Data Description

Numerous factors affect employee salaries, including education, skill level, and years of experience (Chand, 2014). However, these factors should not be pegged on certain demographic factors such as gender and ethnicity. Otherwise, this would be considered to be discrimination. Despite several efforts that have been put in place to curb discrimination, a report by KPMG consulting in 2019 showed that organizations were still reporting cases of salary remuneration discrimination based on demographic factors such as gender and race for employees having the same market characteristics (KPMG, 2019). Discrimination in remuneration reduces labor supply and incapacitates the victim’s ability to meet short-term and long-term financial goals (Becker and Lindsay, 2015). Furthermore, the working morale of the discriminated workers is lowered, thereby affecting their productivity. Therefore, to limit these effects, it is essential that companies continuously examine the potential of salary discrimination and address them effectively. This paper investigates whether gender and ethnic discrimination in salary remuneration in the Manutan International S.A.

The dataset used in the analysis is a random sample of 100 employees collected by the Director of Human Resources Department in Manutan International S.A. The data consists of eight variables describing specific information about the employees. The variable salary is a continuous numeric variable that represents the employees’ annual salary in pound sterling (£), the variable education is a discrete numeric variable that shows employees’ years of education, experience is a discrete numeric variable that represents the employees’ years of work experience, age is a discrete numeric variable that shows employees’ ages in years, gender is a categorical variable measured at the nominal level to ascertain whether an employee is male or female, non-white is a categorical variable measured at the nominal level to show whether an employee is white or not, married is a categorical variable measured at the nominal level to indicate whether an employee is married or not. Lastly, union is a categorical variable measured at the nominal level to show whether an employee is a union member or not.

Frequencies and proportions shown in the table below have been used as the descriptive statistics for the gender variable because it is categorical.

The pie chart below visualizes the data above.

From the summary table and the pie chart, it is evident that the proportion of male employees (53%) was slightly higher than that of female employees (47%).

Non-White

Frequencies and proportions shown in the table below have been used as the descriptive statistics for the non-white variable because it is categorical.

Data Analysis and Results

The pie chart below visualizes the data above.

From the summary table and the pie chart, it is evident that the proportion of white employees in the company (90%) was much higher than the proportion of non-white employees (10%).

Frequencies and proportions shown in the table below have been used as the descriptive statistics for the married variable because it is categorical.

The pie chart below visualizes the data above

From the summary table and the pie chart, it is evident that the proportion of married employees in the company (67%) was much higher than the proportion of single employees (33%).

Frequencies and proportions shown in the table below have been used as the descriptive statistics for the union variable because it is categorical.

The pie chart below visualizes the data above

From the summary table and the pie chart, it is evident that the proportion of union members (18%) is less than the proportion of employees who were not union members (82%).

The summary statistics of age variable are shown below;

The average age of the employees in the company was 39.11 years. 25% of the employees were less than 28 years old, 50% were less than 36.50 years old, and 75% were less than 49.00 years old. However, the majority of the employees were 36.50 years old. The ages of the employees varied quite a bit from the mean (SD =12.572). The youngest employee was 18 years old, while the most aged employee was 64 years old. Therefore, the range of employees’ ages was 46 years.

The histogram below visualizes the distribution of the age variable.

The histogram has a longer tail on the right side of its peak, indicating that the distribution of employees’ ages is positively skewed (Kusleika, 2021).

The summary statistics of education variable are shown below;

The average education years for employees in the company was 12.73 years. 25% of the employees had schooled for less than 12 years, 50% had spent below or above 12 years in school, while 75% had spent less than 14 years in school. However, the majority of employees had spent 12 years in school. The employees’ number of years spent in education had a relatively low variability (SD =2.792) from the mean. The lowest learned employee had spent 4 years in school while the highest learned employee had spent 18 years in school. Therefore, the range of years of education was 14 years.

Descriptive Statistics and Charts

The histogram below visualizes the distribution of the education variable.

The histogram shows that the employees’ years of education are normally distributed since it has approximately equally sized tails on either side of the peak.

The summary statistics of experience variable are shown below;

 The average years of experience for employees in the company was 20.38 years. 25% of the employees had less than 9 years of experience, 50% had below/above 18.50 years of experience, and 75% had less than 29 years of experience. However, the majority of employees had accumulated 6 years of work experience. The employees’ years of experience varied quite a bit from the mean (SD =13.55). The least experienced employee had 0 years of experience, while the most experienced employee had 54 years of experience. Therefore, the range of years of experience was 54 years.

The histogram below visualizes the distribution of the experience variable.

The histogram has a longer tail on the right side of its peak, indicating that the distribution of employees’ years of experience is positively skewed.

The summary statistics of salaries variable are shown below;

The average salary of the employees in the company was £30,833.46. 25% of the employees earned below £17800.75, 50% made below £28815.50, while 75% earned below £35075.25. However, the majority of the employees earned £20852. The employee salaries varied quite a bit from the mean (SD =16947.10). The lowest earner received £9879, while the highest earner received £83601. Therefore, the range of employee salaries was £73722.

The histogram below visualizes the distribution of the salary variable.

The histogram has a longer tail on the right side of its peak, indicating that the distribution of employee salaries is positively skewed.

Whether Salary and Education have a linear correlation

The table below is a correlation matrix for the salary and education variables

The correlation coefficient is , indicating a moderately strong positive linear relationship between salary and years of education. The correlation is significant since the p-value  is less than the 5% significance level. Consequently, it can be said that an employee with more years of education is expected to earn more than an employee with fewer years of schooling.

Whether Salary and Experience have a linear correlation

The table below is a correlation matrix for the salary and experience variables

The correlation coefficient is , indicating a very weak positive linear relationship between salary and years of experience. The correlation is not significant since the p-value  is greater than the 5% significance level. Therefore, it can be said that the weak correlation between the variables is based on chance.

