Bachelor Thesis
The Effect of Task Assignment on Individual
Performance
Evelyn van Duren 10447989 Student at the Faculty of Economics and Business BSc Economics and Business – Finance and Organization Track Supervisor: dr. Silvia Dominguez Martinez Date: January 31th, 2017Abstract
Individual performance is an important determinant for overall organizational success. Organizational structures can be improved by allocating tasks in an effective manner. This thesis examines the effect of different approaches to task assignment on individual performance. Naturally occurring data on test performance from a sample of 471 students is analyzed to determine whether it is more efficient to initially assign low-challenging or high-challenging assignments when performing multiple tasks in a row. The results suggest the beneficial effects of initially assigning high-challenging tasks compared to initially assigning low-challenging tasks.Evelyn van Duren 10447989 Student at the Faculty of Economics and Business BSc Economics and Business – Finance and Organization Track Supervisor: dr. Silvia Dominguez Martinez
Statement of Originality
This document is written by Evelyn van Duren, who declares to take full responsibility for the content 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."
Table of Contents
1. Introduction ...5 2. Literature review ...7 2.1 Psychological effects on performance ...7 2.1.1 Decreasing performance ...7 2.1.2 Enhancing performance ...7 2.2 Task challenge...8 2.3 Sequential tasks... 10 3. Methodology ... 11 3.1 Hypothesis ... 11 3.2 Data... 11 3.3 Measurements... 12 3.3.1 Measurement final examination ... 12 3.3.2 Measurement midterm examination ... 13 3.4 Procedure... 15 3.5 Analyses ... 15 3.6 Control variables... 16 4. Results ... 18 4.1 Descriptive statistics ... 18 4.2 Primary analysis... 19 4.3 Secondary analysis... 21 4.4 Tertiary analysis... 22 5. Discussion ... 24 5.1 task assignment and performance ... 24 5.2 Effort... 25 5.3 Ability... 26 5.4 Limitations and future research... 26 5.5 Practical implications... 27 6. Conclusion... 28 References... 29 Appendices ... 311. Introduction
Motivated individuals and employees are an essential asset for organizations. Enhancing individual performance improves overall organizational performance. Task assignment, the type of assignments and the order in which they are allocated, seems to play a crucial role in the mood effects of employees and their performance (Preenen et al., 2016). For example, assigning interesting and compatible tasks to the right employees could advance their perceived quality of job experiences, which results in motivated and hard-working employees. Task assignment determines which individuals engage in which organizational processes, and consequently their views and performances will frame organizational structure. Therefore, this thesis investigates the effect of two approaches to task assignment on individual performance.
The thesis seeks to examine the effect of task allocation on performance by analyzing naturally occurring data from a sample of 471 students from the University of Amsterdam. In order to explore this effect, the thesis focuses on comparing various difficulty level structures of sequential assignments. The primary purpose of the research is to examine whether individuals who are initially assigned low-challenging tasks perform better than individuals who initially are assigned high-challenging tasks in the case where they have to perform multiple tasks in a row. Based upon the literature, we expect individuals who start with low-challenging tasks to outperform individuals who start with high-challenging tasks.
This study contributes to the current literature in several ways. First, the existing literature is mainly focused on the different difficulty levels of performing tasks only once or twice (Preenen et al., 2014; Taylor, 1981). Incorporating multiple tasks in a row gives new insights into efficient task allocation. Second, the effect of task assignment on performance has mainly been studied through artificial lab-experiments (Kruger and Dunning, 2001; Taylor, 1981). The use of administrative data in the current study provides the opportunity to investigate the effect in a real-life setting.
The empirical results contradict the hypothesis and point to the beneficial effects of initial task challenge on student performance. However, the results are not statistically significant for all analyses carried out. These findings could result in more effective methods to assign tasks in organizations, and would therefore be a promising basis for further study.
The remainder of the thesis is structured as follows. The next section presents an overview of the existing literature on task assignment and individual performance. Section 3 elaborates on the methodology investigating if there is a significant effect of task assignment on performance. Section 4 presents the results of the conducted analyses. The implications of these results are further discussed in section 5. Lastly, a conclusion will be drawn in section 6.
2. Literature review
The relation between task assignment and performance has been the focus of substantial prior research. In this section the existing literature on performance, confidence and task assignment will be analyzed. The literature on performance will be discussed first. Secondly, a comparison will be made between high-challenging assignments and low-challenging assignments. Finally, the literature on sequential task assignment will be examined.
2.1 Psychological effects on performance
Prior research in organizational economics has focused on the different factors that influence individual performance. One of these determinants is a person’s psychological state, which can have a positive as well as a negative effect on performance depending on the type of state (Compte and Postlewaite, 2004). 2.1.1 Decreasing performance Related research in psychology indicates the negative effects of one’s psychological state on performance. Experiments have shown that stress levels (Steele and Aronson, 1995) and depressed moods (Ellis et al., 1997) have a negative impact on individual performance. In the latter study conducted by Ellis et al (1997) subjects were induced either a neutral or a depressed mood. Subjects who were induced a depressed mood performed consistently worse at a grammar task than the subjects who were induced a neutral mood. Another common psychological state that decreases individual performance is choking: a physical response to the fear of failing. Choking could be the response to fear of performing poorly on exams, which compromises results. The response is especially triggered when an individual recalls experiences of performing poorly in the past (Compte and Postlewaite, 2004).
