Group work as a solution for individual procrastination

(1)

Amsterdam Business School July 2015

Master Thesis

Group work as a solution for

individual procrastination

Author: Dao Thi Ngoc Thao – 10824316

Program: MSc Business Economics – Organization Economics Supervisor: Prof. Dr. Randolph Sloof

Abstract

This thesis examines the argument that group work can alleviate the postponement issue by investigating the effect of group work on individual procrastination behavior. Direct and indirect methods are employed to analyze the data from two sources (natural occurring data and survey information) from a sample of 195 students. The results show that people display less procrastination behavior in teamwork than they do in individual tasks. Additionally, the team members less inclined to delay in a group are also proved to have significant influence on the group’s postponement level. A small experiment regarding time preferences is also conducted and yields similar results. The robustness of the findings is tested and some possible explanations for the influence of group work on procrastination are also discussed.

(2)

Statement of Originality

This document is written by Student Dao Thi Ngoc Thao, 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.

(3)

Acknowledgements

I am taking this chance to send my sincere thankfulness to people who have supported me during the completion of this Master Thesis.

Among them, first and foremost, I would like to express my deep sense of gratitude to my supervisor, Prof. Dr. Randolph Sloof, who has provided me valuable scientific guidance and suggestions throughout the time I conduct this research, with his great enthusiasm and profound knowledge.

Second, I am grateful to Drs. Casper Troost and Dr. Claudia Buengeler from the University of Amsterdam (“UvA”) and Ms. Joyce Voogt from the ONCampus Amsterdam for their arrangement so that I could, for the purpose of this research, collect data and conduct the survey with their students in the Bachelor’s Program in Economics and Business at the UvA and the Foundation Program at the ONCampus Amsterdam. Furthermore, the considerable support and encouragement from Dr. Joeri Sol, Prof. Dr. A.J.H.C. Schram, Dr. A.M. (Sander) Onderstal, and other professionals from the UvA are highly appreciated.

My thanks also go to Dr. Jeroen van de Ven, my coordinator, for his contribution in navigating the initial ideas of this research project, and Ms. Péter Noémi, a PhD student introduced to me by my supervisor, for her sharing about the experiment methodology and results of her projects. Additionally, I would like to show appreciation to Viet-Dung Doan, a friend of mine, for his help in proofreading as well as his helpful comments and suggestions to improve my data processing and Stata analysis.

And last but not least, I wish to acknowledge my family for their wholehearted support and endless encouragement to me.

(4)

Procrastination is the practice of postponing an activity under one’s control to a later time, or even not performing it at all (Gafni and Geri, 2010). This tendency has gained increasing attention from academic researchers and has been reported as a fairly common phenomenon for roughly one in every five adults (Harriott and Ferrari, 1996). Postponement leads to the situation where the tasks are not started or must be completed within a short period of time just before the deadlines. In addition to the possible negative effect on quality due to working under time pressure, delaying the tasks until the last minute might decrease the capability to take on spontaneous jobs, because time slots close to the deadlines would be fully devoted to the tasks at hand.

Procrastination appears to be controlled while people work as a team, because it is observed that the typical group work involves dividing the tasks into smaller stages and setting appropriate timeline to ensure timely task completion. Furthermore, team members are expected to discuss and work together to complete the tasks, which requires the combination and harmonization of individuals’ preferences and restrains people from acting based solely on their personal traits. Therefore, it would be important for the organizations in assigning and managing tasks among employees if there is evidence that the procrastination problem can be solved by teamwork. In fact, some previous research papers have studied the relationship between personal procrastination tendency and working in groups. Most of these studies focus on the impact of personal traits on group performance. Whereas, this study aims at investigating whether being put in a group affects one’s personal delay behavior and consequently whether group work could act as a solution to procrastination problem. Put simply, people might have a tendency to delay doing their tasks and this would be reflected when they do their individual jobs. Nonetheless, when people work together in group projects, they would probably adjust their behavior to the group schedule aiming at committing to deadlines. It seems plausible that this effect will be stronger if people with a postponement tendency work in the same group as people showing less inclination for deferment.

In order to explore this aspect, this thesis focuses on directly comparing the behavior in individual and group activities, which is a new approach since the extant literature appears to focus only on personal perceptions and the effect thereof on group activity outcomes. The thesis

(6)

takes an empirical approach using data from both naturally occurring information and self-reported data from a sample of 195 students at the University of Amsterdam, The Netherlands. The empirical results corroborate the hypothesis that group work reduces the personal tendency to postpone, because when working in a team, people appear to adjust their behavior according to the group’s time allocation. Another noteworthy finding is that the postponement level of the group positively depends on the average tendency of its members. In term of time preferences, group discussion also alters the decisions of people with impatience and present-biased preferences, thus tackles the antecedents of procrastination.

In the next section, an overview of the existing literature relating to procrastination is presented. Section 3 explains the empirical methodology and provides a detailed description of the data used. Section 4 provides direct and indirect analysis of the gathered information. Finally, Section 5 concludes by summarizing the main points and discussing the limitations of the research. 2. Literature review

2.1 The occurrence and origins of procrastination

Procrastination has been receiving considerable attention from both psychologists and economists. For this thesis, it is important to know the antecedents of procrastination because it helps to identify the factors influencing postponement actions and allows the research to measure deferment level as well as investigate the influence of teamwork on procrastination behavior via those factors.

The psychological literature provides evidence that such a tendency is widespread and prevalent across diverse populations (Ferrari et al., 2007). Psychologists have also proposed possible explanations for the origin of procrastination tendency, ranging from genetic inheritance (Gustavson et al., 2014) to environmental factors such as parenting style (Ferrari and Olivette, 1994) or social supporting network (Ferrari et al., 1998).

It is worth noting that many psychologists have viewed the individual procrastination as a trait that shows consistency across time and situation (Steel, 2007; Schouwenburg and Lay, 1995). Therefore, this thesis only focuses on the influence of group work on postponement behavior, not the personal preference, or in other words, the research (i) bases on behavior in personal tasks to infer the procrastination traits, then (ii) compares behavior in personal tasks with behavior in

(7)

group work to investigate how each person behave differently in two circumstances, and finally (iii) draws conclusion about how people with certain procrastination tendency change their behavior when working in a team compared to when working alone.

