AFFECTIVE BEHAVIOR OF SCHEDULERS ON PERCEIVED TASK INTERDEPENDENCE AND MENTAL LOAD

(1)

AFFECTIVE BEHAVIOR OF SCHEDULERS ON PERCEIVED

TASK INTERDEPENDENCE AND MENTAL LOAD

Carla Zandvliet

Faculty of Economics and Business Master thesis Technology Management

(2)

2

Abstract

This research investigated the differences in perceived task interdependence and the differences in the level of mental load of schedulers and added insight into their relationships with affective behavior. A scheduling experiment was conducted to identify the effects of the independent variable task design (adapt an existing schedule opposed to make a schedule from scratch and full information transparency opposed to no information transparency) on the dependent variables perceived task interdependence, mental load and affective behavior. Results showed that schedulers without information transparency perceived more task interdependence. Making a schedule from scratch also leads to more perceived task interdependence but only in cases where information is not transparent. This implies that the level of information transparency is an important factor for task interdependence. The results also indicated that, regarding the level of mental load, full information transparency leads to a higher level of mental load but only in case of adapting an existing schedule. This implies that information has to be released carefully in case of rescheduling situations. A correlation analysis was used to investigate the relationship between perceived task interdependence and mental load with affective behavior. Affective behavior was measured by means of analysis of the communication between schedulers. The results indicated that there is a positive relationship between perceived task interdependence and the communication categories social talk, information exchange, analysis/determine strategy, problem indication, acknowledgment, conclusion/ evaluation negative and miscellaneous as well as with the total amount of communication. With regard to the level of mental load, a positive relationship was found with the communication category request for change and a negative relationship with conclusion/ evaluation positive. We also found a positive relationship between perceived task interdependence and the level of mental load, indicating that all our dependent variables affect each other. The findings suggest that perceived task interdependence and the level of mental load can be influenced by task design and that it is plausible that perceived task interdependence and the level of mental load are also partially responsible for the differences in affective behavior of schedulers.

(3)

3

1 Introduction

1.1 Background

Organizations are increasingly confronted with rapidly changing environments and customer demands. It is therefore often impossible to make a production schedule and stick to it until the complete job is finished. As a result, accurate rescheduling is crucial for the successful performance of a production schedule (Jain & Elmaraghy, 1997:282). Rescheduling means that there is not enough time to schedule from scratch or that the current schedule is already in progress (Fang, Ross & Corne, 1993:377). When rescheduling is necessary for one particular job, other jobs might be involved as well. Coordination between schedulers is necessary when a schedule cannot be adapted without violating commitments made to a colleague (De Snoo, 2011:77). This implies that, in case of events, schedulers are often interdependent in order to be able to recreate an optimal schedule for their jobs. Interdependence is defined as “a situation in which the outcomes for individual group members are affected by each other’s actions” (Johnson & Johnson, 1989:23. In: Van der Vegt, Emans & Van de Vliert, 2000:634). When individual employees are interdependent, this is often a reason to form a team (Cartwright & Zander, 1968; Shea & Guzzo, 1987. In: Van der Vegt et al., 2000:634). Already in 1948, Lewin stated that “the essence of a group is the interdependence among its members” (In: Van der Vegt et al., 2000:634). According to Van der Vegt, Emans, & Van de Vliert (1998:124) it is assumable that “the crux of successful team design is in the interdependence of the individual team members”. Also Ilgen (1999) says that task dependence is generally considered as a significant defining characteristic of teams (In: De Jong, 2007:1625). Nevertheless, it happened often that teams were created where they were not appropriate and set up in such way, that a failure could not be avoided (Wageman & Baker, 1997:140). To execute a scheduling or rescheduling task as good as possible, it is important that tasks and teams are designed the right way.

1.2 Aim of this research

(6)

6 performance. Ramamoorthy & Flood (2004:348) stated that task interdependence can have an influence on the behavior of group members. If someone is not able to achieve the goal by working alone, he/she will probably show cooperative behavior in order to reach that goal (Wagner, 1995. In: Ramamoorthy & Flood, 2004:348). According to Wageman & Baker (1997), it is task interdependence that drives cooperative behavior. But Allen, Sargent, & Bradley (2003:734) argue that it is “the perception of the situation by team members, rather than the objective situation” that drives their behavior. This implies that when team members are in fact task interdependent, they will only show cooperative behavior if they feel themselves actually task interdependent. Organizations therefore have to be careful when they make changes in task design, because this could immediately affect the cooperative behavior of team members.

But perceived task interdependence is not the only factor with possible behavioral consequences for schedulers. Earlier research suggests that the need for coordination in teams increases mental load (Bowers, Braun, & Morgan, 1992; Kidd, 1961; Williges, Johnston, & Briggs, 1966. In: Urban, Bowers, Monday, & Morgan, 1995:124). This means that the level of mental load might also play a role in the relationship between perceived task interdependence and rescheduling performance. Mental load is a multidimensional term which includes factors as frustration, mental demand and time pressure. It can have an impact on the behavior of schedulers and thereby affect their performance. An excessively low level of mental load can lead to boredom and reduced attention, whereas a level of mental load that is too high can cause stress and demoralize employees. However, as for perceived task interdependence, also the exact impact of mental load on the affective behavior of schedulers is still unknown.

(7)

7 interdependence and/ or the level of mental load and the way team members communicate. With reference to the aim of this research, the research question is formulated as follows:

What affects perceived task interdependence and the level of mental load of schedulers and what is their relationship with communication behavior?

1.3 Outline

(8)

8

2 Theoretical background: interdependence and mental load

2.1 Types of interdependence

Interdependence can arise from different sources. Wageman (1995:146) made a distinction between task inputs, executing processes, definition of goals and the way performance is rewarded. These sources are combined into two main forms of interdependence; task interdependence (derived from task inputs and executing processes) and outcome interdependence (derived from goals and rewards). 1 Task interdependence means that the performance of one job depends on the successful performance (and thus the efforts or skills of others) of another job (Kiggundu, 1981; Van der Vegt et al., 1998; Wageman & Baker, 1997:141). Thompson (1967:54) made a classification of three types of task interdependence between organizational parts: pooled, sequential and reciprocal. This is a widely used classification to describe differences in interdependence. But there are also researchers who refer to task interdependence at the individual level (Kiggundu, 1981, 1983; Brass, 1981, 1985. In: Van der Vegt et al., 1998). Task interdependence at the individual level is viewed as the affective behavior of individual team members. In other words, it is “the level of interpersonal interaction engaged in by unit members” (Shea & Guzzo, 1989. In: Wageman & Baker, 1997:141). This research focuses on the affective behavior of team members and therefore task interdependence will refer to the individual level from now on.