2.1.2 Enhancing performance
One psychological state that enhances individual performance is over-confidence; holding overly optimistic views of one’s own competence and performance. Social psychologists have found people often misjudge their own competence. Kruger and Dunning (2001) found incompetent people tend to hold misled views of their abilities. The authors conducted four studies in which they tested students on their social and
intellectual abilities. The students performing the worst systematically overestimated their performance in both domains. Kruger and Dunning attributed this over-estimation to the lack in self-monitoring skills incompetent people have. Incompetent people are not capable of accurately assessing their own performance leading to overly favorable views of abilities, hence overconfidence. Once incompetence was eliminated, by training the students to solve the task, their overconfidence decreased. On the contrary, competent people were better in assessing their own abilities but still held some misplaced views. They tended to rate themselves below average when asked to evaluate their own competence compared to others. Similar research in psychology confirms the tendency of people to evaluate themselves more favorably than others. In one research conducted by Guthrie et al. (2001) nearly 90 percent of the participants evaluated their own performance above average.
Overconfidence can have a positive effect on performance by promoting productivity and creativity. Compte and Postlewaite (2004) incorporate biases, such as overconfidence, into a standard rational decision-making model. They determine situations in which over-confidence increases long-term performance. Individuals who are over-confident of their success rate perform better in the long run than individuals who assess their success rate accurately. The mechanism through which biased perceptions lead to positive outcomes is the following. Biased perceptions lead to optimistic beliefs of future outcomes. These optimistic beliefs lead to enhanced performance because negative past occurrences, which could hold an individual back in current performance, play a less important role in decision-making. For example, suppose that a student making a test recalls past events. The student might have failed on the previous test, but attributes this to a circumstance outside his capabilities. The negative effect of this past occurrence is traded-off against the optimistic belief that this was outside his power. Bénabou & Tirole (2002) also incorporate the positive psychological effects of confidence into economic analyses. Confidence is an important tool for information processing and consequently decision-making, affecting rational agents. Confidence and positive mind-sets increase positive outcomes.
2.2 Task challenge
The influential role of one’s psychological state in positive outcomes makes it necessary to look at ways to motivate individuals in work settings in order to positively influence
one’s psychological state. Task assignment is an important motivational determinant for employees within organizations. The literature defines challenging assignments as difficult to achieve, stimulating, new, demanding, and a test of abilities (Preenen et al., 2014). They involve the ability to perform non-routine skills with a high level of independence (Preenen et al., 2011). The literature points to the importance of challenging assignments in organizations.
Berlew and Hall (1966) were the first to introduce the beneficial effects of task challenge on performance. Empirical research over the course of several years on the starting careers of managers showed the importance of initial job challenge for success later on. Managers who were assigned high-challenging tasks in their first year of work were more prone to have successful careers over a five-year period than managers who were assigned less challenging tasks in their first year. However, the researchers contribute two factors to this success. First, the company found a way to initially assign challenging tasks to their most competent employees. Second, assigning high-challenging tasks in the first year leads to enhance individual performance. Later research relates high-challenging assignments to increased employee retention (Preenen et al., 2011), on-the-job-learning (McCauley et al., 1994) and managerial development skills (DeRue & Wellman, 2009). Preenen et al. (2014) extended the research on high- versus low-challenging assignments by incorporating the effect of goal orientation. Students were asked to either give a presentation in front of a camera (perceived to be high-challenging) or to improve a reference list conform the APA-guidelines (perceived to be low-challenging). Task instructions were given to induce either a performance-approach, which focuses on results, or a mastery-approach, which focuses on individual development. The researchers found that the high-challenging assignment led to positive mood effects with a mastery-approach orientation, while the low-challenging assignment led to positive mood effects when inducing a performance-approach. However, challenging assignments do not always lead to positive outcomes. Dong et al. (2013) show the ambivalent effect of job assignment on one’s psychological state. For some individuals high challenging tasks lead to unpleasant feelings of anxiety and distress, which influences their performance negatively as mentioned in the first paragraph.
2.3 Sequential tasks
The research question focuses on the effect of task assignment on performance in case individuals have to perform multiple tasks in a row.
The literature suggests an effect of past task experiences on future task assessment. Past experiences are relevant in assessing the probability of a particular outcome. Whenever tasks are similar people expect the outcome of the past to reoccur. Frequent failure when performing a task in the past makes succeeding in the future more challenging (Compte and Postewaite, 2004). Presumably successful completion of a task in the past makes succeeding in the future more likely.