From an economics point of view, postponement is usually studied together with deadline of the task (i.e. the latest point of time when the assigned task is expected to be finished). Labianca et. al. (2005) view deadlines as environmental stimuli that need to be perceived, interpreted, and remembered by individuals and groups to pace their activities appropriately for the task at hand. Many researchers have shared the conclusion that the ways individuals perceive deadlines and time pressure have considerable influence on their actions towards deadlines, including procrastination (Waller et al., 2001; Gervers et al., 2006; Gafni & Geni 2010)

Economics have also found a close relationship between postponement and the way people discount future value. A particular value at present is viewed differently, normally discounted at a certain discount rate if it is put at a future point of time. Accordingly, a higher discount rate leads to a lower future value, compared to the present value, and thus reflects the impatience level of the subject. Similarly, a lower discount rate is associated with higher future value and infers a patient person. There are also cases that the discount rate changes with the proximity between present and the future point, which are referred to as time inconsistency. Procrastination is hypothesized to be a natural consequence of (i) consistent impatience or (ii) time inconsistency, or more particularly, present-biased preferences.

Firstly, the relationship between consistent impatience and deferment is hereinafter considered. Impatient people give priority to enjoy gratification as soon as possible, instead of having to wait for a period of time before receiving the reward, although the latter payoff is probably greater. The reason is that their high discount rate would make the future reward less attractive than the present well-being. As a consequence, they would consistently prefer leisure activities now to leisure after task completion. Fischer (2001) models time as an exhaustible resource and shows that simple impatience can serve as a reasonable theoretical explanation for dynamically consistent procrastination (i.e. the postponement remaining unchanged over time). The author assumes that time can be allocated between work and leisure, that people like leisure and impatient people prefer it sooner rather than later. Fischer (2001)’s model of impatience offers

(8)

examples of situations where procrastination can be not only consistent over time, but also utility maximizing. Hence it explains procrastination caused by impatience.

Secondly, the following paragraphs explain time inconsistency and then the causal link between present-biased preferences and lagged behavior. If people are fully rational, they would have consistent discount rate and hence unchanged preferences for gratification regardless of the delay. Specifically, if one prefers 100 Euros today to 110 Euros next month, then he or she must also prefer 100 Euros in four months to 110 Euros in five months. However, many people discount the future values with different discount rates over time, resulting in time inconsistency; or put simply, their preferences for well-being at an earlier point of time over a later point of time are different depending on when the question is asked.

Among them, there are some people who suffer from the problems that urge them to choose short-run immediate gratification, with which their long-run self would not agree despite applying the same reasoning. This is referred to as present-biased preferences, which is the tendency, when considering trade-offs between two future moments, to give extra weight to the well-being at earlier point of time as it gets closer (O'Donoghue and Rabin, 1998). Basically, it means that those people are patient in the long-run yet impatient in the short-run. This is triggered by the hyperbolic discount functions, characterized by a relatively high discount rate in the short term and a relatively low discount rate in the long term. This hyperbolic discounting induces a conflict between present preferences and the preferences held in the future (Laibson, 1997). In the above example, a person with present-biased preferences would prefer 110 Euros in five months to 100 Euros in four months but, inconsistently, prefer 100 Euros today to 110 Euros next month.

People with present-biased preferences would prefer immediate gratification over future leisure and thus procrastinate the tasks at hand to enjoy free time at present, despite the fact that they might have to work more in the future. O'Donoghue and Rabin (1999) examine the implications of time inconsistency in a model where a person must do an activity exactly once, and report that if actions involve immediate costs (e.g. writing an assignment), people would procrastinate (i.e. to delay when they should work instantly), while if actions involve immediate rewards (e.g. seeing a movie), people would preproperate (i.e. to start promptly when they should wait).

(9)

However, people with present-biased preferences show different levels of delaying, depending on whether they are sophisticated or naïve. Sophisticated people are aware of their future self-control problems, thus have realistic expectation about their future behavior and self-control their procrastination tendency. In contrast, naïve people are not aware of their future self-control problems and believe that their future self will behave according to their expectation. Consequently, they keep postponing the task because every day they repeatedly believe that they will do the task tomorrow if they do not do it now, which then leads to severe procrastination (O'Donoghue and Rabin, 1998).

Briefly, economists have concluded that deferment is induced by the individual’s deadline perceptions or time preferences (i.e. impatience or naïve present-biased preferences). The origins of procrastination play an important role in determining different ways to measure this tendency, which are presented in detail in Subsection 2.2. Moreover, the difference between naïve and sophisticated present-biased people supports the expectation that subjects with present-biased preferences would change their behavior in teamwork, which is then discussed in Subsection 2.3.

2.2 Procrastination measurement

Since procrastination has been proved to originate from deadline perceptions, or impatience or naïve present-biased preferences, it can be measured by using the deadline of the task or by estimating the time preferences of the subject.

2.2.1 Procrastination measurement using the deadline of the task

The deadline of the task plays an important role in the classification of procrastinators. One’s tendency to delay is considered stronger if (i) the subject starts working on the project when the deadline approaches closer, or (ii) the task completion time is closer to or even after the deadline. In order to identify procrastinators, some articles make use of self-reported data from participants through interviews or surveys. Typically, participants are requested to rate their opinions on a scale from strongly disagree to strongly agree about such statements as: “I often find myself performing tasks that I intended to do days before”, “I usually delay tasks until the last minutes”, “I/my group finished the assignment too late/just in time/in ample time”, etc. (Harriott and Ferrari, 1996; Ferrari et al., 2007; Gevers et al., 2006). Using this method, the authors might obtain a large number of observations, as well as easily replicate other studies by simply using

(10)

the same questionnaires. Nevertheless, since the data are reported by participants and the researchers have difficulty in verifying their responses, such data may suffer from considerable measurement error.

Alternatively, in empirical papers, the proximity between actual completion time and deadline (counted in days, hours or minutes), or in other words the remaining time available to do a task, is usually used to measure the postponement tendency of the participants (Gafni and Geri, 2010; Brown and Previtero, 2014). As the data are collected by the researchers, this method addresses the measurement error of subjective self-reported data. Moreover, the consequences of participants’ actions are real (in case of naturally occurring data) or incentivized (in case of experiments) rather than hypothetical; the external validity of the findings is therefore strengthened.