Outcome interdependence means that the rewards on an individual job depend on the successful performance of a job performed by someone else (Van der Vegt et al., 1998; Wageman & Baker, 1997:142). It can be seen as the level of goal setting, that is, individual versus group goal (e.g., Deutsch, 1973; Thomas, 1957. In: Van der Vegt et al., 2000: 635), or the degree to which employees are provided with group feedback (e.g., Saveedra, Earley, & Van Dyne, 1993; Wageman, 1995). Though some researchers make a distinction between outcome and goal interdependence as well (e.g., Saveedra et al., 1993). The presence of task interdependence does not automatically mean that outcome interdependence is also present, and vice versa (Wageman, 1995) although their interaction level can have an influence on overall performance.2 This research only focuses on affective behavior regarding task interdependence and therefore the concept of outcome interdependence will not be taken into account.3

2.2 Task interdependence

Task interdependence refers to the degree in which “group members have to exchange information and resources or actually work together to complete their jobs” (Kiggundu, 1983; Thompson, 1967.

1

Outcome interdependence and reward interdependence are used interchangeably in literature.

2

“Neither cooperation nor reward interdependence independently affected group performance, but their interaction did so” (Wageman & Baker, 1997:155).

3

(9)

9 In: Van der Vegt et al., 2000: 635). Being a member of a team entails that you are always, to a certain extent, dependent on another team member. The level of task interdependence increases when the job is more difficult and team members need more help from others to perform their jobs (Van der Vegt et al., 2000: 635). It has been demonstrated that task interdependence increases communication, helping behavior and information sharing (Crawford & Haaland, 1972; Johnson, 1973. In: Bachrach, Wang, Bendoly, & Zhang, 2007:257) and may also have an influence on the development of norms between team members (Shaw, 1981. In: Bachrach et al., 2007:257). Besides, task dependence is mentioned as a core requirement to build up trust between team members (Rousseau et al., 1998. In: De Jong, 2007:1625). Team members have to rely on each other when it comes to the accuracy of information so it is very important they can trust each other. But when task interdependence increases, there are more chances for individual team members to withhold personal, informational or organizational resources from other team members (e.g., French & Raven, 1959; Raven, 1992. In: Van der Vegt et al., 2000:637). Steiner (1972) calls this “process losses” which refers to the fact that interaction processes within a team could hinder the performance (In: Hackman & Morris, 1975).4 These process losses can also have a negative influence on satisfaction and commitment (Van der Vegt et al., 2000:637). This shows that task interdependence does not only have positive influences on the affective behavior of team members.

Schedulers communicate every day with many parts of an organization to ensure that they have all relevant and up-to-date information of the production system (McKay, Safayeni, & Buzacott, 1995:81). This information is used for decision making processes and by means of communication, schedulers can also spread relevant information to other departments. The need for accurate and up-to-date information makes schedulers dependent on each other, which can result in process losses. Information can also be found in the information systems of the organizations, but McKay et al. (1995:81) observed that schedulers sometimes assume not all relevant information is directly available to them. As a result, they search for information in other parts of the organization. This can enhance the feeling of schedulers that they are task interdependent. When resources are available to work independently (e.g. when information is transparent), there is less need to coordinate actions (MacMillan, Entin, & Serfaty, 2004). Consequently, schedulers who have to exchange a lot of information will perceive that they are dependent on their colleagues to complete their task. This leads to the following hypothesis:

H1a: Teams without information transparency perceive more task interdependence than teams with information transparency.

4

(10)

10 The coordination of activities will be different for scheduling and rescheduling tasks. At rescheduling tasks, interaction between schedulers is only necessary when there is an unexpected event which cannot be solved at one’s own department and only then when there is enough time available to coordinate actions (De Snoo, 2011:77). For scheduling tasks however, it can be useful to communicate with a team member about each job that has to be scheduled in order to create an optimal schedule for both. In other words, there is more freedom and more jobs to discuss in scheduling than in rescheduling and this could result in more coordination between team members. This leads to the following hypothesis:

H1b: Teams that have to make a schedule from scratch will perceive more task interdependence compared to teams that have to adapt an existing schedule.

2.3 Mental load

In literature there are many definitions of workload which all give different causes, consequences and symptoms for it (Hart, 2006:904). In general, workload is the amount of work that has to be accomplished by an operator in a given time period. However, workload is not only task-specific but it is also person-specific (Leplat, 1978:147; Rousse et al., 1993. In: De Waard, 1996:15; Xie & Salvendy, 2000:220). Capabilities, motivation, strategies applied in task performance, mood and operator state all have an effect on the experienced load (De Waard, 1996:15; Johnson & Widyanti, 2011:509). In this research, the term mental load is used to describe the perceived level of workload by the schedulers, which includes the following factors: mental demand, time pressure, performance, effort and frustration.5

Schedulers do not always have direct access to the information they need but they have to communicate with colleagues to get up-to-date data. Thus, when information is available to all schedulers it is less necessary to communicate about data. According to MacMillan et al. (2004), lower need for communication is associated with lower workload. Beith (1987:67) concluded as well that communication demands affected operator workload and this is also the result from a study about airplane controllers (Porterfield, 1997:180). Controllers who were communicating more (i.e., controllers who had to process and forward more information) did have a higher level of workload. This may be due to communication overload, because not all communicated data will be relevant, but it does take time and effort for the receiver to evaluate if the communicated data is useful and also the sender has to make choices which data he/ she will exchange. Castenon et al. (1986) indeed found that overuse of communication leads to an increase in perceived workload (In: Beith, 1987:68). Although it is useful to have information about the task situation of team members, there may be a cost (i.e. time) to obtain that information (MacMillan, Paley, Levchuk, Entin, Serfaty, & Freeman, 2002:12). Non-critical information that is communicated can be seen as ineffective. Xie & Salvendy (2000:220) argue

5

(11)

11 that this extra information processing leads to higher mental load. Because full information transparency is associated with less communication about data, the following hypothesis is stated:

H2a: Teams without information transparency will have a higher level of mental load compared to teams with information transparency.