Taylor (1981) extended the research on task challenge conducted by Berlew and Hall (1966) by not only looking at initial task challenge, but including the build-up of task challenge. Taylor studied the effect of initial task challenge versus later task challenge when individuals had to perform multiple sequential tasks. An experiment was set up where male students had to perform two or three skilled assignments shortly after each other. The assignment consisted of hidden word puzzles (low-challenging) and logical word problems (challenging). Students who were assigned the high-challenging task later on scored higher on job-satisfaction, accurately assessing one’s own competence and performance expectations compared to students who were initially assigned the high-challenging task. However these process variables did not have a strong effect on performance.
3. Methodology
This section discusses the methods used to conduct this research. The hypothesis following from the literature will be discussed first. Secondly, we will discuss the research design, including the data, the measurements used and the procedure. Lastly, the model of the analyses will be presented.
3.1 Hypothesis
Initially assigning high-challenging tasks could lead to choking due to the fear of failure; assigning low-challenging tasks instead could lead to a confidence-boost. Based upon the importance of confidence stressed by Compte and Postlewaite (2004) and the positive effects of later task challenge on individual performance presented in the study of Taylor (1981), the main hypothesis of this research is the following:
Individuals who are initially assigned low-challenging assignments perform better than individuals who initially are assigned high-challenging assignments when performing multiple small assignments in a row.
When the hypothesis is accepted, there is sufficient evidence to accept the positive effect of initially assigning individuals low-challenging tasks compared to initially assigning individuals high-challenging tasks. In that case task assignment will have a positive effect on individual performance when holding all other variables constant.
3.2 Data
The hypothesis is explored by analyzing a sample of 471 students of the University of Amsterdam. They are students of the Bachelor program Economics and Business. The ages of the Dutch students range from approximately 19 to 25 and the ratio male to female is 1:1. The students were unaware they are subjects of this research. Using students as subjects is common among previous empirical research on task assignment (Preenen et al., 2014; Kruger and Dunning, 2001).
The students are enrolled in the second year course of the Bachelor program, which yields 6 credits (ECTS). This course, given in the first semester during eight weeks, is about the strategic functioning of organizations within the business environment. Successful completion of the course depends on the weighted average of two exams, namely a midterm examination (counting for 30%) and a final examination
(counting for 70%). In addition a full bonus point on top of the weighted grade can be earned by voluntarily participating in a group assignment. Table 1 presents a summary of the assignments.
The coordinator of the course provided the administrative data of the course of the academic year 2014 – 2015. The data gives insight into the performance of the subjects for the three different course examinations. The data is naturally occurring data, which has several advantages over data collected from surveys or experiments. Assessing task assignment on assignments students have to perform during their studies improves the ecological validity, being the extent to which the study matches real world circumstances, of the research (Preenen et al., 2014). Minor adjustments to the data were made. For example, students who registered for the course but did not participate in any assignments were left out. Missing values were completed if possible.
Table 1 – Summary of the course
Midterm Exam Bonus Assignment Final Exam
I 15 multiple-choice questions Two homework assignments 35 multiple-choice questions II Individual Groups of 3 – 4 students Individual
III Participation mandatory Participation voluntary Participation mandatory
IV 30 % Full bonus point 70 %
V Week 4 Week 3 and 7 Week 8
3.3 Measurements
The research studies the final exam consisting of 35 multiple-choice questions and the midterm exam consisting of 15 multiple-choice questions. There are four different versions of the exam with the same questions but in a different order, potentially to avoid cheating during the exam. This results in four versions with each a different difficulty-level distribution.
3.3.1 Measurement final examination
A crucial feature of the research design is devising a reliable measure of the relative difficulty of the four versions. In chapter 1, we defined task challenge as assignments which are difficult to achieve, new, high in responsibility, important, demanding and a test of abilities (Preenen et al., 2014). In this study, one assignment equals one multiple-choice question with four possible answers. Every question differs in content; some
questions are based on definitions of relevant concepts, others include calculations, theorems or interpreting economic figures. An example of the types of questions can be found in appendix 2. Since each subject prefers different types of questions, assessing the questions on their content would not be an objective measurement. Instead, the measurement used to assess task challenge is the percentage of students that answered the question correct. Students who gave the correct answer successfully completed the task. Since there are four possible options there is a chance of 25 percent per question of successful completion of the task if students would not have prior knowledge. To determine the different difficulty distributions of the versions, we grouped the 35 questions in clusters of five. The first three questions and the last two questions are left out of the distribution, since they are the same across all the versions. From the remaining 30 questions, we made 6 clusters and calculated the percentage mean of correctly answered questions. On average 63.2% of the tasks were completed successfully (SD=20.6). The use of this method results in four different difficulty-level distributions, which are presented in figure 1. Version 1 resembles a parabola that opens upward with the high-challenging tasks centered in the middle of the exam and the low-challenging tasks at the beginning and end. Version 2 has a declining difficulty-level structure, meaning the high-challenging tasks are assigned at the beginning and the low-challenging tasks are assigned as the exam progresses. Our hypothesis, a gradually increasing difficulty-level structure, is presented by version 3; the low-challenging tasks are assigned initially and the high-challenging tasks are assigned as the examination progresses. Lastly, version 4 resembles a parabola that opens downward; the beginning and end are relatively high-challenging, while the center of the exam is low-challenging. The ungrouped data shows the same type of distributions, albeit less distinctive. The scatter plot of these distributions can be found in appendix 1.
