Bachelor Thesis
The effect of gender diversity on group performance
Abstract
The effect of gender diversity on team performance will be tested in this thesis. The data is collected from a finance course taught at the University of Amsterdam, where students need to cooperate with each other in groups to accomplish two assignments. In total 92 self-selected groups were formed of 4 or 5 students in different compositions with respect to gender. In this thesis the main hypothesis will be tested, whether mixed teams will outperform homogeneous teams. Also the effect of the presence of female in teams on the performance is studied. The results show that teams with female members perform better than teams without female members. However, for the mixed groups the results were not in line with the hypothesis. Unfortunately, the information about the student’s abilities, personality and motivation was not available, which could have been useful for this study. In the future it might be interesting to do similar research, if this information is available.
Author: Didier Karregat Student Number: 10760474
BSC Programme: Economics & Business Specialization: Finance and Organisation Thesis supervisor: Patrick Stastra
2
Statement of Originality
This document is written by Didier Karregat 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 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
Table of contents
1. INTRODUCTION
4
2. RELATED LITERATURE
5
2.1 GROUP PERFORMANCE
5
2.2 GENDER DIVERSITY IN BOARDS
6
2.3 GENDER DIVERSITY ON GROUP PERFORMANCE
7
3. METHODOLOGY & DATA
8
3.1 DATA
8
3.2 DATA SELECTION
9
3.3 MODEL SETUP
10
4. RESULTS
13
5. CONCLUSION AND DISCUSSION
18
REFERENCES
20
APPENDIX
22
4
1. Introduction
At the moment, teamwork is getting more and more important in daily life. Organizations prefer working in groups instead of individual performance according to Hackman and Morris (1975). One reason for this is that some tasks are far too much work for an individual. Therefore, the individual is not able to handle the task. In other situations the task is complex and difficult to solve, that being the case it is better to solve the task in groups, as each individual member can add his or her own knowledge and skills to the group according to Guzzo & Dickson (1996).
As mentioned above, teamwork is becoming more and more important, but how is the effectiveness of the group enhanced. Rhee et al. (2013) mentions that individual ability and GPA are important tools to reach a good team performance. Others such as Hoogendoorn et al. (2013) and Wegge et al. (2008) conclude that the gender diversity of the groups is an important factor that positively influences the performance of the group. Hoogendoorn et al. (2013) and Rhee et al. (2013) both used undergraduates for their experiment.
Also the diversity of boards is considered to have a positive effect on the firm performance and the value of the firm according to Carter et al. (2003) and Campbell and Minguez – Vera (2009).
Nevertheless, Williams, Harkins, & Latané (1981) suggest that individuals will not exert the same amount of effort in a group as they do with an individual assignment. Therefore the question arises, whether the effect of gender diversity is similar or not to undergraduates, who are bounded to group assignments and cooperation with other undergraduates at university. The effect of gender diversity is studied in this thesis. This study includes three hypotheses, which will be tested. The first hypothesis states that teams with female members will outperform teams without female members. The second and third hypothesis will test the effect of mixed teams, in particular if mixed teams outperform homogenous teams or homogenous men teams.
The dataset for this research is derived from a course taught at the University of Amsterdam. The dataset contained 487 students who were divided over 92 groups in different formations. Therefore a dummy variable for gender or mixed gender is added to the empirical model to estimate the effect of gender diversity on the group performance. The empirical models used are as follows:
(1) Performance = α + β1Gender + β2Age + β3Size + β4Credits + β5Resits + β6Avexam + β7PWF + εi
5 (2) Performance = α + β1MixedGender + β2Age + β3Size + β4Credits + β5Resits +
β6Avexam + β7PWF + εi
Three OLS regressions have been carried out for each of the three hypotheses. The grade for assignment 1, assignment 2 or the average for both of the assignment is taken as performance measure.
The results show evidence that teams with female members perform better than teams without female members. This result is in line with the expectations. Furthermore an effect is found that mixed teams outperform homogenous men teams. Nevertheless, the second hypothesis was not confirmed, as the mixed teams did not significantly perform better than the homogenous teams (male and female).
