkin (KMO) measure verifies the sampling adequacy for the analysis, with KMO = 0.829. On the basis of Bartlett’s test of sphericity χ² (153) = 1059,658, p < 0.000, the correlation between the items is sufficiently large for the PAF analysis. An initial analysis was run to obtain eigenvalues for all items of the scale. Five items had eigenvalues above the Kaiser’s criterion of 1. Combined, those items explain 79.29 % of the variance. In addition, the scree plot reveales a levelling off after the fifth item in agreement with the Kaiser’s criterion. Thus, a rotation of these factors was included with Kaiser normalization. The factor loadings after rotation are shown in Table 2. After rotation, three factors still do not show clear loading to one factor (“To avoid being criticized by others”, “Because I have to prove to myself that I can.”, and “Because it makes me feel proud of myself.”). However, this can be explained by indistinct concepts of introjected and identified motivation (Deci et al., 2017; Gagnè et al., 2014;). However, for further analysis, the unclear factor loading for some sub-scales is not important, as all scales are added up.
Table 2: Rotated factor loadings components total motivation
Regression analysis with PROCESS
Construct level
To test the hypothesized model, the PROCESS macro for SPSS was used (Hayes, 2012). This macro simplifies the procedure required to obtain ordinary least-squares regression analyses.
33
For the proposed multiple-mediation model, the template named Model 6 fits ideally and was applied. With the macro, the total, direct, indirect, and all path coefficients for the model are generated in only few steps. A further advantage is that also the corresponding inferential tests are immediately included by the bias-corrected bootstrap confidence intervals. Hereby, the advantage of bootstrapping is that no assumptions about normality need to be met. In the analysis, 5000 bootstrap resampling was used to obtain 95% bias-corrected confidence intervals (Hayes, 2012).
The regression analysis predicts the dependent variable based on multiple independent variables. It allows us to quantify the intensity and the meaning of the relation between the variables, and thereby also predicts the direction of the relation.
Besides the four variables assembling the model, four control variables were included that are considered to be covariates in PROCESS. These are included to statistically account for interrelations between variables in the final model based on other variables than those of the construct (Hayes, 2012). Two control variables are age and gender; one other control variable, checks whether participants liked the task (TaskCh). And the fourth control variable checks whether participants understood the procedure and were aware of the difference between the individual and collective goals (InstCh1).
The coding of the dichotomous independent variable is also important to the interpretation of the result. (For a similar study using a dichotomous independent variable with Model 6 in PROCESS see Casciano and Massey, 2012). The independent variable represents the manipulation in the treatment group and hence reflects the difference between the control and the treatment groups. The control group is labeled with a 1, and the treatment group with a 2. Consequently, positive associations with a dependent variable are associated with the treatment group.
RESULTS
The experiment was conducted to examine the effect of different situational cues on participants’ decision to contribute to the team goal or not. Two mediating factors, the normative goal frame and joint-production motivation were considered to explain the underlying psychological process. The results of the experiment are presented in the following chapter and put in relation to the hypothesis.
The total effect in the model consists of the direct effect and the sum of three indirect effects. The direct effect is the one which does not pass through any of the mediators. It is the effect on performance of the difference in the participants’ assignment to the control or treatment group with both mediating variables held constant.
The first specific indirect effect links the independent variable to the dependent variable through the first mediating variable. This is equivalent to the product of a1 and b1. The second indirect effect links the independent variable to the dependent variable through the second mediating variable. This is equivalent to the product of a2 and b2. The third indirect effect connects all four variables and is equivalent to the product of a1, a3, and b2. The output generated by PROCESS for these described effects is shown in Table 2.
35
For making statistical inference about the effects, the confidence intervals that are also shown in Table 2 for each effect were considered. Following convention, the effect is significant if the interval does not contain zero.
For the data collected during the experiment, all specific indirect effects, the direct effect, and the total effect include zero in the corresponding bootstrap confidence intervals and are consequently not significant. Though not significant, the effect size and direction are presented briefly, as they reveal surprising results. The direct effect is, contrary to the predictions, negative (c1’ = -0.19, p = 0.31), which implies that participants in the treatment group (labeled with 2 in contrast to the control group that is labeled with 1) contributed less to the team goal. The same direction holds for the total effect (c1 = -0.35) and for the specific indirect effects. However, for the latter, the effect size is extremely small (Ind1 = -0.028; Ind2 = -0.039; Ind3 = -0.089).
The third indirect effect (containing of a1, a3, and b2) reflects the underlying process described in Hypothesis 1. The statistics presented above reveal that this effect is not significant and therefore that the result of the analysis offers insufficient support for the predicted serial mediation predicted in Hypothesis 1.
Though no support for the complete serial mediation was found, different conclusions can be drawn for the separate paths that are reflected by the sub-hypotheses. These path coefficients and p-values are depicted in Figure 3.
Looking only at the path coefficients, a1 is not significant with a p-value of 0.12, which is higher than 0.01. Consequently, Hypothesis 1a is not supported. However, between the two mediators and the outcome variable, both a3 and b2 are significant (p < 0.01). The coefficient for a3 is 0.50, which means that two participants with a difference by one point in the scale measuring normative goal frame are estimated to differ by 0.50 points on the scale measuring joint-production motivation. The coefficient for b2 is smaller (b2 = 0.35), but it is also positive. This means that two participants with a difference of one point in the scale measuring joint-production motivation are estimated to differ by 0.35 in the number of solved matrices. Given the number of matrices is only between one and three, this is a considerable difference. In addition, the values for R2 suggest a significant proportion of the total variance explained by the model to predict joint-production motivation and the number of matrices solved. For the joint-production motivation 58,22% of the total variance are explained and for the number of matrices the model predicts 38,20% (see Table 2). Consequently, for the effect on joint-production motivation and the number of solved matrices, there is enough evidence to reject the null hypothesis and consider hypotheses 1b and 1c.