Another factor that is often mentioned to have a big impact on mental load is time pressure (Beith, 1987:67; Moray, 1982:37; Moray, Dessouky, Kijowski, & Adapathya, 1991; Urban et al., 1995:135; Xie & Salvendy, 2000). Time pressure increases the focus on finishing urgent tasks at the expense of other tasks (De Croon et al., 2000. In: Elfering, Grebner, & Haller, 2012:1464). However, incomplete tasks have to be kept in mind to finish at a later stage and thereby increase cognitive load which can influence the level of mental load (Elfering et al., 2012: 1465). It seems that interruptions while performing a task lead to a higher level of mental load. A study from Wei & Wieringa (1997) showed as well that an unexpected change in the task situation (in that study a drop in the degree of automation) leads to temporally higher levels of mental load. There are differences in time pressure for scheduling and rescheduling tasks. According to De Snoo (2011:109), interdependencies during scheduling tasks can be managed by simple rules and agreements (e.g., schedulers can work sequentially on their schedules). Besides, when problems occur, there is enough time to adapt the schedule, to give feedback and for communication and coordination between team members. But the circumstances are different for rescheduling tasks, because in that case the schedules are already released and (partly) carried out. When problems arise, rapid interventions are required. Therefore communication and deliberation between the schedulers are needed (Van Wezel, 2006. In: De Snoo, 2011: 109), since a change is one schedule can have significant consequences for other schedules. Schedulers have to communicate immediately about the modifications they make and debate the consequences for the rest of the schedules.

(12)

12 H2b: Teams that have to adapt an existing schedule will have a higher level of mental load compared to teams that make a schedule from scratch.

Because rapid decisions and actions are necessary in rescheduling situations, it is expected that teams that have to adapt an existing schedule will suffer more from no information transparency than teams that have to make a schedule from scratch. Due to the necessity to take actions based on accurate and up-to-date information, we expect that the degree of information transparency has a strong influence on the level of mental load for teams that have to adapt an existing schedule. On the other hand, teams that make a schedule from scratch still have to coordinate and communicate in case of information transparency. These teams also have less time pressure so the cost for information processing may not be too high in terms of mental load. Therefore, we do not expect an effect for the degree of information transparency on their level of mental load. This leads to the following hypothesis:

(13)

13

3 Theoretical background: communication behavior

Communication is the instrument for teams to deal with mission execution (Entin, Diedrich, Kleinman, Kemple, Hocevar, Rubineau, & Serfaty , 2003:3) and is also the medium for coordination among team members (Swezey & Salas, 1992. In: Achille et al., 1995:96). The need for communication is related to the need for coordination in which the amount of communication decreases when there is lower need for coordination (MacMillan et al., 2004). This is consistent with the results of several studies which demonstrated that interdependent tasks are generally associated with more communication (e.g. Fulk, 1993:929; Maznewski & Chudoba, 2000. In: Rico & Cohen, 2005:264; Crawford & Haaland, 1972; Johnson, 1973. In: Wageman, 1995:149). Though, coordination also develops as a result of communication (Achille et al., 1995:96). When team members perceive they are task interdependent there will be more coordination between their actions. The distribution of information and capabilities for example, might influence the degree to which team members choose to communicate with each other (Urban et al., 1995:125) and logically also the content of their communication. It is therefore expected that perceived task interdependence has an influence on the content of the messages team members send each other, but the communication can also influence the level of perceived task interdependence.

It is important that team members are provided with the opportunity to communicate with each other. A study from Beith (1987:71) showed that mental load increases almost fifty percent when communication was limited. Several studies (e.g. Bolstad & Endsley, 1999; Kleinman & Serfaty, 1989. In: Urban et al., 1995:126) showed that communication mediates the effects of mental load on performance. Teams with a moderate level of mental load communicated more compared to teams with a low level of mental load, while teams with a high level of mental load communicated the least. Although no main effect of mental load was found on different communication categories, research from Urban et al. (1995) showed that there is an interaction between mental load and performance with respect to communication behavior. This suggests that teams adapt to situational demands by means of communication (Urban et al., 1995:135). Communication is thus affected by the level of mental load, but the level of mental load also has an effect on communication behavior.

(14)

14 H3: There is a significant difference between the scheduling teams and their communication.

As stated in the introduction, we expect a relationship between perceived task interdependence and affective behavior of schedulers as well as between the level of mental load and the affective behavior of schedulers. Previous research (Urban et al., 1995) could not find an effect of mental load on different communication categories. Rasker (2002) showed that team members who were in a situation of unrestricted communication did not adapt their communication to the level of mental load. We nevertheless try again to find one or more significant relationships in order to give some insight in the affective behavior of schedulers on mental load. The next paragraphs describe several types of communication and hypotheses are stated regarding the possible relationships with perceived task interdependence and mental load. The communication categories described in this chapter are based on the findings of earlier research relating to perceived task interdependence and affective behavior.

3.1 Communication: Social talk

Not all communication between team members will relate to the execution of the task. Non task related communication can give a moment of relaxation and can help to develop a bond between team members. They will understand each other better and have more empathy for one another. This ensures that in case of difficulties there is a solid basis of respect and trust between team members (Hackman, 1998: 461). A study on dealer attitudes (Kumar, Scheer, & Steenkamp, 1995:354) demonstrated that greater total interdependence leads to higher levels of trust, more commitment and less conflicts. This is also the case for task interdependence, which has been demonstrated to relate positively to team commitment (e.g., Aube & Rousseau, 2005; Van der Vegt et al., 2000:649). In addition, high interdependence has positive effects on interpersonal relations (Johnson & Johnson, 1989. In: Wageman, 1995:148). Working together to finish a task generally fosters camaraderie and friendship (Van der Vegt et al., 2000:637) which almost certainly results in more social talk. In this research we associate team commitment with interpersonal relations. We assume that team commitment is expressed as non task related communication, because this is seen as a process to strengthen the relationship between team members and thereby enhancing team commitment.