In the remaining part, the literature will be explained and discussed that is related to this study. Furthermore the dataset, the selection of the data and the model setup are described. The expectations for this research are illustrated in three hypotheses. After that the results are announced and finally from those results the conclusions are drawn.
2. Related Literature
2.1 Group Performance
Williams, Harkins & Latané (1981) mention that individuals will not exert the same amount of effort in a group assignment as when they have to do the assignment individually. They did an experiment with undergraduates and concluded that the amount of effort of the students was lower when they thought their output was unidentifiable. After this experiment they did the same experiment, but told the students that their individual output was measurable. This lead to the elimination of social loafing and as a result to an opposing outcome in comparison with the first experiment.
According to Guzzo & Dickson (1996), there are three main characteristics to enhance team effectiveness. These characteristics are: design of the group, diversity of membership and group process. They conclude that each individual member adds his own knowledge and skills to the group, which leads to a good team performance.
Seijts & Latham (2000) studied the effect of self and group efficacy on group performance. They proved that self-set personal goals that were set with an assigned mutual group goal performed better than self-set personal goals without a mutual goal. The individuals were divided into groups of 3 or 7 persons. The size of the groups
6 turned out to have an effect on the performance. Groups with 7 persons had lower collective-efficacy and lower commitment to the group goal than the smaller groups existing of 3 persons. Additionally, the individual performance was better in a smaller group.
Furthermore, age diversity is an important variable by composing a group according to Schlick, Frieling & Wegge, (2013). They suggest that the effectiveness of the group depends on how age diverse the groups are. Age diversity positively correlates with performance. The presence of more relatively elder group members leads to effectiveness of the group performance.
Team performance is also based on the individual abilities and personalities in a group. Rhee, Parent & Basu (2013) found that GPA positively correlated with the grades of the written report and therefore the performance of the group. They also stated that personality correlates positively with the evaluation of group performance by looking at the grades for the oral and written presentations.
2.2 Gender Diversity in Boards
Carter et al. (2003) concluded that a good performance could be reached with a diversified board. They investigated that the performance of US firms is positively correlated with the diversified boards. Also Campbell and Minguez-Vera (2009) found that gender diversity in the boardroom of a financial firm had a significant influence on the value of the firm. They discovered that a higher percentage of women in the boardroom caused an increase of the firm’s value.
Shrader, Blackburn and Iles (1997), however, did not find enough evidence that more women in the board lead to a better firm performance and a higher firm value. In this study some factors are not taken into account, such as leverage and firm age, which can potentially declare the opposing conclusion. Accordingly the first hypothesis of this research is as follows:
H1: Teams with female members will outperform teams without female members.
The hypothesis states that teams with female members will outperform teams without female members. Research in the boardroom proved that diversified boards lead to better firm performance according to Carter et al. (2003) and Campbell and Minguez-Vera (2009). Nevertheless, Shrader, Blackburn and Iles (1997) did not find any effect of gender diversity on the performance, but some factors are not taken into account as
7 mentioned above and therefore the expectation is that teams with female members will outperform teams without female.
2.3 Gender Diversity on group performance
Apesteguia et al. (2012) studied a large business game, where undergraduates in groups had to perform as a general manager. They concluded that teams only consisting of women where outperformed by all other teams. A mixed team, two men and one woman, was the best performing combination. Differences in decision making is a possible explanation according to this study.
Also Hoogendoorn et al. (2013) held an experiment with undergraduates. The teams were randomly created and based on gender. Teams consisting out of a majority, both male and female, were outperformed by mixed teams with an equal gender mix. The more intense mutual monitoring, the male members checked the female members and vice versa, and learning from other team members in mixed teams caused these results.
However, research by Borg and Shapiro (1996) on group performance at high school showed that gender diversity did not have any influence on the performance of the group. This study is relevant as undergraduates are subject, which is similar to this research. They did not find an effect, nevertheless other control variables were used, race and personality variables were included in the empirical model, which probably explain the results.