However, depending solely on task completion time may not help to adequately reflect procrastination traits, because there are probably some people who finish the tasks long before the deadlines but would like to take advantage of the remaining time to review and improve the work if necessary. In this case, a close proximity between finish time and due time would possibly indicate carefulness but not postponement. Addressing this issue, Fischer (2001) argues that procrastination is not always characterized by missing deadlines or abandoning tasks, but can exhibit itself in an increasing workload, as more of the task is performed closer to the deadline. Put differently, having the same task submission times, people who have to work significantly more in the last minutes would be classified as procrastinating while people working less or keeping the same pace as before when the deadline approaches are not.

With the attempt to combine the above methods in analyzing deferment, the current research gathers actual data regarding assignment submission time, as well as conducts a survey with the subjects to study their personal or group tendency and workload allocation. In this approach, actual procrastination is estimated by counting hours between real submission time and task deadline, while personal postponement tendency is measured by subjective self-reported data in response to questions relating to time allocation and increased workload near the deadlines. 2.2.2 Procrastination measurement using time preferences

A person is considered procrastinator if his or her time preferences show impatience or present bias. Among many methods to determine time preferences, choice experiments are commonly

(11)

used to identify differences in patience and present-bias among individuals. A typical choice experiment design is described and applied in Meier and Sprenger (2010), in which participants are asked to make a series of choices, in two different time frames, between a smaller reward at time t and a larger reward one month later. Individual patience is captured by the subject’s discount factor, which in turn is identified by observing the point in a given price list where the individual switches from opting for the smaller earlier payment to opting for the larger later reward. Dynamic time inconsistency (i.e. either present bias or future bias) is identified by using information from two time frames (where t is the present and where t is in six months). An individual is present-biased if he or she shows less patience when the smaller, earlier reward is received in the present (t = 0). Meier and Sprenger (2010) classify 36% of their study participants as present-biased and 9% as future-biased. The results are found quite comparable to other studies using the same method, and are highly correlated with time preferences measures derived from other methodologies. In other studies, Dohmen et al. (2006) and Ashraf et al. (2006) report 28% and 27.5%, respectively, of their sample are present-biased.

The current thesis also takes advantage of choice questions to classify time preferences of its subjects and finds comparable results with nearly 20% of the participants are present-biased with respect to money and about 25% are present-biased in regard of free time.

Additionally, the students in the current research are required to discuss in groups of two or three members on the same choice question regarding free time. This allows people’s behavior when they are put in a group to be observed. More particularly for the purpose of this research, letting participants discuss the choice question regarding leisure time in groups provides information to investigate whether people classified as impatient or present-biased, after discussing with other group members, agree to choose another option, despite their individual preferences.

2.3 Procrastination in group work

As discussed above, two antecedents of lagged behavior are deadline perceptions and time preferences. For that reason, in order to examine whether group work can address the postponement issue, its influences with respect to deadline perceptions and time preferences need to be considered.

(12)

2.3.1 Group work and deadline perceptions

Many existing studies have focused on how team members perceive deadlines and the effects of these perceptions on groups’ performance. Using an empirical approach, Gevers et al. (2006) conduct a longitudinal study of 38 student groups of a business school in the Netherlands with the attempt to explore the antecedents of shared temporal cognitions and investigate whether groups are better able to meet deadlines when group members have shared temporal cognitions. They define temporal cognitions as opinions regarding task execution (for example, how to use time for the tasks), and study two antecedents of shared temporal cognitions, namely the similarity in group members’ pacing styles (time allocating style) and the exchange of temporal reminders (prompting each other to stick to agreements). Then, the authors perform multiple regressions and find that (i) both pacing styles and temporal reminders may benefit the emergence of shared temporal cognitions, and (ii) shared temporal cognitions may either facilitate or impede meeting a deadline, depending on the mean pacing style within the group. These measures are also applied in the current thesis for the purpose of proposing possible explanations to the findings. It is hypothesized that group work solves individual tendency to delay doing assigned tasks, or put differently, team members with procrastination tendency change their behavior when working in groups. If the hypothesis is confirmed, the change in behavior might be explained by either the groups’ mean pacing style being earlier action style, or the subjects receiving appropriate reminders from other members to stick to the agreed schedule. As analyzed in Section 6, the survey results of this study lend support to the notion that a group schedule, a shared pacing style and instant reminders are possible explanations for the influence of work group on the members’ lagged behavior.

Nevertheless, few studies have been conducted to investigate how working in a group may influence the individual behavior towards deadlines. Having a similar attempt to compare individual and group activities with this research, Gafni & Geri (2010) conduct a study on MBA students’ behavior towards their assignments and find that people are usually more punctual in their individual assignments than in general collaborative assignments. Although in the latter type of task, students do not really cooperative to finish the task but rather work individually (i.e. give comment on the analyses of other students), the authors argue that the assignment is considered collaborative because it requires students to rely on other’s work and refer to it. In

(13)

order to see whether working in groups influences personal procrastination behavior, the current research will explore this approach in the context where team members actually work together in the group assignment, and the result has equal impact on all team members, or in other words, every team member receives the same grade from the group work.

There are two possible directions that teamwork could affect the individual procrastination actions. On the positive side, group work can alleviate postponement because the rotten apples learn or are disciplined by the good ones. It is observed that the typical group work involves dividing the tasks into stages and setting an appropriate schedule to ensure task completion before deadlines. Accordingly, group members are prompted to work on the agreed timeline instead of delaying until the last minute. Moreover, members with a tendency to postpone are urged to work at a similar pace with their teammates and/or instantly receive reminders about finishing the tasks before the due time.

On the debit side, group work cannot lessen but even worsen the deferment. The good apples can possibly be spoilt by the bad apples and delay the tasks until the last minute, or the team members fail to solve their conflict of interest or impose discipline and thus cannot cooperate during the task process, which leads to late completion of group tasks. In fact, in respond to the survey of this thesis, some subjects of the current research also cite the difficulty in reaching an agreement with other team members and the fact that some people disobey the group schedule as the main reason for their group’s late assignment submission. Another point on this side is the free-riding problem occasionally encountered in teamwork. Free-riding is the tendency of members in a group to decrease their effort when they expect that their contribution is dispensable for the group’s performance because the success can be obtained through others’ efforts (Kerr and Bruun, 1983). It is likely that the members who want to finish the tasks in time and earn good score would have to do all or most of the jobs because they have to take on the work of those disobeying the group schedule; while other students free-ride on these good apples’ efforts. In such circumstances, through group work, postponement is aggravated as people have the opportunity to delay doing their jobs as much as possible or even to pass them on to other teammates.