No literature has been found about the relationship between mental load and social talk. Though, when schedulers have time for social talk they are not directly working on the task and apparently do not feel time pressure is hindering them. Social talk can also give a moment of relaxation and thereby (temporarily) lower the level of mental load. It is therefore expected that a high level of mental load decreases the amount of social talk. All in all, this leads to the following hypotheses:

(15)

15

3.2 Communication: Information exchange

As noted several times, team members are dependent on up-to-date and accurate information in order to be able to (re)create optimal schedules. According to Kiggundu (1983:147), the extent to which team members have to receive or give information is an element of task interdependence. Perceived task interdependence will be higher when it is necessary to exchange information. Earlier studies have demonstrated that task interdependence increases information sharing (Crawford & Haaland, 1972; Johnson, 1973. In: Bachrach et al., 2007:257) and results in more information processing activities (Fulk, 1993:929). It is therefore plausible to expect an interaction between the level of perceived task interdependence and the amount of information that is exchanged: the higher the perceived task interdependence, the more information will be exchanged and vice versa.

The efficiency and effectiveness of information exchange between team members depends to a certain extent on their shared mental models. When team members are aware of each other’s tasks they are able to predict the information needs of team members and as a result only necessary information will be communicated (Rasker, 2002:213). This leads to a reduction in communication because fewer messages are needed for the exchange of information. This efficiency reduces the level of mental load (MacMillan et al., 2004). The exchange of information costs time and mental effort in order to process this information. This may lead to a higher level of mental load. Based on the above, we propose the following hypotheses:

H5a: There is a positive correlation between perceived task interdependence and information exchange.

H5b: There is a positive correlation between the level of mental load and information exchange.

3.3 Communication: Suggestion

(16)

16 (Anderson & Williams, 1996:285).6 Several studies have demonstrated that a high level of task interdependence results in more helping behavior (e.g., Allen et al., 2003; De Jong, 2007; Wageman, 1995). In this research we define helping behavior as voluntary assistance to the other team member in order to improve the schedule. This means that one provides suggestions about possible changes in the other schedule and gives advice about which tasks the other team member should carry out. We expect a positive relationship between perceived task interdependence and doing suggestions.

No literature has been found about the relationship between mental load and helping behavior. However, considering the factors of mental load (mental demand, time pressure, performance, effort and frustration) it is expected that higher levels of mental load lead to less helping behavior. The focus and available time for successful execution of one’s own task will probably inhibit to be fully aware of the task of a team member. This leads to the following hypotheses:

H6a: There is a positive correlation between perceived task interdependence and suggestion. H6b: There is a negative correlation between the level of mental load and suggestion.

3.4 Communication: Analysis/ Determine strategy

Interdependent tasks are in general associated with a higher level and complexity of the decision making process (Maznewski & Chudoba, 2000. In: Rico & Cohen, 2005: 264). Also Johnson & Johnson (1989) stated that cognitive complexity of thought is a positive result of high interdependence (In: Wageman, 1995:148). This may be due to the level of information sharing. Teams in a high task interdependence setting experience the task as more cognitively complex when a lot of information has to be shared (Allen et al., 2003:734). We relate complexity of thoughts to the analysis of the task and the deliberations about which strategy has to be followed. So when perceived task interdependence is high, we expect team members to coordinate their strategy and that they are more concerned with the analysis of the task situation.

No literature has been found about the level of mental load and whether this is related to the determination of the strategy and analyzing the task situation. When it is clear to team members which strategy has to be followed to carry out their tasks, it can give a better overview of the task situation and make them more relaxed. When the strategy is not determined at the beginning of the task, this can cause a high level of mental load later on, but when the level of mental load is already high at the beginning of the task, it is possible that there is no time to determine the strategy. Thus, the determination of the strategy and the level of mental load affect each other. This is also the case for the analysis of the task situation. When the level of mental load is high, there might be no time to analyze the task situation. On the other hand, we suggest that by means of an analysis, team members get a better understanding of the task and as a result they are better capable of dealing with problem situations. This will make them more relaxed and probably more successful, so the level of mental

6

(17)

17 load will be low. On the other hand, a mentally demanding task generally requires more analysis than a simple task. This implies that when team members analyze more, they also have higher levels of mental load. Therefore we expect that the level of mental load does not correlate with the analysis and determination of the strategy of the task.

H7a: There is a positive correlation between perceived task interdependence and analysis. H7b: There is no correlation between the level of mental load and analysis.

3.5 Communication: Problem indication/ Request for help

Task interdependence implies that someone is dependent on a team member for the execution and successful performance of their job. When team members are task interdependent, it might be easier to explicitly ask for help because they are already involved in the same task and an unsolved problem also affects the performance of the other team member. Anderson & Williams (1996) indeed found a positive correlation between task interdependence and how often employees asked for help. This is in line with Van der Vegt et al. (2000:635), who stated that task interdependence increases when a task becomes more difficult and employees need more assistance. Besides, when perceived task interdependence is high, we assume that team members are more likely to indicate a problem. Again, due to the reason that one’s own problem has an influence on the successful performance of the other team member. A request for help can also

When the elements of mental load are considered, it is assumable that a high level of mental load relates to more problems with the execution of the task. For example, it is possible that tasks cannot be processed within the available time or that someone needs advice in the execution process. In these cases, team members will probably indicate they have a problem with the task. We expect that the amount of indicated problems is related to the level of mental load. Based on the above, we propose the following hypotheses:

H8a: There is a positive correlation between perceived task interdependence and problem indication. H8b: There is a positive correlation between the level of mental load and problem indication.

3.6 Communication: Acknowledgment

(18)

18 Acknowledgments prevent someone for unnecessarily repeating the message which results in a decrease of communication load (Achille et al., 1995:105). It saves time and the receiver does not have to process the message more than once. This will certainly not increase the level of mental load, but we assume that the influence of acknowledgments in the communication process is also not enough for the actual lowering of the level of mental load. Based on the above, the following hypotheses are stated:

H9a: There is a positive correlation between perceived task interdependence and acknowledgments. H9b: There is a no correlation between the level of mental load and acknowledgments.