Wegge et al. (2008) did research on group performance in a public organization. They initiated that not only gender diversity but also age positively correlated with group performance. However, they observed that gender diverse teams outperformed groups with a high proportion of female employees. In line with Hoogendoorn et al. (2013) mixed gender teams are preferred above homogenous female groups. Accordingly the second hypothesis of this research is as follows:
H2a: Mixed teams will outperform the homogenous teams H2b: Mixed teams will outperform the homogenous men teams.
The hypotheses state that the mixed teams will outperform homogenous teams, because Wegge et al. (2008) as well as Hoogendoorn et al. (2013) concluded that mixed gender teams outperformed homogenous teams. Also Apesteguia et al. (2012) stated that mixed teams were the best performing combination. Therefore the expectation is that mixed
8 teams will outperform homogenous teams, despite the research by Borg and Shapiro (1996), where no effect was found.
3. Methodology & Data
In this section the methodologies used to test the hypotheses will be discussed. The variables used in this research will be explained and the way the independent variable and dependent variable are measured will be discussed.
3.1 Data
The data used for this study is collected from a course of the bachelor Economics and Business taught at the University of Amsterdam. Investment & Portfolio Theory 2 is a second year course, which is compulsory for the students that follow the Finance & Economics and Finance & Organization track.
The data is collected from the academic year 2016-2017. The data could be collected by approval of the coordinator of the course. He provided information about case grades, final exam results and the group formations. The total dataset contains information about group formations and size, final exam grade, case grades, total attempts, gender, age and total credits obtained by the subject until the 1st of June 2017. In the academic year 2016-2017, 487 individual students participated the course Investment Portfolio Theory 2. In the first week of the course the individual students were requested to form groups of minimum 4 and maximum 5 persons by themselves. The teams in this course were self-selected, unless the students were unable to find other group members, then the assistance of the coordinator was needed. The amount of self-selected groups and the amount of groups constructed by the coordinator were unknown.
The students were faced with two assignments, which had to be carried out as a group both counting for 15 percent part of the final grade. The grade of the final exam (counts for 70 percent of the final grade) should be at least a 5.0 to pass the course. Furthermore, the total final grade (100 percent) needs to be at least 5.5. The maximum total grade is 10.0 and the minimum 0.
This study supplements previous research, due to the fact that other performance measures are used or measured for a group instead of on individual basis. The group performance is measured on the basis of the grades obtained by the group assignments. Previous research was done at corporation boards, where profit and firm value are
9 measurements for group performance as shown by Campbell and Minguez-Vera (2009). However, Borg and Shapiro (1996) did research at high schools and used the grade of economics as the performance measure. This grade was only an individual performance grade. The boardroom contains similarities with the teams used in this study, as the boardroom consists of several members who have to cooperate to reach their goals. Therefore the literature on boardrooms is useful for this study.
3.2 Data Selection
Due to a lack of attendance, 54 students were removed from the dataset. There were 3 students who were not assigned to a group and left out, the dataset contains the information of 425 students, of whom 129 female and 296 male students. Overall there were 92 groups formed, 35 groups of 4 persons and 57 groups of 5 persons. These groups differ in gender diversity. The share of female is mentioned as a female fraction. Most of the students divided over the groups received grades for both the first and the second assignment, unfortunately some students quit the course early and received a 0 for the second assignment. For those students it is difficult to determine at which point they decided to quit the course, whether this was before, during or after the assignments. Therefore the 5 students who received a 0 for assignment 2 are removed from the dataset. A potential reason could be, that they were not motivated for the first assignment and are expected to have a negative effect on the group performance.
Table 1 Distribution of Groups
Fraction Female Groups with 4 persons Groups with 5 persons
F=0 16 21 F=1/5 0 11 F=1/4 8 0 F=2/5 0 9 F=1/2 3 0 F=3/5 0 11 F=3/4 4 0 F=4/5 0 4 F=1 0 5
Note: As a matter of fact the fraction could be 0, ¼, ½, ¾ or 1 in the smaller groups and 0, 1/5, 2/5, 3/5, 4/5 or 1 for the large groups.