3.3.2 Measurement midterm examination
The midterm examination consists of 15 questions, which are again perceived as 15 tasks with each a different level of difficulty. The percentage of correctly answered questions is used as an objective measurement for task challenge. Grouping the questions is not necessary since the quantity of tasks is considerably lower for the midterm examination. Using this method gives 4 distinctive build-ups, shown in figure 2. Version 1 resembles our hypothesis: initially assigning low-challenging assignments. On
the contrary, version 2 assigns the high-challenging assignments at the beginning. Version 3 and 4 resemble a parabola that opens downward and upward, respectively. Figure 1 – Distribution final examination Version 1: Low-High-Low Challenging Version 2: High-Low Challenging Version 3: Low-High Challenging Version 4: High-Low-High Challenging Figure 2 – Distribution midterm examination Version 1: Low-High Challenging Version 2: High-Low Challenging Version 3: High-Low-High Challenging Version 4: Low-High-Low Challenging 0 20 40 60 80 1 2 3 4 5 6 Pe rc en t Ques3ons 0 20 40 60 80 1 2 3 4 5 6 Pe rc en t Ques3ons 0 20 40 60 80 1 2 3 4 5 6 Pe rc en t Ques3ons 0 20 40 60 80 1 2 3 4 5 6 Pe rc en t Ques3ons 0 20 40 60 80 100 0 2 4 6 8 10 12 14 16 Pe rc en t Ques3ons 0 20 40 60 80 100 0 2 4 6 8 10 12 14 16 Pe rc en t Ques3ons 0 20 40 60 80 100 0 2 4 6 8 10 12 14 16 Pe rc en t Ques3ons 0 20 40 60 80 100 0 2 4 6 8 10 12 14 16 Pe rc en t Ques3ons
3.4 Procedure
The midterm and final examination are taken in the 4th and 8th week of the course in an examination hall. All subjects take the examination in the same time-slot. The given time is two hours for the midterm examination and three hours for the final examination. We assume that the four versions are distributed randomly across students. Since this research is not set up as an experiment but makes use of actual events, the random distribution across students cannot be controlled. There are no assigned seats in the examination hall so the subjects can pick where they want to sit. In general, students who are a group of friends, with presumably a similar work ethic, will sit together. Since the different versions are distributed per row, the random distribution of the versions is not perfect. On top of that, some subjects will make the questions in a different order than the version prescribes. For example, if they do not know a question they can skip it and come back to the question later. We cannot control for this imperfection either. 3.5 Analyses
The relation between task assignment and individual performance is analyzed by running different linear regressions. Ordinary Least Squares (OLS) is used as a method to estimate the unknown parameters. Individual performance is directly measured by the achieved grade of the subjects. Including the different versions as binary variables captures the effect of task assignment. Firstly, the versions are included in the regression to analyze if the complete difficulty level-structure of the test influences performance:
Gradei = β0 + β1 Version1i + β2 Version2i + β3 Version4i + β4 Controlvariablesi + ui (1)
One of the versions is omitted from the regression to avoid perfect multicollinearity. In the analysis using the final test data version 3 is omitted, because this version presents the structure that is most favorable according to the hypothesis: initially assigning the low-challenging tasks. While analyzing the data from the midterm, version 1 will be omitted for the same reason. The second analysis will only focus on the task challenge of the beginning of the test. The versions initially assigning low-challenging tasks and the versions initially assigning high-challenging tasks are paired so only one binary variable presenting initial task challenge is included in the regression:
The study of Kruger and Dunning (2001) show different actions for high-ability and low-ability individuals. Low-ability individuals could react different to high challenging tasks than high-ability individuals. The third analysis examines whether high-ability individuals react different to initial task challenge than low-ability individuals. The model includes a binary variable presenting initial task challenge, a binary variable presenting high-ability (students who perform above average are considered to posses high course-specific ability), and an interaction term between these two binary variables:
Gradei = β0 + β1 High-challengingi + β2 Abilityi + β3 High-challenging*Abilityi + ui (3)
3.6 Control variables
There are more determinants of performance apart from task challenge. Competence, exerted effort, skill ability, preparation, and even the weather could influence the obtained grade. Two control variables are added to the regression to control for these effects: ability and effort.
One’s ability has a positive effect on individual student performance. A related study conducted by Czibor et al. (2014) on the effect of relative grading schemes versus absolute grading schemes on individual student performance, includes ability measures as control variables. The study uses measurements for general ability as well as course-specific ability. Our model only includes a measurement for course-specific ability, since a measurement for general student ability, such as the grade-point average of the students, is not available. The obtained grade for the midterm examination when analyzing the final test scores and vice versa is used as a proxy for ability in the course. Another possible determinant of student performance is the effort exerted in order to successfully pass the course. Bonuses are often used in organizations as an incentive for employees to exert more effort. Whether the student obtained the bonus point or not proxies as a measurement for exerted effort. This measurement is imperfect because it does not take into account all the preparations students exert, such as lecture attendance and self-study time (Czibor et al., 2014).