Generally, group work is expected to have opposite impacts on members’ behavior. This thesis is meant to investigate whether group work can actually alleviate individual procrastination and

(14)

figure out the reasons for that effect. Deadline perceptions and shared temporal cognition are examined, since previous studies have proved their influence on group performance.

2.3.2 Group work and time preferences

With respect to time preferences, it is argued that in order to allow groups to reach consensus on a single group choice, the preferences for decision alternatives across group members need to be aggregated in a process described in Social Decision Scheme theory (Davis, 1973). Hence, in group work, a member would not be able to act according to his or her own preferences if other group members do not share the same opinion. It is also possible that during the discussion and working process with others, impatient and present-biased people would be aware of their self-control problems (or put differently, become more sophisticated), then adjust their expectation of future behavior and thus their delaying temptation. However, there is an equal chance that participants with the taste for immediate gratification or present-bias are grouped together, which leads to no improvement, or they can convince other people to choose their preferred options and hence increase the likelihood of deferment in group work.

In this regard, the current research takes initiative in conducting an experiment where participants firstly make decisions in lists of choices to infer their individual time preferences, and then are asked to work in groups of two or three people to make their shared choices in the same lists in the first stage. Types of subjective time preferences are identified through choice questions as explained in Subsection 2.1 and the difference in each subject’s choices when they answer individually and when they response as a group after discussing with other people is analyzed. If participants, who are classified as procrastinators in the individual questions (i.e. choose options showing impatience or present-biased preferences), change their decision to other options that do not show impatience or present-biased preferences after discussing with other people, then it is possible to draw a conclusion that working in group could influence impatient and present-biased individuals, and their procrastination behavior accordingly. Furthermore, it is also necessary to check other possibilities, i.e. whether after the discussion, procrastinators keep their choices unchanged, or non-procrastinators switch to delaying options.

(15)

3. Methodology 3.1 Main hypothesis

Given the above positive and negative effects of group work on procrastination behavior, this research offers the following main proposition:

Procrastination in individual tasks is stronger than that in group work.

If this hypothesis is confirmed, it is evidenced that people delay their tasks when they are in a group less than when they work individually. Then it can be said that ceteris paribus, teamwork can help to reduce postponement behavior. Otherwise, either there is not enough sufficient information for conclusion, or group work even aggravates postponement issue.

In order to examine the above proposition, this thesis analyzes administrative and survey data to study the differences in behavior of students when dealing with individual and group assignments. Additionally, the survey also includes a small experiment with choice questions to study the behavior of people with present-biased preferences when they act as individuals and as group members.

3.2 Participants

The subjects of this research are 195 students who are studying at the University of Amsterdam. They are students of the following three different programs: Program (I) – Economics and Business Pre-Bachelor program (23%), Program (II) – Bachelor in Economics and Business, first year (61%) and Program (III) – Bachelor in Economics and Business, third year (16%). Different educational levels are chosen in order to get some heterogeneity with respect to age and the number of courses they have to accomplish in the examined academic semester, which act as control variables as later explained in Subsection 3.3.

Participation in the research is voluntary and the subjects are guaranteed confidentiality as well as no influence on their studying results in any way. The students do not know about the research until they are asked to response to the survey, which is conducted after all assignments are completed. Overall, the students are from 17 to 28 years old; follow from two to seven courses (each course yields 5 credits) in the semester when this research is conducted; and approximately a half of them are male.

(16)

During the academic block, there are two types of assignment for the students, namely (i) individual work for which the students are required to work independently, and (ii) group work for which they have to do the assignments in groups of maximum three students. In group tasks, students are allowed to sign up for group membership, which means the students form groups on their own and the teachers do not get involved in group formation process, unless there are students unable to find a group themselves. There are 76 groups formed by the 195 subjects to do the teamwork.

Subjects are required to submit the assignments to the teachers before the announced deadlines via administrative systems (BlackBoard or Turnitin). The submission time is recorded in the systems and only accessed by the teachers, which means the subjects or groups of subjects have no official way to get information about the submission status of other students or groups. This feature eliminates the possible effect that the ability to see everybody’s progress in completing their tasks encourages people to work on their own projects, which was reported by Gafni and Geri (2010). All deadlines have been clearly stated in course descriptions sent to all students at the beginning of the courses, and group composition is decided within the first week of the courses. Essays handed in after the deadlines would not be graded, which means the students do not get any score from late submitted homework. Early submission is accepted but there is no incentive (additional score or more detailed feedback) for the early birds. Generally, the task requirements are comparable for the purpose of this research. A summary of the student tasks are given in Table 1.

Regarding the individual task requirements, students are asked to write essays on particular topics relating to the courses. In an eight-week-block, Program (I) students have to hand in three papers (500 words each) every two weeks starting from week 3 and these assignments make up 20% of the final grade. Students in Program (II) have to follow the course in two consecutive blocks and are expected to submit two 1,200-word-essays in each block (week 3 and 6 of the first block; week 2 and 4 of the second block). In total they have four personal papers for assessment, each of which accounts for 10% of the block final grade. The observed course of Program (III) requires the students to make statements relating to a specific article and support their statements by providing for and against arguments for the statements on a weekly basis. The total word count of each statement and its arguments should be about 200-300 words. All five assignments

(17)

during the course account for 10% of the final grade. The subjects also need to submit a final paper of 2,000-2,500 words at the end of the course for 70% of the final score.

TABLE 1

Summary of the student assignments

The table summarizes the requirements of individual and group assignments set out for 195 students from three different programs, and the proportion of those assignments in the final grade of the courses.