10 To test the first hypothesis, teams with female members perform better than teams without female members, the homogenous group (F=0) will be the reference group. All the other teams contain at least one female member. In contrast to the first hypothesis, both homogenous groups (F=0 and F=1) are used as reference group by adding the dummy variable MixedGender to test if mixed teams outperform homogenous team, the second hypothesis. For the third hypothesis, the female groups (F=1) are omitted to test if mixed teams perform better than homogenous men teams.
Table 2 Descriptive Statistics groups
Variables Observations Mean St. Dev. Minimum Maximum
Age 92 21,46 1,23 19,5 25
Final Exam Grade 92 4,94 1,21 0 7,06
Credits Obtained 92 120,12 24,54 55,2 172,5
Number of attempts 92 1,15 0,26 1 2
Assignment 1 92 7,50 1,10 3,1 9,45
Assignment 2 92 7,15 1,06 3,35 9,4
Female fraction Groups 92 0,30 0,31 0 1
Size 92 4,66 0,48 4 5
3.3 Model Setup
To test the hypothesis three OLS regressions had to be carried out with the use of an empirical model. For every hypothesis another empirical model is used. To measure if female teams perform better than non-female teams, the following model is used:
(1) Performance = α + β1Gender + β2Age + β3Size + β4Credits + β5Resits + β6Avexam + β7PWF + εi
For the mixed teams the gender variable is replaced by a mixed gender variable, which is a dummy variable (1 = mixed teams, 0 = homogenous teams).
(2) Performance = α + β1MixedGender + β2Age + β3Size + β4Credits + β5Resits + β6Avexam + β7PWF + εi
In this study, performance is the dependent variable. Performance is measured by the grades that the groups received for both assignments. All members of the same group got the same grade for the assignments.
11 The control variables are: age, resits, final exam grade, size, number of students that did not participate the final exam, credits obtained until the 1st of June 2017 and gender. These will be further explained in the remaining part.
The first control variable is gender diversity (Gender). To test the hypothesis, this control variable should at least be added to the model to measure the effect of gender diversity on group performance. This variable will be a dummy variable (1=Female, 0=Male) or build up out of fractions as shown in table 1. Due to the fractions, the size of the group is part of this control variable. For some tests the fractions are not used, but the dummy for gender. In this case the seventh control variable size is added to the model (Size).
According to Seijts & Latham (2000) the size of the groups turned out to have an effect on the performance. Groups with 7 persons had lower collective-efficacy, lower outcome expectancies and lower commitment to the group goal than the smaller groups of 3 persons. Although the size of the groups in this course does not vary that much, the expectation is that group size has a negative influence on group performance if the group increases in size.
According to Apesteguia et al. (2012) and Hoogendoorn et al. (2013), mixed teams outperform homogenous teams and therefore gender diversity is expected to have a positive effect on the group performance.
The second control variable is age (Age). As shown in previous literature of Schlick, Frieling & Wegge, (2013), age is an important factor in measuring the performance. They suggest that the effectiveness of the group depends on how age diverse the groups are. Age diversity correlates positively with performance. The presence of more relatively elder group members leads to effectiveness of the group. Therefore the expectation is that the older the students are, the more effective the group is to accomplish the course with a high grade. Age might have a positive influence on the case grades.
The third control variable 3 is resits (Resits). The number of resits that a student needs to accomplish Investment & Portfolio Theory 2 could also influence the performance. This indicates that the student has difficulties with the course and potentially with the case as well or might not be motivated enough to succeed. Therefore attempts could have a negative effect on the performance.
The fourth control variable is total credits obtained (Cred). The motivation of a student is hard to measure and can be seen as a limitation. The total credits obtained by
12 a student until 1 June 2017 is an indication of how motivated a student is. This is chosen as a measure of motivation. The more credits obtained, the older the student is and therefore the more motivated a student is expected to be, as Schlick, Frieling & Wegge, (2013) stated that age had an positive effect on performance. Therefore the control variable credits obtained is expected to have a positive influence on the performance of the group.