3.7 Research model

Now all concepts are explained in the theoretical background and the hypotheses are formulated, we can introduce the research model. The aim of this research is to add some insights into the differences in perceived task interdependence and the differences in the level of mental load between schedulers and the possible relationships with affective behavior. Figure 1 (page 18) shows the research model with all relevant variables and their relationships. As can be seen is (re)scheduling performance the consequence of affective behavior and resulting actions from schedulers. To increase performance, it is therefore important to understand behavior and to find out if perceived task interdependence and/ or the level of mental load have an effect on affective behavior. To manipulate perceived task interdependence and level of mental load we chose to distinguish between teams with and without information transparency and teams that make a schedule from scratch or adapt an existing schedule. The combination of these elements results in four types of scheduling teams which form the independent variable task design. The expectations regarding the differences for perceived task interdependence and the level of mental load between these variables were expressed in hypothesis 1a, 1b and hypothesis 2a, 2b, 2c. Due to differences in the type and design of the task, we also expect a relationship between the variables task design and affective behavior, this was expressed in hypothesis 3. Because the relationships between perceived task interdependence and affective behavior and between mental load and affective behavior are reciprocal, it is not possible to investigate if there are causal relationships between these variables. Only conclusions about possible correlations can be drawn. Hypotheses 4-9 examine the relationships between these variables.

(19)

19

4 Method

The aim of this research is to add some insights into the differences in perceived task interdependence and the differences in the level of mental load between schedulers and the possible relationships with affective behavior. We have used data from an earlier conducted experiment.7 This was a controlled scheduling experiment where subjects filled in a questionnaire after the experiment. Consequently, we had no influence on the design of the experiment and operationalization of the variables scheduling team, perceived task interdependence and mental load, but the processing of this data and everything related to the variable communication category is done in this study. The hypotheses that followed from the investigation of the literature will be tested based on the results and the questionnaires of this experiment. The following paragraphs describe the experimental setting, operationalization of the data and the statistical tests that will be carried out.

4.1 Subjects

Undergraduate Technology Management students (N=189) at the Groningen University participated in the experiment as part of an obliged course. Their age ranged from 18 to 25 (mean = 20,62; SD = 1,57). 125 Of them were male (=66,1%) and 64 female (= 33,9%). All subjects were randomly assigned to a scheduling team and did not know who their partner was before the experiment started.

4.2 Experimental setting8

This research uses data from an earlier conducted controlled scheduling experiment. Subjects worked in teams of two people to solve a scheduling task and were randomly assigned to a team. There were four types of scheduling teams (table 1, page 20) and the goal for all teams was to make the best possible schedule in order to minimize total costs.9 The assignment was set up as a flexible job shop. There were four departments and each subject was responsible for two departments with each two interchangeable machines. Several jobs needed four departments, which created interdependencies between the subjects. Schedulers used a Gantt chart in which tasks could be dragged and dropped while the costs where immediately visible. It was not possible to violate the routing and jobs could not overlap in time. Subjects could see the schedule of the other team member in the transparent configuration, but in the non transparent configuration they could only see their own schedule. Teams who had to make a schedule from scratch did not face unexpected events, but teams who had to adapt an existing schedule were confronted with several unexpected events which required adjustments to the schedule (the list of events can be found in Appendix 1). Team members could communicate with each other by means of a chat program. The assignment took twenty minutes and at the end all subjects filled in a questionnaire which were used to measure perceived task interdependence and the level of mental load. The questionnaire can be found in Appendix 2.

7

Initiated and conducted by Van Wezel, March 2010

8

Information based on a description of the scheduling exercise (Van Wezel, 2010).

9

(20)

20

Table 1: Types of scheduling teams

4.3 Operationalization

4.3.1 Independent variable: Task design

Task design is operationalized by distinguishing four types of scheduling teams (table 1). Subjects had to make a schedule from scratch or had to adapt an existing schedule. Besides, they had instant insight in each other’s schedule or only in their own schedule. Team members could only make changes in their own departments, regardless of their type of scheduling team. The characteristics for each team are briefly explained.

ANT (Adapt, Non Transparent)

- No insight in the schedule of the other team member; - Jobs were already scheduled;

- Several unexpected events occurred. AT (Adapt, Transparent)

- Instant insight in the schedule of the other team member; - Jobs were already scheduled;

- Several unexpected events occurred. MNT (Make, Non Transparent)

- No insight in the schedule of the other team member; - Schedule had to be made from scratch;

- No unexpected events occurred. MT (Make, Transparent)

- Instant insight in the schedule of the other team member; - Schedule had to be made from scratch;

- No unexpected events occurred.

4.3.2 Dependent variable: Perceived task interdependence

The questionnaire contained four questions to measure the perceived task interdependence based on task interdependence items from Van der Vegt et al. (2000:640). The answers were given on a 7-point Likert scale. The items were: “I needed information and advice from the other planner to make the

Type of scheduling tasks

Make schedule from

scratch Adapt existing schedule Level of information transparency

Low (no insight in

other schedule) MNT ANT

High (instant insight

(21)

21 assignment,” “I was dependent on the other planner to make the assignment,” “I had to cooperate closely with the other planner to make the assignment,” and “The other planner needed information and advice from me to make the assignment.” Cronbach’s alpha was used to measure reliability and this value must be between 0.65 and 0.9 (Den Boer, Bouwman, Frissen, & Houben, 1994:62). We found Cronbach’s alpha = 0.574. Therefore the question “I had to cooperate closely with the other planner to make the assignment” was deleted in this research. This results in Cronbach’s alpha = 0.633, which is considered acceptable. As a result, the perceived task interdependence is based on three questions.

4.3.3 Dependent variable: Mental load

There are many ways to assess the level of mental load but the most common is the NASA Task Load Index (NASA-TLX). This is a subjective tool that measures perceived workload on six different scales (mental demand, physical demand, temporal demand, performance, effort and frustration). The answers are given on a 20-point Likert scale. After rating the six scales, subjects have to choose the most taxing element of 15 pair-wise comparisons in order to weight the scales and to give an individualized score on overall workload (Hart & Staveland, 1988).

In this research the “physical demand” scale was deleted.10 This leads to the following items: “How mentally demanding was the task?,” “How much time pressure did you feel?,” “To what extent do you think you was successful?,” “How much effort was needed to get this result?,” and “How uncertain, stressed, and irritated were you as opposed to certain, satisfied, relaxed, and full of confidence?.” After these questions they had to choose between 10 pair-wise comparisons consisting of the five scales. We found Cronbach’s alpha = 0.652 which is considered acceptable.