Program Number of students

Individual assignments Group assignments (Maximum 3 students) (I) Economics and Business Pre-Bachelor program 44 (23%)

3 essays on course’s topics 500 word each essay Every two weeks

All 3 essays: 20% of the final grade

In the same course with the individual assignments Develop and then present a business plan

20% of the final score (II) Bachelor in Economics and Business, first year 119 (61%)

4 essays on the course’s topic 1,200 words each essay

Week 3 and 6 of the first block; Week 2 and 4 of the second block

Each essay: 10% of the final grade of the block

In a different course with the individual assignments An essay and a presentation discussing a specific paper 20% of the final score

(III) Bachelor in Economics and Business, third year 32 (16%)

 5 assignments, each includes a statement relating to a course’s pre-listed article, and supporting arguments for and against the statement

200-300 words for each assignment Every week

All 5 statements: 10% of final score

 A final paper at the end of the course about a course’s topic

2,000-2,500 words

Final paper: 70% of the final score

In the same course with the individual assignments An essay and a presentation discussing a specific paper 20% of the final score

Some general notes:

 Students form groups on their own (maximum 3 students) within the first week of the courses

 All deadlines and requirements are announced at the beginning of the courses

 Late submitted work is strictly not graded. Early submission is allowed but not incentivized

(18)

In respect of group work, subjects in Program (I) work in groups to develop a business plan and later present the plan in the class. In Program (II) and (III), groups hand in their essays discussing one of the pre-listed articles and then also make presentation about the papers. The teamwork accounts for 20% of the final score in all the observed courses. For students from Program (I) and (III), the analyzed personal and group assignments are required by one course. However, for Program (II), there was no course that has both individual and group assignments. Therefore, the information collected are from two different courses.

3.3 Procedures

First of all, with the approval from the courses’ teachers and coordinators, administrative data regarding students’ ID (student number), group composition (the ID of students who work in a team), individual and group assignment submission time, and the corresponding deadlines are collected.

After the subjects finish all the assignments, a survey with two parts is conducted in the classes. There are about 12 to 23 students in a classroom when the survey was carried out. The survey questionnaire is attached in the Appendix 1 of this thesis. The first part of the survey includes (i) questions about their time allocation and workload during the time for the tasks (from assignment release time until the deadline); and (ii) some other choice enquiries to identify their present-biased preferences. After everyone finishes the first part, all the responses are collected and subjects are explained about the second part of the survey. In this part, students are divided into groups of two or three (students sitting next to each other would form a group) and are asked to discuss with their teammates to make joint decisions regarding the same leisure time choices questions that they have already done individually in the first part.

The two-part design is used to avoid that people may come back to their individual questions and change their choices after discussing and answering the group part, which leads to potential measurement problems. Information acquired from the survey is then matched with administrative data received from the teachers by student numbers.

The described method offers many advantages for procrastination research purposes. The most noticeable advantage is that administrative statistics are naturally occurring data and the outcomes are real rather than hypothetical or incentivized with only low-stakes, like in surveys

(19)

or experiments. If students perform poorly, or if they fail to submit the assignments on time, their study results will consequently be negatively affected. Moreover, the submission time is appropriately recorded by the administrative system, which decreases measurement error issues that might often be encountered when using subjective reported data. Nonetheless, submission times may not accurately reflect postponement, because there are probably some students who do not want to submit their work right after finishing, but instead use the remaining time to carefully review and make improvements if necessary. For that reason, additional information from the survey would contribute to a better interpretation.

With respect to the survey contents, the combination of different questions about time allocation and levels of workload when the deadlines approach (in two forms namely multiple choices with visual illustration and rating on a five-point-scale from strongly disagree to strongly agree), increases the confidence in the estimation of procrastination tendency. Additionally, the inclusion of enquiries about whether the subjects set out a schedule when working in groups, and choice questions to infer subjects’ impatience and present-biased preferences, are made for identifying possible explanations for the findings.

In the time preferences questions, subjects are asked to make their decision in choice lists of monetary and leisure day rewards. With regard to time preferences questions for the group, only leisure day rewards are used to avoid arbitration between subjects, which might be observed if money is used instead. For example, suppose subject (I) and (II) are put into a group to choose between:

A. Each receives 100 Euros today; or B. Each receives 110 Euros next month.

Subject (I) prefers 100 Euros now while subject (II) wants to receive 110 Euros next month. Subject (II) realizes the arbitrage opportunity when (II) gives 100 Euros to (I) now and asks (I) to agree with option B, then (II) will be able to receive 220 Euro in one month, which means a net amount of 120 Euros. That being the case, subject (II) might decide to give 100 Euros to subject (I) now to make (I) agree to choose B. As a result, subject (I) changes his or her option, but the real consequence received is not changed.

With questions about leisure days, arbitration is mitigated because the subjects cannot easily offer their benefits to other people.

(20)

3.4 Measures

3.4.1 Assignment data

Actual procrastination is estimated by counting the hours between assignment submission time and the corresponding deadlines. ProcIn and ProcGr signify postponement in individual and group assignments, respectively, and are calculated as follows:

ProcIn = Median (Individual submission time – Corresponding deadline) of different tasks. ProcGr = Group submission time - Corresponding deadline.

Consequently, these proxies are expected to take a non-positive value because the assignments are only graded when they are handed in on time. A larger value (less negative) of these actual procrastination proxies implies stronger procrastination, as the submission time is closer to the deadline. However in fact, the data obtained show some cases of late and non-submission of work. As confirmed, the professors strictly follow the rules stated in the course description, which means the students would not receive any grade if they either fail to send their work or make late submission. Therefore, without loss of generality, those who postpone the tasks the most (i.e. either do not hand in the assignments or present their work after the deadlines) receive an actual procrastination value of 1. Briefly, the proxies ProcIn and ProcGr for actual procrastination receive a negative value when the subject sends the assignment before the deadline, they have a value equal to 0 when the submission is made precisely at the due time. Finally, they get a value equal to 1 if the participant fails to send the work on time or does not hand in anything at all. In addition, people working in the same group would receive the same value of ProcGr.

The students have more than one individual task. In addition, there is a considerable variation in the points of time when a particular student submits his or her different assignments. In this circumstance, the median, compared to the mean, of the actual procrastination values of different assignments better represents the subject’s postponement. Put simply, suppose there are two students who have to complete 3 individual essays during a course. The first student always finishes and sends the assignments one day before the deadlines, whereas the other person submits two essays just a few minutes prior to the cut-off points and presents the third one three days in advance. Because they have the same mean value, the median values would more

(21)

accurately reflect the difference in their behavior. Accordingly, ProcIn is computed by taking the median of the values generated by different tasks.