The fifth control variable is students who were not present at the end term (PWF). This variable will be measured as a percentage. As mentioned above, the exact moment of dropping out is hard to measure, but it can have an effect on the group assignment if the student left before or during the assignments.
The sixth control variable is average final exam grade (avexam). The final exam is an individual test for all of the students and gives an indication about the knowledge of the course and their motivation. The expectation is, the higher the average final exam grade, the higher the average assignment grade for the group. According to Rhee, Parent & Basu (2013) is group performance depending on individual abilities. Therefore this control variable is added to the model.
Table 3 Correlation control variables
ASSGR Gender Age Size Avexam Cred Resits PWF
ASSGR 1 Gender 0.21** 1 Age 0.13 0.1011 1 Size 0.11 0.1647 0.074 1 Avexam 0.47*** -0.046 0.207** 0.1127 1 Cred 0.25** 0.1546 0.548*** 0.0998 0.393*** 1 Resits -0.21** -0.1152 0.194* 0.174* 0.0972 0.0561 1 PWF -0.37*** 0.0211 -0.063 -0.068 -0.764 -0.273*** -0.126 1
Note:***/**/* Denotes significance at 1%/5%/10%-level. Avexam denotes the average individual grade for the final exam. PWF denotes the percentages that did not show up at the final exam.
Table 3 shows moderately strong correlations among the control variables and the performance measure, the average assignment grade. The average individual grade for the final exam is the highest positive correlation coefficient (0.47) for the performance measure. This is in line with the expectations mentioned earlier in this paragraph.
13 According to Pallant (2013), correlations of 0.7 or higher, might lead to potential multicollinearity problems. In table 3, none of the coefficients pass this threshold. Therefore, the data of this research is not likely to suffer from multicollinearity issues.
4. Results
The performance is measured for all the distributed groups, by the average of the assignment grades. These results are shown in table 4. The performance goes with ups and downs as the share of female increases, because the average grade for F=0 and F=1/2 is equal (7.1). Nevertheless, by a distribution of F=4/5 and F=1 the average grades seem to rise in comparison with the lower shares of female (7.8 & 7.9).
Remarkable is that the performance of teams, including at least one female, seems to be better in larger groups then in smaller groups as shown in table 4, because the size of the groups are quite similar. All F=1 distributed teams, are teams consisting of 5 female students. The blue columns in table 4 indicate the performance (average grade both assignments) of the larger teams (5 students) and the grey columns indicate the smaller teams (4 students). These results are in contradiction with Seijts & Latham (2000), who stated that larger groups (7 members) had lower collective-efficacy and lower commitment to the group goal and therefore a worse outcome then smaller group (3 members). However, the size of the groups in this course does not vary that much, only groups of minimum 4 and maximum 5 students are accepted and therefore no hard conclusions can be taken.
Table 4 Performance and distribution of female
6,6 6,8 7,0 7,2 7,4 7,6 7,8 8,0 F=0 F=1/5 F=1/4 F=2/5 F=1/2 F=3/5 F=3/4 F=4/5 F=1
Performance and female distribution
14 The correlations of the performance measures are shown in table 5. All the performance measures are positively correlated and significantly different from zero. A good grade for assignment 1, will most of the time lead to a good performance for assignment 2 as well. The correlation between both assignments and the average is very high, but that’s logical as they are both part of the average. This will not lead to multicollinearity issues, as they are not in the same empirical model and therefore not in the same regression. They can all be used to measure the performance of the groups.
The first hypothesis states that teams with female members would outperform teams without female members. The results show a positive significant effect for the average assignment grade (p=0,058) and for assignment 2 (p=0,053) as shown in table 6 and suggest that teams with female members perform better than teams without female members. However, the p-value is just significant at 10% and therefore no hard conclusions can be drawn.