4.3.4 Dependent variable: Affective behavior

In this research affective behavior is operationalized by means of communication. By classifying the communication of schedulers into different categories, we can distinguish between different aspects of affective behavior. The analysis of communication is complex. There is no commonly used method to measure the content of communication. Almost every research uses their own communication categories to count the frequencies of specific messages (e.g., Rasker, 2002; Urban et al., 1995; Wageman, 1997). Also this research uses its own classification. Initially, all types of communication resulting from the theoretical background in Chapter 3 are distinguished (i.e. social talk, information exchange, suggestion, analysis/ determine strategy, problem indication/ request for help and acknowledgments. Because the experiment was set up as a scheduling task, the subjects would ask for changes in the schedules which resulted in the categories request for change and answer request for change. We also decided to make a distinction between reaction positive on suggestion/ analysis and reaction negative on suggestion/ analysis. These chats can tell us something about the level of

10

(22)

22 agreement between team members regarding the suggestions and analyses. In addition, we wanted to know the feelings of subjects about the execution of the task. This resulted in the categories conclusion/ evaluation positive and conclusion/ evaluation positive. Big differences in the amount of chats in one of these categories could be an indication that the task might have been too easy or too difficult for a specific type of scheduling team. Because not every chat would fit in one of the previous mentioned communication categories, we made one category (miscellaneous) for all remaining chats.

During the experiment team members communicated with each other by means of a chat program. These chats were saved for analysis. Because every chat is different they are coded and this resulted in sixteen communication categories. For the classification of the categories we have chosen to focus on the content of the chats and the message that is aimed to communicate. Complex sentences were not split up and put into different categories but the initial expressed message of the chat was decisive for the communication category. Chats containing separate punctuating marks or corrected typos were categorized as "invalid" and not included in the analysis. The sixteen communication categories are briefly explained. Examples of chats within each category can be found in Appendix 3.

- Social talk: chats were team members introduce themselves and other non task related chats; - Information exchange: all chats in which information is exchanged. This can be a request for

information, an answer on a request for information or giving information without a prior request;

- Request for change: a request for a change in the schedule of the other team member for a specified task and timeslot;

- Answer request for change: answer on the request for change, this can either be a positive or negative answer or a request for change in return;

- Suggestion: suggestions on possible changes in the schedule and recommendations about the task;

- Analysis/ Determine strategy: analysis of the task and chats in which team members coordinate their actions;

- Reaction positive on suggestion/ analysis: chats in which one agrees with the proposed suggestion, analysis or proposed strategy of the other team member;

- Reaction negative on suggestion/ analysis; chats in which one disagrees with the proposed suggestion, analysis or proposed strategy of the other team member;

- Conclusion/ Evaluation positive: expressing positive feelings about the execution of the task in general or expressions of happiness;

- Conclusion/ Evaluation negative: expressing negative feelings about the execution of the task in general or expressions of frustrations;

(23)

23 - Offering help voluntarily: chats in which a team member explicitly offers support without a

prior request for help;

- Acknowledgements: chats in which one makes clear that a message is received or to let the other team member know that they have seen a change in the schedule has been carried out; - Miscellaneous: all chats that do not fit into another category.

4.4 Measurement

All analyses will be done using IBM SPSS Statistics version 20. The testing of the hypotheses can be divided into three steps:

(24)

24

5 Results and analysis

5.1 Introduction

In total, there were 189 complete questionnaires: 45 x ANT (=23,8%); 55 x AT (=29,1%); 40 x MNT (=21,2%) and 49 x MT (=25,9%). Subjects generated in total 6288 valid chats: 1568 x ANT (=24,9%); 1045 x AT (=16,6%); 2418 x MNT (=38,5%) and 1257 x MT (=20%). The mean score for perceived task interdependence on a scale from 1-7 was 4.6208 (SD = 1.203) while the mean level of mental load on a scale from 1-100 was 50.49 (SD = 13.368). No significant differences were found between age or sex with perceived task interdependence, the level of mental load or any of the communication categories.

5.2 H1 – H2: Mann Whitney U Test

5.2.1 H1a: Perceived task interdependence and information transparency

We investigated if there was a difference in perceived task interdependence between teams with or without information transparency. The Mann Whitney U test demonstrated that teams without information transparency perceived statistically significant more task interdependence compared to teams with information transparency (nNT = 85, nT = 104, z = -5.372, p < .000).

5.2.2 H1b: Perceived task interdependence and type of task

We investigated if there was a difference in perceived task interdependence between teams who had to adapt an existing schedule and teams who made a schedule from scratch. The Mann Whitney U test demonstrated a statistically significant difference: teams who made a schedule from scratch perceived more task interdependence compared to teams who had to adapt an existing schedule (nA = 100, nM =

89, z = -2.162, p < .05).

5.2.3 H2a: Mental load and information transparency

We investigated if there was a difference in the level of mental load between teams with or without information transparency. The Mann Whitney U test demonstrated no statistically significant difference: teams without information transparency did not have a higher level of mental load compared to teams with information transparency (nNT = 85, nT = 104, z = -.253, p > .05).

5.2.4 H2b: Mental load and type of task

We investigated if there was a difference in the level of mental load between teams who had to adapt an existing schedule and teams who made a schedule from scratch. The Mann Whitney U test demonstrated no statistically significant difference: teams who had to adapt an existing schedule did not have a higher level of mental load compared to teams who made a schedule from scratch (nA =

(25)

25 5.2.5 H2c: Mental load, type of task, information transparency

We investigated if the degree of information transparency had an effect on the level of mental load for different types of tasks. The Mann Whitney U test demonstrated a statistically significant difference: when teams had to adapt an existing schedule, full information transparency resulted in a higher level of mental load compared to no information transparency (nANT = 45, nAT = 55, z = -2.079, p < .05) but

no statistically significant difference was found for teams who made a schedule from scratch (nMNT =

40, nMT = 49, z = -1.889, p > .05).

5.3 H3: Chi-square test

We investigated if there was a difference between the scheduling teams and their communication. The Chi-square test showed a statistically significant difference (p < .000) between the four scheduling teams and their communication behavior. Table 2 (page 27) shows the complete Chi-square table with the observed counts of chats per scheduling team and communication category. Besides, the distribution within the communication category per scheduling team is displayed as well as the distribution of the communication within each scheduling team. The table can thus provide some insight into the differences in communication behavior between the four types of scheduling teams.