It is also worth noticing that for each assignment unfinished or submitted late, the students receive a value equal to 1; and the variable ProcIn are the median of many tasks’ values. This explains why there are some students with ProcIn falling in the range between 0 and 1. If the subject only has one late or non submission, the proxy still gets non-positive value, since it takes the median of many tasks. The more improper submission, the closer the value gets to 1.

Subjective self-reported tendency is measured based on the participant’s responses to the survey questionnaire.

Individual tendencies: In the survey, students are asked to choose the time allocation that they adopt when doing the individual assignments, selecting one out of 5 options in which the delaying issue ranges from unnoticeable to significant levels (The survey questions can be found in the Appendix 1). Additionally, subjects also rate their opinions on a five-point scale about statements relating to procrastination: “I always postpone doing the tasks to the very end”, and “My workload increases significantly in the due date of the assignments”. All the three questions aim at revealing the participant’s taste for postponement. For each of these enquiries, each individual receives a value from 1 to 5 implying their level of procrastination. The personal delaying tendency TendIn is calculated by taking the mean of these values across questions for each person.

Group work tendencies: Similarly, subjects are asked to indicate the way they allocate their time for the group tasks among 5 options. Moreover, they also rate their agreement on the statement: “My workload when doing the group assignment increased significantly in the due date (compared to other days during the task timeline)”. Similarly to the above, this question is meant to distinguish between people delaying the group tasks and those who carefully review the work. The overall delaying tendency TendGr of subjects in group work is calculated by taking the mean of the responses to the above questions.

It should be noted that people in the same group may have different opinions in response to these queries. It may be the case that the groups agree to divide the jobs into smaller parts for each member, then they can work according to their own time allocation. A person with a preference for early task completion would finish the assigned tasks before the due date, hence his or her

(22)

duties in the last day of the group work mainly include reviewing and putting small parts into complete papers. Whereas a student who delays working until the last day is expected to experience an increased workload in order to finish the assigned parts and then integrate them into the final work of the group.

Furthermore, the response comparison between these question and their counterpart relating to personal tasks may reflect the change in individuals’ behavior if any. For example, a person may have increased workload in the due date when handling personal jobs, but would change behavior when dealing with group work, start working early and experience no huge workload in the last date. Such differences is examined by the sign test in Subsection 4.2 by directly comparing the procrastination in individual and group assignments of the same subject.

Briefly, the two variables TendIn and TendGr represent the self-reported postponement tendency and get value between 1 and 5, in which the former value means invisible deferment and the latter one suggests serious procrastination.

Control variables in the regressions include age, gender, the number of courses the students follow in the previous academic semester, and the program fixed effects (binary variables). In this case, the Pre-Bachelor program does not apply the European Credit Transfer and Accumulation System (ECTS). Also, all the courses of Program (II) and (III) in the observed semester give 5 ECTS. Therefore, the number of courses that the students study in the semester is chosen as a control variable instead of the number of ECTS.

In addition, the paper also includes binary variables to capture the effects from time preferences relative to deferment (i.e. consistent impatient and present-biased preferences) and to check whether these time preferences are indeed the driving factors behind postponement.

3.4.2 Time preferences

Two binary variables PrefMoney and PrefLeisure are created for the examination of subjects’ time preferences regarding money and leisure time awards, respectively. These variables receive a value equal to 1 if the subjects show either time impatience or present-biased preferences. Specifically, for the following couple of questions:

1. A. Receive 100 Euro now; or

(23)

2. A. Receive 100 Euro four months later; or B. Receive 110 Euro five months later.

The subjects are classified as impatient if they choose option A in both questions. If the participants opt for A for question 1 and B for question 2, then they are considered as having present-biased preferences. Regarding group question relative to leisure time, variable PrefGr is determined in a similar way, which receives a value equal to 1 if the group show either time impatience or present-biased preferences.

There are two main purposes of the use of time preferences variables in this research. Firstly, they act as control variables in the regressions explaining individual procrastination. This allows to check whether time preferences indeed have influence on the subjects’ deferment behavior. Secondly, the comparison of their choices in the independent part (indicating their personal preferences) and their group’s decision in the teamwork part (inferring the behavior in group work) support the notion about the impact of teamwork on procrastination.

3.5 Data analysis methodology

In order to examine the main proposition that procrastination in individual tasks is stronger than that in group work, this thesis analyzes the collected data in both direct (using comparison tests) and indirect (via regressions) methods.

3.5.1 Direct methods for data on student assignments

As explained in the above subsections, the collected data are well grounded source to infer procrastination in both individual and group tasks and hence can be used for direct comparison. In order to examine the mentioned main hypothesis, there are two tests applied to directly compare and provide evidence whether procrastination in personal assignments is larger than that in group work. Because a larger of postponement proxies means greater procrastination, the hypothesis is confirmed if individual procrastination’s figures are larger than the numbers inferring postponement in group. Both the actual figures and the subjective reported tendency are put forward for these two tests to see the difference between personal and teamwork.

The first direct tool, the two-sample Kolmogorov-Smirnov test of the equality of distributions (“KS test”), is used to determine whether there are differences in the distribution of procrastination values for personal and group assignments, and which sample has the smaller

(24)

values. The null hypothesis that values from both types of assignment were sampled from populations with identical distributions is expected to be rejected. The advantage of this type of test is that it is non-parametric and makes no assumption about the distribution of data. For that reason, it allows the direct comparison of the natural occurring data between 195 observations in individual tasks (ProcIn) and 76 observations in group work (ProcGr). Since people in the same group have the same value of ProcGr, this feature of the test helps to avoid repetition of the group value, which might lead to measurement error if ProcGr is examined at the individual level.

Secondly, the main proposition can also be directly tested using the sign test of matched pairs. While the KS test considers each group consisting of maximum three students as an observation, the sign test also provides a comparison of natural occurring procrastination figures in the same two cases (i.e. individual and group assignments) but views each students as an observation in all the cases. Put simply, at the individual level, every student has a pair of values representing their postponement, one refers to the independent tasks and the other one is for group work. It should be noted that members in the same group have the same value for procrastination in group work. The equality of each pair of values is tested and the null hypothesis that the median of the differences is equal to zero is expected to be rejected.