Table 5 Correlation performance measurements
ASS1 ASS2 ASSGR
ASS1 1.000 ASS2 0.234** 1.000 ASSGR 0.796*** 0.774*** 1.000
Note: ***/**/* Denotes significance at 1%/5%/10%-level. ASS1 and ASS2 denotes the grades for assignment 1 and 2. ASSGR denotes the average grade for both assignments.
15
Table 6
Regression female groups on group performance
Variables ASS1 ASS2 Average
Gender 0.20 0.42* 0.31* (0.23) (0.21) (0.16) Age 0.01 0.08 0.04 (0.11) (0.1) (0.08) Size 0.30 -0.08 0.11 (0.22) (0.21) (0.15) Credits 0.006 -0.005 0.0006 (0.006) (0.005) (0.004) Resits -0.78* -0.97** -0.87*** (0.44) (0.41) (0.31) PWF -1,21 0.37 -0.43 (1.05) (0.98) (0.74) Avexam 0.15 0.44*** 0.29*** (0.15) (0.14) (0.10) Constant 5.30** 5.03** 5.3*** (2.18) (2.04) (1.53)
Note: The values in brackets denote standard errors. ASS1 and ASS2 denote the grade for assignment 1 and 2 as dependent variable. The average denotes the average grade for both assignments. ***/**/* Denotes significance at the 1%/5%/10% level.
The second hypothesis states that mixed gender teams outperform the homogenous men teams. To test this hypothesis three OLS regressions have been done, for assignment 1, assignment 2 and the average of these assignments. The results in table 7 show a positive effect, mixed gender teams outperform homogenous men teams (P=0.078) for the average of the assignments and (P=0.073) for assignment 2. Similarly as the first hypothesis, no hard conclusions can be drawn, as the p-value is not significant at 5%-level.
16
Table 7
Regression mixed gender groups on team performance
Variables ASS1 ASS2 Average
Mixed Gender 0.19 0.40* 0.29* (0.23) (0.22) (0.16) Age -0.008 0.08 0.03 (0.11) (0.11) (0.08) Size 0.29 -0.10 0.09 (0.23) (0.21) (0.16) Credits 0.006 -0.006 0.0003 (0.006) (0.006) (0.004) Resits -0.84* -1.03** -0.93*** (0.45) (0.42) (0.31) PWF -1.15 0.40 -0.38 (1.09) (1.01) (0.75) Avexam 0.18 0.46*** 0.31*** (0.15) (0.14) (0.11) Constant 5.67** 5.19** 5.54*** (2.29) (2.12) (1.58)
Note: The values in brackets denote standard errors. ***/**/* Denotes significance at the 1%/5%/10% level. ASS1 and ASS2 denote the grade for assignment 1 and 2 as dependent variable. The average denotes the average grade for both assignments.
The result is partly in line with the expectations based on the related literature by Apesteguia et al. (2012) and Hoogendoorn et al. (2013) who stated that mixed teams outperform homogenous teams.
According to Apesteguia et.al. (2012), the best combination will be two male and one female. In this study teams consist of 4 or 5 students therefore the expectation will be that a team with 3 male and 2 female students (one male member more than female) is the best performing combination.
Nevertheless, three OLS regressions have been done to investigate if mixed teams perform better than homogenous teams (male and female). These results, shown in table 8, state that there is no reason to assume that mixed teams perform better than homogenous teams.
17
Table 8
Regression mixed gender groups on team performance
Variables ASS1 ASS2 Average
Mixed Gender 0.14 0.30 0.22 (0.22) (0.21) (0.16) Age 0.01 0.08 0.04 (0.11) (0.10) (0.08) Size 0.32 -0.03 0.14 (0.22) (0.21) (0.15) Credits 0.007 -0.004 0.001 (0.006) (0.005) (0.004) Resits -0.80* -1.01** -0.9*** (0.44) (0.41) (0.31) PWF -1.24 0.3 -0.48 (1.05) (0.99) (0.74) Avexam 0.15 0.42*** 0.28*** (0.15) (0.14) (0.10) Constant 5.3** 5.04*** 5.3*** (2.19) (2.06) (1.55)
Note: The values in brackets denote standard errors. ASS1 and ASS2 denote the grade for assignment 1 and 2 as dependent variable. The average denotes the average grade for both assignments. ***/**/* Denotes significance at the 1%/5%/10% level.