5.4 H6 – H11: Spearman’s rho

5.4.1 H6a – H11a: Correlations concerning perceived task interdependence

Table 3 (page 28) shows the correlation matrix. Each cell represents the correlation coefficient for the intersecting variables. When only correlations regarding the hypotheses are taken into account it can be concluded that the level of perceived task interdependence has a statistically significant correlation with social talk (rs = .238, p < .01), information exchange (rs = .408, p < .01), analysis/ determine

strategy (rs = .179, p < .05), problem indication/ request for help (rs = .147, p < .05) and

acknowledgment (rs = .317, p < .01). No statistically significant correlation was found between

perceived task interdependence and suggestion (rs = -.020, p = .788) or offering help voluntarily (rs =

-.077, p = .292). In addition to the correlations regarding the hypotheses, we found statistically significant correlations for the level of perceived task interdependence with the level of mental load (rs

= .238, p < .01), conclusion/ evaluation negative (rs = .265, p < .01), miscellaneous (rs = .189, p < .01)

and with the total amount of communication (rs = .420, p < .01).

5.4.2 H6b – H11b: Correlations concerning the level of mental load

As can be concluded from table 3 (page 28), no statistically significant differences were found regarding the hypotheses dealing with the correlation of the level of mental load with social talk, information exchange, suggestion, offering help voluntarily, analysis/ determine strategy, problem indication/ request for help and acknowledgment. However, we found statistically significant differences between the level of mental load and request for change (rs = .149, p < .05) and

(26)

26

ANT AT MNT MT Total Social Talk Count ₈₄ ₉₂ ₁₁₆ ₇₈ ₃₇₀

% within communication category _22,7% _24,9% _31,4% _21,1% _100,0% % within team type 5,4% 8,8% 4,8% 6,2% 5,9% Information exchange Count ₈₄₈ ₁₉₈ ₁₄₄₀ ₃₂₄ ₂₈₁₀ % within communication category _30,2% _7,0% _51,2% _11,5% _100,0% % within team type _54,1% _18,9% _59,6% _25,8% _44,7% Request for change Count ₁₀₅ ₁₃₈ ₄₀ ₆₈ ₃₅₁

% within communication category _29,9% _39,3% _11,4% _19,4% _100,0% % within team type _6,7% _13,2% _1,7% _5,4% _5,6% Answer request for change Count ₆₆ ₄₇ ₂₅ ₁₉ ₁₅₇

% within communication category _42,0% _29,9% _15,9% _12,1% _100,0% % within team type 4,2% 4,5% 1,0% 1,5% 2,5% Suggestion Count 21 183 99 181 484

% within communication category 4,3% 37,8% 20,5% 37,4% 100,0% % within team type _1,3% _17,5% _4,1% _14,4% _7,7% Analysis/ Determine

strategy

Count ₁₀₅ ₈₇ ₂₀₅ ₂₃₂ ₆₂₉ % within communication category _16,7% _13,8% _32,6% _36,9% _100,0% % within team type _6,7% _8,3% _8,5% _18,5% _10,0% Reaction postive on

analysis/ suggestion

Count ₂₀ ₄₇ ₈₉ ₇₂ ₂₂₈ % within communication category _8,8% _20,6% _39,0% _31,6% _100,0% % within team type _1,3% _4,5% _3,7% _5,7% _3,6% Reaction negative on

analysis/ suggestion

Count ₁₁ ₂₁ ₁₃ ₁₅ ₆₀ % within communication category 18,3% 35,0% 21,7% 25,0% 100,0% % within team type ,7% 2,0% ,5% 1,2% 1,0% Conclusion/ Evaluation

positive

Count ₃₀ ₃₆ ₂₄ ₄₄ ₁₃₄ % within communication category _22,4% _26,9% _17,9% _32,8% _100,0% % within team type _1,9% _3,4% _1,0% _3,5% _2,1% Conclusion/ Evaluation

negative

Count ₂₀ ₃₁ ₃₃ ₁₂ ₉₆ % within communication category _20,8% _32,3% _34,4% _12,5% _100,0% % within team type _1,3% _3,0% _1,4% _1,0% _1,5% Problem indication/

Request for help

Count ₄₃ ₂₉ ₂₆ ₂₉ ₁₂₇ % within communication category _33,9% _22,8% _20,5% _22,8% _100,0% % within team type 2,7% 2,8% 1,1% 2,3% 2,0% Offering help voluntarily Count ₃₈ ₉ ₁₃ ₄ ₆₄

% within communication category _59,4% _14,1% _20,3% _6,3% _100,0% % within team type _2,4% _,9% _,5% _,3% _1,0% Acknowledgment Count ₁₀₀ ₅₁ ₁₈₉ ₆₃ ₄₀₃

% within communication category _24,8% _12,7% _46,9% _15,6% _100,0% % within team type _6,4% _4,9% _7,8% _5,0% _6,4% Miscellaneous Count ₇₇ ₇₆ ₁₀₆ ₁₁₆ ₃₇₅

(27)

(28)

28

5.5 Analysis of the results

The need for coordination is less when information is available to work independently (MacMillan et al., 2004), so team members who have to exchange crucial information in order to be able to carry out their tasks are very dependent on each other. Our results confirm this because a statistically significant difference was found in perceived task interdependence between teams with and without information transparency. Teams without information transparency perceived more task interdependence and therefore hypothesis 1a is supported. We also found a statistically significant difference in perceived task interdependence between teams who had to adapt an existing schedule and teams who made a schedule from scratch. Perceived task interdependence was higher for those teams who made a schedule from scratch, thus hypothesis 1b is supported. But since the extent to which team members have to receive or give information is an element of task interdependence (Kiggundu, 1983:147), we also tested for differences between teams who had to adapt or make a schedule when the level of information transparency was equal between the teams. This provided an important insight. When information was transparent, there was no statistically significant difference in perceived task interdependence between teams who had to adapt or make a schedule (nAT = 55, nMT = 49, p > .05). In

the Transparent configuration team members will immediately see the changes that are made in the schedules so teams in the Made configuration do not necessarily have to exchange more information about their tasks. However, we did find a statistically significant difference between the Adapt and Make configuration when there was no information transparency (nANT = 45, nMNT = 40, z = -3.074, p <

.01). Teams who made a schedule from scratch perceived more task interdependence compared to teams who had to adapt an existing schedule. This seems reasonable because adapting an existing schedule only requires information sharing in case of events, whereas at making a schedule from scratch the information about all tasks has to be shared.