3.5.2 Indirect method for data on student assignments

Different regressions are proposed to further examine the influence of working in groups on individual behavior. Moreover, some specific hypotheses relative to the regressions are presented in order to eventually substantiate the main proposition of the research.

Firstly, individual tendency and actual procrastination are analyzed to see the relationship between the personal traits and the real data:

ProcIni = β0 + β1 TendIni + β2 Controlvariablesi + ui (1)

Hypothesis 1: Individual postponement tendency has positive influence on actual personal procrastination (β1 > 0).

Confirming hypothesis 1 means that when doing individual tasks, people will behave in accordance with their general tendency, in other words, people with procrastination tendency will choose to postpone doing the assignments.

(25)

Secondly, the influence of individual tendency on group procrastination is investigated in the following regression:

ProcGri = θ0 + θ1 TendIni + θ2 TendGri + θ3 MinTendIni + θ4 MeanTendIni

+ θ5 MaxTendIn + θ6 Controlvariablesi + vi (2)

The above regression is added with three following variables: (i) MinTendIn – the personal self-reported tendency of the teammates who has the least significant procrastination issue, (ii) MaxTendIn – the most serious deferment tendency in the group, and (iii) MeanTendIn – the average postponement tendency of the team members.

Hypothesis 2:

(a) The postponement tendency reported when the subjects work in group has positive effect on group actual procrastination (θ2 > 0), and the influence is stronger than that of the

tendency in individual tasks;

(b) Group procrastination is positively influenced by the personal tendency of the member who has the least significant procrastination issue in the group (θ3 > 0); and/or

(c) Group procrastination is positively influenced by the mean personal postponement tendency of the group members (θ4 > 0).

Variable MeanTendIn are used as independent variables in separate regressions from those using MinTendIn and MaxTendIn because they show high multicollinearity and consequently may lead to inaccurate results. Similarly, TendIn and TendGr are investigated in separate regressions. If it is true that group work can act as a solution for individual postponement, then people working in groups will not procrastinate as much as they would when working alone. Therefore, the effect of personal trait on group deferment is expected to be weaker than the influence observed on individual assignments. Instead, group procrastination is expected to be affected by the tendency that subjects report when they do group work. This, combined with the direct tests’ findings that the tendency to delay in group work is weaker than that in independent jobs, would substantiate the main proposition that teamwork reduces procrastination behavior.

This is a set of regressions at the individual level (195 observations), however, there are only 76 values of ProcGr shared by the teammates. Therefore, in order to correct for autocorrelation, the

(26)

regressions are clustered at the group level (by the variable indicating group ID of the students) to correct the error terms.

Additionally, actual group procrastination is also anticipated to be positively influenced by the personal tendency of the team member who has the least significant procrastination issue among the group, and/or the mean postponement tendency of the group members. The reason for this expectation is that the person who is least likely to delay would act as a controller and constantly remind other members to work on the tasks; and/or members of the group would discuss and set out a schedule for assignment completion that they are all agree with, thus the timeline should reflect the combination of team members’ individual traits. Otherwise, if group deferment is evidenced to be affected by the most serious postponement taste in the team, then the data fail to prove that individual procrastination is alleviated by teamwork.

Alternatively, another regression is proposed to investigate the determinants of postponement in group work at the group level (76 observations):

ProcGri = γ0 + γ1 MinProcIni + γ2 MedProcIni + γ3 MaxProcIni

+ γ4 Controlvariablesi + si (3)

In this regression, the actual procrastination figures are examined and the deferment of a group is projected to depend on its members. The regression results showing who has the most noticeable impact (the one submitting their individual assignments soonest - MinProcIn, latest - MaxProcIn

or the middle one - MedProcIn) would contribute to study the effect of working in a group with other people on individual behavior (which eventually affects and reflected by the time the group hand in their assignments). For groups that only have two members, the variable MedProcIn is missing. The control variable values are calculated by the mean of the team members.

Hypothesis 3: The actual deferment of a group depends on its early bird members.

3.5.3 Time preferences

Regarding the data on time preferences, at first, correlation check is conducted to investigate the relationship between procrastination tendency and time impatience and present-biased preferences. Then, data are examined to see whether subjects opt for other choices which show neither impatience nor present-biased preferences after group discussion.

(27)

As specified in Subsection 3.4, each participant receives a value of variable PrefLeisure indicating their time preferences. If the value is equal to 1, the subject suffers from impatience or present-biased preferences and is expected to be procrastinator. In the second part of the survey, subjects answer the choice question in a group and the value of variable PrefGr inferring whether their group’s choice shows preferences for immediate gratification or present bias or not. Directly comparing the values and distributions of PrefLeisure and PrefGr using the KS test and sign test can draw a conclusion about whether or not people change their decision and the change direction when when they work with other people or.

4. Empirical results regarding student assignments

This section presents an overview of the collected information, and analyzes the results from direct tests as well as regression outcomes to examine the main hypothesis relative to the research question.

4.1 Data overview

4.1.1 Sample statistics and natural occurring data

Table 2 presents summary statistics (age, gender, the number of courses they study in the studied semester) of the observation sample and the collected data relating to actual procrastination. Regarding the individual procrastination, there is one outlier of the variable ProcIn detected with the value of -147.983 hours. In order to minimize the impact of this outlier, variable ProcIn is winsorized (replace the extreme values with upper/lower cutoff) at the 1% level. Thus, this value is replaced by the range’s lower cutoff, which is -73.7 hours. The statistics of ProcIn after the winsorization are also presented in the first panel of Table 2. Data analysis hereinafter use the winsorized values of this variable. The variable ProcGr appears not to contain any significant outlier.

From Table 2, it is noticed that the actual figures representing procrastination in personal work (ProcIn) are less negative than those in group jobs (ProcGr), which supports the idea that people are less likely to delay working when dealing with team tasks than when doing personal jobs.

(28)

TABLE 2

Summary Statistics of the observation sample and the natural occurring data

This table presents summary statistics for the sample used in the analysis of group work’s influence on personal procrastination. The first panel illustrates total data for 195 students from three programs examined, including subjects’ age, the number of courses they follow in the observed semester, and the proximity between task submission time and the corresponding deadlines of individual and group work, which is used as the proxy for personal and group deferment.