The regressions show that the variable for average final exam is positive and significantly different from zero for assignment 2 and the average of the assignments. The individual performance is a good predictor for the performance of the group. The positive coefficient means, the higher the individual result, the higher the grade for the group assignment. The fact that the variable is not significant from zero for assignment 1, could be explained that the unmotivated and less capable students possibly dropped out after assignment 1 and therefore had an negative effect on the motivated an hard working members of the team for assignment 1. This leads to significant results for assignment 2 and the average of the two assignments.
The higher the average resits in the group, the lower the average grade of the group assignments is expected to be. A reason for this is that the students who need to do resits have difficulties with the subject or are less motivated in comparison with other students.
18
5. Conclusion and Discussion
The purpose of this study is to investigate the effect of gender diversity on group performance. For this study undergraduates were used who participated a course at the university of Amsterdam where they had to cooperate with other students to pass two assignments. The performance is measured by the grades obtained for the assignments. Overall, this leads to the following research question:
‘’Does gender diversity have an effect on group performance?’’
There are conclusions that could be drawn from the analyses. Although the most results are not significant, there are some interesting findings. First, the expectation that the performance of groups with female will be better than groups without female was tested. The results are in line with the hypothesis, as they are significant for assignment 2 and the average assignment grade. Teams including at least one female member did have a higher average grade in comparison with teams with only male undergraduates. However, no hard conclusions can be drawn, because the results are significant at 10% level.
Second, the expectation that mixed teams outperform homogenous teams (men and women) is not in line with the results. There are no significant results found. Therefore, mixed teams do not perform better than homogenous teams, which is in contradiction with the hypothesis. However, the results show significant results that mixed teams outperform homogenous male teams, but this is just significant at 10% level and provides no clear evidence.
However, this contraction could be due to the fact that some students received a 0 for their final exam, because they did not show up at the final exam. After the removal of these students and a recalculation of the average final exam grade for each group, the same regression was done (in appendix A). This lead to a significant difference for assignment 2, female groups outperform groups without female at 5% level. Nevertheless, the students that did not show up at their final exam were possibly not confident or motivated enough to pass the exam and should therefore not be excluded from the dataset.
The main contribution of this study is to supplement the existing literature on gender diversity and in particular the effect of gender diversity on group performance. Considering the practical relevance of this research, the relationship between gender diversity and group performance can be valuable to all different settings where
19 teamwork is needed, for example high schools, universities and corporations. Also, for the coordinator of this subject at the university of Amsterdam this study can be useful, as he might implement a rule that teams should be mixed or contain at least one female member to enhance group effectiveness.
Control variables are needed in the regression to clearly identify the relationship between the dependent and independent variable. In this study there are some variables omitted, due to a lack of information availability. Information about the student their personality, race and abilities were not available. This is a limitation of the study, as the relationship cannot be measured precisely as according to Rhee et al. (2013) personality and individual abilities positively correlates with the evaluation of group performance. Also Borg & Shapiro (1996) used race as a control variable in their empirical model to estimate the effect on group performance. They found no effect of gender diversity on group performance and race could possibly explain this result.
The groups are self-selected and this lead to selection bias, because the undergraduates choose friends to cooperate with and might be more motivated and exert more effort, as friendship leads to more cohesiveness and therefore to a better performance according to Jehn, Karen, & Shah (1997). They can also choose to cooperate with other students where they already cooperated with in the past. The information about the formation of groups was not available and therefore a limitation.
Furthermore all the group members contained the same grade for the assignment, except if they dropped out early. Therefore it is not possible to measure the amount of effort the members individually exerted to succeed. Due to the unidentifiable individual performance the groups possibly contain social loafing according to Williams, Harkins & Latané (1981). Students exert less effort if their individual performance is unknown.