The model excludes some explanatory variables, which could influence the regression results. As mentioned, there is no measurement for general ability available, but this is likely to influence individual student performance. Overconfidence is another determinant of performance that is not included in the regression (Czibor et al., 2014;
Kruger and Dunning, 2001). The exclusion of explanatory variables could lead to omitted-variable bias, and consequently the model could overestimate the effect of task assignment on performance.
4. Results
This section will present the results of the research outlined in the previous section. First, descriptive statistics of the data will be provided. Subsequently, the hypothesis will be tested using two analyses. The first analysis presents the results of the effect of task assignment through the entire sequel of assignments. The second analysis focuses on the effect of the initial task challenge on performance. A third analysis examines if competent students react different to initial task challenge than incompetent students.
4.1 Descriptive statistics
Table 2 shows an overview of the administrative data of the midterm and final results. The number of observations differs across sections, since not all students actively participated in every assignment. For example, some subjects made the midterm examination and not the final examination, others only participated in the bonus assignment, and hence the number of observations drops from 471 to 424 and 414.
Table 2 – Summary statistics of the data
Observations Mean Standard
Deviation Minimum Value Maximum Value Correct midterm 424 10.505 2.674 3 15 Grade midterm a 424 6.254 2.229 0 10 Version 1: low 109 6.124 (5.481) c 2.229 0 10 Version 2: high 108 6.551 (5.834) 2.251 0.833 10 Version 3: high 105 6.135 (5.436) 2.138 0.833 10 Version 4: low 102 6.144 (5.754) 2.280 0.833 10 Correct final 414 22.116 4.655 8 32 Grade final b 414 5.598 1.724 0.370 9.259 Version 1: low 110 5.502 (6.449) d 1.615 1.481 9.259 Version 2: high 102 5.726 (6.323) 1.730 1.852 9.259 Version 3: low 100 5.496 (6.339) 1.921 0.370 9.259 Version 4: high 102 5.675 (6.439) 1.640 0.740 9.259 Bonus 471 0.73 0.44 0 1 a. Grade midterm = (correct midterm – 3) / 1.2 b. Grade final = (correct final – 7) / 2.7 c. Final grade of these students in parentheses d. Midterm grade of these students in parentheses
Opposing our expectation, the versions that start with the high-challenging questions score better on average than the versions starting with the low-challenging questions. This holds for the means of the midterm examination as well as for the final examination. However, conducting an independent t-test shows that only the differences in means of the midterm versions compared to version 2 are statistically significant (p<0.10).
4.2 Primary analysis
Table 3 presents the results of the primary analysis, which compared the four different difficulty-level structures. The first regression includes only the different versions as main variables of interest. The second to fourth regressions include control variables. Table 4 presents the same analysis for the midterm examination. Table 3 – Primary regression results final Dependent variable: Grade final I II III IV Grade midterm 0.439 *** (0.035) 0.451 *** (0.034) Bonus – 0.065 (0.191) – 0.349 * (0.180) Version 1: Low-High-Low – 0.007 (0.242) – 0.076 (0.207) 0.005 (0.247) – 0.031 (0.205) Version 2: High-Low 0.217 (0.254) 0.106 (0.202) 0.130 (0.258) 0.089 (0.203) Version 4: High-Low-High 0.167 (0.248) 0.076 (0.201) 0.134 (0.253) 0.084 (0.201) Constant 5.509 *** (0.187) 2.773 *** (0.278) 5.604 *** (0.243) 2.976 *** (0.310) R-squared 0.003 0.302 0.002 0.316 Adjusted R-squared – 0.004 0.295 – 0.008 0.307 Observations 417 400 404 399 Note: Robust standard errors in parentheses * p < .10, ** p< .05, *** p< .01
Table 4 – Primary regression results midterm Dependent variable: Grade midterm I II III IV Grade finale 0.680 *** (0.050) 0.695 *** (0.049) Bonus 1.005 *** (0.255) 0.753 *** (0.232) Version 2: High-Low 0.490 (0.302) 0.397 * (0.235) 0.479 (0.301) 0.388 * (0.236) Version 3: High-Low-High 0.026 (0.297) 0.012 (0.240) 0.049 (0.291) 0.032 (0.234) Version 4: Low-High-Low 0.035 (0.310) – 0.123 (0.70) – 0.073 (0.303) – 0.234 (0.261) Constant 6.109 *** (0.212) 2.525 *** (0.322) 5.340 *** (0.277) 1.850 *** (0.354) R-squared 0.008 0.308 0.043 0.336 Adjusted R-squared 0.002 0.301 0.034 0.328 Observations 432 400 426 399 Note: Robust standard errors in parentheses * p < .10, ** p< .05, *** p< .01
The R-squared of the fourth regression, which includes all the available explanatory variables, is for both analyses relatively high (R2=0.316; R2=0.336). The R-squared gives
information on how well the model approximates the real data. From the results in
Table 3 it can be concluded that regarding to the final there is no significant effect of task assignment on performance when we consider the entire sequel of assignments. Regarding to the midterm results, version 2, whose build-up goes from high-challenging assignments to low-challenging assignments, scores significantly better compared to version 1, whose build-up goes from low-challenging assignments to high-challenging (β=0.388; p<0.10). To examine whether the version variables have a joint effect on performance an F-test is computed on the fourth regression of both the final as the midterm data. The variables presenting the different versions are imposed as three restrictions:
H0 : β3 = 0, β4 = 0, β5 = 0
The value of the test statistic is F = 0.18 for the data on the final and F = 1.95 for data on the midterm. Consequently, for both analyses the H0 cannot be rejected. This implies
that the version variables have no jointly significant effect on performance.