A larger value indicates stronger procrastination and the maximum value of 1 means late or non submission. The data are winsorized to minimize the impact of outlier and thus there is a change in the individual figures. The remaining three panels show data of each of three observed programs.

Total 3 programs Obs Mean Std. Dev. Min Max

Age 195 20.108 1.754 17 28 Courses 195 4.872 0.738 2 7 ProcIn 195 -11.367 15.970 -147.983 1 ProcIn_win 195 -10.986 13.368 -73.700 1 ProcGr 76 -22.615 21.947 -105.483 1 Male: 49.74%

Program (I) – Economics and Business Pre-Bachelor

Obs Mean Std. Dev. Min Max

Age 44 19.068 1.648 17 24

Courses 44 3.909 0.563 2 5

ProcIn 44 -20.063 27.063 -147.983 1

ProcGr 15 -24.161 33.871 -105.483 -1.150

Male: 43.18%

Program (II) – Bachelor in Economics and Business, first year

Age 119 20.017 1.275 18 24

Courses 119 5.017 0.344 4 7

ProcIn 119 -7.664 7.422 -42.342 1

ProcGr 50 -24.627 17.947 -77.817 1

Male: 53.78%

Program (III) – Bachelor in Economics and Business, third year

Age 32 21.875 2.121 20 28

Courses 32 5.656 0.701 4 6

ProcIn 32 -13.180 14.388 -65.827 -0.046

ProcGr 11 -11.362 16.276 -50 1

Male: 43.75%

Obs = Observations; Std. Dev = Standard Deviation;

(29)

FIGURE 1

Distribution of procrastination in individual assignments (Natural occurring data)

The bar chart shows the distribution of procrastination in individual assignments (ProcIn) of 185 subjects, which is computed by the different between task submission time and the relative deadlines. The data have been winsorized to minimize the impact of outliers.

FIGURE 2

Distribution of procrastination in group assignments (Natural occurring data)

The bar chart shows the distribution of procrastination in group assignments (ProcGr) of 76 observations (formed by 195 subjects, maximum 3 students in a group), which is computed by the different between task submission time and the relative deadlines. There is no outlier detected at the 1% level. Stronger deferment is inferred from a larger value (less negative) of ProcIn or ProcGr. The maximum value of 1 means late or non submission.

The height of the bar shows the percentage of subjects sharing the postponement values on the horizontal axis. A higher bar at one value indicates that there are more people submitting assignments at that point of time before the deadline.

0 10 20 30 40 50 Pe rce n t -100 -80 -60 -40 -20 0 ProcGr 0 10 20 30 40 50 Pe rce n t -100 -80 -60 -40 -20 0

ProcIn, Winsorized fraction .01

(30)

Figure 1 and Figure 2 provide a distribution comparison of the two variables ProcIn and ProcGr after winsorization. The two charts highlight a considerable difference in postponement distribution. The vast majority of participants are reported to hand in personal papers within 20 hours before the due time. Whereas for group assignments, this is not the pattern since there are more groups finishing the tasks more than 20 hours before the deadline. This difference in distribution also corroborates the statement that people postpone more when working independently than when working in teams. In term of statistical comparison, the values and distributions of ProcIn and ProcGr are tested using KS test and signed rank test whose results are discussed in Subsection 4.2.

4.1.2 Subjective-reported data

In regard of tendency self-reported in the survey, as mentioned earlier, the two variables TendIn and TendGr act as proxies for self-recognized tendency to delay when working independently and when working in groups. These variables are computed by the mean of values received from different questions in the survey. Therefore, it is necessary that the queries are designed in such a way that the responses from subjects follow the same direction. The correlation of the survey answers may increase the confidence in calculating and using TendIn and TendGr variables. Table 5 displays summary statistics and correlations of the survey information and reveals positive correlations among the responses to the mentioned questions. Thus, TendIn and TendGr can be used as reliable variables to measure reported deferment tendency in personal and group assignments. Additionally, the correlation between the responses to group questions is also positive but weaker than those for individual assignments, which may imply the influence of other team members, in addition to the personal tendency, on one’s behavior in group work. On the one hand, people choose to allocate time and workload according to their own preferences for procrastination. Hence, their replies to the survey questions regarding personal assignments show strong correlations. On the other hand, when working in a group, the subjects have to discuss with their teammates to set a time schedule, then allocate their efforts based on the agreed plan and their own delay traits.

Moreover, a slightly positive correlation between TendIn and TendGr provides evidence that people tend to behave differently when they work as an independent subject and when they act as a team member. Otherwise, a strongly positive correlation between TendIn and TendGr is

(31)

anticipated if the participants treat individual and group assignments equitably. It is also worth noticing that, having the same pattern with the actual data, value of TendGr is smaller than TendIn, which substantiates the opinion that group work solves personal deferment.

Similar to the actual figures, the distribution of these subjective reported data depicted in Figures 3 and 4 reveals the overall procrastination trend of the sample set. Regarding personal traits, most people report their tendency with value of 4 and 5, which indicates strong preferences for deferment. However, in group assignments, the distribution mainly spreads between level 2 and 4, implying less severe procrastination.

TABLE 3

Summary statistics and Inter-correlations of self-reported data

The table exhibits summary and correlations of survey data regarding postponement tendency reported by 195 participants. The first panel shows information when subjects are asked to indicate their time allocation, delay tendency and whether they have increased workload in the due date when handling personal tasks. The answers take value from 1 (no preference for delaying) to 5 (significant preference for delaying). The median of these responses are chosen to proxy procrastination traits in individual work. The second panel presents similar data for group homework.

Individual assignments Mean Std. Dev.

Correlations

1 2 3

1. Time allocation 3.938 1.191 --

2. Delay 3.467 1.245 0.771*** --

3. Increased workload 3.687 0.958 0.697*** 0.681*** -- Personal tendency (TendIn) 3.697 1.021

Group assignments Mean Std. Dev. 4 5 TendIn

4. Time allocation 3.128 1.102 --

5. Increased workload 3.000 0.958 0.467*** --

Tendency in group (TendGr) 3.064 0.884 0.209***

Observations: 195 Std. Dev = Standard Deviation; * p<.1, ** p<.05, *** p<.01

Group work as a solution for individual procrastination

Master Thesis