For further research it can be interesting to do similar research at high schools or universities, where more information is available about the abilities and characteristics of the undergraduates. Also studying another subject instead of economics or finance might lead to different results. Moreover, it can be interesting to do research in other parts of the world. Possibly race and background differences may influence the group performance, because the international students who study economics and business at the university of Amsterdam seem to obtain better grades than the local students.
20
References
Apesteguia, J., Azmat, G., & Iriberri, N., (2012). The Impact of Gender Composition on Team Performance and Decision Making: Evidence from the Field.
Management Science, Vol. 58, 77-95.
Borg, M. O., and Shapiro, S. L., (1996). Personality type and student performance in principles of economics. Journal of Economic Education, Vol. 27, 4-24. Campbell, K., Minguez-Vera, A. (2008). Gender Diversity in the Boardroom and Firm
Financial Performance. Journal of Business Ethics, 83(3), 435-451
Carter, D. A., Simkins, B. J., & Simpson, W. G., (2003). Corporate Governance, Board Diversity and Firm Value. The Financial Review, Vol. 38, 33-53.
Guzzo, R. A., and Dickson, M.W., (1996). Teams in Organizations: Recent Research on Performance and Effectiveness. Review of Psychology, Vol. 47, 307-338.
Hackman, J. R., & Morris, C. G., (1975). Group tasks, group interaction and group performance effectiveness: A review and proposed integration. Elsevier Vol. 8, 45-99.
Hoogendoorn, S., Oosterbeek, H., & Van Praag, M. (2013). The Impact of Gender Diversity on the Performance of Business Teams: Evidence from a Field Experiment. Management Science, Vol. 59(7), 1509-1530.
Jehn, Karen, A., Shah, P.P., (1997). Interpersonal relationships and task performance: An examination of mediation processes in friendship and acquaintance groups.
Journal of Personality and Social Psychology, Vol. 72(4), 775-790.
McCarty, C., Padyham, G., & Bennett, D., (2006). Determinants of student achievement in principles of economics. Journal for economics educators. Vol. 6, 6-8.
Pallant, J. (2013). SPSS survival manual: A step by step guide to data analysis using IBM SPSS , 4th ed. Crows Nest.
21 Rhee, J., Parent, D., Basu, A., (2013). The influence of personality and ability on
undergraduate teamwork and team performance. Springerplus, 5-12.
Seijts, G.H., & Latham, G.P., (2000). The effects of goal setting and group size in a social dilemma. Canadian Journal of behavioural science, Vol. 32(2), 104-116. Shrader, C.B., Blackburn, V. B., Iles, P. (1997). Women in management and firm
financial performance. Journal of managerial issues. 9(3), 356 - 371
Wegge, J., Roth, C., Neubach, B., Schmidt, K., & Kanfer, R., (2008). Age and gender diversity as determinants of performance and health in a public organization: The role of task complexity and group size. Journal of applied psychology, Vol. 93(6), 1301-1313.
Williams, K., Harkins, S. G., & Latané, B. (1981). Identifiability as a deterrant to social loafing: Two cheering experiments. Journal of Personality and Social
22
Appendix
Appendix A: Regression mixed teams on performance Table 7BRegression mixed gender groups on team performance
Variables ASS1 ASS2 Average
Mixed Gender 0.20 0.43** 0.31* (0.23) (0.21) (0.16) Age 0.01 0.08 0.04 (0.11) (0.10) (0.07) Size 0.29 -0.10 0.10 (0.22) (0.20) (0.15) Credits 0.006 -0.005 0.0007 (0.006) (0.005) (0.004) Resits -0.78* -0.96** -0.87*** (0.44) (0.40) (0.31) Avexam 0.11 0.40*** 0.25*** (0.12) (0.11) (0.08) Constant 5.47** 5.39** 5.56*** (2.17) (1.99) (1.51)
Note: The values in brackets denote standard errors. ***/**/* Denotes significance at the 1%/5%/10% level. ASS1 and ASS2 denote the grade for assignment 1 and 2 as dependent variable. The average denotes the average grade for both assignments.