The control variables midterm grade and final grade, which approximate the student’s ability, are highly significant for the final and the midterm results (p<0.01). The control variable bonus is highly significant for the midterm results (p<0.01), as well as statistically significant for the final results (p<0.10). Notable, the coefficient sign flips; the sign is negative for the final results and positive for the midterm results. The discussion section gives an explanation for this effect.
4.2 Secondary analysis
The results of the second analysis, which only focuses on the difficulty level of the beginning of the examination, are presented in table 5. Two regressions were run for the data on the midterm, as well as for the data on the final. The first regression includes only task assignment as independent variable. The second regression includes task assignment and other relevant explanatory variables.
Table 5 – Regression results secondary analysis
Dependent variable: Midterm Final
Grade I II I II High-challenging a 0.224 (0.216) 0.364 ** (0.178) 0.205 (0.168) 0.084 (0.142) Proxy ability b 0.698 *** (0.048) 0.451 *** (0.034) Bonus 0.756 *** (0.235) – 0.351 * (0.180) Constant 6.133 *** (0.155) 1.688 *** (0.349) 5.499 *** (0.122) 2.972 *** (0.289) R-squared 0.003 0.333 0.004 0.315 Adjusted R-squared 0.000 0.328 0.001 0.310 Observations 432 400 426 399 Note: Robust standard errors in parentheses * p < .10, ** p< .05, *** p< .01 a. High-challenging presents the two versions initially assigning high-challenging questions b. Ability presents the midterm grade for the final examination and vice versa
The R-squared, which presents the explained variance, is low for both the first regressions (R2=0.003; R2=0.004). The R-squared for the second regressions is relatively high (R2=0.333; R2=0.315). All the regressions run show a positive beta of the variable high-challenging. That implies a positive effect of initially assigning high-challenging questions on the grade obtained. This result is only statistically significant for the second regression of the midterm results (β=0.364; p<0.05). The expected grade obtained for the midterm is 0.364 higher when a subject is assigned a version that starts with the high-challenging questions versus a version that starts with the low-challenging questions. The results are not in line with the hypothesis, which expected assigning low-challenging tasks initially to be more beneficial than assigning high-challenging tasks initially. 4.3 Tertiary analysis Table 6 presents the results of the tertiary analysis, which examines whether competent individuals react different to initial task challenge than incompetent individuals. Table 6 – Regression results tertiary analysis
Dependent variable: Midterm Final
Grade I II High-challenging 0.123 (0.298) 0.036 (0.243) Ability a 1.017 *** (0.304) 1.316 *** (0.226) Interaction term b 0.161 (0.418) 0.214 (0.314) Constant 5.584 *** (0.228) 4.804 *** (0.160) R-squared 0.063 0.174 Adjusted R-squared 0.056 0.168 Observations 432 417 Note: Robust standard errors in parentheses * p < .10, ** p< .05, *** p< .01 a. Ability presents the students who performed above average b. Interaction term = High-challenging * Ability
In this analysis ability presents a dummy variable, which takes on value 1 for the students who perform above average (grade midterm > 6.254; grade final > 5.598) and value 0 for the students who perform below average. The interaction term is positive for both regressions, which would imply that the beneficial effect of initial task on student performance is higher for competent students than for incompetent students. However, the interaction term is insignificant for the midterm and the final regression, p=0.699 and p=0.489 respectively. This means that from this dataset we cannot conclude that competent individuals react different to initial task challenge than incompetent individuals.
5. Discussion
This section discusses the findings of the research and their implications. The results are positioned into the existing literature reviewed in section 1. The section closes with discussing the limitations of this research and suggestions for future research.
5.1 Task assignment and performance
The summary statistics show the students who were appointed a version that assigned high-challenging tasks initially obtained a higher grade on average than the students who were appointed a version that assigned low-challenging tasks initially. Although the majority of the empirical results of the regression analyses are not statistically significant, they show a tendency towards the beneficial effects of initial task challenge. The effect of task assignment on performance is weaker in the first analysis, when task assignment includes the complete difficulty-level structure of the test, than in the second analysis, when task assignment includes only the difficulty-level of the beginning of the test. These results show that initial task challenge is more relevant for performance than the complete difficulty-level structure of tasks. So the challenge of the beginning of a test or otherwise sequential row of assignments has the most influence on individual performance, what happens after is less decisive.