(29)

29 performance of the task. The type of task was also found to have no effect on the level of mental load and hence hypothesis 2b is rejected. Again, this might be due to the fact that subjects did not perceive time pressure was hindering them. Even when quick responses were needed in the Adapt configuration, time pressure did not hinder the subjects to find adequate solutions. Therefore, it is assumed that subjects could set their own pace and that pace did not contribute significantly to the level of mental load in this experiment. We also stated that the degree of information transparency only affected the level of mental load when teams had to adapt an existing schedule. The results support hypothesis 2c but in the opposite direction of what we would have expected. Teams with information transparency had a higher level of mental load compared to teams without information transparency in the Adapt configuration, while no statistically significant difference was found for teams with and without information transparency in the Make configuration. Because our results are in conflict with earlier findings it is decided to also test for differences when the level of information transparency was equal between teams. We did not find a statistically significant difference between the Adapt and Make configuration when there was no information transparency. Though, when information was transparent, teams who had to adapt an existing schedule had a statistically significant higher level of mental load compared to teams who made a schedule from scratch (nAT = 55, nMT = 49,

z = -3.006, p < .01). This may be caused by the fact that team members should not only pay attention to their own schedule in case of events, but are also expected to analyze the consequences for the other schedules. Team members are less likely to ask for a change in their team members’ schedule when it is obvious that this causes a problem in their schedule. Instead of focusing on one schedule when events occurred, team members in the Adapt/ Transparent configuration had to concentrate on two schedules to find a solution. This can be seen as more mentally demanding and thereby increasing the level of mental load.

(30)

30 existing schedule when the degree of information transparency was equal. Due to the amount of information that is exchanged, there is also a noticeable difference in acknowledgment. More information exchange results in more opportunities to let the other person know a message was received. There is also a clear difference between the amount of suggestions that are given. Teams with full information transparency gave almost twice as much suggestions compared to teams in the Make/ Non Transparent configuration and approximately nine times more suggestions were given compared to teams in the Adapt/ Non Transparent configuration. When we look into detail within the scheduling teams, this difference is even bigger: teams in the Adapt/ Not Transparent configuration spent only 1.3% of their communication on doing suggestions whereas teams in the Adapt/ Transparent configuration have spent 17.5% of their communication on doing suggestions. Another remarkable difference is in the category analysis/ determine strategy. Teams in the Make configuration determine their strategy and analyze twice as much as teams in the Adapt configuration. It is assumable that teams in the Make configuration start with the determination of the strategy to avoid problems in a later stadium and occasionally analyze the task situation to determine whether the strategy still works, while teams in the Adapt situation only analyze the task in case of events. A further striking difference is in offering help voluntarily. Although this category is very small (only 1% of all communication belongs to this category), teams in the Adapt/ Not Transparent configuration offered roughly three times as much help as teams in the Make/ Not Transparent configuration. This shows that in case of events, team members are probably more aware of the fact that the other scheduler might need any form of help. Besides, by offering help it is assumable that in case of future events their team member will also offer help in return. The difference with teams in the Transparent configuration can be explained by the fact that these teams can see in the schedules if any form of help is necessary. If they do not see any problems in the other schedule, it is understandable that no explicit help is offered.

The correlation analysis may explain if the communication differences between the teams have a relationship with perceived task interdependence and the level of mental load. Table 4 shows the valuation of the correlations and strength of the relationship as used in this research.

r < .20 Very low correlation Negligible relationship .20 < r < .40 Low correlation Present but weak relationship .40 < r < .70 Moderate correlation Obvious relationship

.70 < r < .90 High correlation Strong relationship r > .90 Very high correlation Very strong relationship

Table 4: Valuation of correlations according to Guilford (1965. In: Den Boer et al., 1994:245)

(31)

31 supported. The reason that the relationship is weak could be that subjects who did not perceive a high level of task interdependence were still socializing with each other because they might have felt they were amongst friends during the experiment, instead of representing an official job. They probably would have socialized with each other in all circumstances, regardless of their task. The reason we still found a statistically significant difference might be that it was easier to switch to social talk for subjects who communicated already a lot with each other due to task interdependence. We did not find a statistically significant relationship between the level of mental load and social talk, thus hypothesis 4b is rejected. It is already mentioned that time pressure did not hinder subjects to find solutions. This implies that subjects had enough time left for social talk, regardless of their level of mental load. The lowering effect of social talk on mental load is probably only observable in cases of very high levels of mental load, which are not found in this research.11

A statistically significant positive correlation was found between perceived task interdependence and information exchange. Consequently, hypothesis 5a is supported. No statistically significant correlation could be demonstrated between the level of mental load and information exchange. This makes it plausible that the exchanged information was easy to process for subjects and that they were aware of each other’s information needs. Hypothesis 5b is therefore rejected.

We stated that there is a positive relationship between perceived task interdependence and helping behavior. Earlier research already demonstrated this relationship (e.g. Allen et al., 2003; De Jong, 2007; Wageman, 1995). However, neither suggestion, nor offering help voluntarily was found to have a statistically significant relationship with perceived task interdependence so hypothesis 6a is rejected. The reason could be that team members felt not only responsible for the successful performance of the other team member when they were task interdependent. Kiggundu (1981, 1983) has demonstrated that team members do not necessarily have to be task interdependent in order to feel responsibility for each other’s performance, but they can nevertheless choose to help each other (Ferrin et al., 2006. In: De Jong 2007:1627). No statistically significant relationship was found between mental load and helping behavior thus hypothesis 6b is rejected. We suggested that when the level of mental load was high, subjects would not have time to be fully aware of the schedule of a team member and that they needed all their mental capabilities for the successful performance of their own task. As previously mentioned did time pressure not hinder the subjects, which implies that they also had enough time left to help each other. The task was also viewed as medium mentally demanding12 so it is assumed team members also had the mental capabilities to focus on two schedules.

A negligible statistically significant relationship was found between perceived task interdependence and analysis/ determine strategy. This means hypothesis 7a is supported but it is of

11

The maximum observed score on mental load was 79 on a scale from 1-100.

12

AFFECTIVE BEHAVIOR OF SCHEDULERS ON PERCEIVED TASK INTERDEPENDENCE AND MENTAL LOAD