Gender

Chi-square test of independence was performed to investigate whether there is an association between gender and union. The null hypothesis is that there is no association, while the alternative hypothesis is that there is an association between gender and union. i.e.,

 The cross-tabulation table for the test is shown below;

 Assumptions of Chi-Square of independence

The assumptions required to the Chi-Square test to be suitable in this scenario are as follows;

• Both variables should be categorical with two or more mutually exclusive groups: In this case, both gender and union are categorical and have two mutually exclusive groups each (male and female for gender and yes to no for union). Therefore, the assumption is met.
• The minimum expected frequency should be five or more in more than 80% of the cells: In this case, the minimum expected frequency is 8.46. Therefore, the assumption is satisfied.

It is concluded that the assumptions of the Chi-Square test are satisfied. The output of the test is shown below;

The test statistic is  and the p-value is . The p-value is greater than . Therefore, the null hypothesis is not rejected, and the conclusion is that there is not enough evidence to claim that there is an association between gender and union (Stine and Foster, 2017). That is, men and women are equally likely to be union members or not. The bar chart below shows the distribution of men and women who are -and who are not union members.

As can be seen from the plot above, the number of males and females who are not union members is almost the same, and the same is for males and females who are union members.

Creating dummy variables.

The dummy variables for gender, non-white, married, and union were created using the following syntax.

The first five rows of the dummy variables are shown below;

Two sample T-tests

Whether there is a significant difference in annual salary between male and female employees

The hypotheses for the test were as follows;

where  is the average annual salary.

The assumptions required for the test are;

• There should be homogeneity of variances: The assumption is checked using Levene’s test, whose output is shown below;

The p-value of the test  is less than the 5% significance level, indicating that the assumption is violated (Wasserman, 2013). Therefore, the t-test conducted will assume unequal variances.

• The dependent variable (salary) should be approximately normally distributed for each group of the independent variable (male and female): The output of the normality test is shown below;

The p-value of the Shapiro-Wilk test for salaries of males  and females  is less than the 5% significance level, indicating that the assumption is violated. However, considering that the sample sizes are sufficiently large, the violation is assumed, and a parametric t-test is performed.

The output of the independent samples t-test is shown below;

Since the test assumed unequal variance, the test statistic is  and the p-value is . The p-value is less than the 5% significance level. Therefore, the null hypothesis is rejected, and the conclusion is that there is enough evidence to claim that the average annual salary of men is significantly different from that of females. Males earn a higher average wage than females.

Non-White

Whether there is a significant difference in annual salary between male and female employees

The hypotheses for the test were as follows;

where  is the average annual salary.

The assumptions required for the test are;

• There should be homogeneity of variances: The assumption is checked using Levene’s test, whose output is shown below;

The p-value of the test  is greater than the 5% significance level, indicating that the assumption is met. Therefore, the t-test conducted will assume equal variances.

• The dependent variable (salary) should be approximately normally distributed for each group of the independent variable (white and non-white): The output of the normality test is shown below;

The p-value of the Shapiro-Wilk test for salaries of males  is less than the 5% significance level, indicating that the assumption is violated. However, for female wages, the p-value  is greater than the 5% significance level, meaning that the assumption is met. Since the sample sizes are sufficiently large, the violation is assumed, and the parametric independent samples t-test is performed.

The output of the independent samples t-test is shown below;

Since the test assumes equal variance, the test statistic is  and the p-value is . The p-value is greater than the 5% significance level. Therefore, the null hypothesis is not rejected, and the conclusion is that there is not enough evidence to claim that the average annual salary of whites is significantly different from that of non-whites.

Conclusion and Recommendation

The paper’s purpose was to investigate whether there is gender and ethnic discrimination in salary remuneration in the Manutan International S.A. Based on the analysis performed on the data provided, there was a significant difference in the average annual salary of male and female employees, with male employees earning a higher average amount than females. This is an indication of gender discrepancy in salary remuneration, and hence it is recommended that the company should device strict pay policies that promote equal pay for all workers. On the other hand, there was no significant difference in the average annual salary of whites and non-whites. Therefore, the company does not suffer from ethnic discrimination in salary remuneration. However, it was noted that the proportion of white employees in the company was higher than that of non-whites. Therefore, it is recommended that the company should provide equal employment opportunities for people in all ethnic groups. It is also recommended that equal employment opportunity should also be provided for both males and females since preliminary analysis indicated that the proportion of males in the company was higher than that of females. Furthermore, the company should develop ways to encourage members to join unions since most of them tend not to belong to any union. Encouragement of employees to join unions can be done through education, guidance, or enactment of law that encourages permanent union membership.  

References

Becker, E., and Lindsay, C. M. 2015. The limits of the wage impact of discrimination. Managerial and Decision Economics, 26(8), 513-525. doi:10.1002/mde.1238

Chand, S. 2014. 7 factors to consider for determining wage and salary structure of workers. Retrieved from https://www.yourarticlelibrary.com/organization/7-factors-for-determining-wage-and-salary-structure-of-workers/24737

KPMG Consulting 2019. She’s price(d)less: The economics of the gender pay gap. Retrieved from https://home.kpmg/content/dam/kpmg/au/pdf/2019/gender-pay-gap-economics-full-report-2019.pdf

Kusleika, D. 2021. Data visualization with Excel dashboards and reports. John Wiley & Sons.

Stine, R. A., and Foster, D. P. 2017. Statistics for business: Decision making and analysis. London, England: Pearson.

Wasserman, L. 2013. All of statistics: A concise course in statistical inference. Berlin, Germany: Springer Science & Business Media.