There are three possible explanations for the insignificant effect of task assignment when analyzing the final test scores. Firstly, the difference in difficulty-level is relative small across exam questions. The low-challenging questions require a comparable level of education to solve as the high-challenging questions. Therefore, the small difference in task challenge makes the advantages of initially assigned high-challenging tasks smaller. Secondly, our analysis involves performing many sequential tasks; 35 for the final and 15 for the midterm. The effect of initial task challenge could be unpronounced due to the amount of tasks involved. This could explain at the same time why the results for the midterm examination were significant. Lastly, the ungrouped distribution of the final exam shows a less distinctive difficulty-level structure than the grouped data (appendix 1). The insignificant effect of task assignment on performance can be due to the relative small changes in difficulty-level structure of the final exam versions.
The results suggest the beneficial effects of initial task challenge for individual performance in the case individuals have to perform multiple tasks. These findings
contradict the main hypothesis, which expected initial low task challenge to be beneficial, since the literature stressed the importance of confidence (Compte and Postlewaite, 2004) and later task challenge (Taylor, 1981). However, the literature also underlines the importance of high-challenging tasks for performance (Berlew and Hall, 1966), on-the-job-learning (McCauley et al., 1994) and employee satisfaction (Preenen et al., 2011). Considering these positive effects of job challenge, the positive results of initial task challenge are in line with existing literature. The findings contribute to the existing literature by adding the component of multiple sequential assignments. Previous research is limited to the effect of task challenge on performance when subjects have to perform one task (Preenen et al., 2014) or two or three tasks (Taylor, 1981). This study extends the literature since the study analyzed 15 (midterm examination) and 35 (final examination) sequential tasks.
5.2 Effort
The variable bonus presents to some extent the exerted effort of the student in the course, as mentioned in section 3. Having obtained a full-bonus point has a statistically significant positive effect on the midterm grade, but a statistically significant negative effect on the final grade. A plausible explanation for this flip in sign is the dominant culture of mediocrity at Dutch universities (“zesjescultuur”). Dutch students appear to be focused on just passing the course, and less on the grade they hereby obtain (Czibor et al., 2014). At the time of the midterm examination, the bonus grade was not yet obtained, so the students in general exerted a maximum amount of effort. For the midterm examination, the bonus grade presents the amount of effort a student was willing to exert in order to pass the course, and for this reason the sign of the bonus coefficient is positive. At the time of the final examination the students were aware if they obtained the bonus point. This knowledge could have led to strategically adjusting the exerted effort in order to just pass the course. Students who obtained a bonus point would pass with a lower grade for the final compared to students who did not obtain the bonus point. Most likely, this lead to the exertion of less effort on the final examination, and for this reason the sign of the bonus coefficient is negative.
5.3 Ability
The central question in the tertiary analysis is whether competent individuals react different to initial task challenge than incompetent individuals. The research of Kruger and Dunning (2001) show different actions for competent and incompetent individuals. Incompetent individuals do not possess the self-monitoring skills to realize their own incompetence, while competent individuals are more likely to accurately assess their own abilities. Low-ability individuals hold overly optimistic views of their own ability and when facing more challenging tasks they tend to overestimate their abilities even more. This reasoning makes it plausible for low-ability individuals, those scoring below average, to react different to task challenge than high-ability individuals, those scoring above average. However, we did not find a significant effect of the interaction term on student performance, so we cannot conclude that high-ability individuals react different to initial task challenge than low-ability individuals. 5.4 Limitations and future research
The research suffers from some possible methodological limitations. Firstly, the study uses students as subjects, which could be considered a limitation for external validity; the extent to which findings can be generalized to broader settings. It is probable that the influence of task challenge on students differs from the influence of job challenge on employees in work settings. Secondly, using administrative data has many advantages over data gathered from surveys or experiments as mentioned in section 3. However, the use of natural occurring administrative data also has disadvantages, mainly the limited options for control. The regression excludes relevant control variables, such as a measurable variable for general ability and overconfidence. Other uncontrollable factors include the version appointed to the students, which might not be random, and the order in which students made the questions. Research conducted in lab-settings could control for these limitations.
Future research should examine the effect of sequential assignments on performance, since such research is considerably limited to date. Research on sequential assignments is important since this best mimics real life settings. Employees have to perform multiple tasks in a row each day in organizations, and these tasks are usually not limited to only 2 or 3 sequential tasks as previously researched (Taylor, 1981). The research could limit the amount of tasks subjects have to perform in comparison to this
study to a recommended maximum of 10 assignments in a row. Possibly, this could lead to a stronger effect of task assignment on performance, since we observed a stronger effect for the examination involving 15 tasks versus the examination involving 35 tasks. Furthermore, the mechanisms through which initial task challenge enhances performance could be analyzed by adding a survey to the research.
5.5 Practical implications
Despite these limitations, the findings of the study are relevant for the functioning of organizations. Within organizations a diverse range of tasks have to be carried out, which could vary in content and difficulty-level. Organizational performance would benefit if organizations assign the difficult tasks to their employees at first instead of assigning the difficult tasks later on. This efficient way of task assignment motivates employees to exert more effort, which enhances individual performance. In short, task assignment should be perceived as an effective tool to enhance individual performance in organizations.
6. Conclusion
This research examines the effect of task assignment on individual performance in case individuals have to perform multiple tasks in a row by analyzing the test results of 471 students. The use of administrative data offers advantages over methods relying on surveys or experiments. The different versions of the tests are analyzed to determine different difficulty-level structures of multiple sequential assignments. Task difficulty is assessed by the probability of successful completion of the assignment.
Contrary to the hypothesis, individuals who are initially assigned high-challenging tasks perform better than individuals who initially are assigned low-challenging tasks when they have to perform multiple tasks. This effect of initial task challenge on performance is stronger when the amount of assignments is brought back from 35 assignments to 15 assignments. On top of that, the beginning of a sequential row of assignments is more relevant to overall individual performance than the complete difficulty-level structure. Lastly, we examined if competent individuals react different to task challenge than incompetent individuals, but found an insignificant effect.
The findings have noticeable implications for organizations. Organizational performance could benefit from the positive effects of initial task challenge by assigning the high-challenging tasks to their employees first, and the low-challenging tasks later on. Smarter task assignment could enhance individual performance and overall organizational success.
References
Bénabou, R., & Tirole, J. (2002). Self-confidence and personal motivation. Quarterly Journal of Economics, 117 (3), 871 – 915. Berlew, D. E., & Hall, D. T. (1966). The socialization of managers: Effects of expectations on performance. Administrative Science Quarterly, 11 (2), 207-223. Compte, O., & Postlewaite, A. (2004). Confidence enhanced performance. The American Economic Review, 94 (5), 1536 – 1557. Czibor, E., Onderstal, S., Sloof, R., & Van Praag, M. (2014). Does relative grading help male students? Evidence from a field experiment in the classroom, Tinbergen Institute Discussion Paper, TI 2014-116/III.
DeRue, D. S., & Wellman, N. (2009). Developing leaders via experience: the role of developmental challenge, learning orientation, and feedback availability. Journal of Applied Psychology, 94 (4), 859 – 875.
Dong, Y., Seo, M. G., & Bartol, K. M. (2014). No pain, no gain: An affect-based model of developmental job experience and the buffering effects of emotional intelligence. Academy of Management Journal, 57 (4), 1056-1077.
Ellis, H. C., Ottaway, S. A., Varner, L. J., Becker, A. S., & Moore, B. A. (1997). Emotion, Motivation, and Text Comprehension: The Detection of Contradictions in Passages. Journal of Experimental Psychology, 126 (2), 131 – 46. Guthrie, C., Rachlinski, J. J., & Wistrich, A. J. (2000). Inside the judicial mind. Cornell Law Review, 86 (4), 777. Kruger, J., & Dunning, D. (1999). Unskilled and unaware of it: How Difficulties in Recognizing One’s Own Incompetence Lead to Inflated Self-Assessments. Journal of Personality and Social Psychology, 77 (6), 1121 – 1134.
McCauley, C. D., Ruderman, M. N., Ohlott, P. J., & Morrow, J. E. (1994). Assessing the developmental components of managerial jobs. Journal of Applied Psychology, 79 (4), 544 – 560.
Preenen, P. T. Y., Dorenbosch, L., Plantinga, E., & Dhondt, S. (2016). The influence of task challenge on skill utilization, affective wellbeing and intrapreneurial behaviour. Economic and Industrial Democracy, 1 – 22.
Preenen, P. T. Y., De Pater, I. E., Van Vianen, A. E. M., & Keijzer, L. (2011). Managing voluntary turnover through challenging assignments. Group and Organization Managements, 36 (3), 308 – 344.
Preenen, P. T. Y., Van Vianen, A., & De Pater, I. (2014). Challenging tasks: The role of employees' and supervisors' goal orientations. European Journal of Work and Organizational Psychology, 23 (1), 48-61.
Steele, C. M., & Aronson, J. (1995). Stereotype Threat and the Intellectual Test Performance of African-Americans. Journal of Personality and Social Psychology, 69 (5), 797 – 811. Taylor, S. (1981). The Motivational Effects of Task Challenge: A Laboratory Investigation. Organizational Behavior and Human Performance 27, 255 – 278.
Appendices
Appendix 1: Distribution Final Examination – Ungrouped Version 1: Low-High-Low Challenging Version 2: High-Low Challenging Version 3: Low-High Challenging Version 4: High-Low-High Challenging 0 20 40 60 80 100 0 10 20 30 40 Pe rc en t Ques3ons 0 20 40 60 80 100 0 10 20 30 40 Pe rc en t Ques3ons 0 20 40 60 80 100 0 10 20 30 40 Pe rc en t Ques3ons 0 20 40 60 80 100 0 10 20 30 40 Pe rc en t Ques3onsAppendix 2: Question 31 – 35 of the final exam October 22 2014
